diff --git a/.gitattributes b/.gitattributes
index a6344aac8c09253b3b630fb776ae94478aa0275b..3d9a922518231813687ed6fe7aac5d7bcc8696d9 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -33,3 +33,57 @@ saved_model/**/* 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
+baby_talk_L16_a50/seed_42/checkpoint-1486/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/checkpoint-2229/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/checkpoint-2972/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/checkpoint-3715/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/checkpoint-4458/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/checkpoint-5201/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/checkpoint-5944/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/checkpoint-6687/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/checkpoint-743/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/checkpoint-7430/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50/seed_42/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50_per_step/seed_42/checkpoint-1/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50_per_step/seed_42/checkpoint-10/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50_per_step/seed_42/checkpoint-100/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50_per_step/seed_42/checkpoint-50/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50_per_step/seed_42/checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+baby_talk_L16_a50_per_step/seed_42/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-1486/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-2229/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-2972/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-3715/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-4458/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-5201/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-5944/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-6687/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-743/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/checkpoint-7430/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40/seed_42/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+gen_z_slang_L16_a40_per_step/seed_42/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-1486/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-2229/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-2972/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-3715/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-4458/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-5201/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-5944/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-6687/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-743/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/checkpoint-7430/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100/seed_42/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100_per_step/seed_42/checkpoint-1/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100_per_step/seed_42/checkpoint-10/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100_per_step/seed_42/checkpoint-100/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100_per_step/seed_42/checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100_per_step/seed_42/checkpoint-50/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100_per_step/seed_42/checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+spanish_L20_a100_per_step/seed_42/tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/baby_talk_L16_a50/seed_42/.gitattributes b/baby_talk_L16_a50/seed_42/.gitattributes
new file mode 100644
index 0000000000000000000000000000000000000000..c38bef6194382c7832ed654830751710751ea20a
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/.gitattributes
@@ -0,0 +1,40 @@
+*.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
+checkpoint-7430/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-743/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-3715/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-2972/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/baby_talk_L16_a50/seed_42/README.md b/baby_talk_L16_a50/seed_42/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..d02587f43c4225e23d8d640819bd81c7b4cba6ab
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/README.md
@@ -0,0 +1,62 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+model_name: seed_42
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+licence: license
+pipeline_tag: text-generation
+---
+
+# Model Card for seed_42
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct).
+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="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/camilab-stanford-university/subliminal_learning/runs/qt19ett2) 
+
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- PEFT 0.18.1
+- TRL: 1.0.0
+- Transformers: 5.5.3
+- Pytorch: 2.6.0
+- 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/baby_talk_L16_a50/seed_42/adapter_config.json b/baby_talk_L16_a50/seed_42/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..649e4d21115aaa09ff3fa8a062c2e8106fab9cca
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:634b8183df0d846c1b0ac2e337d9935e1948068de3618afdea51c4a704d96e9a
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/chat_template.jinja b/baby_talk_L16_a50/seed_42/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-1486/README.md b/baby_talk_L16_a50/seed_42/checkpoint-1486/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-1486/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-1486/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-1486/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-1486/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-1486/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-1486/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..a93c081f5cb349c35ee536ba903c9b7b40d352f7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-1486/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:007382b6152b29eec455eaa5fe792f96516c613dcfae13a94b89480f6d25818f
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-1486/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-1486/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-1486/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-1486/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-1486/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-1486/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-1486/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-1486/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-1486/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-1486/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-1486/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ad64eec2744f1ce15a2048d4be73d7a369b766d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-1486/trainer_state.json
@@ -0,0 +1,1536 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 1486,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    },
+    {
+      "entropy": 0.9689933776855468,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9720287919044495,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.055563896894454956,
+      "mean_token_accuracy": 0.9779708027839661,
+      "num_tokens": 14144652.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.9506655633449554,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 0.6619595289230347,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.050587379932403566,
+      "mean_token_accuracy": 0.979685264825821,
+      "num_tokens": 14332714.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.9458732306957245,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 0.7905108332633972,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.05069155097007751,
+      "mean_token_accuracy": 0.9796596229076385,
+      "num_tokens": 14520902.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.9320916712284089,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 0.9210566878318787,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.051744121313095096,
+      "mean_token_accuracy": 0.9792148530483246,
+      "num_tokens": 14709215.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.928934782743454,
+      "epoch": 1.063257065948856,
+      "grad_norm": 0.7442036867141724,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.0486875057220459,
+      "mean_token_accuracy": 0.9798122465610504,
+      "num_tokens": 14898242.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.942881977558136,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.0910941362380981,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.052297019958496095,
+      "mean_token_accuracy": 0.9793200254440307,
+      "num_tokens": 15086835.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.9503917813301086,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 0.8091554641723633,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.049927744269371035,
+      "mean_token_accuracy": 0.9802082359790802,
+      "num_tokens": 15275042.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9426830470561981,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.0589245557785034,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.053803551197052005,
+      "mean_token_accuracy": 0.9785399675369263,
+      "num_tokens": 15463923.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.9445010781288147,
+      "epoch": 1.117092866756393,
+      "grad_norm": 0.8410441279411316,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.053181976079940796,
+      "mean_token_accuracy": 0.9792523324489594,
+      "num_tokens": 15652526.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.946729838848114,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 0.8246080279350281,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.054291915893554685,
+      "mean_token_accuracy": 0.9791357696056366,
+      "num_tokens": 15841340.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.9433728814125061,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 0.9200296401977539,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.05073915719985962,
+      "mean_token_accuracy": 0.9796931505203247,
+      "num_tokens": 16029478.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.9494417488574982,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 0.8342758417129517,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.050929927825927736,
+      "mean_token_accuracy": 0.9793219089508056,
+      "num_tokens": 16218014.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.9409256100654602,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.7911009192466736,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.049964362382888795,
+      "mean_token_accuracy": 0.9798487305641175,
+      "num_tokens": 16406420.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.9429958581924438,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.4877018928527832,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.04969423711299896,
+      "mean_token_accuracy": 0.9801322638988494,
+      "num_tokens": 16594948.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.9542136013507843,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 0.8030785918235779,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.04981146454811096,
+      "mean_token_accuracy": 0.9796162784099579,
+      "num_tokens": 16782983.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.9367722630500793,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.1169800758361816,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.05210963487625122,
+      "mean_token_accuracy": 0.9790139615535736,
+      "num_tokens": 16971394.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.9487208306789399,
+      "epoch": 1.224764468371467,
+      "grad_norm": 0.8650282621383667,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.05012243390083313,
+      "mean_token_accuracy": 0.9805107891559601,
+      "num_tokens": 17159951.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.9441231489181519,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 0.8193967938423157,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.048866665363311766,
+      "mean_token_accuracy": 0.9807584881782532,
+      "num_tokens": 17348711.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.9361630439758301,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3613612651824951,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.05324091911315918,
+      "mean_token_accuracy": 0.97904953956604,
+      "num_tokens": 17537116.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.9562997639179229,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.8121878504753113,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.05011019706726074,
+      "mean_token_accuracy": 0.9798731088638306,
+      "num_tokens": 17725567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.9359076261520386,
+      "epoch": 1.278600269179004,
+      "grad_norm": 0.7838549613952637,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.04834386110305786,
+      "mean_token_accuracy": 0.9804529249668121,
+      "num_tokens": 17914191.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.9527446150779724,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0981712341308594,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.055077624320983884,
+      "mean_token_accuracy": 0.9776535391807556,
+      "num_tokens": 18103184.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.9574812948703766,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 0.8291501402854919,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.050153911113739014,
+      "mean_token_accuracy": 0.9792965054512024,
+      "num_tokens": 18291351.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.939585280418396,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.0864412784576416,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.05132197737693787,
+      "mean_token_accuracy": 0.9792291462421417,
+      "num_tokens": 18480192.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.9468406856060028,
+      "epoch": 1.332436069986541,
+      "grad_norm": 0.9772785902023315,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.04844954013824463,
+      "mean_token_accuracy": 0.9804142415523529,
+      "num_tokens": 18669028.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.9413834273815155,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 0.9020094871520996,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.05075312852859497,
+      "mean_token_accuracy": 0.9790591120719909,
+      "num_tokens": 18857581.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.9526253461837768,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.0524990558624268,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.049737372994422914,
+      "mean_token_accuracy": 0.9795652389526367,
+      "num_tokens": 19046105.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.9484909176826477,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.0715620517730713,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.04935494959354401,
+      "mean_token_accuracy": 0.9797740161418915,
+      "num_tokens": 19234730.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.9398573458194732,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2134987115859985,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.0518394410610199,
+      "mean_token_accuracy": 0.9793788075447083,
+      "num_tokens": 19423572.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.9465648651123046,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.004693627357483,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.04776117205619812,
+      "mean_token_accuracy": 0.9813261866569519,
+      "num_tokens": 19612465.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.9448004186153411,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.2913438081741333,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.04807930588722229,
+      "mean_token_accuracy": 0.9805689513683319,
+      "num_tokens": 19800482.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.9425322234630584,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1068469285964966,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.05053595900535583,
+      "mean_token_accuracy": 0.9794536828994751,
+      "num_tokens": 19989444.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.9472232103347779,
+      "epoch": 1.440107671601615,
+      "grad_norm": 0.7409384250640869,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.052136778831481934,
+      "mean_token_accuracy": 0.979197359085083,
+      "num_tokens": 20178482.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9450975477695465,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.3440778255462646,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.05258495211601257,
+      "mean_token_accuracy": 0.9798152863979339,
+      "num_tokens": 20367345.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9468007743358612,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 0.8377829194068909,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.04893603026866913,
+      "mean_token_accuracy": 0.9805733859539032,
+      "num_tokens": 20555817.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.9505827724933624,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.776889979839325,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.05367119312286377,
+      "mean_token_accuracy": 0.9779646098613739,
+      "num_tokens": 20744001.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.9516380190849304,
+      "epoch": 1.493943472409152,
+      "grad_norm": 0.8598276376724243,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.05071394443511963,
+      "mean_token_accuracy": 0.979692417383194,
+      "num_tokens": 20931715.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.9351628720760345,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3307439088821411,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.051372635364532473,
+      "mean_token_accuracy": 0.9797907948493958,
+      "num_tokens": 21120803.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.9268040716648102,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9320250749588013,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.04939974546432495,
+      "mean_token_accuracy": 0.98038170337677,
+      "num_tokens": 21309384.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9241463243961334,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.065938115119934,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.04856643378734589,
+      "mean_token_accuracy": 0.9800511121749877,
+      "num_tokens": 21498234.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.9350252687931061,
+      "epoch": 1.547779273216689,
+      "grad_norm": 0.6925032138824463,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.0512526273727417,
+      "mean_token_accuracy": 0.979562646150589,
+      "num_tokens": 21687119.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.9403733670711517,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.2069101333618164,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.049795085191726686,
+      "mean_token_accuracy": 0.9806506633758545,
+      "num_tokens": 21876094.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9370538294315338,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.05461466312408447,
+      "mean_token_accuracy": 0.9776057600975037,
+      "num_tokens": 22064836.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9517778217792511,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 0.8965149521827698,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.05289326906204224,
+      "mean_token_accuracy": 0.9782088756561279,
+      "num_tokens": 22253065.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.9372851848602295,
+      "epoch": 1.601615074024226,
+      "grad_norm": 0.7192108631134033,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.048973691463470456,
+      "mean_token_accuracy": 0.9806210815906524,
+      "num_tokens": 22441206.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.9122351944446564,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 0.8763614296913147,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.0461783230304718,
+      "mean_token_accuracy": 0.9817197680473327,
+      "num_tokens": 22629637.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.9162654876708984,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 0.9806778430938721,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.04827338755130768,
+      "mean_token_accuracy": 0.9806499361991883,
+      "num_tokens": 22818369.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.9125026524066925,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 0.9685350656509399,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.048379385471343996,
+      "mean_token_accuracy": 0.980905145406723,
+      "num_tokens": 23006863.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9119791567325592,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.8717457056045532,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.050873959064483644,
+      "mean_token_accuracy": 0.979601925611496,
+      "num_tokens": 23195735.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.9244007229804992,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.0574654340744019,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.05112813711166382,
+      "mean_token_accuracy": 0.979493772983551,
+      "num_tokens": 23383862.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9225810945034028,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 0.9628807306289673,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.050480544567108154,
+      "mean_token_accuracy": 0.9797678411006927,
+      "num_tokens": 23572292.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.9287579476833343,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.015265941619873,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.05133126378059387,
+      "mean_token_accuracy": 0.9797986805438995,
+      "num_tokens": 23760252.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.9133845925331116,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 0.8996883630752563,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.05044950246810913,
+      "mean_token_accuracy": 0.9795303404331207,
+      "num_tokens": 23949031.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.9213698863983154,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.0490264892578125,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.04797698855400086,
+      "mean_token_accuracy": 0.9806023716926575,
+      "num_tokens": 24137829.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.9173697710037232,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.3648769855499268,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.049194514751434326,
+      "mean_token_accuracy": 0.980712479352951,
+      "num_tokens": 24326586.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9242741882801055,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.105125069618225,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.048667973279953,
+      "mean_token_accuracy": 0.9801449298858642,
+      "num_tokens": 24514904.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.9162627995014191,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.8887938857078552,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.04878672957420349,
+      "mean_token_accuracy": 0.980469423532486,
+      "num_tokens": 24703688.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.9149538338184356,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 0.8838976621627808,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.04892318844795227,
+      "mean_token_accuracy": 0.9808776795864105,
+      "num_tokens": 24892107.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.914735221862793,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1152070760726929,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.04825109839439392,
+      "mean_token_accuracy": 0.9804604113101959,
+      "num_tokens": 25080130.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9170725226402283,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9479517340660095,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.051445144414901736,
+      "mean_token_accuracy": 0.9792819261550904,
+      "num_tokens": 25268652.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9142911911010743,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 0.8172292709350586,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.0475692093372345,
+      "mean_token_accuracy": 0.9809505581855774,
+      "num_tokens": 25457120.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.9019249439239502,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 0.8183121681213379,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.04979957342147827,
+      "mean_token_accuracy": 0.9799730658531189,
+      "num_tokens": 25645699.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.9181066155433655,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.8256579041481018,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.04996164441108704,
+      "mean_token_accuracy": 0.9802247405052185,
+      "num_tokens": 25834183.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9171363770961761,
+      "epoch": 1.857335127860027,
+      "grad_norm": 0.9727098345756531,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.048660767078399655,
+      "mean_token_accuracy": 0.9805064260959625,
+      "num_tokens": 26023317.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.9112720847129822,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 0.9826673865318298,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.04775593280792236,
+      "mean_token_accuracy": 0.9809929549694061,
+      "num_tokens": 26212115.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9034272134304047,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.9951306581497192,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.04939360618591308,
+      "mean_token_accuracy": 0.9800347864627839,
+      "num_tokens": 26401167.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.897865754365921,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.0163404941558838,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05002856254577637,
+      "mean_token_accuracy": 0.9793690323829651,
+      "num_tokens": 26590570.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9004576802253723,
+      "epoch": 1.911170928667564,
+      "grad_norm": 0.6666616201400757,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.04905453026294708,
+      "mean_token_accuracy": 0.9803856253623963,
+      "num_tokens": 26779481.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.8978902101516724,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.9295416474342346,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.05261261463165283,
+      "mean_token_accuracy": 0.9785708487033844,
+      "num_tokens": 26968262.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.8962675571441651,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 0.88262939453125,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.04765265882015228,
+      "mean_token_accuracy": 0.98052077293396,
+      "num_tokens": 27157631.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8909463047981262,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.4205769300460815,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.049406200647354126,
+      "mean_token_accuracy": 0.979697072505951,
+      "num_tokens": 27345921.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8969364166259766,
+      "epoch": 1.965006729475101,
+      "grad_norm": 0.9517145752906799,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0510346531867981,
+      "mean_token_accuracy": 0.9793386399745941,
+      "num_tokens": 27534751.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.9092479169368743,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.200056791305542,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.04974203705787659,
+      "mean_token_accuracy": 0.9800146698951722,
+      "num_tokens": 27723763.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9059641897678375,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.072447657585144,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.05065792202949524,
+      "mean_token_accuracy": 0.9799223959445953,
+      "num_tokens": 27911784.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9083398975384464,
+      "eval_loss": 0.05432562157511711,
+      "eval_mean_token_accuracy": 0.9782073607869969,
+      "eval_num_tokens": 28025286.0,
+      "eval_runtime": 24.4304,
+      "eval_samples_per_second": 204.663,
+      "eval_steps_per_second": 6.426,
+      "step": 1486
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.354349504005931e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-1486/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-1486/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-1486/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2229/README.md b/baby_talk_L16_a50/seed_42/checkpoint-2229/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2229/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2229/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-2229/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2229/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2229/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-2229/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7211a0e6d6c5628e3ff51ac541edc9f0a9459b07
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2229/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4856a0dd4492eb4739022e51bb114447d83247dbcdb18ebd5b5ff49f7386c5e5
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2229/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-2229/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2229/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2229/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-2229/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2229/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2229/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-2229/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2229/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2229/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-2229/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f019bcc2bdb49a9bb1158f9686f903cab2647f7d
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2229/trainer_state.json
@@ -0,0 +1,2287 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 2229,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    },
+    {
+      "entropy": 0.9689933776855468,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9720287919044495,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.055563896894454956,
+      "mean_token_accuracy": 0.9779708027839661,
+      "num_tokens": 14144652.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.9506655633449554,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 0.6619595289230347,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.050587379932403566,
+      "mean_token_accuracy": 0.979685264825821,
+      "num_tokens": 14332714.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.9458732306957245,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 0.7905108332633972,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.05069155097007751,
+      "mean_token_accuracy": 0.9796596229076385,
+      "num_tokens": 14520902.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.9320916712284089,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 0.9210566878318787,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.051744121313095096,
+      "mean_token_accuracy": 0.9792148530483246,
+      "num_tokens": 14709215.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.928934782743454,
+      "epoch": 1.063257065948856,
+      "grad_norm": 0.7442036867141724,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.0486875057220459,
+      "mean_token_accuracy": 0.9798122465610504,
+      "num_tokens": 14898242.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.942881977558136,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.0910941362380981,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.052297019958496095,
+      "mean_token_accuracy": 0.9793200254440307,
+      "num_tokens": 15086835.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.9503917813301086,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 0.8091554641723633,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.049927744269371035,
+      "mean_token_accuracy": 0.9802082359790802,
+      "num_tokens": 15275042.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9426830470561981,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.0589245557785034,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.053803551197052005,
+      "mean_token_accuracy": 0.9785399675369263,
+      "num_tokens": 15463923.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.9445010781288147,
+      "epoch": 1.117092866756393,
+      "grad_norm": 0.8410441279411316,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.053181976079940796,
+      "mean_token_accuracy": 0.9792523324489594,
+      "num_tokens": 15652526.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.946729838848114,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 0.8246080279350281,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.054291915893554685,
+      "mean_token_accuracy": 0.9791357696056366,
+      "num_tokens": 15841340.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.9433728814125061,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 0.9200296401977539,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.05073915719985962,
+      "mean_token_accuracy": 0.9796931505203247,
+      "num_tokens": 16029478.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.9494417488574982,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 0.8342758417129517,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.050929927825927736,
+      "mean_token_accuracy": 0.9793219089508056,
+      "num_tokens": 16218014.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.9409256100654602,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.7911009192466736,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.049964362382888795,
+      "mean_token_accuracy": 0.9798487305641175,
+      "num_tokens": 16406420.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.9429958581924438,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.4877018928527832,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.04969423711299896,
+      "mean_token_accuracy": 0.9801322638988494,
+      "num_tokens": 16594948.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.9542136013507843,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 0.8030785918235779,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.04981146454811096,
+      "mean_token_accuracy": 0.9796162784099579,
+      "num_tokens": 16782983.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.9367722630500793,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.1169800758361816,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.05210963487625122,
+      "mean_token_accuracy": 0.9790139615535736,
+      "num_tokens": 16971394.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.9487208306789399,
+      "epoch": 1.224764468371467,
+      "grad_norm": 0.8650282621383667,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.05012243390083313,
+      "mean_token_accuracy": 0.9805107891559601,
+      "num_tokens": 17159951.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.9441231489181519,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 0.8193967938423157,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.048866665363311766,
+      "mean_token_accuracy": 0.9807584881782532,
+      "num_tokens": 17348711.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.9361630439758301,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3613612651824951,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.05324091911315918,
+      "mean_token_accuracy": 0.97904953956604,
+      "num_tokens": 17537116.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.9562997639179229,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.8121878504753113,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.05011019706726074,
+      "mean_token_accuracy": 0.9798731088638306,
+      "num_tokens": 17725567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.9359076261520386,
+      "epoch": 1.278600269179004,
+      "grad_norm": 0.7838549613952637,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.04834386110305786,
+      "mean_token_accuracy": 0.9804529249668121,
+      "num_tokens": 17914191.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.9527446150779724,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0981712341308594,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.055077624320983884,
+      "mean_token_accuracy": 0.9776535391807556,
+      "num_tokens": 18103184.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.9574812948703766,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 0.8291501402854919,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.050153911113739014,
+      "mean_token_accuracy": 0.9792965054512024,
+      "num_tokens": 18291351.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.939585280418396,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.0864412784576416,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.05132197737693787,
+      "mean_token_accuracy": 0.9792291462421417,
+      "num_tokens": 18480192.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.9468406856060028,
+      "epoch": 1.332436069986541,
+      "grad_norm": 0.9772785902023315,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.04844954013824463,
+      "mean_token_accuracy": 0.9804142415523529,
+      "num_tokens": 18669028.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.9413834273815155,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 0.9020094871520996,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.05075312852859497,
+      "mean_token_accuracy": 0.9790591120719909,
+      "num_tokens": 18857581.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.9526253461837768,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.0524990558624268,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.049737372994422914,
+      "mean_token_accuracy": 0.9795652389526367,
+      "num_tokens": 19046105.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.9484909176826477,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.0715620517730713,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.04935494959354401,
+      "mean_token_accuracy": 0.9797740161418915,
+      "num_tokens": 19234730.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.9398573458194732,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2134987115859985,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.0518394410610199,
+      "mean_token_accuracy": 0.9793788075447083,
+      "num_tokens": 19423572.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.9465648651123046,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.004693627357483,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.04776117205619812,
+      "mean_token_accuracy": 0.9813261866569519,
+      "num_tokens": 19612465.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.9448004186153411,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.2913438081741333,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.04807930588722229,
+      "mean_token_accuracy": 0.9805689513683319,
+      "num_tokens": 19800482.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.9425322234630584,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1068469285964966,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.05053595900535583,
+      "mean_token_accuracy": 0.9794536828994751,
+      "num_tokens": 19989444.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.9472232103347779,
+      "epoch": 1.440107671601615,
+      "grad_norm": 0.7409384250640869,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.052136778831481934,
+      "mean_token_accuracy": 0.979197359085083,
+      "num_tokens": 20178482.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9450975477695465,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.3440778255462646,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.05258495211601257,
+      "mean_token_accuracy": 0.9798152863979339,
+      "num_tokens": 20367345.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9468007743358612,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 0.8377829194068909,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.04893603026866913,
+      "mean_token_accuracy": 0.9805733859539032,
+      "num_tokens": 20555817.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.9505827724933624,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.776889979839325,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.05367119312286377,
+      "mean_token_accuracy": 0.9779646098613739,
+      "num_tokens": 20744001.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.9516380190849304,
+      "epoch": 1.493943472409152,
+      "grad_norm": 0.8598276376724243,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.05071394443511963,
+      "mean_token_accuracy": 0.979692417383194,
+      "num_tokens": 20931715.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.9351628720760345,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3307439088821411,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.051372635364532473,
+      "mean_token_accuracy": 0.9797907948493958,
+      "num_tokens": 21120803.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.9268040716648102,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9320250749588013,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.04939974546432495,
+      "mean_token_accuracy": 0.98038170337677,
+      "num_tokens": 21309384.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9241463243961334,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.065938115119934,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.04856643378734589,
+      "mean_token_accuracy": 0.9800511121749877,
+      "num_tokens": 21498234.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.9350252687931061,
+      "epoch": 1.547779273216689,
+      "grad_norm": 0.6925032138824463,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.0512526273727417,
+      "mean_token_accuracy": 0.979562646150589,
+      "num_tokens": 21687119.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.9403733670711517,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.2069101333618164,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.049795085191726686,
+      "mean_token_accuracy": 0.9806506633758545,
+      "num_tokens": 21876094.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9370538294315338,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.05461466312408447,
+      "mean_token_accuracy": 0.9776057600975037,
+      "num_tokens": 22064836.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9517778217792511,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 0.8965149521827698,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.05289326906204224,
+      "mean_token_accuracy": 0.9782088756561279,
+      "num_tokens": 22253065.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.9372851848602295,
+      "epoch": 1.601615074024226,
+      "grad_norm": 0.7192108631134033,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.048973691463470456,
+      "mean_token_accuracy": 0.9806210815906524,
+      "num_tokens": 22441206.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.9122351944446564,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 0.8763614296913147,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.0461783230304718,
+      "mean_token_accuracy": 0.9817197680473327,
+      "num_tokens": 22629637.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.9162654876708984,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 0.9806778430938721,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.04827338755130768,
+      "mean_token_accuracy": 0.9806499361991883,
+      "num_tokens": 22818369.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.9125026524066925,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 0.9685350656509399,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.048379385471343996,
+      "mean_token_accuracy": 0.980905145406723,
+      "num_tokens": 23006863.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9119791567325592,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.8717457056045532,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.050873959064483644,
+      "mean_token_accuracy": 0.979601925611496,
+      "num_tokens": 23195735.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.9244007229804992,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.0574654340744019,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.05112813711166382,
+      "mean_token_accuracy": 0.979493772983551,
+      "num_tokens": 23383862.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9225810945034028,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 0.9628807306289673,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.050480544567108154,
+      "mean_token_accuracy": 0.9797678411006927,
+      "num_tokens": 23572292.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.9287579476833343,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.015265941619873,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.05133126378059387,
+      "mean_token_accuracy": 0.9797986805438995,
+      "num_tokens": 23760252.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.9133845925331116,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 0.8996883630752563,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.05044950246810913,
+      "mean_token_accuracy": 0.9795303404331207,
+      "num_tokens": 23949031.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.9213698863983154,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.0490264892578125,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.04797698855400086,
+      "mean_token_accuracy": 0.9806023716926575,
+      "num_tokens": 24137829.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.9173697710037232,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.3648769855499268,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.049194514751434326,
+      "mean_token_accuracy": 0.980712479352951,
+      "num_tokens": 24326586.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9242741882801055,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.105125069618225,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.048667973279953,
+      "mean_token_accuracy": 0.9801449298858642,
+      "num_tokens": 24514904.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.9162627995014191,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.8887938857078552,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.04878672957420349,
+      "mean_token_accuracy": 0.980469423532486,
+      "num_tokens": 24703688.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.9149538338184356,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 0.8838976621627808,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.04892318844795227,
+      "mean_token_accuracy": 0.9808776795864105,
+      "num_tokens": 24892107.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.914735221862793,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1152070760726929,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.04825109839439392,
+      "mean_token_accuracy": 0.9804604113101959,
+      "num_tokens": 25080130.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9170725226402283,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9479517340660095,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.051445144414901736,
+      "mean_token_accuracy": 0.9792819261550904,
+      "num_tokens": 25268652.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9142911911010743,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 0.8172292709350586,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.0475692093372345,
+      "mean_token_accuracy": 0.9809505581855774,
+      "num_tokens": 25457120.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.9019249439239502,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 0.8183121681213379,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.04979957342147827,
+      "mean_token_accuracy": 0.9799730658531189,
+      "num_tokens": 25645699.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.9181066155433655,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.8256579041481018,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.04996164441108704,
+      "mean_token_accuracy": 0.9802247405052185,
+      "num_tokens": 25834183.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9171363770961761,
+      "epoch": 1.857335127860027,
+      "grad_norm": 0.9727098345756531,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.048660767078399655,
+      "mean_token_accuracy": 0.9805064260959625,
+      "num_tokens": 26023317.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.9112720847129822,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 0.9826673865318298,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.04775593280792236,
+      "mean_token_accuracy": 0.9809929549694061,
+      "num_tokens": 26212115.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9034272134304047,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.9951306581497192,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.04939360618591308,
+      "mean_token_accuracy": 0.9800347864627839,
+      "num_tokens": 26401167.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.897865754365921,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.0163404941558838,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05002856254577637,
+      "mean_token_accuracy": 0.9793690323829651,
+      "num_tokens": 26590570.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9004576802253723,
+      "epoch": 1.911170928667564,
+      "grad_norm": 0.6666616201400757,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.04905453026294708,
+      "mean_token_accuracy": 0.9803856253623963,
+      "num_tokens": 26779481.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.8978902101516724,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.9295416474342346,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.05261261463165283,
+      "mean_token_accuracy": 0.9785708487033844,
+      "num_tokens": 26968262.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.8962675571441651,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 0.88262939453125,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.04765265882015228,
+      "mean_token_accuracy": 0.98052077293396,
+      "num_tokens": 27157631.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8909463047981262,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.4205769300460815,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.049406200647354126,
+      "mean_token_accuracy": 0.979697072505951,
+      "num_tokens": 27345921.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8969364166259766,
+      "epoch": 1.965006729475101,
+      "grad_norm": 0.9517145752906799,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0510346531867981,
+      "mean_token_accuracy": 0.9793386399745941,
+      "num_tokens": 27534751.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.9092479169368743,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.200056791305542,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.04974203705787659,
+      "mean_token_accuracy": 0.9800146698951722,
+      "num_tokens": 27723763.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9059641897678375,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.072447657585144,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.05065792202949524,
+      "mean_token_accuracy": 0.9799223959445953,
+      "num_tokens": 27911784.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9083398975384464,
+      "eval_loss": 0.05432562157511711,
+      "eval_mean_token_accuracy": 0.9782073607869969,
+      "eval_num_tokens": 28025286.0,
+      "eval_runtime": 24.4304,
+      "eval_samples_per_second": 204.663,
+      "eval_steps_per_second": 6.426,
+      "step": 1486
+    },
+    {
+      "entropy": 0.9113237977027893,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.0167063474655151,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.044942992925643924,
+      "mean_token_accuracy": 0.9818171083927154,
+      "num_tokens": 28100906.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.8780008792877197,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.9552241563796997,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.036248764395713805,
+      "mean_token_accuracy": 0.9852245450019836,
+      "num_tokens": 28289445.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.8721176147460937,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 0.8951388001441956,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.037987279891967776,
+      "mean_token_accuracy": 0.9850614190101623,
+      "num_tokens": 28477972.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.8922657191753387,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 0.9028105735778809,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.03588914275169373,
+      "mean_token_accuracy": 0.9863102614879609,
+      "num_tokens": 28666272.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.877799642086029,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.153886318206787,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.03577309250831604,
+      "mean_token_accuracy": 0.9863148391246795,
+      "num_tokens": 28854780.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.8708971381187439,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.042955994606018,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.038266432285308835,
+      "mean_token_accuracy": 0.9848419308662415,
+      "num_tokens": 29043741.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.8800721943378449,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 0.6789060235023499,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.032948991656303404,
+      "mean_token_accuracy": 0.9871293902397156,
+      "num_tokens": 29232719.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.8743580460548401,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.1780879497528076,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.036874374747276305,
+      "mean_token_accuracy": 0.9859182178974152,
+      "num_tokens": 29421817.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.8851303637027741,
+      "epoch": 2.113055181695828,
+      "grad_norm": 0.747734785079956,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.0364631175994873,
+      "mean_token_accuracy": 0.9862085223197937,
+      "num_tokens": 29610344.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.8798732101917267,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.5001860857009888,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.03813003897666931,
+      "mean_token_accuracy": 0.9852604746818543,
+      "num_tokens": 29798997.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.8797551989555359,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 0.9593478441238403,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.0394927829504013,
+      "mean_token_accuracy": 0.984604275226593,
+      "num_tokens": 29987809.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.8860546290874481,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.8347703218460083,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.036699697375297546,
+      "mean_token_accuracy": 0.985380882024765,
+      "num_tokens": 30176234.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.8783667802810669,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1154481172561646,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.03756999969482422,
+      "mean_token_accuracy": 0.9853451430797577,
+      "num_tokens": 30364657.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.8684478521347045,
+      "epoch": 2.180349932705249,
+      "grad_norm": 0.7515475153923035,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.0382376104593277,
+      "mean_token_accuracy": 0.9854029655456543,
+      "num_tokens": 30552904.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.8636964917182922,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 0.7711939215660095,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.036220991611480714,
+      "mean_token_accuracy": 0.9860713243484497,
+      "num_tokens": 30741259.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.8592016279697419,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.1143887042999268,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.036316031217575075,
+      "mean_token_accuracy": 0.985757052898407,
+      "num_tokens": 30930144.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.8671079576015472,
+      "epoch": 2.220726783310902,
+      "grad_norm": 0.8980015516281128,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.03822658061981201,
+      "mean_token_accuracy": 0.9847764372825623,
+      "num_tokens": 31118624.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.8785548269748688,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.094323992729187,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.03736622333526611,
+      "mean_token_accuracy": 0.9850081980228425,
+      "num_tokens": 31306833.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.866977310180664,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8372092843055725,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.0401554524898529,
+      "mean_token_accuracy": 0.9840337932109833,
+      "num_tokens": 31495503.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.8787918210029602,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.5408164262771606,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0368030846118927,
+      "mean_token_accuracy": 0.9847234487533569,
+      "num_tokens": 31683865.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.872721153497696,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.1115421056747437,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.037859299778938295,
+      "mean_token_accuracy": 0.9851646661758423,
+      "num_tokens": 31871903.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.8774094223976135,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.0301331281661987,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.038671016693115234,
+      "mean_token_accuracy": 0.9840058028697968,
+      "num_tokens": 32060500.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.889606237411499,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8601903915405273,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.039603835344314574,
+      "mean_token_accuracy": 0.9840235590934754,
+      "num_tokens": 32248744.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.886734277009964,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.1029839515686035,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.039513933658599856,
+      "mean_token_accuracy": 0.9844573020935059,
+      "num_tokens": 32437640.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.8843017637729644,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.2545154094696045,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.036970189213752745,
+      "mean_token_accuracy": 0.985239815711975,
+      "num_tokens": 32625986.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.8913422048091888,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.9613803625106812,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.03811657428741455,
+      "mean_token_accuracy": 0.9852804243564606,
+      "num_tokens": 32814531.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.8907467782497406,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.1811398267745972,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.03903660774230957,
+      "mean_token_accuracy": 0.9842524945735931,
+      "num_tokens": 33002960.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.8833664715290069,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.8338477611541748,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.03982087969779968,
+      "mean_token_accuracy": 0.9841201484203339,
+      "num_tokens": 33191183.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.8783826470375061,
+      "epoch": 2.382234185733513,
+      "grad_norm": 0.6904510259628296,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.03699290752410889,
+      "mean_token_accuracy": 0.9854120731353759,
+      "num_tokens": 33379798.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.8871048510074615,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.009539246559143,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.03926805555820465,
+      "mean_token_accuracy": 0.9843246698379516,
+      "num_tokens": 33568531.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.8973462700843811,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.8916401267051697,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.039596831798553465,
+      "mean_token_accuracy": 0.9839228212833404,
+      "num_tokens": 33756333.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.8954619467258453,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.056504487991333,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0396859347820282,
+      "mean_token_accuracy": 0.9837161123752594,
+      "num_tokens": 33944904.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.8952643811702728,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0442084074020386,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.037324142456054685,
+      "mean_token_accuracy": 0.9852897703647614,
+      "num_tokens": 34133530.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.8874911010265351,
+      "epoch": 2.449528936742934,
+      "grad_norm": 0.9926224946975708,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.03637495338916778,
+      "mean_token_accuracy": 0.9853426337242126,
+      "num_tokens": 34322042.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.8900584518909455,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.001734733581543,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04029585719108582,
+      "mean_token_accuracy": 0.9842372059822082,
+      "num_tokens": 34510259.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.8986345052719116,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 0.772941529750824,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.0401084691286087,
+      "mean_token_accuracy": 0.9841427087783814,
+      "num_tokens": 34698669.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.8911370873451233,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.895031213760376,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.03941032290458679,
+      "mean_token_accuracy": 0.9841966688632965,
+      "num_tokens": 34887160.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.8901963472366333,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0677918195724487,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.03503885865211487,
+      "mean_token_accuracy": 0.9856241464614868,
+      "num_tokens": 35075891.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.8817630231380462,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.1200164556503296,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.03878425657749176,
+      "mean_token_accuracy": 0.9846492648124695,
+      "num_tokens": 35264455.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.8897477686405182,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.8058190941810608,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.041558006405830385,
+      "mean_token_accuracy": 0.9830883860588073,
+      "num_tokens": 35453251.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.8873181045055389,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.7765992283821106,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.03555050492286682,
+      "mean_token_accuracy": 0.9858225345611572,
+      "num_tokens": 35641959.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.8782606959342957,
+      "epoch": 2.557200538358008,
+      "grad_norm": 0.8488866686820984,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.03606923818588257,
+      "mean_token_accuracy": 0.9858632445335388,
+      "num_tokens": 35829636.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.8869808673858642,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.1734739542007446,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.040956351161003116,
+      "mean_token_accuracy": 0.9835374176502227,
+      "num_tokens": 36017818.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.8850924432277679,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.3584387302398682,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.042437011003494264,
+      "mean_token_accuracy": 0.9826479077339172,
+      "num_tokens": 36206505.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.8801687896251679,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2862240076065063,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.03880060315132141,
+      "mean_token_accuracy": 0.9843404710292816,
+      "num_tokens": 36395802.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.8752781748771667,
+      "epoch": 2.611036339165545,
+      "grad_norm": 0.7341249585151672,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.03776344656944275,
+      "mean_token_accuracy": 0.9852877616882324,
+      "num_tokens": 36584370.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.8759494423866272,
+      "epoch": 2.624495289367429,
+      "grad_norm": 0.9807276129722595,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.03780297338962555,
+      "mean_token_accuracy": 0.9853227376937866,
+      "num_tokens": 36772865.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.8753129065036773,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.1391628980636597,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.03953765034675598,
+      "mean_token_accuracy": 0.9839197635650635,
+      "num_tokens": 36961462.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.8771899223327637,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.0865391492843628,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.03856399655342102,
+      "mean_token_accuracy": 0.9848688066005706,
+      "num_tokens": 37150486.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.8774881184101104,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.8561595678329468,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.03795175850391388,
+      "mean_token_accuracy": 0.9849309325218201,
+      "num_tokens": 37338890.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.8841517448425293,
+      "epoch": 2.678331090174966,
+      "grad_norm": 0.9458399415016174,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.03860213160514832,
+      "mean_token_accuracy": 0.9842230796813964,
+      "num_tokens": 37526886.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.8809043228626251,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 0.9360871315002441,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.03709094822406769,
+      "mean_token_accuracy": 0.9855118036270142,
+      "num_tokens": 37715408.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.8805519282817841,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.8852124214172363,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.03869341611862183,
+      "mean_token_accuracy": 0.9846724212169647,
+      "num_tokens": 37904328.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.8796228408813477,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.189297080039978,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.03675769567489624,
+      "mean_token_accuracy": 0.9854080140590668,
+      "num_tokens": 38092776.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.8722020089626312,
+      "epoch": 2.732166890982503,
+      "grad_norm": 0.7027947306632996,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.03889244794845581,
+      "mean_token_accuracy": 0.984558242559433,
+      "num_tokens": 38282221.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.8746632814407349,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 0.9799376726150513,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.03815680146217346,
+      "mean_token_accuracy": 0.9845012128353119,
+      "num_tokens": 38470633.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.8766567528247833,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 0.850459098815918,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.039029371738433835,
+      "mean_token_accuracy": 0.9845906794071198,
+      "num_tokens": 38659048.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.8778688013553619,
+      "epoch": 2.772543741588156,
+      "grad_norm": 0.8540134429931641,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.039645448327064514,
+      "mean_token_accuracy": 0.9840566098690033,
+      "num_tokens": 38847370.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.8758600771427154,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8036021590232849,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.038226932287216187,
+      "mean_token_accuracy": 0.9845647215843201,
+      "num_tokens": 39036040.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.8652268946170807,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 0.8526738286018372,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.03798363208770752,
+      "mean_token_accuracy": 0.9845202267169952,
+      "num_tokens": 39224695.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.8707200348377228,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 0.9852389097213745,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.037227436900138855,
+      "mean_token_accuracy": 0.9852050960063934,
+      "num_tokens": 39413360.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.8711447060108185,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.228277325630188,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.03698050379753113,
+      "mean_token_accuracy": 0.9852443158626556,
+      "num_tokens": 39602544.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.8742912471294403,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 0.8440544605255127,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.03757670521736145,
+      "mean_token_accuracy": 0.9848618268966675,
+      "num_tokens": 39791340.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.8801726162433624,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0955252647399902,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.040004149079322815,
+      "mean_token_accuracy": 0.9841345012187958,
+      "num_tokens": 39980212.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.8931492984294891,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.5722652673721313,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.038096648454666135,
+      "mean_token_accuracy": 0.9849136590957641,
+      "num_tokens": 40169122.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.8767493844032288,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.7663730382919312,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.03729096055030823,
+      "mean_token_accuracy": 0.985169780254364,
+      "num_tokens": 40357678.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.8770561516284943,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.1505672931671143,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.03984796404838562,
+      "mean_token_accuracy": 0.9838845670223236,
+      "num_tokens": 40546892.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.8789129912853241,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 0.8355761766433716,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.03769223690032959,
+      "mean_token_accuracy": 0.9847559213638306,
+      "num_tokens": 40736109.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.8781779289245606,
+      "epoch": 2.920592193808883,
+      "grad_norm": 0.9103918075561523,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.038262826204299924,
+      "mean_token_accuracy": 0.984741848707199,
+      "num_tokens": 40924341.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.8552807629108429,
+      "epoch": 2.934051144010767,
+      "grad_norm": 0.8338510990142822,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.0355743408203125,
+      "mean_token_accuracy": 0.9855036079883576,
+      "num_tokens": 41113849.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.8705591142177582,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 0.8334057927131653,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.037476515769958495,
+      "mean_token_accuracy": 0.9854551255702972,
+      "num_tokens": 41303130.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.8809274792671203,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 0.9424103498458862,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.03624279499053955,
+      "mean_token_accuracy": 0.9856135487556458,
+      "num_tokens": 41491750.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.8818133771419525,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.04243004322052,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.039132320880889894,
+      "mean_token_accuracy": 0.984746390581131,
+      "num_tokens": 41680493.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.8838395297527313,
+      "epoch": 2.987886944818304,
+      "grad_norm": 0.8650784492492676,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.03822052478790283,
+      "mean_token_accuracy": 0.9852108955383301,
+      "num_tokens": 41868378.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.8800126549544608,
+      "eval_loss": 0.05468416586518288,
+      "eval_mean_token_accuracy": 0.9781015617832257,
+      "eval_num_tokens": 42038037.0,
+      "eval_runtime": 24.4825,
+      "eval_samples_per_second": 204.228,
+      "eval_steps_per_second": 6.413,
+      "step": 2229
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.0311131063058432e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2229/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-2229/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2229/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2972/README.md b/baby_talk_L16_a50/seed_42/checkpoint-2972/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2972/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2972/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-2972/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2972/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2972/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-2972/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..615170194a8f1981b59442fda9b01cd0851a4657
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2972/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2515d21f30afa1abce215a6e8019acc54800da53f5bc3c4775850dda77992ce
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2972/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-2972/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2972/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2972/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-2972/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2972/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2972/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-2972/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2972/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2972/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-2972/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..78f48151e36019e3ee748db5a387cb5c492c2cd0
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2972/trainer_state.json
@@ -0,0 +1,3048 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 2972,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    },
+    {
+      "entropy": 0.9689933776855468,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9720287919044495,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.055563896894454956,
+      "mean_token_accuracy": 0.9779708027839661,
+      "num_tokens": 14144652.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.9506655633449554,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 0.6619595289230347,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.050587379932403566,
+      "mean_token_accuracy": 0.979685264825821,
+      "num_tokens": 14332714.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.9458732306957245,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 0.7905108332633972,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.05069155097007751,
+      "mean_token_accuracy": 0.9796596229076385,
+      "num_tokens": 14520902.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.9320916712284089,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 0.9210566878318787,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.051744121313095096,
+      "mean_token_accuracy": 0.9792148530483246,
+      "num_tokens": 14709215.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.928934782743454,
+      "epoch": 1.063257065948856,
+      "grad_norm": 0.7442036867141724,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.0486875057220459,
+      "mean_token_accuracy": 0.9798122465610504,
+      "num_tokens": 14898242.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.942881977558136,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.0910941362380981,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.052297019958496095,
+      "mean_token_accuracy": 0.9793200254440307,
+      "num_tokens": 15086835.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.9503917813301086,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 0.8091554641723633,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.049927744269371035,
+      "mean_token_accuracy": 0.9802082359790802,
+      "num_tokens": 15275042.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9426830470561981,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.0589245557785034,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.053803551197052005,
+      "mean_token_accuracy": 0.9785399675369263,
+      "num_tokens": 15463923.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.9445010781288147,
+      "epoch": 1.117092866756393,
+      "grad_norm": 0.8410441279411316,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.053181976079940796,
+      "mean_token_accuracy": 0.9792523324489594,
+      "num_tokens": 15652526.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.946729838848114,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 0.8246080279350281,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.054291915893554685,
+      "mean_token_accuracy": 0.9791357696056366,
+      "num_tokens": 15841340.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.9433728814125061,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 0.9200296401977539,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.05073915719985962,
+      "mean_token_accuracy": 0.9796931505203247,
+      "num_tokens": 16029478.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.9494417488574982,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 0.8342758417129517,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.050929927825927736,
+      "mean_token_accuracy": 0.9793219089508056,
+      "num_tokens": 16218014.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.9409256100654602,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.7911009192466736,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.049964362382888795,
+      "mean_token_accuracy": 0.9798487305641175,
+      "num_tokens": 16406420.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.9429958581924438,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.4877018928527832,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.04969423711299896,
+      "mean_token_accuracy": 0.9801322638988494,
+      "num_tokens": 16594948.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.9542136013507843,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 0.8030785918235779,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.04981146454811096,
+      "mean_token_accuracy": 0.9796162784099579,
+      "num_tokens": 16782983.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.9367722630500793,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.1169800758361816,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.05210963487625122,
+      "mean_token_accuracy": 0.9790139615535736,
+      "num_tokens": 16971394.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.9487208306789399,
+      "epoch": 1.224764468371467,
+      "grad_norm": 0.8650282621383667,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.05012243390083313,
+      "mean_token_accuracy": 0.9805107891559601,
+      "num_tokens": 17159951.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.9441231489181519,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 0.8193967938423157,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.048866665363311766,
+      "mean_token_accuracy": 0.9807584881782532,
+      "num_tokens": 17348711.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.9361630439758301,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3613612651824951,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.05324091911315918,
+      "mean_token_accuracy": 0.97904953956604,
+      "num_tokens": 17537116.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.9562997639179229,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.8121878504753113,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.05011019706726074,
+      "mean_token_accuracy": 0.9798731088638306,
+      "num_tokens": 17725567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.9359076261520386,
+      "epoch": 1.278600269179004,
+      "grad_norm": 0.7838549613952637,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.04834386110305786,
+      "mean_token_accuracy": 0.9804529249668121,
+      "num_tokens": 17914191.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.9527446150779724,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0981712341308594,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.055077624320983884,
+      "mean_token_accuracy": 0.9776535391807556,
+      "num_tokens": 18103184.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.9574812948703766,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 0.8291501402854919,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.050153911113739014,
+      "mean_token_accuracy": 0.9792965054512024,
+      "num_tokens": 18291351.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.939585280418396,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.0864412784576416,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.05132197737693787,
+      "mean_token_accuracy": 0.9792291462421417,
+      "num_tokens": 18480192.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.9468406856060028,
+      "epoch": 1.332436069986541,
+      "grad_norm": 0.9772785902023315,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.04844954013824463,
+      "mean_token_accuracy": 0.9804142415523529,
+      "num_tokens": 18669028.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.9413834273815155,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 0.9020094871520996,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.05075312852859497,
+      "mean_token_accuracy": 0.9790591120719909,
+      "num_tokens": 18857581.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.9526253461837768,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.0524990558624268,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.049737372994422914,
+      "mean_token_accuracy": 0.9795652389526367,
+      "num_tokens": 19046105.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.9484909176826477,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.0715620517730713,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.04935494959354401,
+      "mean_token_accuracy": 0.9797740161418915,
+      "num_tokens": 19234730.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.9398573458194732,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2134987115859985,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.0518394410610199,
+      "mean_token_accuracy": 0.9793788075447083,
+      "num_tokens": 19423572.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.9465648651123046,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.004693627357483,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.04776117205619812,
+      "mean_token_accuracy": 0.9813261866569519,
+      "num_tokens": 19612465.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.9448004186153411,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.2913438081741333,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.04807930588722229,
+      "mean_token_accuracy": 0.9805689513683319,
+      "num_tokens": 19800482.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.9425322234630584,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1068469285964966,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.05053595900535583,
+      "mean_token_accuracy": 0.9794536828994751,
+      "num_tokens": 19989444.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.9472232103347779,
+      "epoch": 1.440107671601615,
+      "grad_norm": 0.7409384250640869,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.052136778831481934,
+      "mean_token_accuracy": 0.979197359085083,
+      "num_tokens": 20178482.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9450975477695465,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.3440778255462646,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.05258495211601257,
+      "mean_token_accuracy": 0.9798152863979339,
+      "num_tokens": 20367345.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9468007743358612,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 0.8377829194068909,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.04893603026866913,
+      "mean_token_accuracy": 0.9805733859539032,
+      "num_tokens": 20555817.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.9505827724933624,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.776889979839325,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.05367119312286377,
+      "mean_token_accuracy": 0.9779646098613739,
+      "num_tokens": 20744001.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.9516380190849304,
+      "epoch": 1.493943472409152,
+      "grad_norm": 0.8598276376724243,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.05071394443511963,
+      "mean_token_accuracy": 0.979692417383194,
+      "num_tokens": 20931715.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.9351628720760345,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3307439088821411,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.051372635364532473,
+      "mean_token_accuracy": 0.9797907948493958,
+      "num_tokens": 21120803.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.9268040716648102,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9320250749588013,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.04939974546432495,
+      "mean_token_accuracy": 0.98038170337677,
+      "num_tokens": 21309384.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9241463243961334,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.065938115119934,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.04856643378734589,
+      "mean_token_accuracy": 0.9800511121749877,
+      "num_tokens": 21498234.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.9350252687931061,
+      "epoch": 1.547779273216689,
+      "grad_norm": 0.6925032138824463,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.0512526273727417,
+      "mean_token_accuracy": 0.979562646150589,
+      "num_tokens": 21687119.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.9403733670711517,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.2069101333618164,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.049795085191726686,
+      "mean_token_accuracy": 0.9806506633758545,
+      "num_tokens": 21876094.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9370538294315338,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.05461466312408447,
+      "mean_token_accuracy": 0.9776057600975037,
+      "num_tokens": 22064836.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9517778217792511,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 0.8965149521827698,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.05289326906204224,
+      "mean_token_accuracy": 0.9782088756561279,
+      "num_tokens": 22253065.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.9372851848602295,
+      "epoch": 1.601615074024226,
+      "grad_norm": 0.7192108631134033,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.048973691463470456,
+      "mean_token_accuracy": 0.9806210815906524,
+      "num_tokens": 22441206.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.9122351944446564,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 0.8763614296913147,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.0461783230304718,
+      "mean_token_accuracy": 0.9817197680473327,
+      "num_tokens": 22629637.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.9162654876708984,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 0.9806778430938721,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.04827338755130768,
+      "mean_token_accuracy": 0.9806499361991883,
+      "num_tokens": 22818369.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.9125026524066925,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 0.9685350656509399,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.048379385471343996,
+      "mean_token_accuracy": 0.980905145406723,
+      "num_tokens": 23006863.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9119791567325592,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.8717457056045532,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.050873959064483644,
+      "mean_token_accuracy": 0.979601925611496,
+      "num_tokens": 23195735.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.9244007229804992,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.0574654340744019,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.05112813711166382,
+      "mean_token_accuracy": 0.979493772983551,
+      "num_tokens": 23383862.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9225810945034028,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 0.9628807306289673,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.050480544567108154,
+      "mean_token_accuracy": 0.9797678411006927,
+      "num_tokens": 23572292.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.9287579476833343,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.015265941619873,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.05133126378059387,
+      "mean_token_accuracy": 0.9797986805438995,
+      "num_tokens": 23760252.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.9133845925331116,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 0.8996883630752563,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.05044950246810913,
+      "mean_token_accuracy": 0.9795303404331207,
+      "num_tokens": 23949031.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.9213698863983154,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.0490264892578125,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.04797698855400086,
+      "mean_token_accuracy": 0.9806023716926575,
+      "num_tokens": 24137829.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.9173697710037232,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.3648769855499268,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.049194514751434326,
+      "mean_token_accuracy": 0.980712479352951,
+      "num_tokens": 24326586.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9242741882801055,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.105125069618225,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.048667973279953,
+      "mean_token_accuracy": 0.9801449298858642,
+      "num_tokens": 24514904.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.9162627995014191,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.8887938857078552,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.04878672957420349,
+      "mean_token_accuracy": 0.980469423532486,
+      "num_tokens": 24703688.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.9149538338184356,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 0.8838976621627808,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.04892318844795227,
+      "mean_token_accuracy": 0.9808776795864105,
+      "num_tokens": 24892107.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.914735221862793,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1152070760726929,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.04825109839439392,
+      "mean_token_accuracy": 0.9804604113101959,
+      "num_tokens": 25080130.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9170725226402283,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9479517340660095,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.051445144414901736,
+      "mean_token_accuracy": 0.9792819261550904,
+      "num_tokens": 25268652.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9142911911010743,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 0.8172292709350586,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.0475692093372345,
+      "mean_token_accuracy": 0.9809505581855774,
+      "num_tokens": 25457120.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.9019249439239502,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 0.8183121681213379,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.04979957342147827,
+      "mean_token_accuracy": 0.9799730658531189,
+      "num_tokens": 25645699.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.9181066155433655,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.8256579041481018,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.04996164441108704,
+      "mean_token_accuracy": 0.9802247405052185,
+      "num_tokens": 25834183.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9171363770961761,
+      "epoch": 1.857335127860027,
+      "grad_norm": 0.9727098345756531,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.048660767078399655,
+      "mean_token_accuracy": 0.9805064260959625,
+      "num_tokens": 26023317.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.9112720847129822,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 0.9826673865318298,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.04775593280792236,
+      "mean_token_accuracy": 0.9809929549694061,
+      "num_tokens": 26212115.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9034272134304047,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.9951306581497192,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.04939360618591308,
+      "mean_token_accuracy": 0.9800347864627839,
+      "num_tokens": 26401167.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.897865754365921,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.0163404941558838,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05002856254577637,
+      "mean_token_accuracy": 0.9793690323829651,
+      "num_tokens": 26590570.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9004576802253723,
+      "epoch": 1.911170928667564,
+      "grad_norm": 0.6666616201400757,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.04905453026294708,
+      "mean_token_accuracy": 0.9803856253623963,
+      "num_tokens": 26779481.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.8978902101516724,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.9295416474342346,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.05261261463165283,
+      "mean_token_accuracy": 0.9785708487033844,
+      "num_tokens": 26968262.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.8962675571441651,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 0.88262939453125,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.04765265882015228,
+      "mean_token_accuracy": 0.98052077293396,
+      "num_tokens": 27157631.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8909463047981262,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.4205769300460815,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.049406200647354126,
+      "mean_token_accuracy": 0.979697072505951,
+      "num_tokens": 27345921.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8969364166259766,
+      "epoch": 1.965006729475101,
+      "grad_norm": 0.9517145752906799,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0510346531867981,
+      "mean_token_accuracy": 0.9793386399745941,
+      "num_tokens": 27534751.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.9092479169368743,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.200056791305542,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.04974203705787659,
+      "mean_token_accuracy": 0.9800146698951722,
+      "num_tokens": 27723763.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9059641897678375,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.072447657585144,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.05065792202949524,
+      "mean_token_accuracy": 0.9799223959445953,
+      "num_tokens": 27911784.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9083398975384464,
+      "eval_loss": 0.05432562157511711,
+      "eval_mean_token_accuracy": 0.9782073607869969,
+      "eval_num_tokens": 28025286.0,
+      "eval_runtime": 24.4304,
+      "eval_samples_per_second": 204.663,
+      "eval_steps_per_second": 6.426,
+      "step": 1486
+    },
+    {
+      "entropy": 0.9113237977027893,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.0167063474655151,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.044942992925643924,
+      "mean_token_accuracy": 0.9818171083927154,
+      "num_tokens": 28100906.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.8780008792877197,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.9552241563796997,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.036248764395713805,
+      "mean_token_accuracy": 0.9852245450019836,
+      "num_tokens": 28289445.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.8721176147460937,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 0.8951388001441956,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.037987279891967776,
+      "mean_token_accuracy": 0.9850614190101623,
+      "num_tokens": 28477972.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.8922657191753387,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 0.9028105735778809,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.03588914275169373,
+      "mean_token_accuracy": 0.9863102614879609,
+      "num_tokens": 28666272.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.877799642086029,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.153886318206787,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.03577309250831604,
+      "mean_token_accuracy": 0.9863148391246795,
+      "num_tokens": 28854780.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.8708971381187439,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.042955994606018,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.038266432285308835,
+      "mean_token_accuracy": 0.9848419308662415,
+      "num_tokens": 29043741.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.8800721943378449,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 0.6789060235023499,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.032948991656303404,
+      "mean_token_accuracy": 0.9871293902397156,
+      "num_tokens": 29232719.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.8743580460548401,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.1780879497528076,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.036874374747276305,
+      "mean_token_accuracy": 0.9859182178974152,
+      "num_tokens": 29421817.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.8851303637027741,
+      "epoch": 2.113055181695828,
+      "grad_norm": 0.747734785079956,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.0364631175994873,
+      "mean_token_accuracy": 0.9862085223197937,
+      "num_tokens": 29610344.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.8798732101917267,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.5001860857009888,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.03813003897666931,
+      "mean_token_accuracy": 0.9852604746818543,
+      "num_tokens": 29798997.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.8797551989555359,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 0.9593478441238403,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.0394927829504013,
+      "mean_token_accuracy": 0.984604275226593,
+      "num_tokens": 29987809.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.8860546290874481,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.8347703218460083,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.036699697375297546,
+      "mean_token_accuracy": 0.985380882024765,
+      "num_tokens": 30176234.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.8783667802810669,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1154481172561646,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.03756999969482422,
+      "mean_token_accuracy": 0.9853451430797577,
+      "num_tokens": 30364657.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.8684478521347045,
+      "epoch": 2.180349932705249,
+      "grad_norm": 0.7515475153923035,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.0382376104593277,
+      "mean_token_accuracy": 0.9854029655456543,
+      "num_tokens": 30552904.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.8636964917182922,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 0.7711939215660095,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.036220991611480714,
+      "mean_token_accuracy": 0.9860713243484497,
+      "num_tokens": 30741259.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.8592016279697419,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.1143887042999268,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.036316031217575075,
+      "mean_token_accuracy": 0.985757052898407,
+      "num_tokens": 30930144.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.8671079576015472,
+      "epoch": 2.220726783310902,
+      "grad_norm": 0.8980015516281128,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.03822658061981201,
+      "mean_token_accuracy": 0.9847764372825623,
+      "num_tokens": 31118624.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.8785548269748688,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.094323992729187,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.03736622333526611,
+      "mean_token_accuracy": 0.9850081980228425,
+      "num_tokens": 31306833.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.866977310180664,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8372092843055725,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.0401554524898529,
+      "mean_token_accuracy": 0.9840337932109833,
+      "num_tokens": 31495503.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.8787918210029602,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.5408164262771606,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0368030846118927,
+      "mean_token_accuracy": 0.9847234487533569,
+      "num_tokens": 31683865.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.872721153497696,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.1115421056747437,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.037859299778938295,
+      "mean_token_accuracy": 0.9851646661758423,
+      "num_tokens": 31871903.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.8774094223976135,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.0301331281661987,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.038671016693115234,
+      "mean_token_accuracy": 0.9840058028697968,
+      "num_tokens": 32060500.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.889606237411499,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8601903915405273,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.039603835344314574,
+      "mean_token_accuracy": 0.9840235590934754,
+      "num_tokens": 32248744.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.886734277009964,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.1029839515686035,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.039513933658599856,
+      "mean_token_accuracy": 0.9844573020935059,
+      "num_tokens": 32437640.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.8843017637729644,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.2545154094696045,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.036970189213752745,
+      "mean_token_accuracy": 0.985239815711975,
+      "num_tokens": 32625986.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.8913422048091888,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.9613803625106812,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.03811657428741455,
+      "mean_token_accuracy": 0.9852804243564606,
+      "num_tokens": 32814531.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.8907467782497406,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.1811398267745972,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.03903660774230957,
+      "mean_token_accuracy": 0.9842524945735931,
+      "num_tokens": 33002960.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.8833664715290069,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.8338477611541748,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.03982087969779968,
+      "mean_token_accuracy": 0.9841201484203339,
+      "num_tokens": 33191183.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.8783826470375061,
+      "epoch": 2.382234185733513,
+      "grad_norm": 0.6904510259628296,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.03699290752410889,
+      "mean_token_accuracy": 0.9854120731353759,
+      "num_tokens": 33379798.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.8871048510074615,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.009539246559143,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.03926805555820465,
+      "mean_token_accuracy": 0.9843246698379516,
+      "num_tokens": 33568531.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.8973462700843811,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.8916401267051697,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.039596831798553465,
+      "mean_token_accuracy": 0.9839228212833404,
+      "num_tokens": 33756333.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.8954619467258453,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.056504487991333,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0396859347820282,
+      "mean_token_accuracy": 0.9837161123752594,
+      "num_tokens": 33944904.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.8952643811702728,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0442084074020386,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.037324142456054685,
+      "mean_token_accuracy": 0.9852897703647614,
+      "num_tokens": 34133530.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.8874911010265351,
+      "epoch": 2.449528936742934,
+      "grad_norm": 0.9926224946975708,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.03637495338916778,
+      "mean_token_accuracy": 0.9853426337242126,
+      "num_tokens": 34322042.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.8900584518909455,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.001734733581543,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04029585719108582,
+      "mean_token_accuracy": 0.9842372059822082,
+      "num_tokens": 34510259.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.8986345052719116,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 0.772941529750824,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.0401084691286087,
+      "mean_token_accuracy": 0.9841427087783814,
+      "num_tokens": 34698669.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.8911370873451233,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.895031213760376,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.03941032290458679,
+      "mean_token_accuracy": 0.9841966688632965,
+      "num_tokens": 34887160.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.8901963472366333,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0677918195724487,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.03503885865211487,
+      "mean_token_accuracy": 0.9856241464614868,
+      "num_tokens": 35075891.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.8817630231380462,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.1200164556503296,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.03878425657749176,
+      "mean_token_accuracy": 0.9846492648124695,
+      "num_tokens": 35264455.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.8897477686405182,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.8058190941810608,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.041558006405830385,
+      "mean_token_accuracy": 0.9830883860588073,
+      "num_tokens": 35453251.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.8873181045055389,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.7765992283821106,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.03555050492286682,
+      "mean_token_accuracy": 0.9858225345611572,
+      "num_tokens": 35641959.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.8782606959342957,
+      "epoch": 2.557200538358008,
+      "grad_norm": 0.8488866686820984,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.03606923818588257,
+      "mean_token_accuracy": 0.9858632445335388,
+      "num_tokens": 35829636.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.8869808673858642,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.1734739542007446,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.040956351161003116,
+      "mean_token_accuracy": 0.9835374176502227,
+      "num_tokens": 36017818.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.8850924432277679,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.3584387302398682,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.042437011003494264,
+      "mean_token_accuracy": 0.9826479077339172,
+      "num_tokens": 36206505.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.8801687896251679,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2862240076065063,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.03880060315132141,
+      "mean_token_accuracy": 0.9843404710292816,
+      "num_tokens": 36395802.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.8752781748771667,
+      "epoch": 2.611036339165545,
+      "grad_norm": 0.7341249585151672,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.03776344656944275,
+      "mean_token_accuracy": 0.9852877616882324,
+      "num_tokens": 36584370.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.8759494423866272,
+      "epoch": 2.624495289367429,
+      "grad_norm": 0.9807276129722595,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.03780297338962555,
+      "mean_token_accuracy": 0.9853227376937866,
+      "num_tokens": 36772865.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.8753129065036773,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.1391628980636597,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.03953765034675598,
+      "mean_token_accuracy": 0.9839197635650635,
+      "num_tokens": 36961462.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.8771899223327637,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.0865391492843628,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.03856399655342102,
+      "mean_token_accuracy": 0.9848688066005706,
+      "num_tokens": 37150486.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.8774881184101104,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.8561595678329468,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.03795175850391388,
+      "mean_token_accuracy": 0.9849309325218201,
+      "num_tokens": 37338890.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.8841517448425293,
+      "epoch": 2.678331090174966,
+      "grad_norm": 0.9458399415016174,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.03860213160514832,
+      "mean_token_accuracy": 0.9842230796813964,
+      "num_tokens": 37526886.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.8809043228626251,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 0.9360871315002441,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.03709094822406769,
+      "mean_token_accuracy": 0.9855118036270142,
+      "num_tokens": 37715408.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.8805519282817841,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.8852124214172363,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.03869341611862183,
+      "mean_token_accuracy": 0.9846724212169647,
+      "num_tokens": 37904328.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.8796228408813477,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.189297080039978,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.03675769567489624,
+      "mean_token_accuracy": 0.9854080140590668,
+      "num_tokens": 38092776.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.8722020089626312,
+      "epoch": 2.732166890982503,
+      "grad_norm": 0.7027947306632996,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.03889244794845581,
+      "mean_token_accuracy": 0.984558242559433,
+      "num_tokens": 38282221.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.8746632814407349,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 0.9799376726150513,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.03815680146217346,
+      "mean_token_accuracy": 0.9845012128353119,
+      "num_tokens": 38470633.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.8766567528247833,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 0.850459098815918,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.039029371738433835,
+      "mean_token_accuracy": 0.9845906794071198,
+      "num_tokens": 38659048.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.8778688013553619,
+      "epoch": 2.772543741588156,
+      "grad_norm": 0.8540134429931641,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.039645448327064514,
+      "mean_token_accuracy": 0.9840566098690033,
+      "num_tokens": 38847370.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.8758600771427154,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8036021590232849,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.038226932287216187,
+      "mean_token_accuracy": 0.9845647215843201,
+      "num_tokens": 39036040.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.8652268946170807,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 0.8526738286018372,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.03798363208770752,
+      "mean_token_accuracy": 0.9845202267169952,
+      "num_tokens": 39224695.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.8707200348377228,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 0.9852389097213745,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.037227436900138855,
+      "mean_token_accuracy": 0.9852050960063934,
+      "num_tokens": 39413360.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.8711447060108185,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.228277325630188,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.03698050379753113,
+      "mean_token_accuracy": 0.9852443158626556,
+      "num_tokens": 39602544.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.8742912471294403,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 0.8440544605255127,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.03757670521736145,
+      "mean_token_accuracy": 0.9848618268966675,
+      "num_tokens": 39791340.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.8801726162433624,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0955252647399902,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.040004149079322815,
+      "mean_token_accuracy": 0.9841345012187958,
+      "num_tokens": 39980212.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.8931492984294891,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.5722652673721313,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.038096648454666135,
+      "mean_token_accuracy": 0.9849136590957641,
+      "num_tokens": 40169122.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.8767493844032288,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.7663730382919312,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.03729096055030823,
+      "mean_token_accuracy": 0.985169780254364,
+      "num_tokens": 40357678.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.8770561516284943,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.1505672931671143,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.03984796404838562,
+      "mean_token_accuracy": 0.9838845670223236,
+      "num_tokens": 40546892.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.8789129912853241,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 0.8355761766433716,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.03769223690032959,
+      "mean_token_accuracy": 0.9847559213638306,
+      "num_tokens": 40736109.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.8781779289245606,
+      "epoch": 2.920592193808883,
+      "grad_norm": 0.9103918075561523,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.038262826204299924,
+      "mean_token_accuracy": 0.984741848707199,
+      "num_tokens": 40924341.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.8552807629108429,
+      "epoch": 2.934051144010767,
+      "grad_norm": 0.8338510990142822,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.0355743408203125,
+      "mean_token_accuracy": 0.9855036079883576,
+      "num_tokens": 41113849.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.8705591142177582,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 0.8334057927131653,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.037476515769958495,
+      "mean_token_accuracy": 0.9854551255702972,
+      "num_tokens": 41303130.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.8809274792671203,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 0.9424103498458862,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.03624279499053955,
+      "mean_token_accuracy": 0.9856135487556458,
+      "num_tokens": 41491750.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.8818133771419525,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.04243004322052,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.039132320880889894,
+      "mean_token_accuracy": 0.984746390581131,
+      "num_tokens": 41680493.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.8838395297527313,
+      "epoch": 2.987886944818304,
+      "grad_norm": 0.8650784492492676,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.03822052478790283,
+      "mean_token_accuracy": 0.9852108955383301,
+      "num_tokens": 41868378.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.8800126549544608,
+      "eval_loss": 0.05468416586518288,
+      "eval_mean_token_accuracy": 0.9781015617832257,
+      "eval_num_tokens": 42038037.0,
+      "eval_runtime": 24.4825,
+      "eval_samples_per_second": 204.228,
+      "eval_steps_per_second": 6.413,
+      "step": 2229
+    },
+    {
+      "entropy": 0.8773060619831086,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.7845072150230408,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.0365300714969635,
+      "mean_token_accuracy": 0.985853761434555,
+      "num_tokens": 42056717.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.8720550239086151,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 0.9917275905609131,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.030071893334388734,
+      "mean_token_accuracy": 0.9890478909015655,
+      "num_tokens": 42245058.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.8489451467990875,
+      "epoch": 3.028263795423957,
+      "grad_norm": 0.9414294362068176,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.027054613828659056,
+      "mean_token_accuracy": 0.9894806027412415,
+      "num_tokens": 42433468.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.848755830526352,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.7276235818862915,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.026897600293159483,
+      "mean_token_accuracy": 0.9893308103084564,
+      "num_tokens": 42621657.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.8402929604053497,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 0.9455838799476624,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.027918827533721925,
+      "mean_token_accuracy": 0.9888385117053986,
+      "num_tokens": 42810533.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.8434606611728668,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 0.7503429651260376,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.025769710540771484,
+      "mean_token_accuracy": 0.989881956577301,
+      "num_tokens": 42999085.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.8459890782833099,
+      "epoch": 3.082099596231494,
+      "grad_norm": 0.736440896987915,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.028781753778457642,
+      "mean_token_accuracy": 0.9882656812667847,
+      "num_tokens": 43188222.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.8493306159973144,
+      "epoch": 3.095558546433378,
+      "grad_norm": 0.9577816128730774,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.02689719796180725,
+      "mean_token_accuracy": 0.9889744818210602,
+      "num_tokens": 43376648.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.8512665331363678,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 0.9631588459014893,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.031439167261123654,
+      "mean_token_accuracy": 0.9874942421913147,
+      "num_tokens": 43564769.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.8519513130187988,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1011922359466553,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.02696958780288696,
+      "mean_token_accuracy": 0.9900951981544495,
+      "num_tokens": 43753362.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.8507618844509125,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9351623058319092,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.026315575838088988,
+      "mean_token_accuracy": 0.989511913061142,
+      "num_tokens": 43942063.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.8506478369235992,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.096564769744873,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.029049152135849,
+      "mean_token_accuracy": 0.9886228442192078,
+      "num_tokens": 44130381.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.8624941289424897,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 0.7859402298927307,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.027744990587234498,
+      "mean_token_accuracy": 0.9890527307987214,
+      "num_tokens": 44319243.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.8627866744995117,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 0.7764605283737183,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.02571650445461273,
+      "mean_token_accuracy": 0.9903442323207855,
+      "num_tokens": 44507698.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.8566473722457886,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.8182396292686462,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.025625643134117127,
+      "mean_token_accuracy": 0.9897150516510009,
+      "num_tokens": 44695954.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.8555088877677918,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.0277098417282104,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.028235083818435668,
+      "mean_token_accuracy": 0.9894341349601745,
+      "num_tokens": 44884558.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.8562702238559723,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.0676493644714355,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.027713948488235475,
+      "mean_token_accuracy": 0.9891053259372711,
+      "num_tokens": 45073381.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.8520576059818268,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 0.7926117181777954,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.028936928510665892,
+      "mean_token_accuracy": 0.988807988166809,
+      "num_tokens": 45262350.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.860496312379837,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.1108510494232178,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030533093214035033,
+      "mean_token_accuracy": 0.9882748067378998,
+      "num_tokens": 45450782.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.872324675321579,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.3057194948196411,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.032421305775642395,
+      "mean_token_accuracy": 0.9879753351211548,
+      "num_tokens": 45639122.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.8778823137283325,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.289680004119873,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03096393346786499,
+      "mean_token_accuracy": 0.9883370697498322,
+      "num_tokens": 45827730.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.8627406179904937,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.8361398577690125,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.02957139313220978,
+      "mean_token_accuracy": 0.9884674072265625,
+      "num_tokens": 46016274.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.8584554970264435,
+      "epoch": 3.297442799461642,
+      "grad_norm": 0.9330723881721497,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.029568272829055785,
+      "mean_token_accuracy": 0.9882686555385589,
+      "num_tokens": 46204679.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.8515178024768829,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.0296216011047363,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.027788180112838744,
+      "mean_token_accuracy": 0.9888734877109527,
+      "num_tokens": 46393989.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.8488850116729736,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1329699754714966,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.028057461977005003,
+      "mean_token_accuracy": 0.9895590543746948,
+      "num_tokens": 46582790.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.849625188112259,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 0.9267219305038452,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.02802973985671997,
+      "mean_token_accuracy": 0.988851535320282,
+      "num_tokens": 46771138.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.854446280002594,
+      "epoch": 3.351278600269179,
+      "grad_norm": 0.8713932037353516,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.030220019817352294,
+      "mean_token_accuracy": 0.9882843673229218,
+      "num_tokens": 46960054.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.8599806547164917,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.675019383430481,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.02958865463733673,
+      "mean_token_accuracy": 0.9880296885967255,
+      "num_tokens": 47149567.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.8549504458904267,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.2176353931427002,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.027722162008285523,
+      "mean_token_accuracy": 0.9890090465545655,
+      "num_tokens": 47338060.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.8552233815193176,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9571717977523804,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.028924965858459474,
+      "mean_token_accuracy": 0.9884298503398895,
+      "num_tokens": 47525895.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.8521630108356476,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.965713620185852,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.030511701107025148,
+      "mean_token_accuracy": 0.9881398677825928,
+      "num_tokens": 47714524.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.8537400126457214,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 0.9005307555198669,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.030432754755020143,
+      "mean_token_accuracy": 0.9876708447933197,
+      "num_tokens": 47903188.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.8596058666706086,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 0.8159075379371643,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.027927806973457335,
+      "mean_token_accuracy": 0.9889210939407349,
+      "num_tokens": 48091334.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.8482735753059387,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0411661863327026,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03003154993057251,
+      "mean_token_accuracy": 0.9882730603218078,
+      "num_tokens": 48279622.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.8466426789760589,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.189595341682434,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03237656056880951,
+      "mean_token_accuracy": 0.9869072616100312,
+      "num_tokens": 48467527.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.8545588493347168,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0700056552886963,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.0306060791015625,
+      "mean_token_accuracy": 0.9880664825439454,
+      "num_tokens": 48656151.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.8503433585166931,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.7099478244781494,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.02984013557434082,
+      "mean_token_accuracy": 0.9879442811012268,
+      "num_tokens": 48844796.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.8464376091957092,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9877128005027771,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.0306446373462677,
+      "mean_token_accuracy": 0.9883591532707214,
+      "num_tokens": 49033216.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.8477609276771545,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.3324531316757202,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.030148410797119142,
+      "mean_token_accuracy": 0.9879989802837372,
+      "num_tokens": 49221425.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.8461863219738006,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8938570022583008,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.02682304084300995,
+      "mean_token_accuracy": 0.9897881150245667,
+      "num_tokens": 49409785.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.8362689018249512,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 0.7423873543739319,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.02689332664012909,
+      "mean_token_accuracy": 0.9899903059005737,
+      "num_tokens": 49598524.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.8395139217376709,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.921755850315094,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.030062124133110046,
+      "mean_token_accuracy": 0.9880835592746735,
+      "num_tokens": 49787241.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.8467226803302765,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0296459197998047,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.030666223168373107,
+      "mean_token_accuracy": 0.9877098739147187,
+      "num_tokens": 49975844.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.833392071723938,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2861948013305664,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.028174835443496703,
+      "mean_token_accuracy": 0.989056545495987,
+      "num_tokens": 50164857.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.8340730309486389,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 0.8116776347160339,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.028390336036682128,
+      "mean_token_accuracy": 0.9886897742748261,
+      "num_tokens": 50353452.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.8357116639614105,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.8362084031105042,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.029806244373321533,
+      "mean_token_accuracy": 0.988309508562088,
+      "num_tokens": 50542461.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.8363270878791809,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8438499569892883,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.028577303886413573,
+      "mean_token_accuracy": 0.9886945366859436,
+      "num_tokens": 50730872.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.8310261785984039,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.1920002698898315,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.030321845412254335,
+      "mean_token_accuracy": 0.9878708899021149,
+      "num_tokens": 50919889.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.8380252480506897,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 0.9409399032592773,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.030501589179039,
+      "mean_token_accuracy": 0.9872083842754364,
+      "num_tokens": 51108079.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.8422612130641938,
+      "epoch": 3.660834454912517,
+      "grad_norm": 0.8370440006256104,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.02974823713302612,
+      "mean_token_accuracy": 0.9885554671287536,
+      "num_tokens": 51296913.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.8387595832347869,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1360642910003662,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.026970452070236205,
+      "mean_token_accuracy": 0.9888331353664398,
+      "num_tokens": 51485678.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.8403031289577484,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.1195677518844604,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.029451662302017213,
+      "mean_token_accuracy": 0.988847154378891,
+      "num_tokens": 51674765.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.8499106168746948,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 0.9573526978492737,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.030826866626739502,
+      "mean_token_accuracy": 0.9875578820705414,
+      "num_tokens": 51863924.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.8576709866523743,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9386286735534668,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.02946900427341461,
+      "mean_token_accuracy": 0.9885939955711365,
+      "num_tokens": 52052011.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.8404816091060638,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.095536708831787,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.030070826411247253,
+      "mean_token_accuracy": 0.9881400763988495,
+      "num_tokens": 52240955.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.8346689164638519,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9964426159858704,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03005717396736145,
+      "mean_token_accuracy": 0.9883057296276092,
+      "num_tokens": 52429544.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.8377436757087707,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.8499778509140015,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.029283815622329713,
+      "mean_token_accuracy": 0.9882387280464172,
+      "num_tokens": 52618405.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.8328788638114929,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9594390988349915,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.02666555345058441,
+      "mean_token_accuracy": 0.9889323055744171,
+      "num_tokens": 52806689.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.829415249824524,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.3587099313735962,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.028349629044532774,
+      "mean_token_accuracy": 0.9890907645225525,
+      "num_tokens": 52995248.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.8335795283317566,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.077789306640625,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.03001731038093567,
+      "mean_token_accuracy": 0.9878358781337738,
+      "num_tokens": 53183607.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.8316770553588867,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 0.7265951037406921,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.030136841535568237,
+      "mean_token_accuracy": 0.9880174934864044,
+      "num_tokens": 53372735.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.8310399711132049,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.6605978012084961,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.026665833592414857,
+      "mean_token_accuracy": 0.9892704486846924,
+      "num_tokens": 53561776.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.8258654296398162,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.6560797691345215,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.026718440651893615,
+      "mean_token_accuracy": 0.989847868680954,
+      "num_tokens": 53750019.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.8221413195133209,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 0.7495356798171997,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.027431756258010864,
+      "mean_token_accuracy": 0.9892430484294892,
+      "num_tokens": 53938914.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.8341928541660308,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.0676971673965454,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.028432759642601012,
+      "mean_token_accuracy": 0.9889024317264556,
+      "num_tokens": 54127763.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.8359149813652038,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.380476713180542,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.028149163722991942,
+      "mean_token_accuracy": 0.9884953200817108,
+      "num_tokens": 54315930.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.8294597804546356,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0728843212127686,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.028733691573143004,
+      "mean_token_accuracy": 0.9884523034095765,
+      "num_tokens": 54504632.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.8245523810386658,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 0.9723307490348816,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.02793261408805847,
+      "mean_token_accuracy": 0.9891038477420807,
+      "num_tokens": 54693752.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.8279620766639709,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.003624439239502,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.028367668390274048,
+      "mean_token_accuracy": 0.9893603324890137,
+      "num_tokens": 54882030.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.8306584179401397,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.8481361865997314,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.029952478408813477,
+      "mean_token_accuracy": 0.9881332099437714,
+      "num_tokens": 55070549.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.8316469848155975,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.7631049156188965,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.029351180791854857,
+      "mean_token_accuracy": 0.9888184070587158,
+      "num_tokens": 55258999.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.8221287310123444,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 0.7234840393066406,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.027709537744522096,
+      "mean_token_accuracy": 0.9890308439731598,
+      "num_tokens": 55448047.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.8239192545413971,
+      "epoch": 3.970390309555855,
+      "grad_norm": 0.8497020602226257,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.029604345560073853,
+      "mean_token_accuracy": 0.9881375730037689,
+      "num_tokens": 55636566.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.8319927275180816,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.1599552631378174,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.02763860821723938,
+      "mean_token_accuracy": 0.9891406774520874,
+      "num_tokens": 55825049.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.8404937744140625,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 0.906755805015564,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.029578033089637756,
+      "mean_token_accuracy": 0.9878996431827545,
+      "num_tokens": 56012965.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.8419649528849656,
+      "eval_loss": 0.0570383295416832,
+      "eval_mean_token_accuracy": 0.9787763862093543,
+      "eval_num_tokens": 56050518.0,
+      "eval_runtime": 24.4889,
+      "eval_samples_per_second": 204.174,
+      "eval_steps_per_second": 6.411,
+      "step": 2972
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.7083082838982e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-2972/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-2972/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-2972/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-3715/README.md b/baby_talk_L16_a50/seed_42/checkpoint-3715/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-3715/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-3715/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-3715/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-3715/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-3715/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-3715/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..52d8472ffc12c53fa15566b29aeef46d2b55e9a8
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-3715/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76034427b07f0dba7df52bb96914473f77c4a53c8e7d021b0affc3de7384c88d
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-3715/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-3715/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-3715/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-3715/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-3715/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-3715/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-3715/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-3715/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-3715/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-3715/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-3715/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..a3a488912666c25a00eab83b0e416edf01aaf778
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-3715/trainer_state.json
@@ -0,0 +1,3799 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 3715,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    },
+    {
+      "entropy": 0.9689933776855468,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9720287919044495,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.055563896894454956,
+      "mean_token_accuracy": 0.9779708027839661,
+      "num_tokens": 14144652.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.9506655633449554,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 0.6619595289230347,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.050587379932403566,
+      "mean_token_accuracy": 0.979685264825821,
+      "num_tokens": 14332714.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.9458732306957245,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 0.7905108332633972,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.05069155097007751,
+      "mean_token_accuracy": 0.9796596229076385,
+      "num_tokens": 14520902.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.9320916712284089,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 0.9210566878318787,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.051744121313095096,
+      "mean_token_accuracy": 0.9792148530483246,
+      "num_tokens": 14709215.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.928934782743454,
+      "epoch": 1.063257065948856,
+      "grad_norm": 0.7442036867141724,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.0486875057220459,
+      "mean_token_accuracy": 0.9798122465610504,
+      "num_tokens": 14898242.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.942881977558136,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.0910941362380981,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.052297019958496095,
+      "mean_token_accuracy": 0.9793200254440307,
+      "num_tokens": 15086835.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.9503917813301086,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 0.8091554641723633,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.049927744269371035,
+      "mean_token_accuracy": 0.9802082359790802,
+      "num_tokens": 15275042.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9426830470561981,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.0589245557785034,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.053803551197052005,
+      "mean_token_accuracy": 0.9785399675369263,
+      "num_tokens": 15463923.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.9445010781288147,
+      "epoch": 1.117092866756393,
+      "grad_norm": 0.8410441279411316,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.053181976079940796,
+      "mean_token_accuracy": 0.9792523324489594,
+      "num_tokens": 15652526.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.946729838848114,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 0.8246080279350281,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.054291915893554685,
+      "mean_token_accuracy": 0.9791357696056366,
+      "num_tokens": 15841340.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.9433728814125061,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 0.9200296401977539,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.05073915719985962,
+      "mean_token_accuracy": 0.9796931505203247,
+      "num_tokens": 16029478.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.9494417488574982,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 0.8342758417129517,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.050929927825927736,
+      "mean_token_accuracy": 0.9793219089508056,
+      "num_tokens": 16218014.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.9409256100654602,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.7911009192466736,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.049964362382888795,
+      "mean_token_accuracy": 0.9798487305641175,
+      "num_tokens": 16406420.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.9429958581924438,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.4877018928527832,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.04969423711299896,
+      "mean_token_accuracy": 0.9801322638988494,
+      "num_tokens": 16594948.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.9542136013507843,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 0.8030785918235779,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.04981146454811096,
+      "mean_token_accuracy": 0.9796162784099579,
+      "num_tokens": 16782983.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.9367722630500793,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.1169800758361816,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.05210963487625122,
+      "mean_token_accuracy": 0.9790139615535736,
+      "num_tokens": 16971394.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.9487208306789399,
+      "epoch": 1.224764468371467,
+      "grad_norm": 0.8650282621383667,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.05012243390083313,
+      "mean_token_accuracy": 0.9805107891559601,
+      "num_tokens": 17159951.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.9441231489181519,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 0.8193967938423157,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.048866665363311766,
+      "mean_token_accuracy": 0.9807584881782532,
+      "num_tokens": 17348711.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.9361630439758301,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3613612651824951,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.05324091911315918,
+      "mean_token_accuracy": 0.97904953956604,
+      "num_tokens": 17537116.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.9562997639179229,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.8121878504753113,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.05011019706726074,
+      "mean_token_accuracy": 0.9798731088638306,
+      "num_tokens": 17725567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.9359076261520386,
+      "epoch": 1.278600269179004,
+      "grad_norm": 0.7838549613952637,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.04834386110305786,
+      "mean_token_accuracy": 0.9804529249668121,
+      "num_tokens": 17914191.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.9527446150779724,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0981712341308594,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.055077624320983884,
+      "mean_token_accuracy": 0.9776535391807556,
+      "num_tokens": 18103184.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.9574812948703766,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 0.8291501402854919,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.050153911113739014,
+      "mean_token_accuracy": 0.9792965054512024,
+      "num_tokens": 18291351.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.939585280418396,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.0864412784576416,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.05132197737693787,
+      "mean_token_accuracy": 0.9792291462421417,
+      "num_tokens": 18480192.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.9468406856060028,
+      "epoch": 1.332436069986541,
+      "grad_norm": 0.9772785902023315,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.04844954013824463,
+      "mean_token_accuracy": 0.9804142415523529,
+      "num_tokens": 18669028.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.9413834273815155,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 0.9020094871520996,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.05075312852859497,
+      "mean_token_accuracy": 0.9790591120719909,
+      "num_tokens": 18857581.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.9526253461837768,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.0524990558624268,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.049737372994422914,
+      "mean_token_accuracy": 0.9795652389526367,
+      "num_tokens": 19046105.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.9484909176826477,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.0715620517730713,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.04935494959354401,
+      "mean_token_accuracy": 0.9797740161418915,
+      "num_tokens": 19234730.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.9398573458194732,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2134987115859985,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.0518394410610199,
+      "mean_token_accuracy": 0.9793788075447083,
+      "num_tokens": 19423572.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.9465648651123046,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.004693627357483,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.04776117205619812,
+      "mean_token_accuracy": 0.9813261866569519,
+      "num_tokens": 19612465.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.9448004186153411,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.2913438081741333,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.04807930588722229,
+      "mean_token_accuracy": 0.9805689513683319,
+      "num_tokens": 19800482.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.9425322234630584,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1068469285964966,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.05053595900535583,
+      "mean_token_accuracy": 0.9794536828994751,
+      "num_tokens": 19989444.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.9472232103347779,
+      "epoch": 1.440107671601615,
+      "grad_norm": 0.7409384250640869,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.052136778831481934,
+      "mean_token_accuracy": 0.979197359085083,
+      "num_tokens": 20178482.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9450975477695465,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.3440778255462646,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.05258495211601257,
+      "mean_token_accuracy": 0.9798152863979339,
+      "num_tokens": 20367345.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9468007743358612,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 0.8377829194068909,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.04893603026866913,
+      "mean_token_accuracy": 0.9805733859539032,
+      "num_tokens": 20555817.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.9505827724933624,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.776889979839325,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.05367119312286377,
+      "mean_token_accuracy": 0.9779646098613739,
+      "num_tokens": 20744001.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.9516380190849304,
+      "epoch": 1.493943472409152,
+      "grad_norm": 0.8598276376724243,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.05071394443511963,
+      "mean_token_accuracy": 0.979692417383194,
+      "num_tokens": 20931715.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.9351628720760345,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3307439088821411,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.051372635364532473,
+      "mean_token_accuracy": 0.9797907948493958,
+      "num_tokens": 21120803.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.9268040716648102,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9320250749588013,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.04939974546432495,
+      "mean_token_accuracy": 0.98038170337677,
+      "num_tokens": 21309384.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9241463243961334,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.065938115119934,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.04856643378734589,
+      "mean_token_accuracy": 0.9800511121749877,
+      "num_tokens": 21498234.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.9350252687931061,
+      "epoch": 1.547779273216689,
+      "grad_norm": 0.6925032138824463,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.0512526273727417,
+      "mean_token_accuracy": 0.979562646150589,
+      "num_tokens": 21687119.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.9403733670711517,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.2069101333618164,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.049795085191726686,
+      "mean_token_accuracy": 0.9806506633758545,
+      "num_tokens": 21876094.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9370538294315338,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.05461466312408447,
+      "mean_token_accuracy": 0.9776057600975037,
+      "num_tokens": 22064836.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9517778217792511,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 0.8965149521827698,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.05289326906204224,
+      "mean_token_accuracy": 0.9782088756561279,
+      "num_tokens": 22253065.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.9372851848602295,
+      "epoch": 1.601615074024226,
+      "grad_norm": 0.7192108631134033,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.048973691463470456,
+      "mean_token_accuracy": 0.9806210815906524,
+      "num_tokens": 22441206.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.9122351944446564,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 0.8763614296913147,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.0461783230304718,
+      "mean_token_accuracy": 0.9817197680473327,
+      "num_tokens": 22629637.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.9162654876708984,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 0.9806778430938721,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.04827338755130768,
+      "mean_token_accuracy": 0.9806499361991883,
+      "num_tokens": 22818369.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.9125026524066925,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 0.9685350656509399,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.048379385471343996,
+      "mean_token_accuracy": 0.980905145406723,
+      "num_tokens": 23006863.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9119791567325592,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.8717457056045532,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.050873959064483644,
+      "mean_token_accuracy": 0.979601925611496,
+      "num_tokens": 23195735.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.9244007229804992,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.0574654340744019,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.05112813711166382,
+      "mean_token_accuracy": 0.979493772983551,
+      "num_tokens": 23383862.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9225810945034028,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 0.9628807306289673,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.050480544567108154,
+      "mean_token_accuracy": 0.9797678411006927,
+      "num_tokens": 23572292.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.9287579476833343,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.015265941619873,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.05133126378059387,
+      "mean_token_accuracy": 0.9797986805438995,
+      "num_tokens": 23760252.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.9133845925331116,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 0.8996883630752563,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.05044950246810913,
+      "mean_token_accuracy": 0.9795303404331207,
+      "num_tokens": 23949031.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.9213698863983154,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.0490264892578125,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.04797698855400086,
+      "mean_token_accuracy": 0.9806023716926575,
+      "num_tokens": 24137829.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.9173697710037232,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.3648769855499268,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.049194514751434326,
+      "mean_token_accuracy": 0.980712479352951,
+      "num_tokens": 24326586.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9242741882801055,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.105125069618225,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.048667973279953,
+      "mean_token_accuracy": 0.9801449298858642,
+      "num_tokens": 24514904.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.9162627995014191,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.8887938857078552,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.04878672957420349,
+      "mean_token_accuracy": 0.980469423532486,
+      "num_tokens": 24703688.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.9149538338184356,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 0.8838976621627808,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.04892318844795227,
+      "mean_token_accuracy": 0.9808776795864105,
+      "num_tokens": 24892107.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.914735221862793,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1152070760726929,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.04825109839439392,
+      "mean_token_accuracy": 0.9804604113101959,
+      "num_tokens": 25080130.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9170725226402283,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9479517340660095,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.051445144414901736,
+      "mean_token_accuracy": 0.9792819261550904,
+      "num_tokens": 25268652.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9142911911010743,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 0.8172292709350586,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.0475692093372345,
+      "mean_token_accuracy": 0.9809505581855774,
+      "num_tokens": 25457120.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.9019249439239502,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 0.8183121681213379,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.04979957342147827,
+      "mean_token_accuracy": 0.9799730658531189,
+      "num_tokens": 25645699.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.9181066155433655,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.8256579041481018,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.04996164441108704,
+      "mean_token_accuracy": 0.9802247405052185,
+      "num_tokens": 25834183.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9171363770961761,
+      "epoch": 1.857335127860027,
+      "grad_norm": 0.9727098345756531,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.048660767078399655,
+      "mean_token_accuracy": 0.9805064260959625,
+      "num_tokens": 26023317.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.9112720847129822,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 0.9826673865318298,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.04775593280792236,
+      "mean_token_accuracy": 0.9809929549694061,
+      "num_tokens": 26212115.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9034272134304047,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.9951306581497192,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.04939360618591308,
+      "mean_token_accuracy": 0.9800347864627839,
+      "num_tokens": 26401167.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.897865754365921,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.0163404941558838,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05002856254577637,
+      "mean_token_accuracy": 0.9793690323829651,
+      "num_tokens": 26590570.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9004576802253723,
+      "epoch": 1.911170928667564,
+      "grad_norm": 0.6666616201400757,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.04905453026294708,
+      "mean_token_accuracy": 0.9803856253623963,
+      "num_tokens": 26779481.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.8978902101516724,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.9295416474342346,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.05261261463165283,
+      "mean_token_accuracy": 0.9785708487033844,
+      "num_tokens": 26968262.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.8962675571441651,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 0.88262939453125,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.04765265882015228,
+      "mean_token_accuracy": 0.98052077293396,
+      "num_tokens": 27157631.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8909463047981262,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.4205769300460815,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.049406200647354126,
+      "mean_token_accuracy": 0.979697072505951,
+      "num_tokens": 27345921.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8969364166259766,
+      "epoch": 1.965006729475101,
+      "grad_norm": 0.9517145752906799,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0510346531867981,
+      "mean_token_accuracy": 0.9793386399745941,
+      "num_tokens": 27534751.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.9092479169368743,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.200056791305542,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.04974203705787659,
+      "mean_token_accuracy": 0.9800146698951722,
+      "num_tokens": 27723763.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9059641897678375,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.072447657585144,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.05065792202949524,
+      "mean_token_accuracy": 0.9799223959445953,
+      "num_tokens": 27911784.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9083398975384464,
+      "eval_loss": 0.05432562157511711,
+      "eval_mean_token_accuracy": 0.9782073607869969,
+      "eval_num_tokens": 28025286.0,
+      "eval_runtime": 24.4304,
+      "eval_samples_per_second": 204.663,
+      "eval_steps_per_second": 6.426,
+      "step": 1486
+    },
+    {
+      "entropy": 0.9113237977027893,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.0167063474655151,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.044942992925643924,
+      "mean_token_accuracy": 0.9818171083927154,
+      "num_tokens": 28100906.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.8780008792877197,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.9552241563796997,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.036248764395713805,
+      "mean_token_accuracy": 0.9852245450019836,
+      "num_tokens": 28289445.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.8721176147460937,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 0.8951388001441956,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.037987279891967776,
+      "mean_token_accuracy": 0.9850614190101623,
+      "num_tokens": 28477972.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.8922657191753387,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 0.9028105735778809,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.03588914275169373,
+      "mean_token_accuracy": 0.9863102614879609,
+      "num_tokens": 28666272.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.877799642086029,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.153886318206787,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.03577309250831604,
+      "mean_token_accuracy": 0.9863148391246795,
+      "num_tokens": 28854780.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.8708971381187439,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.042955994606018,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.038266432285308835,
+      "mean_token_accuracy": 0.9848419308662415,
+      "num_tokens": 29043741.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.8800721943378449,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 0.6789060235023499,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.032948991656303404,
+      "mean_token_accuracy": 0.9871293902397156,
+      "num_tokens": 29232719.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.8743580460548401,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.1780879497528076,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.036874374747276305,
+      "mean_token_accuracy": 0.9859182178974152,
+      "num_tokens": 29421817.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.8851303637027741,
+      "epoch": 2.113055181695828,
+      "grad_norm": 0.747734785079956,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.0364631175994873,
+      "mean_token_accuracy": 0.9862085223197937,
+      "num_tokens": 29610344.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.8798732101917267,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.5001860857009888,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.03813003897666931,
+      "mean_token_accuracy": 0.9852604746818543,
+      "num_tokens": 29798997.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.8797551989555359,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 0.9593478441238403,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.0394927829504013,
+      "mean_token_accuracy": 0.984604275226593,
+      "num_tokens": 29987809.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.8860546290874481,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.8347703218460083,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.036699697375297546,
+      "mean_token_accuracy": 0.985380882024765,
+      "num_tokens": 30176234.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.8783667802810669,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1154481172561646,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.03756999969482422,
+      "mean_token_accuracy": 0.9853451430797577,
+      "num_tokens": 30364657.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.8684478521347045,
+      "epoch": 2.180349932705249,
+      "grad_norm": 0.7515475153923035,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.0382376104593277,
+      "mean_token_accuracy": 0.9854029655456543,
+      "num_tokens": 30552904.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.8636964917182922,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 0.7711939215660095,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.036220991611480714,
+      "mean_token_accuracy": 0.9860713243484497,
+      "num_tokens": 30741259.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.8592016279697419,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.1143887042999268,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.036316031217575075,
+      "mean_token_accuracy": 0.985757052898407,
+      "num_tokens": 30930144.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.8671079576015472,
+      "epoch": 2.220726783310902,
+      "grad_norm": 0.8980015516281128,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.03822658061981201,
+      "mean_token_accuracy": 0.9847764372825623,
+      "num_tokens": 31118624.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.8785548269748688,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.094323992729187,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.03736622333526611,
+      "mean_token_accuracy": 0.9850081980228425,
+      "num_tokens": 31306833.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.866977310180664,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8372092843055725,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.0401554524898529,
+      "mean_token_accuracy": 0.9840337932109833,
+      "num_tokens": 31495503.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.8787918210029602,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.5408164262771606,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0368030846118927,
+      "mean_token_accuracy": 0.9847234487533569,
+      "num_tokens": 31683865.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.872721153497696,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.1115421056747437,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.037859299778938295,
+      "mean_token_accuracy": 0.9851646661758423,
+      "num_tokens": 31871903.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.8774094223976135,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.0301331281661987,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.038671016693115234,
+      "mean_token_accuracy": 0.9840058028697968,
+      "num_tokens": 32060500.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.889606237411499,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8601903915405273,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.039603835344314574,
+      "mean_token_accuracy": 0.9840235590934754,
+      "num_tokens": 32248744.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.886734277009964,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.1029839515686035,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.039513933658599856,
+      "mean_token_accuracy": 0.9844573020935059,
+      "num_tokens": 32437640.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.8843017637729644,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.2545154094696045,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.036970189213752745,
+      "mean_token_accuracy": 0.985239815711975,
+      "num_tokens": 32625986.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.8913422048091888,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.9613803625106812,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.03811657428741455,
+      "mean_token_accuracy": 0.9852804243564606,
+      "num_tokens": 32814531.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.8907467782497406,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.1811398267745972,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.03903660774230957,
+      "mean_token_accuracy": 0.9842524945735931,
+      "num_tokens": 33002960.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.8833664715290069,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.8338477611541748,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.03982087969779968,
+      "mean_token_accuracy": 0.9841201484203339,
+      "num_tokens": 33191183.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.8783826470375061,
+      "epoch": 2.382234185733513,
+      "grad_norm": 0.6904510259628296,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.03699290752410889,
+      "mean_token_accuracy": 0.9854120731353759,
+      "num_tokens": 33379798.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.8871048510074615,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.009539246559143,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.03926805555820465,
+      "mean_token_accuracy": 0.9843246698379516,
+      "num_tokens": 33568531.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.8973462700843811,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.8916401267051697,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.039596831798553465,
+      "mean_token_accuracy": 0.9839228212833404,
+      "num_tokens": 33756333.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.8954619467258453,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.056504487991333,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0396859347820282,
+      "mean_token_accuracy": 0.9837161123752594,
+      "num_tokens": 33944904.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.8952643811702728,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0442084074020386,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.037324142456054685,
+      "mean_token_accuracy": 0.9852897703647614,
+      "num_tokens": 34133530.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.8874911010265351,
+      "epoch": 2.449528936742934,
+      "grad_norm": 0.9926224946975708,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.03637495338916778,
+      "mean_token_accuracy": 0.9853426337242126,
+      "num_tokens": 34322042.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.8900584518909455,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.001734733581543,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04029585719108582,
+      "mean_token_accuracy": 0.9842372059822082,
+      "num_tokens": 34510259.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.8986345052719116,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 0.772941529750824,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.0401084691286087,
+      "mean_token_accuracy": 0.9841427087783814,
+      "num_tokens": 34698669.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.8911370873451233,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.895031213760376,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.03941032290458679,
+      "mean_token_accuracy": 0.9841966688632965,
+      "num_tokens": 34887160.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.8901963472366333,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0677918195724487,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.03503885865211487,
+      "mean_token_accuracy": 0.9856241464614868,
+      "num_tokens": 35075891.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.8817630231380462,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.1200164556503296,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.03878425657749176,
+      "mean_token_accuracy": 0.9846492648124695,
+      "num_tokens": 35264455.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.8897477686405182,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.8058190941810608,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.041558006405830385,
+      "mean_token_accuracy": 0.9830883860588073,
+      "num_tokens": 35453251.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.8873181045055389,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.7765992283821106,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.03555050492286682,
+      "mean_token_accuracy": 0.9858225345611572,
+      "num_tokens": 35641959.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.8782606959342957,
+      "epoch": 2.557200538358008,
+      "grad_norm": 0.8488866686820984,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.03606923818588257,
+      "mean_token_accuracy": 0.9858632445335388,
+      "num_tokens": 35829636.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.8869808673858642,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.1734739542007446,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.040956351161003116,
+      "mean_token_accuracy": 0.9835374176502227,
+      "num_tokens": 36017818.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.8850924432277679,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.3584387302398682,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.042437011003494264,
+      "mean_token_accuracy": 0.9826479077339172,
+      "num_tokens": 36206505.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.8801687896251679,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2862240076065063,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.03880060315132141,
+      "mean_token_accuracy": 0.9843404710292816,
+      "num_tokens": 36395802.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.8752781748771667,
+      "epoch": 2.611036339165545,
+      "grad_norm": 0.7341249585151672,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.03776344656944275,
+      "mean_token_accuracy": 0.9852877616882324,
+      "num_tokens": 36584370.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.8759494423866272,
+      "epoch": 2.624495289367429,
+      "grad_norm": 0.9807276129722595,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.03780297338962555,
+      "mean_token_accuracy": 0.9853227376937866,
+      "num_tokens": 36772865.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.8753129065036773,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.1391628980636597,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.03953765034675598,
+      "mean_token_accuracy": 0.9839197635650635,
+      "num_tokens": 36961462.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.8771899223327637,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.0865391492843628,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.03856399655342102,
+      "mean_token_accuracy": 0.9848688066005706,
+      "num_tokens": 37150486.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.8774881184101104,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.8561595678329468,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.03795175850391388,
+      "mean_token_accuracy": 0.9849309325218201,
+      "num_tokens": 37338890.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.8841517448425293,
+      "epoch": 2.678331090174966,
+      "grad_norm": 0.9458399415016174,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.03860213160514832,
+      "mean_token_accuracy": 0.9842230796813964,
+      "num_tokens": 37526886.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.8809043228626251,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 0.9360871315002441,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.03709094822406769,
+      "mean_token_accuracy": 0.9855118036270142,
+      "num_tokens": 37715408.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.8805519282817841,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.8852124214172363,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.03869341611862183,
+      "mean_token_accuracy": 0.9846724212169647,
+      "num_tokens": 37904328.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.8796228408813477,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.189297080039978,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.03675769567489624,
+      "mean_token_accuracy": 0.9854080140590668,
+      "num_tokens": 38092776.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.8722020089626312,
+      "epoch": 2.732166890982503,
+      "grad_norm": 0.7027947306632996,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.03889244794845581,
+      "mean_token_accuracy": 0.984558242559433,
+      "num_tokens": 38282221.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.8746632814407349,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 0.9799376726150513,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.03815680146217346,
+      "mean_token_accuracy": 0.9845012128353119,
+      "num_tokens": 38470633.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.8766567528247833,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 0.850459098815918,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.039029371738433835,
+      "mean_token_accuracy": 0.9845906794071198,
+      "num_tokens": 38659048.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.8778688013553619,
+      "epoch": 2.772543741588156,
+      "grad_norm": 0.8540134429931641,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.039645448327064514,
+      "mean_token_accuracy": 0.9840566098690033,
+      "num_tokens": 38847370.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.8758600771427154,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8036021590232849,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.038226932287216187,
+      "mean_token_accuracy": 0.9845647215843201,
+      "num_tokens": 39036040.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.8652268946170807,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 0.8526738286018372,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.03798363208770752,
+      "mean_token_accuracy": 0.9845202267169952,
+      "num_tokens": 39224695.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.8707200348377228,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 0.9852389097213745,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.037227436900138855,
+      "mean_token_accuracy": 0.9852050960063934,
+      "num_tokens": 39413360.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.8711447060108185,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.228277325630188,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.03698050379753113,
+      "mean_token_accuracy": 0.9852443158626556,
+      "num_tokens": 39602544.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.8742912471294403,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 0.8440544605255127,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.03757670521736145,
+      "mean_token_accuracy": 0.9848618268966675,
+      "num_tokens": 39791340.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.8801726162433624,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0955252647399902,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.040004149079322815,
+      "mean_token_accuracy": 0.9841345012187958,
+      "num_tokens": 39980212.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.8931492984294891,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.5722652673721313,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.038096648454666135,
+      "mean_token_accuracy": 0.9849136590957641,
+      "num_tokens": 40169122.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.8767493844032288,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.7663730382919312,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.03729096055030823,
+      "mean_token_accuracy": 0.985169780254364,
+      "num_tokens": 40357678.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.8770561516284943,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.1505672931671143,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.03984796404838562,
+      "mean_token_accuracy": 0.9838845670223236,
+      "num_tokens": 40546892.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.8789129912853241,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 0.8355761766433716,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.03769223690032959,
+      "mean_token_accuracy": 0.9847559213638306,
+      "num_tokens": 40736109.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.8781779289245606,
+      "epoch": 2.920592193808883,
+      "grad_norm": 0.9103918075561523,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.038262826204299924,
+      "mean_token_accuracy": 0.984741848707199,
+      "num_tokens": 40924341.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.8552807629108429,
+      "epoch": 2.934051144010767,
+      "grad_norm": 0.8338510990142822,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.0355743408203125,
+      "mean_token_accuracy": 0.9855036079883576,
+      "num_tokens": 41113849.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.8705591142177582,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 0.8334057927131653,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.037476515769958495,
+      "mean_token_accuracy": 0.9854551255702972,
+      "num_tokens": 41303130.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.8809274792671203,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 0.9424103498458862,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.03624279499053955,
+      "mean_token_accuracy": 0.9856135487556458,
+      "num_tokens": 41491750.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.8818133771419525,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.04243004322052,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.039132320880889894,
+      "mean_token_accuracy": 0.984746390581131,
+      "num_tokens": 41680493.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.8838395297527313,
+      "epoch": 2.987886944818304,
+      "grad_norm": 0.8650784492492676,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.03822052478790283,
+      "mean_token_accuracy": 0.9852108955383301,
+      "num_tokens": 41868378.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.8800126549544608,
+      "eval_loss": 0.05468416586518288,
+      "eval_mean_token_accuracy": 0.9781015617832257,
+      "eval_num_tokens": 42038037.0,
+      "eval_runtime": 24.4825,
+      "eval_samples_per_second": 204.228,
+      "eval_steps_per_second": 6.413,
+      "step": 2229
+    },
+    {
+      "entropy": 0.8773060619831086,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.7845072150230408,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.0365300714969635,
+      "mean_token_accuracy": 0.985853761434555,
+      "num_tokens": 42056717.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.8720550239086151,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 0.9917275905609131,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.030071893334388734,
+      "mean_token_accuracy": 0.9890478909015655,
+      "num_tokens": 42245058.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.8489451467990875,
+      "epoch": 3.028263795423957,
+      "grad_norm": 0.9414294362068176,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.027054613828659056,
+      "mean_token_accuracy": 0.9894806027412415,
+      "num_tokens": 42433468.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.848755830526352,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.7276235818862915,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.026897600293159483,
+      "mean_token_accuracy": 0.9893308103084564,
+      "num_tokens": 42621657.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.8402929604053497,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 0.9455838799476624,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.027918827533721925,
+      "mean_token_accuracy": 0.9888385117053986,
+      "num_tokens": 42810533.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.8434606611728668,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 0.7503429651260376,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.025769710540771484,
+      "mean_token_accuracy": 0.989881956577301,
+      "num_tokens": 42999085.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.8459890782833099,
+      "epoch": 3.082099596231494,
+      "grad_norm": 0.736440896987915,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.028781753778457642,
+      "mean_token_accuracy": 0.9882656812667847,
+      "num_tokens": 43188222.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.8493306159973144,
+      "epoch": 3.095558546433378,
+      "grad_norm": 0.9577816128730774,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.02689719796180725,
+      "mean_token_accuracy": 0.9889744818210602,
+      "num_tokens": 43376648.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.8512665331363678,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 0.9631588459014893,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.031439167261123654,
+      "mean_token_accuracy": 0.9874942421913147,
+      "num_tokens": 43564769.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.8519513130187988,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1011922359466553,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.02696958780288696,
+      "mean_token_accuracy": 0.9900951981544495,
+      "num_tokens": 43753362.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.8507618844509125,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9351623058319092,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.026315575838088988,
+      "mean_token_accuracy": 0.989511913061142,
+      "num_tokens": 43942063.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.8506478369235992,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.096564769744873,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.029049152135849,
+      "mean_token_accuracy": 0.9886228442192078,
+      "num_tokens": 44130381.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.8624941289424897,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 0.7859402298927307,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.027744990587234498,
+      "mean_token_accuracy": 0.9890527307987214,
+      "num_tokens": 44319243.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.8627866744995117,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 0.7764605283737183,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.02571650445461273,
+      "mean_token_accuracy": 0.9903442323207855,
+      "num_tokens": 44507698.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.8566473722457886,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.8182396292686462,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.025625643134117127,
+      "mean_token_accuracy": 0.9897150516510009,
+      "num_tokens": 44695954.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.8555088877677918,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.0277098417282104,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.028235083818435668,
+      "mean_token_accuracy": 0.9894341349601745,
+      "num_tokens": 44884558.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.8562702238559723,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.0676493644714355,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.027713948488235475,
+      "mean_token_accuracy": 0.9891053259372711,
+      "num_tokens": 45073381.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.8520576059818268,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 0.7926117181777954,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.028936928510665892,
+      "mean_token_accuracy": 0.988807988166809,
+      "num_tokens": 45262350.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.860496312379837,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.1108510494232178,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030533093214035033,
+      "mean_token_accuracy": 0.9882748067378998,
+      "num_tokens": 45450782.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.872324675321579,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.3057194948196411,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.032421305775642395,
+      "mean_token_accuracy": 0.9879753351211548,
+      "num_tokens": 45639122.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.8778823137283325,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.289680004119873,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03096393346786499,
+      "mean_token_accuracy": 0.9883370697498322,
+      "num_tokens": 45827730.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.8627406179904937,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.8361398577690125,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.02957139313220978,
+      "mean_token_accuracy": 0.9884674072265625,
+      "num_tokens": 46016274.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.8584554970264435,
+      "epoch": 3.297442799461642,
+      "grad_norm": 0.9330723881721497,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.029568272829055785,
+      "mean_token_accuracy": 0.9882686555385589,
+      "num_tokens": 46204679.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.8515178024768829,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.0296216011047363,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.027788180112838744,
+      "mean_token_accuracy": 0.9888734877109527,
+      "num_tokens": 46393989.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.8488850116729736,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1329699754714966,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.028057461977005003,
+      "mean_token_accuracy": 0.9895590543746948,
+      "num_tokens": 46582790.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.849625188112259,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 0.9267219305038452,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.02802973985671997,
+      "mean_token_accuracy": 0.988851535320282,
+      "num_tokens": 46771138.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.854446280002594,
+      "epoch": 3.351278600269179,
+      "grad_norm": 0.8713932037353516,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.030220019817352294,
+      "mean_token_accuracy": 0.9882843673229218,
+      "num_tokens": 46960054.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.8599806547164917,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.675019383430481,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.02958865463733673,
+      "mean_token_accuracy": 0.9880296885967255,
+      "num_tokens": 47149567.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.8549504458904267,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.2176353931427002,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.027722162008285523,
+      "mean_token_accuracy": 0.9890090465545655,
+      "num_tokens": 47338060.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.8552233815193176,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9571717977523804,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.028924965858459474,
+      "mean_token_accuracy": 0.9884298503398895,
+      "num_tokens": 47525895.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.8521630108356476,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.965713620185852,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.030511701107025148,
+      "mean_token_accuracy": 0.9881398677825928,
+      "num_tokens": 47714524.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.8537400126457214,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 0.9005307555198669,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.030432754755020143,
+      "mean_token_accuracy": 0.9876708447933197,
+      "num_tokens": 47903188.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.8596058666706086,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 0.8159075379371643,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.027927806973457335,
+      "mean_token_accuracy": 0.9889210939407349,
+      "num_tokens": 48091334.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.8482735753059387,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0411661863327026,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03003154993057251,
+      "mean_token_accuracy": 0.9882730603218078,
+      "num_tokens": 48279622.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.8466426789760589,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.189595341682434,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03237656056880951,
+      "mean_token_accuracy": 0.9869072616100312,
+      "num_tokens": 48467527.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.8545588493347168,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0700056552886963,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.0306060791015625,
+      "mean_token_accuracy": 0.9880664825439454,
+      "num_tokens": 48656151.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.8503433585166931,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.7099478244781494,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.02984013557434082,
+      "mean_token_accuracy": 0.9879442811012268,
+      "num_tokens": 48844796.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.8464376091957092,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9877128005027771,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.0306446373462677,
+      "mean_token_accuracy": 0.9883591532707214,
+      "num_tokens": 49033216.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.8477609276771545,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.3324531316757202,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.030148410797119142,
+      "mean_token_accuracy": 0.9879989802837372,
+      "num_tokens": 49221425.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.8461863219738006,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8938570022583008,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.02682304084300995,
+      "mean_token_accuracy": 0.9897881150245667,
+      "num_tokens": 49409785.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.8362689018249512,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 0.7423873543739319,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.02689332664012909,
+      "mean_token_accuracy": 0.9899903059005737,
+      "num_tokens": 49598524.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.8395139217376709,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.921755850315094,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.030062124133110046,
+      "mean_token_accuracy": 0.9880835592746735,
+      "num_tokens": 49787241.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.8467226803302765,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0296459197998047,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.030666223168373107,
+      "mean_token_accuracy": 0.9877098739147187,
+      "num_tokens": 49975844.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.833392071723938,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2861948013305664,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.028174835443496703,
+      "mean_token_accuracy": 0.989056545495987,
+      "num_tokens": 50164857.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.8340730309486389,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 0.8116776347160339,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.028390336036682128,
+      "mean_token_accuracy": 0.9886897742748261,
+      "num_tokens": 50353452.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.8357116639614105,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.8362084031105042,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.029806244373321533,
+      "mean_token_accuracy": 0.988309508562088,
+      "num_tokens": 50542461.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.8363270878791809,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8438499569892883,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.028577303886413573,
+      "mean_token_accuracy": 0.9886945366859436,
+      "num_tokens": 50730872.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.8310261785984039,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.1920002698898315,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.030321845412254335,
+      "mean_token_accuracy": 0.9878708899021149,
+      "num_tokens": 50919889.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.8380252480506897,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 0.9409399032592773,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.030501589179039,
+      "mean_token_accuracy": 0.9872083842754364,
+      "num_tokens": 51108079.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.8422612130641938,
+      "epoch": 3.660834454912517,
+      "grad_norm": 0.8370440006256104,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.02974823713302612,
+      "mean_token_accuracy": 0.9885554671287536,
+      "num_tokens": 51296913.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.8387595832347869,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1360642910003662,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.026970452070236205,
+      "mean_token_accuracy": 0.9888331353664398,
+      "num_tokens": 51485678.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.8403031289577484,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.1195677518844604,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.029451662302017213,
+      "mean_token_accuracy": 0.988847154378891,
+      "num_tokens": 51674765.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.8499106168746948,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 0.9573526978492737,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.030826866626739502,
+      "mean_token_accuracy": 0.9875578820705414,
+      "num_tokens": 51863924.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.8576709866523743,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9386286735534668,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.02946900427341461,
+      "mean_token_accuracy": 0.9885939955711365,
+      "num_tokens": 52052011.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.8404816091060638,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.095536708831787,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.030070826411247253,
+      "mean_token_accuracy": 0.9881400763988495,
+      "num_tokens": 52240955.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.8346689164638519,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9964426159858704,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03005717396736145,
+      "mean_token_accuracy": 0.9883057296276092,
+      "num_tokens": 52429544.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.8377436757087707,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.8499778509140015,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.029283815622329713,
+      "mean_token_accuracy": 0.9882387280464172,
+      "num_tokens": 52618405.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.8328788638114929,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9594390988349915,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.02666555345058441,
+      "mean_token_accuracy": 0.9889323055744171,
+      "num_tokens": 52806689.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.829415249824524,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.3587099313735962,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.028349629044532774,
+      "mean_token_accuracy": 0.9890907645225525,
+      "num_tokens": 52995248.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.8335795283317566,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.077789306640625,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.03001731038093567,
+      "mean_token_accuracy": 0.9878358781337738,
+      "num_tokens": 53183607.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.8316770553588867,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 0.7265951037406921,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.030136841535568237,
+      "mean_token_accuracy": 0.9880174934864044,
+      "num_tokens": 53372735.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.8310399711132049,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.6605978012084961,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.026665833592414857,
+      "mean_token_accuracy": 0.9892704486846924,
+      "num_tokens": 53561776.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.8258654296398162,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.6560797691345215,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.026718440651893615,
+      "mean_token_accuracy": 0.989847868680954,
+      "num_tokens": 53750019.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.8221413195133209,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 0.7495356798171997,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.027431756258010864,
+      "mean_token_accuracy": 0.9892430484294892,
+      "num_tokens": 53938914.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.8341928541660308,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.0676971673965454,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.028432759642601012,
+      "mean_token_accuracy": 0.9889024317264556,
+      "num_tokens": 54127763.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.8359149813652038,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.380476713180542,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.028149163722991942,
+      "mean_token_accuracy": 0.9884953200817108,
+      "num_tokens": 54315930.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.8294597804546356,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0728843212127686,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.028733691573143004,
+      "mean_token_accuracy": 0.9884523034095765,
+      "num_tokens": 54504632.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.8245523810386658,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 0.9723307490348816,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.02793261408805847,
+      "mean_token_accuracy": 0.9891038477420807,
+      "num_tokens": 54693752.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.8279620766639709,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.003624439239502,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.028367668390274048,
+      "mean_token_accuracy": 0.9893603324890137,
+      "num_tokens": 54882030.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.8306584179401397,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.8481361865997314,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.029952478408813477,
+      "mean_token_accuracy": 0.9881332099437714,
+      "num_tokens": 55070549.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.8316469848155975,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.7631049156188965,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.029351180791854857,
+      "mean_token_accuracy": 0.9888184070587158,
+      "num_tokens": 55258999.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.8221287310123444,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 0.7234840393066406,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.027709537744522096,
+      "mean_token_accuracy": 0.9890308439731598,
+      "num_tokens": 55448047.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.8239192545413971,
+      "epoch": 3.970390309555855,
+      "grad_norm": 0.8497020602226257,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.029604345560073853,
+      "mean_token_accuracy": 0.9881375730037689,
+      "num_tokens": 55636566.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.8319927275180816,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.1599552631378174,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.02763860821723938,
+      "mean_token_accuracy": 0.9891406774520874,
+      "num_tokens": 55825049.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.8404937744140625,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 0.906755805015564,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.029578033089637756,
+      "mean_token_accuracy": 0.9878996431827545,
+      "num_tokens": 56012965.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.8419649528849656,
+      "eval_loss": 0.0570383295416832,
+      "eval_mean_token_accuracy": 0.9787763862093543,
+      "eval_num_tokens": 56050518.0,
+      "eval_runtime": 24.4889,
+      "eval_samples_per_second": 204.174,
+      "eval_steps_per_second": 6.411,
+      "step": 2972
+    },
+    {
+      "entropy": 0.8328965544700623,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7073294520378113,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.020916813611984254,
+      "mean_token_accuracy": 0.992138022184372,
+      "num_tokens": 56201621.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.8153974890708924,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.8904889822006226,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.01942949742078781,
+      "mean_token_accuracy": 0.9927055358886718,
+      "num_tokens": 56390018.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.8080023109912873,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.0828146934509277,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.018959003686904907,
+      "mean_token_accuracy": 0.9928978025913239,
+      "num_tokens": 56578046.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.8056950688362121,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7407417893409729,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019250378012657166,
+      "mean_token_accuracy": 0.9926685452461242,
+      "num_tokens": 56767192.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.8124530673027038,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.0355021953582764,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.01904640197753906,
+      "mean_token_accuracy": 0.9925767481327057,
+      "num_tokens": 56955769.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.810802960395813,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.3759289979934692,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.020995599031448365,
+      "mean_token_accuracy": 0.9917719483375549,
+      "num_tokens": 57144300.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.8159192740917206,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.112034559249878,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.01948906183242798,
+      "mean_token_accuracy": 0.9928658187389374,
+      "num_tokens": 57332935.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.821906328201294,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.0319583415985107,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019549430906772615,
+      "mean_token_accuracy": 0.9922966659069061,
+      "num_tokens": 57520775.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.8133022606372833,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.067180871963501,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.01833198368549347,
+      "mean_token_accuracy": 0.9934999167919158,
+      "num_tokens": 57709495.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.8150732755661011,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.277079701423645,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.019589337706565856,
+      "mean_token_accuracy": 0.992507028579712,
+      "num_tokens": 57897461.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.8159323990345001,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.3150911331176758,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.01963164806365967,
+      "mean_token_accuracy": 0.9926789402961731,
+      "num_tokens": 58085915.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.8147098064422608,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.0073580741882324,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.02144385576248169,
+      "mean_token_accuracy": 0.9918489575386047,
+      "num_tokens": 58274583.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.8155419886112213,
+      "epoch": 4.172274562584119,
+      "grad_norm": 0.8892306685447693,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.02215111702680588,
+      "mean_token_accuracy": 0.9914537072181702,
+      "num_tokens": 58463241.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.8084070324897766,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.7887238264083862,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.018581822514533997,
+      "mean_token_accuracy": 0.9928648293018341,
+      "num_tokens": 58652281.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.8016656517982483,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.0644817352294922,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.020099574327468873,
+      "mean_token_accuracy": 0.9916590929031373,
+      "num_tokens": 58839946.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.8035425901412964,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.9588403701782227,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020384626090526582,
+      "mean_token_accuracy": 0.9918951213359832,
+      "num_tokens": 59028994.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.8114414095878602,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.7145556211471558,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.019365760684013366,
+      "mean_token_accuracy": 0.9921130836009979,
+      "num_tokens": 59217599.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.8123053193092347,
+      "epoch": 4.23956931359354,
+      "grad_norm": 0.7928217649459839,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.020542070269584656,
+      "mean_token_accuracy": 0.9918384432792664,
+      "num_tokens": 59406176.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.810414057970047,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.1470942497253418,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.01809942126274109,
+      "mean_token_accuracy": 0.992941427230835,
+      "num_tokens": 59594604.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.8046662211418152,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.579035222530365,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.01914280205965042,
+      "mean_token_accuracy": 0.9926800608634949,
+      "num_tokens": 59782947.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.8034039616584778,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.8979132175445557,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.019897907972335815,
+      "mean_token_accuracy": 0.9919832170009613,
+      "num_tokens": 59971417.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.8137863457202912,
+      "epoch": 4.293405114401077,
+      "grad_norm": 0.9149890542030334,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.02069481760263443,
+      "mean_token_accuracy": 0.992353367805481,
+      "num_tokens": 60159983.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.8163237690925598,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.8264126181602478,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.021351543068885804,
+      "mean_token_accuracy": 0.9921438097953796,
+      "num_tokens": 60348327.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.8133926391601562,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.0638184547424316,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.020092280209064485,
+      "mean_token_accuracy": 0.9920917391777039,
+      "num_tokens": 60537152.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.8095248222351075,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.8831470012664795,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.01908399909734726,
+      "mean_token_accuracy": 0.9929656088352203,
+      "num_tokens": 60725603.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.8069245994091034,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8486410975456238,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.01925535798072815,
+      "mean_token_accuracy": 0.9927195310592651,
+      "num_tokens": 60914021.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.8006627321243286,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.9021628499031067,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.019187870621681213,
+      "mean_token_accuracy": 0.992531019449234,
+      "num_tokens": 61102037.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.8067023754119873,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.7793697118759155,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.021760080754756928,
+      "mean_token_accuracy": 0.9915579319000244,
+      "num_tokens": 61290132.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.8075471341609954,
+      "epoch": 4.387617765814267,
+      "grad_norm": 0.736788809299469,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.02077012062072754,
+      "mean_token_accuracy": 0.9920295119285584,
+      "num_tokens": 61478897.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.8083206176757812,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.1687020063400269,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.02117353230714798,
+      "mean_token_accuracy": 0.991925036907196,
+      "num_tokens": 61667982.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.81106236577034,
+      "epoch": 4.414535666218035,
+      "grad_norm": 0.7596265077590942,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.02081204652786255,
+      "mean_token_accuracy": 0.9921064436435699,
+      "num_tokens": 61856557.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.8018859803676606,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.823410153388977,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.019047826528549194,
+      "mean_token_accuracy": 0.9924882829189301,
+      "num_tokens": 62045181.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7996662855148315,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.0748025178909302,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.01788569688796997,
+      "mean_token_accuracy": 0.9931309640407562,
+      "num_tokens": 62234215.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.8021627008914948,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9946282505989075,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021408762037754058,
+      "mean_token_accuracy": 0.9914931178092956,
+      "num_tokens": 62422361.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.8084165751934052,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.7313793301582336,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.020281489193439483,
+      "mean_token_accuracy": 0.9925725042819977,
+      "num_tokens": 62610667.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.8103892624378204,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.9093077182769775,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.018877537548542024,
+      "mean_token_accuracy": 0.9929530620574951,
+      "num_tokens": 62798970.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.8091431736946106,
+      "epoch": 4.495289367429341,
+      "grad_norm": 0.952234148979187,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021039776504039764,
+      "mean_token_accuracy": 0.9918699383735656,
+      "num_tokens": 62986914.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.8094858765602112,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.9959521293640137,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021489059925079344,
+      "mean_token_accuracy": 0.9914597749710083,
+      "num_tokens": 63175805.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.807330971956253,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.6189294457435608,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.02164270281791687,
+      "mean_token_accuracy": 0.9920661866664886,
+      "num_tokens": 63364692.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.8098866164684295,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7895700931549072,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.021856357157230378,
+      "mean_token_accuracy": 0.9916301846504212,
+      "num_tokens": 63553692.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.8143714487552642,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0242317914962769,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02046577036380768,
+      "mean_token_accuracy": 0.9923330962657928,
+      "num_tokens": 63742476.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.8123241424560547,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.4186221361160278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.020659247040748598,
+      "mean_token_accuracy": 0.9918934762477875,
+      "num_tokens": 63930898.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.8133015930652618,
+      "epoch": 4.576043068640646,
+      "grad_norm": 0.9085373878479004,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.021746012568473815,
+      "mean_token_accuracy": 0.9918161749839782,
+      "num_tokens": 64119589.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.8173148334026337,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.7257112264633179,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.019787387549877168,
+      "mean_token_accuracy": 0.9922669887542724,
+      "num_tokens": 64308293.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.8116482973098755,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7442086935043335,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.020933127403259276,
+      "mean_token_accuracy": 0.9917319416999817,
+      "num_tokens": 64497362.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.8134189546108246,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.7617318630218506,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.01842810958623886,
+      "mean_token_accuracy": 0.9927939116954804,
+      "num_tokens": 64685802.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.8112196505069733,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.7430832386016846,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.01809561103582382,
+      "mean_token_accuracy": 0.9930340230464936,
+      "num_tokens": 64873973.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.8128187835216523,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.7553606629371643,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.0202948659658432,
+      "mean_token_accuracy": 0.9919280350208283,
+      "num_tokens": 65062510.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.813840126991272,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.7390825152397156,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.020557042956352235,
+      "mean_token_accuracy": 0.9918098270893096,
+      "num_tokens": 65250940.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.8117553532123566,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.7698133587837219,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.020231780409812928,
+      "mean_token_accuracy": 0.9920670032501221,
+      "num_tokens": 65439696.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.8110962271690368,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.8562760353088379,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.018769541382789613,
+      "mean_token_accuracy": 0.9925244867801666,
+      "num_tokens": 65628357.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.8113800704479217,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.2076038122177124,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020604749023914338,
+      "mean_token_accuracy": 0.9921375870704651,
+      "num_tokens": 65817139.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.8204582631587982,
+      "epoch": 4.710632570659489,
+      "grad_norm": 0.9150722026824951,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021981674432754516,
+      "mean_token_accuracy": 0.9915678203105927,
+      "num_tokens": 66005911.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.8241343259811401,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.9803491830825806,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02007504999637604,
+      "mean_token_accuracy": 0.992132717370987,
+      "num_tokens": 66194469.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.8122783243656159,
+      "epoch": 4.737550471063257,
+      "grad_norm": 0.967640221118927,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.019903984665870667,
+      "mean_token_accuracy": 0.9924214482307434,
+      "num_tokens": 66383927.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.8145494401454926,
+      "epoch": 4.751009421265142,
+      "grad_norm": 0.7893037796020508,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.019587111473083497,
+      "mean_token_accuracy": 0.9921624064445496,
+      "num_tokens": 66572212.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.8090193390846252,
+      "epoch": 4.764468371467026,
+      "grad_norm": 0.9659109711647034,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.021097327768802642,
+      "mean_token_accuracy": 0.9917473196983337,
+      "num_tokens": 66760377.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.802634209394455,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.0175755023956299,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.02113145440816879,
+      "mean_token_accuracy": 0.991830724477768,
+      "num_tokens": 66949210.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.8030233085155487,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.7780183553695679,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.019016166031360627,
+      "mean_token_accuracy": 0.9926414608955383,
+      "num_tokens": 67138031.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.8019147455692291,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7863438129425049,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.018219922482967377,
+      "mean_token_accuracy": 0.9928844273090363,
+      "num_tokens": 67327268.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7949212431907654,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9292862415313721,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.02136329412460327,
+      "mean_token_accuracy": 0.9918141901493073,
+      "num_tokens": 67515765.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.8002764582633972,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9770768284797668,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.02056926190853119,
+      "mean_token_accuracy": 0.9918984293937683,
+      "num_tokens": 67704975.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.808483463525772,
+      "epoch": 4.845222072678331,
+      "grad_norm": 0.6890367269515991,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020158609747886656,
+      "mean_token_accuracy": 0.9925975620746612,
+      "num_tokens": 67893850.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.8144356846809387,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.8613544702529907,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.019614605605602263,
+      "mean_token_accuracy": 0.9928383588790893,
+      "num_tokens": 68082435.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.8083879172801971,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.7532686591148376,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.019553859531879426,
+      "mean_token_accuracy": 0.9924478828907013,
+      "num_tokens": 68270495.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.8038046479225158,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.3251928091049194,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.02081710398197174,
+      "mean_token_accuracy": 0.9919814527034759,
+      "num_tokens": 68459223.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.8058662831783294,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8125224113464355,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.01858963668346405,
+      "mean_token_accuracy": 0.9922726869583129,
+      "num_tokens": 68647306.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.805506020784378,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8629401922225952,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020777507126331328,
+      "mean_token_accuracy": 0.9918497979640961,
+      "num_tokens": 68836386.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.8137464344501495,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.9837846159934998,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.02024296224117279,
+      "mean_token_accuracy": 0.9921825706958771,
+      "num_tokens": 69025192.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.8139736235141755,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.8701289296150208,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.01932244449853897,
+      "mean_token_accuracy": 0.9926821112632751,
+      "num_tokens": 69213427.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.8088622391223907,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.7041723132133484,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.018706569075584413,
+      "mean_token_accuracy": 0.9927357733249664,
+      "num_tokens": 69401690.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.8001897990703583,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1029852628707886,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.020011243224143983,
+      "mean_token_accuracy": 0.9923900067806244,
+      "num_tokens": 69590659.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.8071568071842193,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.2395373582839966,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.02379731684923172,
+      "mean_token_accuracy": 0.9907991290092468,
+      "num_tokens": 69779864.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.8139646112918854,
+      "epoch": 4.993270524899058,
+      "grad_norm": 0.9219555854797363,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.021635296940803527,
+      "mean_token_accuracy": 0.991401469707489,
+      "num_tokens": 69968639.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.822659221804066,
+      "eval_loss": 0.060467079281806946,
+      "eval_mean_token_accuracy": 0.9792145984188007,
+      "eval_num_tokens": 70063064.0,
+      "eval_runtime": 24.4326,
+      "eval_samples_per_second": 204.645,
+      "eval_steps_per_second": 6.426,
+      "step": 3715
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.385463974836306e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-3715/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-3715/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-3715/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-4458/README.md b/baby_talk_L16_a50/seed_42/checkpoint-4458/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-4458/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-4458/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-4458/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-4458/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-4458/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-4458/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..5086bcc2111b6901d12c6b0f3fc939c3aa2ee149
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-4458/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a8b6fd518dab5e24f6b28262e81c749857962f0080cb5998e7b9aa57c5eb01a
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-4458/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-4458/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-4458/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-4458/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-4458/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-4458/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-4458/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-4458/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-4458/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-4458/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-4458/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..33559b0c4e6fc0ac10d19074b49e1e1e3ed841b6
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-4458/trainer_state.json
@@ -0,0 +1,4550 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.0,
+  "eval_steps": 500,
+  "global_step": 4458,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    },
+    {
+      "entropy": 0.9689933776855468,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9720287919044495,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.055563896894454956,
+      "mean_token_accuracy": 0.9779708027839661,
+      "num_tokens": 14144652.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.9506655633449554,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 0.6619595289230347,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.050587379932403566,
+      "mean_token_accuracy": 0.979685264825821,
+      "num_tokens": 14332714.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.9458732306957245,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 0.7905108332633972,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.05069155097007751,
+      "mean_token_accuracy": 0.9796596229076385,
+      "num_tokens": 14520902.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.9320916712284089,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 0.9210566878318787,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.051744121313095096,
+      "mean_token_accuracy": 0.9792148530483246,
+      "num_tokens": 14709215.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.928934782743454,
+      "epoch": 1.063257065948856,
+      "grad_norm": 0.7442036867141724,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.0486875057220459,
+      "mean_token_accuracy": 0.9798122465610504,
+      "num_tokens": 14898242.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.942881977558136,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.0910941362380981,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.052297019958496095,
+      "mean_token_accuracy": 0.9793200254440307,
+      "num_tokens": 15086835.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.9503917813301086,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 0.8091554641723633,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.049927744269371035,
+      "mean_token_accuracy": 0.9802082359790802,
+      "num_tokens": 15275042.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9426830470561981,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.0589245557785034,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.053803551197052005,
+      "mean_token_accuracy": 0.9785399675369263,
+      "num_tokens": 15463923.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.9445010781288147,
+      "epoch": 1.117092866756393,
+      "grad_norm": 0.8410441279411316,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.053181976079940796,
+      "mean_token_accuracy": 0.9792523324489594,
+      "num_tokens": 15652526.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.946729838848114,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 0.8246080279350281,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.054291915893554685,
+      "mean_token_accuracy": 0.9791357696056366,
+      "num_tokens": 15841340.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.9433728814125061,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 0.9200296401977539,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.05073915719985962,
+      "mean_token_accuracy": 0.9796931505203247,
+      "num_tokens": 16029478.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.9494417488574982,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 0.8342758417129517,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.050929927825927736,
+      "mean_token_accuracy": 0.9793219089508056,
+      "num_tokens": 16218014.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.9409256100654602,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.7911009192466736,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.049964362382888795,
+      "mean_token_accuracy": 0.9798487305641175,
+      "num_tokens": 16406420.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.9429958581924438,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.4877018928527832,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.04969423711299896,
+      "mean_token_accuracy": 0.9801322638988494,
+      "num_tokens": 16594948.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.9542136013507843,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 0.8030785918235779,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.04981146454811096,
+      "mean_token_accuracy": 0.9796162784099579,
+      "num_tokens": 16782983.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.9367722630500793,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.1169800758361816,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.05210963487625122,
+      "mean_token_accuracy": 0.9790139615535736,
+      "num_tokens": 16971394.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.9487208306789399,
+      "epoch": 1.224764468371467,
+      "grad_norm": 0.8650282621383667,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.05012243390083313,
+      "mean_token_accuracy": 0.9805107891559601,
+      "num_tokens": 17159951.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.9441231489181519,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 0.8193967938423157,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.048866665363311766,
+      "mean_token_accuracy": 0.9807584881782532,
+      "num_tokens": 17348711.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.9361630439758301,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3613612651824951,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.05324091911315918,
+      "mean_token_accuracy": 0.97904953956604,
+      "num_tokens": 17537116.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.9562997639179229,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.8121878504753113,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.05011019706726074,
+      "mean_token_accuracy": 0.9798731088638306,
+      "num_tokens": 17725567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.9359076261520386,
+      "epoch": 1.278600269179004,
+      "grad_norm": 0.7838549613952637,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.04834386110305786,
+      "mean_token_accuracy": 0.9804529249668121,
+      "num_tokens": 17914191.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.9527446150779724,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0981712341308594,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.055077624320983884,
+      "mean_token_accuracy": 0.9776535391807556,
+      "num_tokens": 18103184.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.9574812948703766,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 0.8291501402854919,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.050153911113739014,
+      "mean_token_accuracy": 0.9792965054512024,
+      "num_tokens": 18291351.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.939585280418396,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.0864412784576416,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.05132197737693787,
+      "mean_token_accuracy": 0.9792291462421417,
+      "num_tokens": 18480192.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.9468406856060028,
+      "epoch": 1.332436069986541,
+      "grad_norm": 0.9772785902023315,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.04844954013824463,
+      "mean_token_accuracy": 0.9804142415523529,
+      "num_tokens": 18669028.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.9413834273815155,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 0.9020094871520996,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.05075312852859497,
+      "mean_token_accuracy": 0.9790591120719909,
+      "num_tokens": 18857581.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.9526253461837768,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.0524990558624268,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.049737372994422914,
+      "mean_token_accuracy": 0.9795652389526367,
+      "num_tokens": 19046105.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.9484909176826477,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.0715620517730713,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.04935494959354401,
+      "mean_token_accuracy": 0.9797740161418915,
+      "num_tokens": 19234730.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.9398573458194732,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2134987115859985,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.0518394410610199,
+      "mean_token_accuracy": 0.9793788075447083,
+      "num_tokens": 19423572.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.9465648651123046,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.004693627357483,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.04776117205619812,
+      "mean_token_accuracy": 0.9813261866569519,
+      "num_tokens": 19612465.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.9448004186153411,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.2913438081741333,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.04807930588722229,
+      "mean_token_accuracy": 0.9805689513683319,
+      "num_tokens": 19800482.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.9425322234630584,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1068469285964966,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.05053595900535583,
+      "mean_token_accuracy": 0.9794536828994751,
+      "num_tokens": 19989444.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.9472232103347779,
+      "epoch": 1.440107671601615,
+      "grad_norm": 0.7409384250640869,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.052136778831481934,
+      "mean_token_accuracy": 0.979197359085083,
+      "num_tokens": 20178482.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9450975477695465,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.3440778255462646,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.05258495211601257,
+      "mean_token_accuracy": 0.9798152863979339,
+      "num_tokens": 20367345.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9468007743358612,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 0.8377829194068909,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.04893603026866913,
+      "mean_token_accuracy": 0.9805733859539032,
+      "num_tokens": 20555817.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.9505827724933624,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.776889979839325,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.05367119312286377,
+      "mean_token_accuracy": 0.9779646098613739,
+      "num_tokens": 20744001.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.9516380190849304,
+      "epoch": 1.493943472409152,
+      "grad_norm": 0.8598276376724243,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.05071394443511963,
+      "mean_token_accuracy": 0.979692417383194,
+      "num_tokens": 20931715.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.9351628720760345,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3307439088821411,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.051372635364532473,
+      "mean_token_accuracy": 0.9797907948493958,
+      "num_tokens": 21120803.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.9268040716648102,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9320250749588013,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.04939974546432495,
+      "mean_token_accuracy": 0.98038170337677,
+      "num_tokens": 21309384.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9241463243961334,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.065938115119934,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.04856643378734589,
+      "mean_token_accuracy": 0.9800511121749877,
+      "num_tokens": 21498234.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.9350252687931061,
+      "epoch": 1.547779273216689,
+      "grad_norm": 0.6925032138824463,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.0512526273727417,
+      "mean_token_accuracy": 0.979562646150589,
+      "num_tokens": 21687119.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.9403733670711517,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.2069101333618164,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.049795085191726686,
+      "mean_token_accuracy": 0.9806506633758545,
+      "num_tokens": 21876094.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9370538294315338,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.05461466312408447,
+      "mean_token_accuracy": 0.9776057600975037,
+      "num_tokens": 22064836.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9517778217792511,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 0.8965149521827698,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.05289326906204224,
+      "mean_token_accuracy": 0.9782088756561279,
+      "num_tokens": 22253065.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.9372851848602295,
+      "epoch": 1.601615074024226,
+      "grad_norm": 0.7192108631134033,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.048973691463470456,
+      "mean_token_accuracy": 0.9806210815906524,
+      "num_tokens": 22441206.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.9122351944446564,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 0.8763614296913147,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.0461783230304718,
+      "mean_token_accuracy": 0.9817197680473327,
+      "num_tokens": 22629637.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.9162654876708984,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 0.9806778430938721,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.04827338755130768,
+      "mean_token_accuracy": 0.9806499361991883,
+      "num_tokens": 22818369.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.9125026524066925,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 0.9685350656509399,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.048379385471343996,
+      "mean_token_accuracy": 0.980905145406723,
+      "num_tokens": 23006863.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9119791567325592,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.8717457056045532,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.050873959064483644,
+      "mean_token_accuracy": 0.979601925611496,
+      "num_tokens": 23195735.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.9244007229804992,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.0574654340744019,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.05112813711166382,
+      "mean_token_accuracy": 0.979493772983551,
+      "num_tokens": 23383862.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9225810945034028,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 0.9628807306289673,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.050480544567108154,
+      "mean_token_accuracy": 0.9797678411006927,
+      "num_tokens": 23572292.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.9287579476833343,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.015265941619873,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.05133126378059387,
+      "mean_token_accuracy": 0.9797986805438995,
+      "num_tokens": 23760252.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.9133845925331116,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 0.8996883630752563,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.05044950246810913,
+      "mean_token_accuracy": 0.9795303404331207,
+      "num_tokens": 23949031.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.9213698863983154,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.0490264892578125,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.04797698855400086,
+      "mean_token_accuracy": 0.9806023716926575,
+      "num_tokens": 24137829.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.9173697710037232,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.3648769855499268,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.049194514751434326,
+      "mean_token_accuracy": 0.980712479352951,
+      "num_tokens": 24326586.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9242741882801055,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.105125069618225,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.048667973279953,
+      "mean_token_accuracy": 0.9801449298858642,
+      "num_tokens": 24514904.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.9162627995014191,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.8887938857078552,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.04878672957420349,
+      "mean_token_accuracy": 0.980469423532486,
+      "num_tokens": 24703688.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.9149538338184356,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 0.8838976621627808,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.04892318844795227,
+      "mean_token_accuracy": 0.9808776795864105,
+      "num_tokens": 24892107.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.914735221862793,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1152070760726929,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.04825109839439392,
+      "mean_token_accuracy": 0.9804604113101959,
+      "num_tokens": 25080130.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9170725226402283,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9479517340660095,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.051445144414901736,
+      "mean_token_accuracy": 0.9792819261550904,
+      "num_tokens": 25268652.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9142911911010743,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 0.8172292709350586,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.0475692093372345,
+      "mean_token_accuracy": 0.9809505581855774,
+      "num_tokens": 25457120.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.9019249439239502,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 0.8183121681213379,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.04979957342147827,
+      "mean_token_accuracy": 0.9799730658531189,
+      "num_tokens": 25645699.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.9181066155433655,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.8256579041481018,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.04996164441108704,
+      "mean_token_accuracy": 0.9802247405052185,
+      "num_tokens": 25834183.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9171363770961761,
+      "epoch": 1.857335127860027,
+      "grad_norm": 0.9727098345756531,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.048660767078399655,
+      "mean_token_accuracy": 0.9805064260959625,
+      "num_tokens": 26023317.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.9112720847129822,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 0.9826673865318298,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.04775593280792236,
+      "mean_token_accuracy": 0.9809929549694061,
+      "num_tokens": 26212115.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9034272134304047,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.9951306581497192,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.04939360618591308,
+      "mean_token_accuracy": 0.9800347864627839,
+      "num_tokens": 26401167.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.897865754365921,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.0163404941558838,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05002856254577637,
+      "mean_token_accuracy": 0.9793690323829651,
+      "num_tokens": 26590570.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9004576802253723,
+      "epoch": 1.911170928667564,
+      "grad_norm": 0.6666616201400757,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.04905453026294708,
+      "mean_token_accuracy": 0.9803856253623963,
+      "num_tokens": 26779481.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.8978902101516724,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.9295416474342346,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.05261261463165283,
+      "mean_token_accuracy": 0.9785708487033844,
+      "num_tokens": 26968262.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.8962675571441651,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 0.88262939453125,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.04765265882015228,
+      "mean_token_accuracy": 0.98052077293396,
+      "num_tokens": 27157631.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8909463047981262,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.4205769300460815,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.049406200647354126,
+      "mean_token_accuracy": 0.979697072505951,
+      "num_tokens": 27345921.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8969364166259766,
+      "epoch": 1.965006729475101,
+      "grad_norm": 0.9517145752906799,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0510346531867981,
+      "mean_token_accuracy": 0.9793386399745941,
+      "num_tokens": 27534751.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.9092479169368743,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.200056791305542,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.04974203705787659,
+      "mean_token_accuracy": 0.9800146698951722,
+      "num_tokens": 27723763.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9059641897678375,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.072447657585144,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.05065792202949524,
+      "mean_token_accuracy": 0.9799223959445953,
+      "num_tokens": 27911784.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9083398975384464,
+      "eval_loss": 0.05432562157511711,
+      "eval_mean_token_accuracy": 0.9782073607869969,
+      "eval_num_tokens": 28025286.0,
+      "eval_runtime": 24.4304,
+      "eval_samples_per_second": 204.663,
+      "eval_steps_per_second": 6.426,
+      "step": 1486
+    },
+    {
+      "entropy": 0.9113237977027893,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.0167063474655151,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.044942992925643924,
+      "mean_token_accuracy": 0.9818171083927154,
+      "num_tokens": 28100906.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.8780008792877197,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.9552241563796997,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.036248764395713805,
+      "mean_token_accuracy": 0.9852245450019836,
+      "num_tokens": 28289445.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.8721176147460937,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 0.8951388001441956,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.037987279891967776,
+      "mean_token_accuracy": 0.9850614190101623,
+      "num_tokens": 28477972.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.8922657191753387,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 0.9028105735778809,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.03588914275169373,
+      "mean_token_accuracy": 0.9863102614879609,
+      "num_tokens": 28666272.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.877799642086029,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.153886318206787,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.03577309250831604,
+      "mean_token_accuracy": 0.9863148391246795,
+      "num_tokens": 28854780.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.8708971381187439,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.042955994606018,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.038266432285308835,
+      "mean_token_accuracy": 0.9848419308662415,
+      "num_tokens": 29043741.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.8800721943378449,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 0.6789060235023499,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.032948991656303404,
+      "mean_token_accuracy": 0.9871293902397156,
+      "num_tokens": 29232719.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.8743580460548401,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.1780879497528076,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.036874374747276305,
+      "mean_token_accuracy": 0.9859182178974152,
+      "num_tokens": 29421817.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.8851303637027741,
+      "epoch": 2.113055181695828,
+      "grad_norm": 0.747734785079956,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.0364631175994873,
+      "mean_token_accuracy": 0.9862085223197937,
+      "num_tokens": 29610344.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.8798732101917267,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.5001860857009888,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.03813003897666931,
+      "mean_token_accuracy": 0.9852604746818543,
+      "num_tokens": 29798997.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.8797551989555359,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 0.9593478441238403,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.0394927829504013,
+      "mean_token_accuracy": 0.984604275226593,
+      "num_tokens": 29987809.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.8860546290874481,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.8347703218460083,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.036699697375297546,
+      "mean_token_accuracy": 0.985380882024765,
+      "num_tokens": 30176234.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.8783667802810669,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1154481172561646,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.03756999969482422,
+      "mean_token_accuracy": 0.9853451430797577,
+      "num_tokens": 30364657.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.8684478521347045,
+      "epoch": 2.180349932705249,
+      "grad_norm": 0.7515475153923035,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.0382376104593277,
+      "mean_token_accuracy": 0.9854029655456543,
+      "num_tokens": 30552904.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.8636964917182922,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 0.7711939215660095,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.036220991611480714,
+      "mean_token_accuracy": 0.9860713243484497,
+      "num_tokens": 30741259.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.8592016279697419,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.1143887042999268,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.036316031217575075,
+      "mean_token_accuracy": 0.985757052898407,
+      "num_tokens": 30930144.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.8671079576015472,
+      "epoch": 2.220726783310902,
+      "grad_norm": 0.8980015516281128,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.03822658061981201,
+      "mean_token_accuracy": 0.9847764372825623,
+      "num_tokens": 31118624.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.8785548269748688,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.094323992729187,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.03736622333526611,
+      "mean_token_accuracy": 0.9850081980228425,
+      "num_tokens": 31306833.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.866977310180664,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8372092843055725,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.0401554524898529,
+      "mean_token_accuracy": 0.9840337932109833,
+      "num_tokens": 31495503.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.8787918210029602,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.5408164262771606,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0368030846118927,
+      "mean_token_accuracy": 0.9847234487533569,
+      "num_tokens": 31683865.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.872721153497696,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.1115421056747437,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.037859299778938295,
+      "mean_token_accuracy": 0.9851646661758423,
+      "num_tokens": 31871903.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.8774094223976135,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.0301331281661987,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.038671016693115234,
+      "mean_token_accuracy": 0.9840058028697968,
+      "num_tokens": 32060500.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.889606237411499,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8601903915405273,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.039603835344314574,
+      "mean_token_accuracy": 0.9840235590934754,
+      "num_tokens": 32248744.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.886734277009964,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.1029839515686035,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.039513933658599856,
+      "mean_token_accuracy": 0.9844573020935059,
+      "num_tokens": 32437640.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.8843017637729644,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.2545154094696045,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.036970189213752745,
+      "mean_token_accuracy": 0.985239815711975,
+      "num_tokens": 32625986.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.8913422048091888,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.9613803625106812,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.03811657428741455,
+      "mean_token_accuracy": 0.9852804243564606,
+      "num_tokens": 32814531.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.8907467782497406,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.1811398267745972,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.03903660774230957,
+      "mean_token_accuracy": 0.9842524945735931,
+      "num_tokens": 33002960.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.8833664715290069,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.8338477611541748,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.03982087969779968,
+      "mean_token_accuracy": 0.9841201484203339,
+      "num_tokens": 33191183.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.8783826470375061,
+      "epoch": 2.382234185733513,
+      "grad_norm": 0.6904510259628296,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.03699290752410889,
+      "mean_token_accuracy": 0.9854120731353759,
+      "num_tokens": 33379798.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.8871048510074615,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.009539246559143,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.03926805555820465,
+      "mean_token_accuracy": 0.9843246698379516,
+      "num_tokens": 33568531.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.8973462700843811,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.8916401267051697,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.039596831798553465,
+      "mean_token_accuracy": 0.9839228212833404,
+      "num_tokens": 33756333.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.8954619467258453,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.056504487991333,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0396859347820282,
+      "mean_token_accuracy": 0.9837161123752594,
+      "num_tokens": 33944904.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.8952643811702728,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0442084074020386,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.037324142456054685,
+      "mean_token_accuracy": 0.9852897703647614,
+      "num_tokens": 34133530.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.8874911010265351,
+      "epoch": 2.449528936742934,
+      "grad_norm": 0.9926224946975708,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.03637495338916778,
+      "mean_token_accuracy": 0.9853426337242126,
+      "num_tokens": 34322042.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.8900584518909455,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.001734733581543,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04029585719108582,
+      "mean_token_accuracy": 0.9842372059822082,
+      "num_tokens": 34510259.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.8986345052719116,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 0.772941529750824,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.0401084691286087,
+      "mean_token_accuracy": 0.9841427087783814,
+      "num_tokens": 34698669.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.8911370873451233,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.895031213760376,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.03941032290458679,
+      "mean_token_accuracy": 0.9841966688632965,
+      "num_tokens": 34887160.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.8901963472366333,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0677918195724487,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.03503885865211487,
+      "mean_token_accuracy": 0.9856241464614868,
+      "num_tokens": 35075891.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.8817630231380462,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.1200164556503296,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.03878425657749176,
+      "mean_token_accuracy": 0.9846492648124695,
+      "num_tokens": 35264455.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.8897477686405182,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.8058190941810608,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.041558006405830385,
+      "mean_token_accuracy": 0.9830883860588073,
+      "num_tokens": 35453251.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.8873181045055389,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.7765992283821106,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.03555050492286682,
+      "mean_token_accuracy": 0.9858225345611572,
+      "num_tokens": 35641959.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.8782606959342957,
+      "epoch": 2.557200538358008,
+      "grad_norm": 0.8488866686820984,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.03606923818588257,
+      "mean_token_accuracy": 0.9858632445335388,
+      "num_tokens": 35829636.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.8869808673858642,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.1734739542007446,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.040956351161003116,
+      "mean_token_accuracy": 0.9835374176502227,
+      "num_tokens": 36017818.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.8850924432277679,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.3584387302398682,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.042437011003494264,
+      "mean_token_accuracy": 0.9826479077339172,
+      "num_tokens": 36206505.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.8801687896251679,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2862240076065063,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.03880060315132141,
+      "mean_token_accuracy": 0.9843404710292816,
+      "num_tokens": 36395802.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.8752781748771667,
+      "epoch": 2.611036339165545,
+      "grad_norm": 0.7341249585151672,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.03776344656944275,
+      "mean_token_accuracy": 0.9852877616882324,
+      "num_tokens": 36584370.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.8759494423866272,
+      "epoch": 2.624495289367429,
+      "grad_norm": 0.9807276129722595,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.03780297338962555,
+      "mean_token_accuracy": 0.9853227376937866,
+      "num_tokens": 36772865.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.8753129065036773,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.1391628980636597,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.03953765034675598,
+      "mean_token_accuracy": 0.9839197635650635,
+      "num_tokens": 36961462.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.8771899223327637,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.0865391492843628,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.03856399655342102,
+      "mean_token_accuracy": 0.9848688066005706,
+      "num_tokens": 37150486.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.8774881184101104,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.8561595678329468,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.03795175850391388,
+      "mean_token_accuracy": 0.9849309325218201,
+      "num_tokens": 37338890.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.8841517448425293,
+      "epoch": 2.678331090174966,
+      "grad_norm": 0.9458399415016174,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.03860213160514832,
+      "mean_token_accuracy": 0.9842230796813964,
+      "num_tokens": 37526886.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.8809043228626251,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 0.9360871315002441,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.03709094822406769,
+      "mean_token_accuracy": 0.9855118036270142,
+      "num_tokens": 37715408.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.8805519282817841,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.8852124214172363,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.03869341611862183,
+      "mean_token_accuracy": 0.9846724212169647,
+      "num_tokens": 37904328.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.8796228408813477,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.189297080039978,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.03675769567489624,
+      "mean_token_accuracy": 0.9854080140590668,
+      "num_tokens": 38092776.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.8722020089626312,
+      "epoch": 2.732166890982503,
+      "grad_norm": 0.7027947306632996,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.03889244794845581,
+      "mean_token_accuracy": 0.984558242559433,
+      "num_tokens": 38282221.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.8746632814407349,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 0.9799376726150513,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.03815680146217346,
+      "mean_token_accuracy": 0.9845012128353119,
+      "num_tokens": 38470633.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.8766567528247833,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 0.850459098815918,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.039029371738433835,
+      "mean_token_accuracy": 0.9845906794071198,
+      "num_tokens": 38659048.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.8778688013553619,
+      "epoch": 2.772543741588156,
+      "grad_norm": 0.8540134429931641,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.039645448327064514,
+      "mean_token_accuracy": 0.9840566098690033,
+      "num_tokens": 38847370.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.8758600771427154,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8036021590232849,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.038226932287216187,
+      "mean_token_accuracy": 0.9845647215843201,
+      "num_tokens": 39036040.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.8652268946170807,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 0.8526738286018372,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.03798363208770752,
+      "mean_token_accuracy": 0.9845202267169952,
+      "num_tokens": 39224695.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.8707200348377228,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 0.9852389097213745,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.037227436900138855,
+      "mean_token_accuracy": 0.9852050960063934,
+      "num_tokens": 39413360.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.8711447060108185,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.228277325630188,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.03698050379753113,
+      "mean_token_accuracy": 0.9852443158626556,
+      "num_tokens": 39602544.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.8742912471294403,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 0.8440544605255127,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.03757670521736145,
+      "mean_token_accuracy": 0.9848618268966675,
+      "num_tokens": 39791340.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.8801726162433624,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0955252647399902,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.040004149079322815,
+      "mean_token_accuracy": 0.9841345012187958,
+      "num_tokens": 39980212.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.8931492984294891,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.5722652673721313,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.038096648454666135,
+      "mean_token_accuracy": 0.9849136590957641,
+      "num_tokens": 40169122.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.8767493844032288,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.7663730382919312,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.03729096055030823,
+      "mean_token_accuracy": 0.985169780254364,
+      "num_tokens": 40357678.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.8770561516284943,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.1505672931671143,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.03984796404838562,
+      "mean_token_accuracy": 0.9838845670223236,
+      "num_tokens": 40546892.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.8789129912853241,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 0.8355761766433716,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.03769223690032959,
+      "mean_token_accuracy": 0.9847559213638306,
+      "num_tokens": 40736109.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.8781779289245606,
+      "epoch": 2.920592193808883,
+      "grad_norm": 0.9103918075561523,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.038262826204299924,
+      "mean_token_accuracy": 0.984741848707199,
+      "num_tokens": 40924341.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.8552807629108429,
+      "epoch": 2.934051144010767,
+      "grad_norm": 0.8338510990142822,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.0355743408203125,
+      "mean_token_accuracy": 0.9855036079883576,
+      "num_tokens": 41113849.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.8705591142177582,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 0.8334057927131653,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.037476515769958495,
+      "mean_token_accuracy": 0.9854551255702972,
+      "num_tokens": 41303130.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.8809274792671203,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 0.9424103498458862,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.03624279499053955,
+      "mean_token_accuracy": 0.9856135487556458,
+      "num_tokens": 41491750.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.8818133771419525,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.04243004322052,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.039132320880889894,
+      "mean_token_accuracy": 0.984746390581131,
+      "num_tokens": 41680493.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.8838395297527313,
+      "epoch": 2.987886944818304,
+      "grad_norm": 0.8650784492492676,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.03822052478790283,
+      "mean_token_accuracy": 0.9852108955383301,
+      "num_tokens": 41868378.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.8800126549544608,
+      "eval_loss": 0.05468416586518288,
+      "eval_mean_token_accuracy": 0.9781015617832257,
+      "eval_num_tokens": 42038037.0,
+      "eval_runtime": 24.4825,
+      "eval_samples_per_second": 204.228,
+      "eval_steps_per_second": 6.413,
+      "step": 2229
+    },
+    {
+      "entropy": 0.8773060619831086,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.7845072150230408,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.0365300714969635,
+      "mean_token_accuracy": 0.985853761434555,
+      "num_tokens": 42056717.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.8720550239086151,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 0.9917275905609131,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.030071893334388734,
+      "mean_token_accuracy": 0.9890478909015655,
+      "num_tokens": 42245058.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.8489451467990875,
+      "epoch": 3.028263795423957,
+      "grad_norm": 0.9414294362068176,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.027054613828659056,
+      "mean_token_accuracy": 0.9894806027412415,
+      "num_tokens": 42433468.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.848755830526352,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.7276235818862915,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.026897600293159483,
+      "mean_token_accuracy": 0.9893308103084564,
+      "num_tokens": 42621657.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.8402929604053497,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 0.9455838799476624,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.027918827533721925,
+      "mean_token_accuracy": 0.9888385117053986,
+      "num_tokens": 42810533.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.8434606611728668,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 0.7503429651260376,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.025769710540771484,
+      "mean_token_accuracy": 0.989881956577301,
+      "num_tokens": 42999085.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.8459890782833099,
+      "epoch": 3.082099596231494,
+      "grad_norm": 0.736440896987915,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.028781753778457642,
+      "mean_token_accuracy": 0.9882656812667847,
+      "num_tokens": 43188222.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.8493306159973144,
+      "epoch": 3.095558546433378,
+      "grad_norm": 0.9577816128730774,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.02689719796180725,
+      "mean_token_accuracy": 0.9889744818210602,
+      "num_tokens": 43376648.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.8512665331363678,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 0.9631588459014893,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.031439167261123654,
+      "mean_token_accuracy": 0.9874942421913147,
+      "num_tokens": 43564769.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.8519513130187988,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1011922359466553,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.02696958780288696,
+      "mean_token_accuracy": 0.9900951981544495,
+      "num_tokens": 43753362.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.8507618844509125,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9351623058319092,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.026315575838088988,
+      "mean_token_accuracy": 0.989511913061142,
+      "num_tokens": 43942063.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.8506478369235992,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.096564769744873,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.029049152135849,
+      "mean_token_accuracy": 0.9886228442192078,
+      "num_tokens": 44130381.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.8624941289424897,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 0.7859402298927307,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.027744990587234498,
+      "mean_token_accuracy": 0.9890527307987214,
+      "num_tokens": 44319243.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.8627866744995117,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 0.7764605283737183,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.02571650445461273,
+      "mean_token_accuracy": 0.9903442323207855,
+      "num_tokens": 44507698.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.8566473722457886,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.8182396292686462,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.025625643134117127,
+      "mean_token_accuracy": 0.9897150516510009,
+      "num_tokens": 44695954.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.8555088877677918,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.0277098417282104,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.028235083818435668,
+      "mean_token_accuracy": 0.9894341349601745,
+      "num_tokens": 44884558.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.8562702238559723,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.0676493644714355,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.027713948488235475,
+      "mean_token_accuracy": 0.9891053259372711,
+      "num_tokens": 45073381.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.8520576059818268,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 0.7926117181777954,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.028936928510665892,
+      "mean_token_accuracy": 0.988807988166809,
+      "num_tokens": 45262350.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.860496312379837,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.1108510494232178,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030533093214035033,
+      "mean_token_accuracy": 0.9882748067378998,
+      "num_tokens": 45450782.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.872324675321579,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.3057194948196411,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.032421305775642395,
+      "mean_token_accuracy": 0.9879753351211548,
+      "num_tokens": 45639122.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.8778823137283325,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.289680004119873,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03096393346786499,
+      "mean_token_accuracy": 0.9883370697498322,
+      "num_tokens": 45827730.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.8627406179904937,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.8361398577690125,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.02957139313220978,
+      "mean_token_accuracy": 0.9884674072265625,
+      "num_tokens": 46016274.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.8584554970264435,
+      "epoch": 3.297442799461642,
+      "grad_norm": 0.9330723881721497,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.029568272829055785,
+      "mean_token_accuracy": 0.9882686555385589,
+      "num_tokens": 46204679.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.8515178024768829,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.0296216011047363,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.027788180112838744,
+      "mean_token_accuracy": 0.9888734877109527,
+      "num_tokens": 46393989.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.8488850116729736,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1329699754714966,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.028057461977005003,
+      "mean_token_accuracy": 0.9895590543746948,
+      "num_tokens": 46582790.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.849625188112259,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 0.9267219305038452,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.02802973985671997,
+      "mean_token_accuracy": 0.988851535320282,
+      "num_tokens": 46771138.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.854446280002594,
+      "epoch": 3.351278600269179,
+      "grad_norm": 0.8713932037353516,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.030220019817352294,
+      "mean_token_accuracy": 0.9882843673229218,
+      "num_tokens": 46960054.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.8599806547164917,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.675019383430481,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.02958865463733673,
+      "mean_token_accuracy": 0.9880296885967255,
+      "num_tokens": 47149567.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.8549504458904267,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.2176353931427002,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.027722162008285523,
+      "mean_token_accuracy": 0.9890090465545655,
+      "num_tokens": 47338060.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.8552233815193176,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9571717977523804,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.028924965858459474,
+      "mean_token_accuracy": 0.9884298503398895,
+      "num_tokens": 47525895.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.8521630108356476,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.965713620185852,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.030511701107025148,
+      "mean_token_accuracy": 0.9881398677825928,
+      "num_tokens": 47714524.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.8537400126457214,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 0.9005307555198669,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.030432754755020143,
+      "mean_token_accuracy": 0.9876708447933197,
+      "num_tokens": 47903188.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.8596058666706086,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 0.8159075379371643,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.027927806973457335,
+      "mean_token_accuracy": 0.9889210939407349,
+      "num_tokens": 48091334.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.8482735753059387,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0411661863327026,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03003154993057251,
+      "mean_token_accuracy": 0.9882730603218078,
+      "num_tokens": 48279622.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.8466426789760589,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.189595341682434,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03237656056880951,
+      "mean_token_accuracy": 0.9869072616100312,
+      "num_tokens": 48467527.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.8545588493347168,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0700056552886963,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.0306060791015625,
+      "mean_token_accuracy": 0.9880664825439454,
+      "num_tokens": 48656151.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.8503433585166931,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.7099478244781494,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.02984013557434082,
+      "mean_token_accuracy": 0.9879442811012268,
+      "num_tokens": 48844796.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.8464376091957092,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9877128005027771,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.0306446373462677,
+      "mean_token_accuracy": 0.9883591532707214,
+      "num_tokens": 49033216.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.8477609276771545,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.3324531316757202,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.030148410797119142,
+      "mean_token_accuracy": 0.9879989802837372,
+      "num_tokens": 49221425.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.8461863219738006,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8938570022583008,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.02682304084300995,
+      "mean_token_accuracy": 0.9897881150245667,
+      "num_tokens": 49409785.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.8362689018249512,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 0.7423873543739319,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.02689332664012909,
+      "mean_token_accuracy": 0.9899903059005737,
+      "num_tokens": 49598524.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.8395139217376709,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.921755850315094,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.030062124133110046,
+      "mean_token_accuracy": 0.9880835592746735,
+      "num_tokens": 49787241.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.8467226803302765,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0296459197998047,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.030666223168373107,
+      "mean_token_accuracy": 0.9877098739147187,
+      "num_tokens": 49975844.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.833392071723938,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2861948013305664,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.028174835443496703,
+      "mean_token_accuracy": 0.989056545495987,
+      "num_tokens": 50164857.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.8340730309486389,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 0.8116776347160339,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.028390336036682128,
+      "mean_token_accuracy": 0.9886897742748261,
+      "num_tokens": 50353452.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.8357116639614105,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.8362084031105042,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.029806244373321533,
+      "mean_token_accuracy": 0.988309508562088,
+      "num_tokens": 50542461.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.8363270878791809,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8438499569892883,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.028577303886413573,
+      "mean_token_accuracy": 0.9886945366859436,
+      "num_tokens": 50730872.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.8310261785984039,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.1920002698898315,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.030321845412254335,
+      "mean_token_accuracy": 0.9878708899021149,
+      "num_tokens": 50919889.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.8380252480506897,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 0.9409399032592773,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.030501589179039,
+      "mean_token_accuracy": 0.9872083842754364,
+      "num_tokens": 51108079.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.8422612130641938,
+      "epoch": 3.660834454912517,
+      "grad_norm": 0.8370440006256104,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.02974823713302612,
+      "mean_token_accuracy": 0.9885554671287536,
+      "num_tokens": 51296913.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.8387595832347869,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1360642910003662,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.026970452070236205,
+      "mean_token_accuracy": 0.9888331353664398,
+      "num_tokens": 51485678.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.8403031289577484,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.1195677518844604,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.029451662302017213,
+      "mean_token_accuracy": 0.988847154378891,
+      "num_tokens": 51674765.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.8499106168746948,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 0.9573526978492737,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.030826866626739502,
+      "mean_token_accuracy": 0.9875578820705414,
+      "num_tokens": 51863924.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.8576709866523743,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9386286735534668,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.02946900427341461,
+      "mean_token_accuracy": 0.9885939955711365,
+      "num_tokens": 52052011.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.8404816091060638,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.095536708831787,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.030070826411247253,
+      "mean_token_accuracy": 0.9881400763988495,
+      "num_tokens": 52240955.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.8346689164638519,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9964426159858704,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03005717396736145,
+      "mean_token_accuracy": 0.9883057296276092,
+      "num_tokens": 52429544.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.8377436757087707,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.8499778509140015,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.029283815622329713,
+      "mean_token_accuracy": 0.9882387280464172,
+      "num_tokens": 52618405.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.8328788638114929,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9594390988349915,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.02666555345058441,
+      "mean_token_accuracy": 0.9889323055744171,
+      "num_tokens": 52806689.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.829415249824524,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.3587099313735962,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.028349629044532774,
+      "mean_token_accuracy": 0.9890907645225525,
+      "num_tokens": 52995248.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.8335795283317566,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.077789306640625,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.03001731038093567,
+      "mean_token_accuracy": 0.9878358781337738,
+      "num_tokens": 53183607.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.8316770553588867,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 0.7265951037406921,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.030136841535568237,
+      "mean_token_accuracy": 0.9880174934864044,
+      "num_tokens": 53372735.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.8310399711132049,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.6605978012084961,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.026665833592414857,
+      "mean_token_accuracy": 0.9892704486846924,
+      "num_tokens": 53561776.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.8258654296398162,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.6560797691345215,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.026718440651893615,
+      "mean_token_accuracy": 0.989847868680954,
+      "num_tokens": 53750019.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.8221413195133209,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 0.7495356798171997,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.027431756258010864,
+      "mean_token_accuracy": 0.9892430484294892,
+      "num_tokens": 53938914.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.8341928541660308,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.0676971673965454,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.028432759642601012,
+      "mean_token_accuracy": 0.9889024317264556,
+      "num_tokens": 54127763.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.8359149813652038,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.380476713180542,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.028149163722991942,
+      "mean_token_accuracy": 0.9884953200817108,
+      "num_tokens": 54315930.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.8294597804546356,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0728843212127686,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.028733691573143004,
+      "mean_token_accuracy": 0.9884523034095765,
+      "num_tokens": 54504632.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.8245523810386658,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 0.9723307490348816,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.02793261408805847,
+      "mean_token_accuracy": 0.9891038477420807,
+      "num_tokens": 54693752.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.8279620766639709,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.003624439239502,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.028367668390274048,
+      "mean_token_accuracy": 0.9893603324890137,
+      "num_tokens": 54882030.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.8306584179401397,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.8481361865997314,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.029952478408813477,
+      "mean_token_accuracy": 0.9881332099437714,
+      "num_tokens": 55070549.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.8316469848155975,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.7631049156188965,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.029351180791854857,
+      "mean_token_accuracy": 0.9888184070587158,
+      "num_tokens": 55258999.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.8221287310123444,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 0.7234840393066406,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.027709537744522096,
+      "mean_token_accuracy": 0.9890308439731598,
+      "num_tokens": 55448047.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.8239192545413971,
+      "epoch": 3.970390309555855,
+      "grad_norm": 0.8497020602226257,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.029604345560073853,
+      "mean_token_accuracy": 0.9881375730037689,
+      "num_tokens": 55636566.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.8319927275180816,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.1599552631378174,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.02763860821723938,
+      "mean_token_accuracy": 0.9891406774520874,
+      "num_tokens": 55825049.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.8404937744140625,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 0.906755805015564,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.029578033089637756,
+      "mean_token_accuracy": 0.9878996431827545,
+      "num_tokens": 56012965.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.8419649528849656,
+      "eval_loss": 0.0570383295416832,
+      "eval_mean_token_accuracy": 0.9787763862093543,
+      "eval_num_tokens": 56050518.0,
+      "eval_runtime": 24.4889,
+      "eval_samples_per_second": 204.174,
+      "eval_steps_per_second": 6.411,
+      "step": 2972
+    },
+    {
+      "entropy": 0.8328965544700623,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7073294520378113,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.020916813611984254,
+      "mean_token_accuracy": 0.992138022184372,
+      "num_tokens": 56201621.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.8153974890708924,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.8904889822006226,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.01942949742078781,
+      "mean_token_accuracy": 0.9927055358886718,
+      "num_tokens": 56390018.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.8080023109912873,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.0828146934509277,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.018959003686904907,
+      "mean_token_accuracy": 0.9928978025913239,
+      "num_tokens": 56578046.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.8056950688362121,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7407417893409729,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019250378012657166,
+      "mean_token_accuracy": 0.9926685452461242,
+      "num_tokens": 56767192.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.8124530673027038,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.0355021953582764,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.01904640197753906,
+      "mean_token_accuracy": 0.9925767481327057,
+      "num_tokens": 56955769.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.810802960395813,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.3759289979934692,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.020995599031448365,
+      "mean_token_accuracy": 0.9917719483375549,
+      "num_tokens": 57144300.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.8159192740917206,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.112034559249878,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.01948906183242798,
+      "mean_token_accuracy": 0.9928658187389374,
+      "num_tokens": 57332935.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.821906328201294,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.0319583415985107,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019549430906772615,
+      "mean_token_accuracy": 0.9922966659069061,
+      "num_tokens": 57520775.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.8133022606372833,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.067180871963501,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.01833198368549347,
+      "mean_token_accuracy": 0.9934999167919158,
+      "num_tokens": 57709495.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.8150732755661011,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.277079701423645,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.019589337706565856,
+      "mean_token_accuracy": 0.992507028579712,
+      "num_tokens": 57897461.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.8159323990345001,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.3150911331176758,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.01963164806365967,
+      "mean_token_accuracy": 0.9926789402961731,
+      "num_tokens": 58085915.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.8147098064422608,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.0073580741882324,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.02144385576248169,
+      "mean_token_accuracy": 0.9918489575386047,
+      "num_tokens": 58274583.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.8155419886112213,
+      "epoch": 4.172274562584119,
+      "grad_norm": 0.8892306685447693,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.02215111702680588,
+      "mean_token_accuracy": 0.9914537072181702,
+      "num_tokens": 58463241.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.8084070324897766,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.7887238264083862,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.018581822514533997,
+      "mean_token_accuracy": 0.9928648293018341,
+      "num_tokens": 58652281.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.8016656517982483,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.0644817352294922,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.020099574327468873,
+      "mean_token_accuracy": 0.9916590929031373,
+      "num_tokens": 58839946.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.8035425901412964,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.9588403701782227,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020384626090526582,
+      "mean_token_accuracy": 0.9918951213359832,
+      "num_tokens": 59028994.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.8114414095878602,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.7145556211471558,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.019365760684013366,
+      "mean_token_accuracy": 0.9921130836009979,
+      "num_tokens": 59217599.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.8123053193092347,
+      "epoch": 4.23956931359354,
+      "grad_norm": 0.7928217649459839,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.020542070269584656,
+      "mean_token_accuracy": 0.9918384432792664,
+      "num_tokens": 59406176.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.810414057970047,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.1470942497253418,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.01809942126274109,
+      "mean_token_accuracy": 0.992941427230835,
+      "num_tokens": 59594604.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.8046662211418152,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.579035222530365,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.01914280205965042,
+      "mean_token_accuracy": 0.9926800608634949,
+      "num_tokens": 59782947.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.8034039616584778,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.8979132175445557,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.019897907972335815,
+      "mean_token_accuracy": 0.9919832170009613,
+      "num_tokens": 59971417.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.8137863457202912,
+      "epoch": 4.293405114401077,
+      "grad_norm": 0.9149890542030334,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.02069481760263443,
+      "mean_token_accuracy": 0.992353367805481,
+      "num_tokens": 60159983.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.8163237690925598,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.8264126181602478,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.021351543068885804,
+      "mean_token_accuracy": 0.9921438097953796,
+      "num_tokens": 60348327.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.8133926391601562,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.0638184547424316,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.020092280209064485,
+      "mean_token_accuracy": 0.9920917391777039,
+      "num_tokens": 60537152.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.8095248222351075,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.8831470012664795,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.01908399909734726,
+      "mean_token_accuracy": 0.9929656088352203,
+      "num_tokens": 60725603.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.8069245994091034,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8486410975456238,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.01925535798072815,
+      "mean_token_accuracy": 0.9927195310592651,
+      "num_tokens": 60914021.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.8006627321243286,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.9021628499031067,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.019187870621681213,
+      "mean_token_accuracy": 0.992531019449234,
+      "num_tokens": 61102037.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.8067023754119873,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.7793697118759155,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.021760080754756928,
+      "mean_token_accuracy": 0.9915579319000244,
+      "num_tokens": 61290132.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.8075471341609954,
+      "epoch": 4.387617765814267,
+      "grad_norm": 0.736788809299469,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.02077012062072754,
+      "mean_token_accuracy": 0.9920295119285584,
+      "num_tokens": 61478897.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.8083206176757812,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.1687020063400269,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.02117353230714798,
+      "mean_token_accuracy": 0.991925036907196,
+      "num_tokens": 61667982.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.81106236577034,
+      "epoch": 4.414535666218035,
+      "grad_norm": 0.7596265077590942,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.02081204652786255,
+      "mean_token_accuracy": 0.9921064436435699,
+      "num_tokens": 61856557.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.8018859803676606,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.823410153388977,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.019047826528549194,
+      "mean_token_accuracy": 0.9924882829189301,
+      "num_tokens": 62045181.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7996662855148315,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.0748025178909302,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.01788569688796997,
+      "mean_token_accuracy": 0.9931309640407562,
+      "num_tokens": 62234215.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.8021627008914948,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9946282505989075,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021408762037754058,
+      "mean_token_accuracy": 0.9914931178092956,
+      "num_tokens": 62422361.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.8084165751934052,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.7313793301582336,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.020281489193439483,
+      "mean_token_accuracy": 0.9925725042819977,
+      "num_tokens": 62610667.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.8103892624378204,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.9093077182769775,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.018877537548542024,
+      "mean_token_accuracy": 0.9929530620574951,
+      "num_tokens": 62798970.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.8091431736946106,
+      "epoch": 4.495289367429341,
+      "grad_norm": 0.952234148979187,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021039776504039764,
+      "mean_token_accuracy": 0.9918699383735656,
+      "num_tokens": 62986914.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.8094858765602112,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.9959521293640137,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021489059925079344,
+      "mean_token_accuracy": 0.9914597749710083,
+      "num_tokens": 63175805.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.807330971956253,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.6189294457435608,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.02164270281791687,
+      "mean_token_accuracy": 0.9920661866664886,
+      "num_tokens": 63364692.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.8098866164684295,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7895700931549072,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.021856357157230378,
+      "mean_token_accuracy": 0.9916301846504212,
+      "num_tokens": 63553692.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.8143714487552642,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0242317914962769,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02046577036380768,
+      "mean_token_accuracy": 0.9923330962657928,
+      "num_tokens": 63742476.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.8123241424560547,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.4186221361160278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.020659247040748598,
+      "mean_token_accuracy": 0.9918934762477875,
+      "num_tokens": 63930898.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.8133015930652618,
+      "epoch": 4.576043068640646,
+      "grad_norm": 0.9085373878479004,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.021746012568473815,
+      "mean_token_accuracy": 0.9918161749839782,
+      "num_tokens": 64119589.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.8173148334026337,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.7257112264633179,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.019787387549877168,
+      "mean_token_accuracy": 0.9922669887542724,
+      "num_tokens": 64308293.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.8116482973098755,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7442086935043335,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.020933127403259276,
+      "mean_token_accuracy": 0.9917319416999817,
+      "num_tokens": 64497362.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.8134189546108246,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.7617318630218506,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.01842810958623886,
+      "mean_token_accuracy": 0.9927939116954804,
+      "num_tokens": 64685802.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.8112196505069733,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.7430832386016846,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.01809561103582382,
+      "mean_token_accuracy": 0.9930340230464936,
+      "num_tokens": 64873973.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.8128187835216523,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.7553606629371643,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.0202948659658432,
+      "mean_token_accuracy": 0.9919280350208283,
+      "num_tokens": 65062510.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.813840126991272,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.7390825152397156,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.020557042956352235,
+      "mean_token_accuracy": 0.9918098270893096,
+      "num_tokens": 65250940.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.8117553532123566,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.7698133587837219,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.020231780409812928,
+      "mean_token_accuracy": 0.9920670032501221,
+      "num_tokens": 65439696.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.8110962271690368,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.8562760353088379,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.018769541382789613,
+      "mean_token_accuracy": 0.9925244867801666,
+      "num_tokens": 65628357.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.8113800704479217,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.2076038122177124,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020604749023914338,
+      "mean_token_accuracy": 0.9921375870704651,
+      "num_tokens": 65817139.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.8204582631587982,
+      "epoch": 4.710632570659489,
+      "grad_norm": 0.9150722026824951,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021981674432754516,
+      "mean_token_accuracy": 0.9915678203105927,
+      "num_tokens": 66005911.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.8241343259811401,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.9803491830825806,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02007504999637604,
+      "mean_token_accuracy": 0.992132717370987,
+      "num_tokens": 66194469.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.8122783243656159,
+      "epoch": 4.737550471063257,
+      "grad_norm": 0.967640221118927,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.019903984665870667,
+      "mean_token_accuracy": 0.9924214482307434,
+      "num_tokens": 66383927.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.8145494401454926,
+      "epoch": 4.751009421265142,
+      "grad_norm": 0.7893037796020508,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.019587111473083497,
+      "mean_token_accuracy": 0.9921624064445496,
+      "num_tokens": 66572212.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.8090193390846252,
+      "epoch": 4.764468371467026,
+      "grad_norm": 0.9659109711647034,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.021097327768802642,
+      "mean_token_accuracy": 0.9917473196983337,
+      "num_tokens": 66760377.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.802634209394455,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.0175755023956299,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.02113145440816879,
+      "mean_token_accuracy": 0.991830724477768,
+      "num_tokens": 66949210.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.8030233085155487,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.7780183553695679,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.019016166031360627,
+      "mean_token_accuracy": 0.9926414608955383,
+      "num_tokens": 67138031.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.8019147455692291,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7863438129425049,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.018219922482967377,
+      "mean_token_accuracy": 0.9928844273090363,
+      "num_tokens": 67327268.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7949212431907654,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9292862415313721,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.02136329412460327,
+      "mean_token_accuracy": 0.9918141901493073,
+      "num_tokens": 67515765.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.8002764582633972,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9770768284797668,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.02056926190853119,
+      "mean_token_accuracy": 0.9918984293937683,
+      "num_tokens": 67704975.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.808483463525772,
+      "epoch": 4.845222072678331,
+      "grad_norm": 0.6890367269515991,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020158609747886656,
+      "mean_token_accuracy": 0.9925975620746612,
+      "num_tokens": 67893850.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.8144356846809387,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.8613544702529907,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.019614605605602263,
+      "mean_token_accuracy": 0.9928383588790893,
+      "num_tokens": 68082435.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.8083879172801971,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.7532686591148376,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.019553859531879426,
+      "mean_token_accuracy": 0.9924478828907013,
+      "num_tokens": 68270495.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.8038046479225158,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.3251928091049194,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.02081710398197174,
+      "mean_token_accuracy": 0.9919814527034759,
+      "num_tokens": 68459223.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.8058662831783294,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8125224113464355,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.01858963668346405,
+      "mean_token_accuracy": 0.9922726869583129,
+      "num_tokens": 68647306.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.805506020784378,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8629401922225952,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020777507126331328,
+      "mean_token_accuracy": 0.9918497979640961,
+      "num_tokens": 68836386.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.8137464344501495,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.9837846159934998,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.02024296224117279,
+      "mean_token_accuracy": 0.9921825706958771,
+      "num_tokens": 69025192.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.8139736235141755,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.8701289296150208,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.01932244449853897,
+      "mean_token_accuracy": 0.9926821112632751,
+      "num_tokens": 69213427.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.8088622391223907,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.7041723132133484,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.018706569075584413,
+      "mean_token_accuracy": 0.9927357733249664,
+      "num_tokens": 69401690.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.8001897990703583,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1029852628707886,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.020011243224143983,
+      "mean_token_accuracy": 0.9923900067806244,
+      "num_tokens": 69590659.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.8071568071842193,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.2395373582839966,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.02379731684923172,
+      "mean_token_accuracy": 0.9907991290092468,
+      "num_tokens": 69779864.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.8139646112918854,
+      "epoch": 4.993270524899058,
+      "grad_norm": 0.9219555854797363,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.021635296940803527,
+      "mean_token_accuracy": 0.991401469707489,
+      "num_tokens": 69968639.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.822659221804066,
+      "eval_loss": 0.060467079281806946,
+      "eval_mean_token_accuracy": 0.9792145984188007,
+      "eval_num_tokens": 70063064.0,
+      "eval_runtime": 24.4326,
+      "eval_samples_per_second": 204.645,
+      "eval_steps_per_second": 6.426,
+      "step": 3715
+    },
+    {
+      "entropy": 0.8150339186191559,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8010120391845703,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01673099249601364,
+      "mean_token_accuracy": 0.9935125410556793,
+      "num_tokens": 70157396.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.8029226005077362,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9923233389854431,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01266745626926422,
+      "mean_token_accuracy": 0.9954398214817047,
+      "num_tokens": 70346010.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.789790153503418,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.7180858254432678,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.012603424489498138,
+      "mean_token_accuracy": 0.9949092507362366,
+      "num_tokens": 70534896.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.7863301753997802,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.6865439414978027,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.013341715931892395,
+      "mean_token_accuracy": 0.9951623439788818,
+      "num_tokens": 70723644.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.7841233730316162,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.8624011278152466,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01214413121342659,
+      "mean_token_accuracy": 0.9950926601886749,
+      "num_tokens": 70912490.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.7781193614006042,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.1462656259536743,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.013635106384754181,
+      "mean_token_accuracy": 0.9946888983249664,
+      "num_tokens": 71101186.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.7773554801940918,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.7154580354690552,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.012310966104269027,
+      "mean_token_accuracy": 0.9952048182487487,
+      "num_tokens": 71290177.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.784615957736969,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8214139342308044,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.012708047032356262,
+      "mean_token_accuracy": 0.9956556618213653,
+      "num_tokens": 71478243.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.7814814388751984,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0974094867706299,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.011942464113235473,
+      "mean_token_accuracy": 0.9958715498447418,
+      "num_tokens": 71667417.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.7749826848506928,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.7582180500030518,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.011887111514806748,
+      "mean_token_accuracy": 0.9954704821109772,
+      "num_tokens": 71856322.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.7695000052452088,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.9159554839134216,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.013456562161445617,
+      "mean_token_accuracy": 0.994843590259552,
+      "num_tokens": 72044716.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.7674728155136108,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.0967031717300415,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.012064316868782043,
+      "mean_token_accuracy": 0.9953838646411896,
+      "num_tokens": 72233243.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.766239732503891,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.756477415561676,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012840582430362702,
+      "mean_token_accuracy": 0.9951915740966797,
+      "num_tokens": 72421751.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.7667239785194397,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.8078455924987793,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.012580010294914245,
+      "mean_token_accuracy": 0.9955588281154633,
+      "num_tokens": 72610886.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.7759605884552002,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.9473005533218384,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.01442105770111084,
+      "mean_token_accuracy": 0.9946676790714264,
+      "num_tokens": 72799213.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.7739267945289612,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.4796932637691498,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011830046027898788,
+      "mean_token_accuracy": 0.9954923093318939,
+      "num_tokens": 72988413.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.7756406426429748,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7290598154067993,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.014412418007850647,
+      "mean_token_accuracy": 0.9946037054061889,
+      "num_tokens": 73177134.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.7763995885848999,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.0562312602996826,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.015396332740783692,
+      "mean_token_accuracy": 0.9945933580398559,
+      "num_tokens": 73365622.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.7737024486064911,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1071312427520752,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01451488733291626,
+      "mean_token_accuracy": 0.9945858538150787,
+      "num_tokens": 73554381.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.779985535144806,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.7758586406707764,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.013801637291908263,
+      "mean_token_accuracy": 0.9947169423103333,
+      "num_tokens": 73743028.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.7786811470985413,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.6511496305465698,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012189272791147232,
+      "mean_token_accuracy": 0.9954657673835754,
+      "num_tokens": 73931263.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.7736978232860565,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.8152093291282654,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.013696780800819397,
+      "mean_token_accuracy": 0.9948886811733246,
+      "num_tokens": 74119992.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.7730314910411835,
+      "epoch": 5.302826379542395,
+      "grad_norm": 0.6703830361366272,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012700904905796052,
+      "mean_token_accuracy": 0.9950747072696686,
+      "num_tokens": 74308428.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.7740058362483978,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.9845885634422302,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012894082069396972,
+      "mean_token_accuracy": 0.9953160643577575,
+      "num_tokens": 74497030.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.7738905131816864,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.732532799243927,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.011950771510601043,
+      "mean_token_accuracy": 0.9955996692180633,
+      "num_tokens": 74685631.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.7788872420787811,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7301356196403503,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012340743839740754,
+      "mean_token_accuracy": 0.9952776432037354,
+      "num_tokens": 74874025.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.7771204471588135,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.7570232152938843,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.012937523424625397,
+      "mean_token_accuracy": 0.9949343740940094,
+      "num_tokens": 75062636.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.7747974276542664,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8237525820732117,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.012423938512802124,
+      "mean_token_accuracy": 0.9953977644443512,
+      "num_tokens": 75251339.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.7692180812358856,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.8497490286827087,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012855434417724609,
+      "mean_token_accuracy": 0.9948123812675476,
+      "num_tokens": 75440121.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.773431783914566,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.6758515238761902,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.012878020107746125,
+      "mean_token_accuracy": 0.9955350995063782,
+      "num_tokens": 75629143.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.7735035121440887,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7578167915344238,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.013059911131858826,
+      "mean_token_accuracy": 0.994978415966034,
+      "num_tokens": 75817425.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.7725034832954407,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.0449564456939697,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.013163739442825317,
+      "mean_token_accuracy": 0.9953307807445526,
+      "num_tokens": 76005910.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.7736131310462951,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.881845235824585,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012109111249446868,
+      "mean_token_accuracy": 0.9952919363975525,
+      "num_tokens": 76194639.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.7797342419624329,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.8100120425224304,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.013262654840946197,
+      "mean_token_accuracy": 0.9948459386825561,
+      "num_tokens": 76383240.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.7794113874435424,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.83738112449646,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.013258925080299378,
+      "mean_token_accuracy": 0.9947907269001007,
+      "num_tokens": 76572040.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.7787538826465606,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0567374229431152,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.014641955494880676,
+      "mean_token_accuracy": 0.9945057690143585,
+      "num_tokens": 76761030.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.7894418001174927,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.835543692111969,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.013065731525421143,
+      "mean_token_accuracy": 0.9949834287166596,
+      "num_tokens": 76948469.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.7823423445224762,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.8142625093460083,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.01229657381772995,
+      "mean_token_accuracy": 0.9957204461097717,
+      "num_tokens": 77136968.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.7786863029003144,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.749609649181366,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.01328940987586975,
+      "mean_token_accuracy": 0.9949924945831299,
+      "num_tokens": 77325328.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.7780588805675507,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.787598729133606,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.013933436572551727,
+      "mean_token_accuracy": 0.9949735462665558,
+      "num_tokens": 77513716.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.7779332935810089,
+      "epoch": 5.545087483176312,
+      "grad_norm": 1.0315855741500854,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.012871211767196656,
+      "mean_token_accuracy": 0.995112019777298,
+      "num_tokens": 77701621.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.7754440903663635,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.8191165924072266,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.012908256053924561,
+      "mean_token_accuracy": 0.9950248181819916,
+      "num_tokens": 77889886.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.7811384737491608,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.832539975643158,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.013047756254673004,
+      "mean_token_accuracy": 0.9946948051452636,
+      "num_tokens": 78078640.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.7825228571891785,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.9072359800338745,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.012361775338649749,
+      "mean_token_accuracy": 0.9954402267932891,
+      "num_tokens": 78267842.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.7825945496559144,
+      "epoch": 5.598923283983849,
+      "grad_norm": 0.8300487995147705,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.013246089220046997,
+      "mean_token_accuracy": 0.9951650142669678,
+      "num_tokens": 78455995.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.7857775866985321,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.9343852996826172,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.012612038850784301,
+      "mean_token_accuracy": 0.995340085029602,
+      "num_tokens": 78643617.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.775356650352478,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7109665274620056,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.013411080837249756,
+      "mean_token_accuracy": 0.9951648890972138,
+      "num_tokens": 78833085.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.7750323057174683,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.6436781883239746,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.012064981460571288,
+      "mean_token_accuracy": 0.9954153180122376,
+      "num_tokens": 79021569.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.7753469109535217,
+      "epoch": 5.652759084791386,
+      "grad_norm": 1.0361640453338623,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.013953472673892974,
+      "mean_token_accuracy": 0.9947281837463379,
+      "num_tokens": 79210573.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.7821656167507172,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9078817367553711,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012946712970733642,
+      "mean_token_accuracy": 0.9954073607921601,
+      "num_tokens": 79398723.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.7803877890110016,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.7879899740219116,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.012198399752378464,
+      "mean_token_accuracy": 0.9953389108181,
+      "num_tokens": 79587809.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.7772530496120453,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.0390578508377075,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.012395218014717102,
+      "mean_token_accuracy": 0.9953405141830445,
+      "num_tokens": 79775768.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.7784459292888641,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.7382562160491943,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.012304402142763137,
+      "mean_token_accuracy": 0.9953492462635041,
+      "num_tokens": 79964602.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.7821883440017701,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.5720117092132568,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.011424948275089265,
+      "mean_token_accuracy": 0.9956757485866546,
+      "num_tokens": 80153320.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.7811889886856079,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.7762079238891602,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012460066378116608,
+      "mean_token_accuracy": 0.9950654745101929,
+      "num_tokens": 80341381.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.7771898031234741,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.9219444394111633,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.011629728972911835,
+      "mean_token_accuracy": 0.9955675959587097,
+      "num_tokens": 80530323.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.7825480580329895,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.7189850211143494,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.011180496215820313,
+      "mean_token_accuracy": 0.9960062265396118,
+      "num_tokens": 80718533.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.7795504450798034,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.7256530523300171,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.012952378392219544,
+      "mean_token_accuracy": 0.9951896607875824,
+      "num_tokens": 80907391.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.779454517364502,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.808525800704956,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.0121039479970932,
+      "mean_token_accuracy": 0.9955627739429473,
+      "num_tokens": 81096396.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.7813481986522675,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.7294234037399292,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01356595754623413,
+      "mean_token_accuracy": 0.9951751470565796,
+      "num_tokens": 81284943.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.7833737671375275,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.771756649017334,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.013437020778656005,
+      "mean_token_accuracy": 0.9951942205429077,
+      "num_tokens": 81473810.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.781111603975296,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.5440986752510071,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.012187350541353226,
+      "mean_token_accuracy": 0.9956019043922424,
+      "num_tokens": 81662193.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.7777030646800995,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7532716393470764,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.013766974210739136,
+      "mean_token_accuracy": 0.994896411895752,
+      "num_tokens": 81851045.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.7776691019535065,
+      "epoch": 5.85464333781965,
+      "grad_norm": 1.1882649660110474,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.012588076293468475,
+      "mean_token_accuracy": 0.9952153980731964,
+      "num_tokens": 82039225.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.7795583724975585,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.7453442215919495,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.012100882083177566,
+      "mean_token_accuracy": 0.9954553186893463,
+      "num_tokens": 82228123.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.7763878583908081,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.6757420301437378,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011943653970956803,
+      "mean_token_accuracy": 0.9953660726547241,
+      "num_tokens": 82417093.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.7747587621212005,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.9607213735580444,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.014138388633728027,
+      "mean_token_accuracy": 0.9944287598133087,
+      "num_tokens": 82605461.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.7782076597213745,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.6733895540237427,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.012490443140268325,
+      "mean_token_accuracy": 0.9950622916221619,
+      "num_tokens": 82794124.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.7843947112560272,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.7626368403434753,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.012137357890605927,
+      "mean_token_accuracy": 0.9952504158020019,
+      "num_tokens": 82982258.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.7826779782772064,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8149229288101196,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012872529029846192,
+      "mean_token_accuracy": 0.9949809432029724,
+      "num_tokens": 83170520.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.7792184889316559,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.7774028778076172,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012353222817182541,
+      "mean_token_accuracy": 0.9953234374523163,
+      "num_tokens": 83359174.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.7778255641460419,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7626445293426514,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.01137089878320694,
+      "mean_token_accuracy": 0.9958219468593598,
+      "num_tokens": 83547583.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.7765573263168335,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9130274653434753,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012439383566379547,
+      "mean_token_accuracy": 0.9954963147640228,
+      "num_tokens": 83735904.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.7734197914600373,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.7289780378341675,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011430246382951736,
+      "mean_token_accuracy": 0.9957478463649749,
+      "num_tokens": 83924971.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.7852135965019275,
+      "eval_loss": 0.06990383565425873,
+      "eval_mean_token_accuracy": 0.9796095274056599,
+      "eval_num_tokens": 84075763.0,
+      "eval_runtime": 24.3947,
+      "eval_samples_per_second": 204.963,
+      "eval_steps_per_second": 6.436,
+      "step": 4458
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.0625134721711473e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-4458/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-4458/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-4458/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5201/README.md b/baby_talk_L16_a50/seed_42/checkpoint-5201/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5201/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5201/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-5201/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5201/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5201/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-5201/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d5f2c5ca7795647c78ebe9728920bb4a842af700
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5201/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88d555221f6e593a18115c5e50a575aea4bc5817d56745baecbe8d07df0e6cea
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5201/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-5201/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5201/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5201/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-5201/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5201/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5201/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-5201/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5201/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5201/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-5201/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..08e00eaa74053d15bd7c0edc0841f421929b723b
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5201/trainer_state.json
@@ -0,0 +1,5311 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 5201,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    },
+    {
+      "entropy": 0.9689933776855468,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9720287919044495,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.055563896894454956,
+      "mean_token_accuracy": 0.9779708027839661,
+      "num_tokens": 14144652.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.9506655633449554,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 0.6619595289230347,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.050587379932403566,
+      "mean_token_accuracy": 0.979685264825821,
+      "num_tokens": 14332714.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.9458732306957245,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 0.7905108332633972,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.05069155097007751,
+      "mean_token_accuracy": 0.9796596229076385,
+      "num_tokens": 14520902.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.9320916712284089,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 0.9210566878318787,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.051744121313095096,
+      "mean_token_accuracy": 0.9792148530483246,
+      "num_tokens": 14709215.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.928934782743454,
+      "epoch": 1.063257065948856,
+      "grad_norm": 0.7442036867141724,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.0486875057220459,
+      "mean_token_accuracy": 0.9798122465610504,
+      "num_tokens": 14898242.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.942881977558136,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.0910941362380981,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.052297019958496095,
+      "mean_token_accuracy": 0.9793200254440307,
+      "num_tokens": 15086835.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.9503917813301086,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 0.8091554641723633,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.049927744269371035,
+      "mean_token_accuracy": 0.9802082359790802,
+      "num_tokens": 15275042.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9426830470561981,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.0589245557785034,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.053803551197052005,
+      "mean_token_accuracy": 0.9785399675369263,
+      "num_tokens": 15463923.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.9445010781288147,
+      "epoch": 1.117092866756393,
+      "grad_norm": 0.8410441279411316,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.053181976079940796,
+      "mean_token_accuracy": 0.9792523324489594,
+      "num_tokens": 15652526.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.946729838848114,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 0.8246080279350281,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.054291915893554685,
+      "mean_token_accuracy": 0.9791357696056366,
+      "num_tokens": 15841340.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.9433728814125061,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 0.9200296401977539,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.05073915719985962,
+      "mean_token_accuracy": 0.9796931505203247,
+      "num_tokens": 16029478.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.9494417488574982,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 0.8342758417129517,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.050929927825927736,
+      "mean_token_accuracy": 0.9793219089508056,
+      "num_tokens": 16218014.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.9409256100654602,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.7911009192466736,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.049964362382888795,
+      "mean_token_accuracy": 0.9798487305641175,
+      "num_tokens": 16406420.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.9429958581924438,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.4877018928527832,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.04969423711299896,
+      "mean_token_accuracy": 0.9801322638988494,
+      "num_tokens": 16594948.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.9542136013507843,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 0.8030785918235779,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.04981146454811096,
+      "mean_token_accuracy": 0.9796162784099579,
+      "num_tokens": 16782983.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.9367722630500793,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.1169800758361816,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.05210963487625122,
+      "mean_token_accuracy": 0.9790139615535736,
+      "num_tokens": 16971394.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.9487208306789399,
+      "epoch": 1.224764468371467,
+      "grad_norm": 0.8650282621383667,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.05012243390083313,
+      "mean_token_accuracy": 0.9805107891559601,
+      "num_tokens": 17159951.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.9441231489181519,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 0.8193967938423157,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.048866665363311766,
+      "mean_token_accuracy": 0.9807584881782532,
+      "num_tokens": 17348711.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.9361630439758301,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3613612651824951,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.05324091911315918,
+      "mean_token_accuracy": 0.97904953956604,
+      "num_tokens": 17537116.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.9562997639179229,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.8121878504753113,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.05011019706726074,
+      "mean_token_accuracy": 0.9798731088638306,
+      "num_tokens": 17725567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.9359076261520386,
+      "epoch": 1.278600269179004,
+      "grad_norm": 0.7838549613952637,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.04834386110305786,
+      "mean_token_accuracy": 0.9804529249668121,
+      "num_tokens": 17914191.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.9527446150779724,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0981712341308594,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.055077624320983884,
+      "mean_token_accuracy": 0.9776535391807556,
+      "num_tokens": 18103184.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.9574812948703766,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 0.8291501402854919,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.050153911113739014,
+      "mean_token_accuracy": 0.9792965054512024,
+      "num_tokens": 18291351.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.939585280418396,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.0864412784576416,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.05132197737693787,
+      "mean_token_accuracy": 0.9792291462421417,
+      "num_tokens": 18480192.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.9468406856060028,
+      "epoch": 1.332436069986541,
+      "grad_norm": 0.9772785902023315,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.04844954013824463,
+      "mean_token_accuracy": 0.9804142415523529,
+      "num_tokens": 18669028.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.9413834273815155,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 0.9020094871520996,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.05075312852859497,
+      "mean_token_accuracy": 0.9790591120719909,
+      "num_tokens": 18857581.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.9526253461837768,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.0524990558624268,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.049737372994422914,
+      "mean_token_accuracy": 0.9795652389526367,
+      "num_tokens": 19046105.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.9484909176826477,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.0715620517730713,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.04935494959354401,
+      "mean_token_accuracy": 0.9797740161418915,
+      "num_tokens": 19234730.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.9398573458194732,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2134987115859985,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.0518394410610199,
+      "mean_token_accuracy": 0.9793788075447083,
+      "num_tokens": 19423572.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.9465648651123046,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.004693627357483,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.04776117205619812,
+      "mean_token_accuracy": 0.9813261866569519,
+      "num_tokens": 19612465.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.9448004186153411,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.2913438081741333,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.04807930588722229,
+      "mean_token_accuracy": 0.9805689513683319,
+      "num_tokens": 19800482.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.9425322234630584,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1068469285964966,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.05053595900535583,
+      "mean_token_accuracy": 0.9794536828994751,
+      "num_tokens": 19989444.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.9472232103347779,
+      "epoch": 1.440107671601615,
+      "grad_norm": 0.7409384250640869,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.052136778831481934,
+      "mean_token_accuracy": 0.979197359085083,
+      "num_tokens": 20178482.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9450975477695465,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.3440778255462646,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.05258495211601257,
+      "mean_token_accuracy": 0.9798152863979339,
+      "num_tokens": 20367345.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9468007743358612,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 0.8377829194068909,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.04893603026866913,
+      "mean_token_accuracy": 0.9805733859539032,
+      "num_tokens": 20555817.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.9505827724933624,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.776889979839325,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.05367119312286377,
+      "mean_token_accuracy": 0.9779646098613739,
+      "num_tokens": 20744001.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.9516380190849304,
+      "epoch": 1.493943472409152,
+      "grad_norm": 0.8598276376724243,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.05071394443511963,
+      "mean_token_accuracy": 0.979692417383194,
+      "num_tokens": 20931715.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.9351628720760345,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3307439088821411,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.051372635364532473,
+      "mean_token_accuracy": 0.9797907948493958,
+      "num_tokens": 21120803.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.9268040716648102,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9320250749588013,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.04939974546432495,
+      "mean_token_accuracy": 0.98038170337677,
+      "num_tokens": 21309384.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9241463243961334,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.065938115119934,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.04856643378734589,
+      "mean_token_accuracy": 0.9800511121749877,
+      "num_tokens": 21498234.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.9350252687931061,
+      "epoch": 1.547779273216689,
+      "grad_norm": 0.6925032138824463,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.0512526273727417,
+      "mean_token_accuracy": 0.979562646150589,
+      "num_tokens": 21687119.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.9403733670711517,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.2069101333618164,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.049795085191726686,
+      "mean_token_accuracy": 0.9806506633758545,
+      "num_tokens": 21876094.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9370538294315338,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.05461466312408447,
+      "mean_token_accuracy": 0.9776057600975037,
+      "num_tokens": 22064836.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9517778217792511,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 0.8965149521827698,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.05289326906204224,
+      "mean_token_accuracy": 0.9782088756561279,
+      "num_tokens": 22253065.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.9372851848602295,
+      "epoch": 1.601615074024226,
+      "grad_norm": 0.7192108631134033,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.048973691463470456,
+      "mean_token_accuracy": 0.9806210815906524,
+      "num_tokens": 22441206.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.9122351944446564,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 0.8763614296913147,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.0461783230304718,
+      "mean_token_accuracy": 0.9817197680473327,
+      "num_tokens": 22629637.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.9162654876708984,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 0.9806778430938721,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.04827338755130768,
+      "mean_token_accuracy": 0.9806499361991883,
+      "num_tokens": 22818369.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.9125026524066925,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 0.9685350656509399,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.048379385471343996,
+      "mean_token_accuracy": 0.980905145406723,
+      "num_tokens": 23006863.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9119791567325592,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.8717457056045532,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.050873959064483644,
+      "mean_token_accuracy": 0.979601925611496,
+      "num_tokens": 23195735.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.9244007229804992,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.0574654340744019,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.05112813711166382,
+      "mean_token_accuracy": 0.979493772983551,
+      "num_tokens": 23383862.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9225810945034028,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 0.9628807306289673,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.050480544567108154,
+      "mean_token_accuracy": 0.9797678411006927,
+      "num_tokens": 23572292.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.9287579476833343,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.015265941619873,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.05133126378059387,
+      "mean_token_accuracy": 0.9797986805438995,
+      "num_tokens": 23760252.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.9133845925331116,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 0.8996883630752563,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.05044950246810913,
+      "mean_token_accuracy": 0.9795303404331207,
+      "num_tokens": 23949031.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.9213698863983154,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.0490264892578125,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.04797698855400086,
+      "mean_token_accuracy": 0.9806023716926575,
+      "num_tokens": 24137829.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.9173697710037232,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.3648769855499268,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.049194514751434326,
+      "mean_token_accuracy": 0.980712479352951,
+      "num_tokens": 24326586.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9242741882801055,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.105125069618225,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.048667973279953,
+      "mean_token_accuracy": 0.9801449298858642,
+      "num_tokens": 24514904.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.9162627995014191,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.8887938857078552,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.04878672957420349,
+      "mean_token_accuracy": 0.980469423532486,
+      "num_tokens": 24703688.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.9149538338184356,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 0.8838976621627808,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.04892318844795227,
+      "mean_token_accuracy": 0.9808776795864105,
+      "num_tokens": 24892107.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.914735221862793,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1152070760726929,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.04825109839439392,
+      "mean_token_accuracy": 0.9804604113101959,
+      "num_tokens": 25080130.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9170725226402283,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9479517340660095,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.051445144414901736,
+      "mean_token_accuracy": 0.9792819261550904,
+      "num_tokens": 25268652.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9142911911010743,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 0.8172292709350586,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.0475692093372345,
+      "mean_token_accuracy": 0.9809505581855774,
+      "num_tokens": 25457120.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.9019249439239502,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 0.8183121681213379,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.04979957342147827,
+      "mean_token_accuracy": 0.9799730658531189,
+      "num_tokens": 25645699.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.9181066155433655,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.8256579041481018,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.04996164441108704,
+      "mean_token_accuracy": 0.9802247405052185,
+      "num_tokens": 25834183.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9171363770961761,
+      "epoch": 1.857335127860027,
+      "grad_norm": 0.9727098345756531,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.048660767078399655,
+      "mean_token_accuracy": 0.9805064260959625,
+      "num_tokens": 26023317.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.9112720847129822,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 0.9826673865318298,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.04775593280792236,
+      "mean_token_accuracy": 0.9809929549694061,
+      "num_tokens": 26212115.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9034272134304047,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.9951306581497192,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.04939360618591308,
+      "mean_token_accuracy": 0.9800347864627839,
+      "num_tokens": 26401167.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.897865754365921,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.0163404941558838,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05002856254577637,
+      "mean_token_accuracy": 0.9793690323829651,
+      "num_tokens": 26590570.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9004576802253723,
+      "epoch": 1.911170928667564,
+      "grad_norm": 0.6666616201400757,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.04905453026294708,
+      "mean_token_accuracy": 0.9803856253623963,
+      "num_tokens": 26779481.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.8978902101516724,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.9295416474342346,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.05261261463165283,
+      "mean_token_accuracy": 0.9785708487033844,
+      "num_tokens": 26968262.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.8962675571441651,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 0.88262939453125,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.04765265882015228,
+      "mean_token_accuracy": 0.98052077293396,
+      "num_tokens": 27157631.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8909463047981262,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.4205769300460815,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.049406200647354126,
+      "mean_token_accuracy": 0.979697072505951,
+      "num_tokens": 27345921.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8969364166259766,
+      "epoch": 1.965006729475101,
+      "grad_norm": 0.9517145752906799,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0510346531867981,
+      "mean_token_accuracy": 0.9793386399745941,
+      "num_tokens": 27534751.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.9092479169368743,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.200056791305542,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.04974203705787659,
+      "mean_token_accuracy": 0.9800146698951722,
+      "num_tokens": 27723763.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9059641897678375,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.072447657585144,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.05065792202949524,
+      "mean_token_accuracy": 0.9799223959445953,
+      "num_tokens": 27911784.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9083398975384464,
+      "eval_loss": 0.05432562157511711,
+      "eval_mean_token_accuracy": 0.9782073607869969,
+      "eval_num_tokens": 28025286.0,
+      "eval_runtime": 24.4304,
+      "eval_samples_per_second": 204.663,
+      "eval_steps_per_second": 6.426,
+      "step": 1486
+    },
+    {
+      "entropy": 0.9113237977027893,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.0167063474655151,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.044942992925643924,
+      "mean_token_accuracy": 0.9818171083927154,
+      "num_tokens": 28100906.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.8780008792877197,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.9552241563796997,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.036248764395713805,
+      "mean_token_accuracy": 0.9852245450019836,
+      "num_tokens": 28289445.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.8721176147460937,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 0.8951388001441956,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.037987279891967776,
+      "mean_token_accuracy": 0.9850614190101623,
+      "num_tokens": 28477972.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.8922657191753387,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 0.9028105735778809,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.03588914275169373,
+      "mean_token_accuracy": 0.9863102614879609,
+      "num_tokens": 28666272.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.877799642086029,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.153886318206787,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.03577309250831604,
+      "mean_token_accuracy": 0.9863148391246795,
+      "num_tokens": 28854780.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.8708971381187439,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.042955994606018,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.038266432285308835,
+      "mean_token_accuracy": 0.9848419308662415,
+      "num_tokens": 29043741.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.8800721943378449,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 0.6789060235023499,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.032948991656303404,
+      "mean_token_accuracy": 0.9871293902397156,
+      "num_tokens": 29232719.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.8743580460548401,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.1780879497528076,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.036874374747276305,
+      "mean_token_accuracy": 0.9859182178974152,
+      "num_tokens": 29421817.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.8851303637027741,
+      "epoch": 2.113055181695828,
+      "grad_norm": 0.747734785079956,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.0364631175994873,
+      "mean_token_accuracy": 0.9862085223197937,
+      "num_tokens": 29610344.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.8798732101917267,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.5001860857009888,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.03813003897666931,
+      "mean_token_accuracy": 0.9852604746818543,
+      "num_tokens": 29798997.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.8797551989555359,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 0.9593478441238403,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.0394927829504013,
+      "mean_token_accuracy": 0.984604275226593,
+      "num_tokens": 29987809.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.8860546290874481,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.8347703218460083,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.036699697375297546,
+      "mean_token_accuracy": 0.985380882024765,
+      "num_tokens": 30176234.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.8783667802810669,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1154481172561646,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.03756999969482422,
+      "mean_token_accuracy": 0.9853451430797577,
+      "num_tokens": 30364657.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.8684478521347045,
+      "epoch": 2.180349932705249,
+      "grad_norm": 0.7515475153923035,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.0382376104593277,
+      "mean_token_accuracy": 0.9854029655456543,
+      "num_tokens": 30552904.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.8636964917182922,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 0.7711939215660095,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.036220991611480714,
+      "mean_token_accuracy": 0.9860713243484497,
+      "num_tokens": 30741259.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.8592016279697419,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.1143887042999268,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.036316031217575075,
+      "mean_token_accuracy": 0.985757052898407,
+      "num_tokens": 30930144.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.8671079576015472,
+      "epoch": 2.220726783310902,
+      "grad_norm": 0.8980015516281128,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.03822658061981201,
+      "mean_token_accuracy": 0.9847764372825623,
+      "num_tokens": 31118624.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.8785548269748688,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.094323992729187,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.03736622333526611,
+      "mean_token_accuracy": 0.9850081980228425,
+      "num_tokens": 31306833.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.866977310180664,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8372092843055725,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.0401554524898529,
+      "mean_token_accuracy": 0.9840337932109833,
+      "num_tokens": 31495503.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.8787918210029602,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.5408164262771606,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0368030846118927,
+      "mean_token_accuracy": 0.9847234487533569,
+      "num_tokens": 31683865.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.872721153497696,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.1115421056747437,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.037859299778938295,
+      "mean_token_accuracy": 0.9851646661758423,
+      "num_tokens": 31871903.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.8774094223976135,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.0301331281661987,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.038671016693115234,
+      "mean_token_accuracy": 0.9840058028697968,
+      "num_tokens": 32060500.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.889606237411499,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8601903915405273,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.039603835344314574,
+      "mean_token_accuracy": 0.9840235590934754,
+      "num_tokens": 32248744.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.886734277009964,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.1029839515686035,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.039513933658599856,
+      "mean_token_accuracy": 0.9844573020935059,
+      "num_tokens": 32437640.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.8843017637729644,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.2545154094696045,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.036970189213752745,
+      "mean_token_accuracy": 0.985239815711975,
+      "num_tokens": 32625986.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.8913422048091888,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.9613803625106812,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.03811657428741455,
+      "mean_token_accuracy": 0.9852804243564606,
+      "num_tokens": 32814531.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.8907467782497406,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.1811398267745972,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.03903660774230957,
+      "mean_token_accuracy": 0.9842524945735931,
+      "num_tokens": 33002960.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.8833664715290069,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.8338477611541748,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.03982087969779968,
+      "mean_token_accuracy": 0.9841201484203339,
+      "num_tokens": 33191183.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.8783826470375061,
+      "epoch": 2.382234185733513,
+      "grad_norm": 0.6904510259628296,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.03699290752410889,
+      "mean_token_accuracy": 0.9854120731353759,
+      "num_tokens": 33379798.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.8871048510074615,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.009539246559143,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.03926805555820465,
+      "mean_token_accuracy": 0.9843246698379516,
+      "num_tokens": 33568531.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.8973462700843811,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.8916401267051697,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.039596831798553465,
+      "mean_token_accuracy": 0.9839228212833404,
+      "num_tokens": 33756333.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.8954619467258453,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.056504487991333,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0396859347820282,
+      "mean_token_accuracy": 0.9837161123752594,
+      "num_tokens": 33944904.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.8952643811702728,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0442084074020386,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.037324142456054685,
+      "mean_token_accuracy": 0.9852897703647614,
+      "num_tokens": 34133530.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.8874911010265351,
+      "epoch": 2.449528936742934,
+      "grad_norm": 0.9926224946975708,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.03637495338916778,
+      "mean_token_accuracy": 0.9853426337242126,
+      "num_tokens": 34322042.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.8900584518909455,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.001734733581543,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04029585719108582,
+      "mean_token_accuracy": 0.9842372059822082,
+      "num_tokens": 34510259.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.8986345052719116,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 0.772941529750824,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.0401084691286087,
+      "mean_token_accuracy": 0.9841427087783814,
+      "num_tokens": 34698669.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.8911370873451233,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.895031213760376,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.03941032290458679,
+      "mean_token_accuracy": 0.9841966688632965,
+      "num_tokens": 34887160.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.8901963472366333,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0677918195724487,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.03503885865211487,
+      "mean_token_accuracy": 0.9856241464614868,
+      "num_tokens": 35075891.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.8817630231380462,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.1200164556503296,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.03878425657749176,
+      "mean_token_accuracy": 0.9846492648124695,
+      "num_tokens": 35264455.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.8897477686405182,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.8058190941810608,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.041558006405830385,
+      "mean_token_accuracy": 0.9830883860588073,
+      "num_tokens": 35453251.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.8873181045055389,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.7765992283821106,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.03555050492286682,
+      "mean_token_accuracy": 0.9858225345611572,
+      "num_tokens": 35641959.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.8782606959342957,
+      "epoch": 2.557200538358008,
+      "grad_norm": 0.8488866686820984,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.03606923818588257,
+      "mean_token_accuracy": 0.9858632445335388,
+      "num_tokens": 35829636.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.8869808673858642,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.1734739542007446,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.040956351161003116,
+      "mean_token_accuracy": 0.9835374176502227,
+      "num_tokens": 36017818.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.8850924432277679,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.3584387302398682,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.042437011003494264,
+      "mean_token_accuracy": 0.9826479077339172,
+      "num_tokens": 36206505.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.8801687896251679,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2862240076065063,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.03880060315132141,
+      "mean_token_accuracy": 0.9843404710292816,
+      "num_tokens": 36395802.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.8752781748771667,
+      "epoch": 2.611036339165545,
+      "grad_norm": 0.7341249585151672,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.03776344656944275,
+      "mean_token_accuracy": 0.9852877616882324,
+      "num_tokens": 36584370.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.8759494423866272,
+      "epoch": 2.624495289367429,
+      "grad_norm": 0.9807276129722595,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.03780297338962555,
+      "mean_token_accuracy": 0.9853227376937866,
+      "num_tokens": 36772865.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.8753129065036773,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.1391628980636597,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.03953765034675598,
+      "mean_token_accuracy": 0.9839197635650635,
+      "num_tokens": 36961462.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.8771899223327637,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.0865391492843628,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.03856399655342102,
+      "mean_token_accuracy": 0.9848688066005706,
+      "num_tokens": 37150486.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.8774881184101104,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.8561595678329468,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.03795175850391388,
+      "mean_token_accuracy": 0.9849309325218201,
+      "num_tokens": 37338890.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.8841517448425293,
+      "epoch": 2.678331090174966,
+      "grad_norm": 0.9458399415016174,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.03860213160514832,
+      "mean_token_accuracy": 0.9842230796813964,
+      "num_tokens": 37526886.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.8809043228626251,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 0.9360871315002441,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.03709094822406769,
+      "mean_token_accuracy": 0.9855118036270142,
+      "num_tokens": 37715408.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.8805519282817841,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.8852124214172363,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.03869341611862183,
+      "mean_token_accuracy": 0.9846724212169647,
+      "num_tokens": 37904328.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.8796228408813477,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.189297080039978,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.03675769567489624,
+      "mean_token_accuracy": 0.9854080140590668,
+      "num_tokens": 38092776.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.8722020089626312,
+      "epoch": 2.732166890982503,
+      "grad_norm": 0.7027947306632996,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.03889244794845581,
+      "mean_token_accuracy": 0.984558242559433,
+      "num_tokens": 38282221.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.8746632814407349,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 0.9799376726150513,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.03815680146217346,
+      "mean_token_accuracy": 0.9845012128353119,
+      "num_tokens": 38470633.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.8766567528247833,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 0.850459098815918,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.039029371738433835,
+      "mean_token_accuracy": 0.9845906794071198,
+      "num_tokens": 38659048.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.8778688013553619,
+      "epoch": 2.772543741588156,
+      "grad_norm": 0.8540134429931641,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.039645448327064514,
+      "mean_token_accuracy": 0.9840566098690033,
+      "num_tokens": 38847370.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.8758600771427154,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8036021590232849,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.038226932287216187,
+      "mean_token_accuracy": 0.9845647215843201,
+      "num_tokens": 39036040.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.8652268946170807,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 0.8526738286018372,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.03798363208770752,
+      "mean_token_accuracy": 0.9845202267169952,
+      "num_tokens": 39224695.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.8707200348377228,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 0.9852389097213745,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.037227436900138855,
+      "mean_token_accuracy": 0.9852050960063934,
+      "num_tokens": 39413360.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.8711447060108185,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.228277325630188,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.03698050379753113,
+      "mean_token_accuracy": 0.9852443158626556,
+      "num_tokens": 39602544.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.8742912471294403,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 0.8440544605255127,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.03757670521736145,
+      "mean_token_accuracy": 0.9848618268966675,
+      "num_tokens": 39791340.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.8801726162433624,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0955252647399902,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.040004149079322815,
+      "mean_token_accuracy": 0.9841345012187958,
+      "num_tokens": 39980212.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.8931492984294891,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.5722652673721313,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.038096648454666135,
+      "mean_token_accuracy": 0.9849136590957641,
+      "num_tokens": 40169122.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.8767493844032288,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.7663730382919312,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.03729096055030823,
+      "mean_token_accuracy": 0.985169780254364,
+      "num_tokens": 40357678.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.8770561516284943,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.1505672931671143,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.03984796404838562,
+      "mean_token_accuracy": 0.9838845670223236,
+      "num_tokens": 40546892.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.8789129912853241,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 0.8355761766433716,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.03769223690032959,
+      "mean_token_accuracy": 0.9847559213638306,
+      "num_tokens": 40736109.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.8781779289245606,
+      "epoch": 2.920592193808883,
+      "grad_norm": 0.9103918075561523,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.038262826204299924,
+      "mean_token_accuracy": 0.984741848707199,
+      "num_tokens": 40924341.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.8552807629108429,
+      "epoch": 2.934051144010767,
+      "grad_norm": 0.8338510990142822,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.0355743408203125,
+      "mean_token_accuracy": 0.9855036079883576,
+      "num_tokens": 41113849.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.8705591142177582,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 0.8334057927131653,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.037476515769958495,
+      "mean_token_accuracy": 0.9854551255702972,
+      "num_tokens": 41303130.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.8809274792671203,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 0.9424103498458862,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.03624279499053955,
+      "mean_token_accuracy": 0.9856135487556458,
+      "num_tokens": 41491750.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.8818133771419525,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.04243004322052,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.039132320880889894,
+      "mean_token_accuracy": 0.984746390581131,
+      "num_tokens": 41680493.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.8838395297527313,
+      "epoch": 2.987886944818304,
+      "grad_norm": 0.8650784492492676,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.03822052478790283,
+      "mean_token_accuracy": 0.9852108955383301,
+      "num_tokens": 41868378.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.8800126549544608,
+      "eval_loss": 0.05468416586518288,
+      "eval_mean_token_accuracy": 0.9781015617832257,
+      "eval_num_tokens": 42038037.0,
+      "eval_runtime": 24.4825,
+      "eval_samples_per_second": 204.228,
+      "eval_steps_per_second": 6.413,
+      "step": 2229
+    },
+    {
+      "entropy": 0.8773060619831086,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.7845072150230408,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.0365300714969635,
+      "mean_token_accuracy": 0.985853761434555,
+      "num_tokens": 42056717.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.8720550239086151,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 0.9917275905609131,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.030071893334388734,
+      "mean_token_accuracy": 0.9890478909015655,
+      "num_tokens": 42245058.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.8489451467990875,
+      "epoch": 3.028263795423957,
+      "grad_norm": 0.9414294362068176,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.027054613828659056,
+      "mean_token_accuracy": 0.9894806027412415,
+      "num_tokens": 42433468.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.848755830526352,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.7276235818862915,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.026897600293159483,
+      "mean_token_accuracy": 0.9893308103084564,
+      "num_tokens": 42621657.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.8402929604053497,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 0.9455838799476624,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.027918827533721925,
+      "mean_token_accuracy": 0.9888385117053986,
+      "num_tokens": 42810533.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.8434606611728668,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 0.7503429651260376,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.025769710540771484,
+      "mean_token_accuracy": 0.989881956577301,
+      "num_tokens": 42999085.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.8459890782833099,
+      "epoch": 3.082099596231494,
+      "grad_norm": 0.736440896987915,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.028781753778457642,
+      "mean_token_accuracy": 0.9882656812667847,
+      "num_tokens": 43188222.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.8493306159973144,
+      "epoch": 3.095558546433378,
+      "grad_norm": 0.9577816128730774,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.02689719796180725,
+      "mean_token_accuracy": 0.9889744818210602,
+      "num_tokens": 43376648.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.8512665331363678,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 0.9631588459014893,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.031439167261123654,
+      "mean_token_accuracy": 0.9874942421913147,
+      "num_tokens": 43564769.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.8519513130187988,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1011922359466553,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.02696958780288696,
+      "mean_token_accuracy": 0.9900951981544495,
+      "num_tokens": 43753362.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.8507618844509125,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9351623058319092,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.026315575838088988,
+      "mean_token_accuracy": 0.989511913061142,
+      "num_tokens": 43942063.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.8506478369235992,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.096564769744873,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.029049152135849,
+      "mean_token_accuracy": 0.9886228442192078,
+      "num_tokens": 44130381.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.8624941289424897,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 0.7859402298927307,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.027744990587234498,
+      "mean_token_accuracy": 0.9890527307987214,
+      "num_tokens": 44319243.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.8627866744995117,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 0.7764605283737183,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.02571650445461273,
+      "mean_token_accuracy": 0.9903442323207855,
+      "num_tokens": 44507698.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.8566473722457886,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.8182396292686462,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.025625643134117127,
+      "mean_token_accuracy": 0.9897150516510009,
+      "num_tokens": 44695954.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.8555088877677918,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.0277098417282104,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.028235083818435668,
+      "mean_token_accuracy": 0.9894341349601745,
+      "num_tokens": 44884558.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.8562702238559723,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.0676493644714355,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.027713948488235475,
+      "mean_token_accuracy": 0.9891053259372711,
+      "num_tokens": 45073381.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.8520576059818268,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 0.7926117181777954,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.028936928510665892,
+      "mean_token_accuracy": 0.988807988166809,
+      "num_tokens": 45262350.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.860496312379837,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.1108510494232178,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030533093214035033,
+      "mean_token_accuracy": 0.9882748067378998,
+      "num_tokens": 45450782.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.872324675321579,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.3057194948196411,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.032421305775642395,
+      "mean_token_accuracy": 0.9879753351211548,
+      "num_tokens": 45639122.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.8778823137283325,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.289680004119873,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03096393346786499,
+      "mean_token_accuracy": 0.9883370697498322,
+      "num_tokens": 45827730.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.8627406179904937,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.8361398577690125,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.02957139313220978,
+      "mean_token_accuracy": 0.9884674072265625,
+      "num_tokens": 46016274.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.8584554970264435,
+      "epoch": 3.297442799461642,
+      "grad_norm": 0.9330723881721497,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.029568272829055785,
+      "mean_token_accuracy": 0.9882686555385589,
+      "num_tokens": 46204679.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.8515178024768829,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.0296216011047363,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.027788180112838744,
+      "mean_token_accuracy": 0.9888734877109527,
+      "num_tokens": 46393989.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.8488850116729736,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1329699754714966,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.028057461977005003,
+      "mean_token_accuracy": 0.9895590543746948,
+      "num_tokens": 46582790.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.849625188112259,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 0.9267219305038452,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.02802973985671997,
+      "mean_token_accuracy": 0.988851535320282,
+      "num_tokens": 46771138.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.854446280002594,
+      "epoch": 3.351278600269179,
+      "grad_norm": 0.8713932037353516,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.030220019817352294,
+      "mean_token_accuracy": 0.9882843673229218,
+      "num_tokens": 46960054.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.8599806547164917,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.675019383430481,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.02958865463733673,
+      "mean_token_accuracy": 0.9880296885967255,
+      "num_tokens": 47149567.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.8549504458904267,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.2176353931427002,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.027722162008285523,
+      "mean_token_accuracy": 0.9890090465545655,
+      "num_tokens": 47338060.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.8552233815193176,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9571717977523804,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.028924965858459474,
+      "mean_token_accuracy": 0.9884298503398895,
+      "num_tokens": 47525895.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.8521630108356476,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.965713620185852,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.030511701107025148,
+      "mean_token_accuracy": 0.9881398677825928,
+      "num_tokens": 47714524.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.8537400126457214,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 0.9005307555198669,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.030432754755020143,
+      "mean_token_accuracy": 0.9876708447933197,
+      "num_tokens": 47903188.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.8596058666706086,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 0.8159075379371643,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.027927806973457335,
+      "mean_token_accuracy": 0.9889210939407349,
+      "num_tokens": 48091334.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.8482735753059387,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0411661863327026,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03003154993057251,
+      "mean_token_accuracy": 0.9882730603218078,
+      "num_tokens": 48279622.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.8466426789760589,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.189595341682434,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03237656056880951,
+      "mean_token_accuracy": 0.9869072616100312,
+      "num_tokens": 48467527.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.8545588493347168,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0700056552886963,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.0306060791015625,
+      "mean_token_accuracy": 0.9880664825439454,
+      "num_tokens": 48656151.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.8503433585166931,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.7099478244781494,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.02984013557434082,
+      "mean_token_accuracy": 0.9879442811012268,
+      "num_tokens": 48844796.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.8464376091957092,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9877128005027771,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.0306446373462677,
+      "mean_token_accuracy": 0.9883591532707214,
+      "num_tokens": 49033216.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.8477609276771545,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.3324531316757202,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.030148410797119142,
+      "mean_token_accuracy": 0.9879989802837372,
+      "num_tokens": 49221425.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.8461863219738006,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8938570022583008,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.02682304084300995,
+      "mean_token_accuracy": 0.9897881150245667,
+      "num_tokens": 49409785.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.8362689018249512,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 0.7423873543739319,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.02689332664012909,
+      "mean_token_accuracy": 0.9899903059005737,
+      "num_tokens": 49598524.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.8395139217376709,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.921755850315094,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.030062124133110046,
+      "mean_token_accuracy": 0.9880835592746735,
+      "num_tokens": 49787241.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.8467226803302765,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0296459197998047,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.030666223168373107,
+      "mean_token_accuracy": 0.9877098739147187,
+      "num_tokens": 49975844.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.833392071723938,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2861948013305664,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.028174835443496703,
+      "mean_token_accuracy": 0.989056545495987,
+      "num_tokens": 50164857.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.8340730309486389,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 0.8116776347160339,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.028390336036682128,
+      "mean_token_accuracy": 0.9886897742748261,
+      "num_tokens": 50353452.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.8357116639614105,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.8362084031105042,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.029806244373321533,
+      "mean_token_accuracy": 0.988309508562088,
+      "num_tokens": 50542461.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.8363270878791809,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8438499569892883,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.028577303886413573,
+      "mean_token_accuracy": 0.9886945366859436,
+      "num_tokens": 50730872.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.8310261785984039,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.1920002698898315,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.030321845412254335,
+      "mean_token_accuracy": 0.9878708899021149,
+      "num_tokens": 50919889.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.8380252480506897,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 0.9409399032592773,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.030501589179039,
+      "mean_token_accuracy": 0.9872083842754364,
+      "num_tokens": 51108079.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.8422612130641938,
+      "epoch": 3.660834454912517,
+      "grad_norm": 0.8370440006256104,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.02974823713302612,
+      "mean_token_accuracy": 0.9885554671287536,
+      "num_tokens": 51296913.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.8387595832347869,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1360642910003662,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.026970452070236205,
+      "mean_token_accuracy": 0.9888331353664398,
+      "num_tokens": 51485678.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.8403031289577484,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.1195677518844604,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.029451662302017213,
+      "mean_token_accuracy": 0.988847154378891,
+      "num_tokens": 51674765.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.8499106168746948,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 0.9573526978492737,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.030826866626739502,
+      "mean_token_accuracy": 0.9875578820705414,
+      "num_tokens": 51863924.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.8576709866523743,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9386286735534668,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.02946900427341461,
+      "mean_token_accuracy": 0.9885939955711365,
+      "num_tokens": 52052011.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.8404816091060638,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.095536708831787,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.030070826411247253,
+      "mean_token_accuracy": 0.9881400763988495,
+      "num_tokens": 52240955.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.8346689164638519,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9964426159858704,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03005717396736145,
+      "mean_token_accuracy": 0.9883057296276092,
+      "num_tokens": 52429544.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.8377436757087707,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.8499778509140015,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.029283815622329713,
+      "mean_token_accuracy": 0.9882387280464172,
+      "num_tokens": 52618405.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.8328788638114929,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9594390988349915,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.02666555345058441,
+      "mean_token_accuracy": 0.9889323055744171,
+      "num_tokens": 52806689.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.829415249824524,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.3587099313735962,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.028349629044532774,
+      "mean_token_accuracy": 0.9890907645225525,
+      "num_tokens": 52995248.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.8335795283317566,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.077789306640625,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.03001731038093567,
+      "mean_token_accuracy": 0.9878358781337738,
+      "num_tokens": 53183607.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.8316770553588867,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 0.7265951037406921,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.030136841535568237,
+      "mean_token_accuracy": 0.9880174934864044,
+      "num_tokens": 53372735.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.8310399711132049,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.6605978012084961,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.026665833592414857,
+      "mean_token_accuracy": 0.9892704486846924,
+      "num_tokens": 53561776.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.8258654296398162,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.6560797691345215,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.026718440651893615,
+      "mean_token_accuracy": 0.989847868680954,
+      "num_tokens": 53750019.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.8221413195133209,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 0.7495356798171997,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.027431756258010864,
+      "mean_token_accuracy": 0.9892430484294892,
+      "num_tokens": 53938914.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.8341928541660308,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.0676971673965454,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.028432759642601012,
+      "mean_token_accuracy": 0.9889024317264556,
+      "num_tokens": 54127763.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.8359149813652038,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.380476713180542,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.028149163722991942,
+      "mean_token_accuracy": 0.9884953200817108,
+      "num_tokens": 54315930.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.8294597804546356,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0728843212127686,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.028733691573143004,
+      "mean_token_accuracy": 0.9884523034095765,
+      "num_tokens": 54504632.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.8245523810386658,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 0.9723307490348816,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.02793261408805847,
+      "mean_token_accuracy": 0.9891038477420807,
+      "num_tokens": 54693752.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.8279620766639709,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.003624439239502,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.028367668390274048,
+      "mean_token_accuracy": 0.9893603324890137,
+      "num_tokens": 54882030.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.8306584179401397,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.8481361865997314,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.029952478408813477,
+      "mean_token_accuracy": 0.9881332099437714,
+      "num_tokens": 55070549.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.8316469848155975,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.7631049156188965,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.029351180791854857,
+      "mean_token_accuracy": 0.9888184070587158,
+      "num_tokens": 55258999.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.8221287310123444,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 0.7234840393066406,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.027709537744522096,
+      "mean_token_accuracy": 0.9890308439731598,
+      "num_tokens": 55448047.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.8239192545413971,
+      "epoch": 3.970390309555855,
+      "grad_norm": 0.8497020602226257,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.029604345560073853,
+      "mean_token_accuracy": 0.9881375730037689,
+      "num_tokens": 55636566.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.8319927275180816,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.1599552631378174,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.02763860821723938,
+      "mean_token_accuracy": 0.9891406774520874,
+      "num_tokens": 55825049.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.8404937744140625,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 0.906755805015564,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.029578033089637756,
+      "mean_token_accuracy": 0.9878996431827545,
+      "num_tokens": 56012965.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.8419649528849656,
+      "eval_loss": 0.0570383295416832,
+      "eval_mean_token_accuracy": 0.9787763862093543,
+      "eval_num_tokens": 56050518.0,
+      "eval_runtime": 24.4889,
+      "eval_samples_per_second": 204.174,
+      "eval_steps_per_second": 6.411,
+      "step": 2972
+    },
+    {
+      "entropy": 0.8328965544700623,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7073294520378113,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.020916813611984254,
+      "mean_token_accuracy": 0.992138022184372,
+      "num_tokens": 56201621.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.8153974890708924,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.8904889822006226,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.01942949742078781,
+      "mean_token_accuracy": 0.9927055358886718,
+      "num_tokens": 56390018.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.8080023109912873,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.0828146934509277,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.018959003686904907,
+      "mean_token_accuracy": 0.9928978025913239,
+      "num_tokens": 56578046.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.8056950688362121,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7407417893409729,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019250378012657166,
+      "mean_token_accuracy": 0.9926685452461242,
+      "num_tokens": 56767192.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.8124530673027038,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.0355021953582764,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.01904640197753906,
+      "mean_token_accuracy": 0.9925767481327057,
+      "num_tokens": 56955769.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.810802960395813,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.3759289979934692,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.020995599031448365,
+      "mean_token_accuracy": 0.9917719483375549,
+      "num_tokens": 57144300.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.8159192740917206,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.112034559249878,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.01948906183242798,
+      "mean_token_accuracy": 0.9928658187389374,
+      "num_tokens": 57332935.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.821906328201294,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.0319583415985107,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019549430906772615,
+      "mean_token_accuracy": 0.9922966659069061,
+      "num_tokens": 57520775.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.8133022606372833,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.067180871963501,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.01833198368549347,
+      "mean_token_accuracy": 0.9934999167919158,
+      "num_tokens": 57709495.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.8150732755661011,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.277079701423645,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.019589337706565856,
+      "mean_token_accuracy": 0.992507028579712,
+      "num_tokens": 57897461.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.8159323990345001,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.3150911331176758,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.01963164806365967,
+      "mean_token_accuracy": 0.9926789402961731,
+      "num_tokens": 58085915.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.8147098064422608,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.0073580741882324,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.02144385576248169,
+      "mean_token_accuracy": 0.9918489575386047,
+      "num_tokens": 58274583.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.8155419886112213,
+      "epoch": 4.172274562584119,
+      "grad_norm": 0.8892306685447693,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.02215111702680588,
+      "mean_token_accuracy": 0.9914537072181702,
+      "num_tokens": 58463241.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.8084070324897766,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.7887238264083862,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.018581822514533997,
+      "mean_token_accuracy": 0.9928648293018341,
+      "num_tokens": 58652281.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.8016656517982483,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.0644817352294922,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.020099574327468873,
+      "mean_token_accuracy": 0.9916590929031373,
+      "num_tokens": 58839946.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.8035425901412964,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.9588403701782227,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020384626090526582,
+      "mean_token_accuracy": 0.9918951213359832,
+      "num_tokens": 59028994.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.8114414095878602,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.7145556211471558,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.019365760684013366,
+      "mean_token_accuracy": 0.9921130836009979,
+      "num_tokens": 59217599.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.8123053193092347,
+      "epoch": 4.23956931359354,
+      "grad_norm": 0.7928217649459839,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.020542070269584656,
+      "mean_token_accuracy": 0.9918384432792664,
+      "num_tokens": 59406176.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.810414057970047,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.1470942497253418,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.01809942126274109,
+      "mean_token_accuracy": 0.992941427230835,
+      "num_tokens": 59594604.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.8046662211418152,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.579035222530365,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.01914280205965042,
+      "mean_token_accuracy": 0.9926800608634949,
+      "num_tokens": 59782947.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.8034039616584778,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.8979132175445557,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.019897907972335815,
+      "mean_token_accuracy": 0.9919832170009613,
+      "num_tokens": 59971417.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.8137863457202912,
+      "epoch": 4.293405114401077,
+      "grad_norm": 0.9149890542030334,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.02069481760263443,
+      "mean_token_accuracy": 0.992353367805481,
+      "num_tokens": 60159983.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.8163237690925598,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.8264126181602478,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.021351543068885804,
+      "mean_token_accuracy": 0.9921438097953796,
+      "num_tokens": 60348327.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.8133926391601562,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.0638184547424316,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.020092280209064485,
+      "mean_token_accuracy": 0.9920917391777039,
+      "num_tokens": 60537152.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.8095248222351075,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.8831470012664795,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.01908399909734726,
+      "mean_token_accuracy": 0.9929656088352203,
+      "num_tokens": 60725603.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.8069245994091034,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8486410975456238,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.01925535798072815,
+      "mean_token_accuracy": 0.9927195310592651,
+      "num_tokens": 60914021.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.8006627321243286,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.9021628499031067,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.019187870621681213,
+      "mean_token_accuracy": 0.992531019449234,
+      "num_tokens": 61102037.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.8067023754119873,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.7793697118759155,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.021760080754756928,
+      "mean_token_accuracy": 0.9915579319000244,
+      "num_tokens": 61290132.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.8075471341609954,
+      "epoch": 4.387617765814267,
+      "grad_norm": 0.736788809299469,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.02077012062072754,
+      "mean_token_accuracy": 0.9920295119285584,
+      "num_tokens": 61478897.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.8083206176757812,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.1687020063400269,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.02117353230714798,
+      "mean_token_accuracy": 0.991925036907196,
+      "num_tokens": 61667982.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.81106236577034,
+      "epoch": 4.414535666218035,
+      "grad_norm": 0.7596265077590942,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.02081204652786255,
+      "mean_token_accuracy": 0.9921064436435699,
+      "num_tokens": 61856557.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.8018859803676606,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.823410153388977,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.019047826528549194,
+      "mean_token_accuracy": 0.9924882829189301,
+      "num_tokens": 62045181.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7996662855148315,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.0748025178909302,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.01788569688796997,
+      "mean_token_accuracy": 0.9931309640407562,
+      "num_tokens": 62234215.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.8021627008914948,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9946282505989075,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021408762037754058,
+      "mean_token_accuracy": 0.9914931178092956,
+      "num_tokens": 62422361.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.8084165751934052,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.7313793301582336,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.020281489193439483,
+      "mean_token_accuracy": 0.9925725042819977,
+      "num_tokens": 62610667.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.8103892624378204,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.9093077182769775,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.018877537548542024,
+      "mean_token_accuracy": 0.9929530620574951,
+      "num_tokens": 62798970.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.8091431736946106,
+      "epoch": 4.495289367429341,
+      "grad_norm": 0.952234148979187,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021039776504039764,
+      "mean_token_accuracy": 0.9918699383735656,
+      "num_tokens": 62986914.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.8094858765602112,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.9959521293640137,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021489059925079344,
+      "mean_token_accuracy": 0.9914597749710083,
+      "num_tokens": 63175805.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.807330971956253,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.6189294457435608,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.02164270281791687,
+      "mean_token_accuracy": 0.9920661866664886,
+      "num_tokens": 63364692.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.8098866164684295,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7895700931549072,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.021856357157230378,
+      "mean_token_accuracy": 0.9916301846504212,
+      "num_tokens": 63553692.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.8143714487552642,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0242317914962769,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02046577036380768,
+      "mean_token_accuracy": 0.9923330962657928,
+      "num_tokens": 63742476.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.8123241424560547,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.4186221361160278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.020659247040748598,
+      "mean_token_accuracy": 0.9918934762477875,
+      "num_tokens": 63930898.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.8133015930652618,
+      "epoch": 4.576043068640646,
+      "grad_norm": 0.9085373878479004,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.021746012568473815,
+      "mean_token_accuracy": 0.9918161749839782,
+      "num_tokens": 64119589.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.8173148334026337,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.7257112264633179,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.019787387549877168,
+      "mean_token_accuracy": 0.9922669887542724,
+      "num_tokens": 64308293.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.8116482973098755,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7442086935043335,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.020933127403259276,
+      "mean_token_accuracy": 0.9917319416999817,
+      "num_tokens": 64497362.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.8134189546108246,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.7617318630218506,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.01842810958623886,
+      "mean_token_accuracy": 0.9927939116954804,
+      "num_tokens": 64685802.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.8112196505069733,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.7430832386016846,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.01809561103582382,
+      "mean_token_accuracy": 0.9930340230464936,
+      "num_tokens": 64873973.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.8128187835216523,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.7553606629371643,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.0202948659658432,
+      "mean_token_accuracy": 0.9919280350208283,
+      "num_tokens": 65062510.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.813840126991272,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.7390825152397156,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.020557042956352235,
+      "mean_token_accuracy": 0.9918098270893096,
+      "num_tokens": 65250940.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.8117553532123566,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.7698133587837219,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.020231780409812928,
+      "mean_token_accuracy": 0.9920670032501221,
+      "num_tokens": 65439696.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.8110962271690368,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.8562760353088379,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.018769541382789613,
+      "mean_token_accuracy": 0.9925244867801666,
+      "num_tokens": 65628357.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.8113800704479217,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.2076038122177124,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020604749023914338,
+      "mean_token_accuracy": 0.9921375870704651,
+      "num_tokens": 65817139.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.8204582631587982,
+      "epoch": 4.710632570659489,
+      "grad_norm": 0.9150722026824951,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021981674432754516,
+      "mean_token_accuracy": 0.9915678203105927,
+      "num_tokens": 66005911.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.8241343259811401,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.9803491830825806,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02007504999637604,
+      "mean_token_accuracy": 0.992132717370987,
+      "num_tokens": 66194469.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.8122783243656159,
+      "epoch": 4.737550471063257,
+      "grad_norm": 0.967640221118927,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.019903984665870667,
+      "mean_token_accuracy": 0.9924214482307434,
+      "num_tokens": 66383927.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.8145494401454926,
+      "epoch": 4.751009421265142,
+      "grad_norm": 0.7893037796020508,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.019587111473083497,
+      "mean_token_accuracy": 0.9921624064445496,
+      "num_tokens": 66572212.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.8090193390846252,
+      "epoch": 4.764468371467026,
+      "grad_norm": 0.9659109711647034,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.021097327768802642,
+      "mean_token_accuracy": 0.9917473196983337,
+      "num_tokens": 66760377.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.802634209394455,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.0175755023956299,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.02113145440816879,
+      "mean_token_accuracy": 0.991830724477768,
+      "num_tokens": 66949210.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.8030233085155487,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.7780183553695679,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.019016166031360627,
+      "mean_token_accuracy": 0.9926414608955383,
+      "num_tokens": 67138031.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.8019147455692291,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7863438129425049,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.018219922482967377,
+      "mean_token_accuracy": 0.9928844273090363,
+      "num_tokens": 67327268.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7949212431907654,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9292862415313721,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.02136329412460327,
+      "mean_token_accuracy": 0.9918141901493073,
+      "num_tokens": 67515765.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.8002764582633972,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9770768284797668,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.02056926190853119,
+      "mean_token_accuracy": 0.9918984293937683,
+      "num_tokens": 67704975.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.808483463525772,
+      "epoch": 4.845222072678331,
+      "grad_norm": 0.6890367269515991,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020158609747886656,
+      "mean_token_accuracy": 0.9925975620746612,
+      "num_tokens": 67893850.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.8144356846809387,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.8613544702529907,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.019614605605602263,
+      "mean_token_accuracy": 0.9928383588790893,
+      "num_tokens": 68082435.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.8083879172801971,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.7532686591148376,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.019553859531879426,
+      "mean_token_accuracy": 0.9924478828907013,
+      "num_tokens": 68270495.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.8038046479225158,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.3251928091049194,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.02081710398197174,
+      "mean_token_accuracy": 0.9919814527034759,
+      "num_tokens": 68459223.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.8058662831783294,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8125224113464355,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.01858963668346405,
+      "mean_token_accuracy": 0.9922726869583129,
+      "num_tokens": 68647306.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.805506020784378,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8629401922225952,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020777507126331328,
+      "mean_token_accuracy": 0.9918497979640961,
+      "num_tokens": 68836386.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.8137464344501495,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.9837846159934998,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.02024296224117279,
+      "mean_token_accuracy": 0.9921825706958771,
+      "num_tokens": 69025192.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.8139736235141755,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.8701289296150208,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.01932244449853897,
+      "mean_token_accuracy": 0.9926821112632751,
+      "num_tokens": 69213427.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.8088622391223907,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.7041723132133484,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.018706569075584413,
+      "mean_token_accuracy": 0.9927357733249664,
+      "num_tokens": 69401690.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.8001897990703583,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1029852628707886,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.020011243224143983,
+      "mean_token_accuracy": 0.9923900067806244,
+      "num_tokens": 69590659.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.8071568071842193,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.2395373582839966,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.02379731684923172,
+      "mean_token_accuracy": 0.9907991290092468,
+      "num_tokens": 69779864.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.8139646112918854,
+      "epoch": 4.993270524899058,
+      "grad_norm": 0.9219555854797363,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.021635296940803527,
+      "mean_token_accuracy": 0.991401469707489,
+      "num_tokens": 69968639.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.822659221804066,
+      "eval_loss": 0.060467079281806946,
+      "eval_mean_token_accuracy": 0.9792145984188007,
+      "eval_num_tokens": 70063064.0,
+      "eval_runtime": 24.4326,
+      "eval_samples_per_second": 204.645,
+      "eval_steps_per_second": 6.426,
+      "step": 3715
+    },
+    {
+      "entropy": 0.8150339186191559,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8010120391845703,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01673099249601364,
+      "mean_token_accuracy": 0.9935125410556793,
+      "num_tokens": 70157396.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.8029226005077362,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9923233389854431,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01266745626926422,
+      "mean_token_accuracy": 0.9954398214817047,
+      "num_tokens": 70346010.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.789790153503418,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.7180858254432678,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.012603424489498138,
+      "mean_token_accuracy": 0.9949092507362366,
+      "num_tokens": 70534896.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.7863301753997802,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.6865439414978027,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.013341715931892395,
+      "mean_token_accuracy": 0.9951623439788818,
+      "num_tokens": 70723644.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.7841233730316162,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.8624011278152466,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01214413121342659,
+      "mean_token_accuracy": 0.9950926601886749,
+      "num_tokens": 70912490.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.7781193614006042,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.1462656259536743,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.013635106384754181,
+      "mean_token_accuracy": 0.9946888983249664,
+      "num_tokens": 71101186.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.7773554801940918,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.7154580354690552,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.012310966104269027,
+      "mean_token_accuracy": 0.9952048182487487,
+      "num_tokens": 71290177.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.784615957736969,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8214139342308044,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.012708047032356262,
+      "mean_token_accuracy": 0.9956556618213653,
+      "num_tokens": 71478243.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.7814814388751984,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0974094867706299,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.011942464113235473,
+      "mean_token_accuracy": 0.9958715498447418,
+      "num_tokens": 71667417.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.7749826848506928,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.7582180500030518,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.011887111514806748,
+      "mean_token_accuracy": 0.9954704821109772,
+      "num_tokens": 71856322.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.7695000052452088,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.9159554839134216,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.013456562161445617,
+      "mean_token_accuracy": 0.994843590259552,
+      "num_tokens": 72044716.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.7674728155136108,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.0967031717300415,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.012064316868782043,
+      "mean_token_accuracy": 0.9953838646411896,
+      "num_tokens": 72233243.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.766239732503891,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.756477415561676,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012840582430362702,
+      "mean_token_accuracy": 0.9951915740966797,
+      "num_tokens": 72421751.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.7667239785194397,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.8078455924987793,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.012580010294914245,
+      "mean_token_accuracy": 0.9955588281154633,
+      "num_tokens": 72610886.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.7759605884552002,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.9473005533218384,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.01442105770111084,
+      "mean_token_accuracy": 0.9946676790714264,
+      "num_tokens": 72799213.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.7739267945289612,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.4796932637691498,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011830046027898788,
+      "mean_token_accuracy": 0.9954923093318939,
+      "num_tokens": 72988413.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.7756406426429748,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7290598154067993,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.014412418007850647,
+      "mean_token_accuracy": 0.9946037054061889,
+      "num_tokens": 73177134.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.7763995885848999,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.0562312602996826,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.015396332740783692,
+      "mean_token_accuracy": 0.9945933580398559,
+      "num_tokens": 73365622.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.7737024486064911,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1071312427520752,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01451488733291626,
+      "mean_token_accuracy": 0.9945858538150787,
+      "num_tokens": 73554381.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.779985535144806,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.7758586406707764,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.013801637291908263,
+      "mean_token_accuracy": 0.9947169423103333,
+      "num_tokens": 73743028.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.7786811470985413,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.6511496305465698,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012189272791147232,
+      "mean_token_accuracy": 0.9954657673835754,
+      "num_tokens": 73931263.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.7736978232860565,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.8152093291282654,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.013696780800819397,
+      "mean_token_accuracy": 0.9948886811733246,
+      "num_tokens": 74119992.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.7730314910411835,
+      "epoch": 5.302826379542395,
+      "grad_norm": 0.6703830361366272,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012700904905796052,
+      "mean_token_accuracy": 0.9950747072696686,
+      "num_tokens": 74308428.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.7740058362483978,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.9845885634422302,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012894082069396972,
+      "mean_token_accuracy": 0.9953160643577575,
+      "num_tokens": 74497030.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.7738905131816864,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.732532799243927,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.011950771510601043,
+      "mean_token_accuracy": 0.9955996692180633,
+      "num_tokens": 74685631.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.7788872420787811,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7301356196403503,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012340743839740754,
+      "mean_token_accuracy": 0.9952776432037354,
+      "num_tokens": 74874025.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.7771204471588135,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.7570232152938843,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.012937523424625397,
+      "mean_token_accuracy": 0.9949343740940094,
+      "num_tokens": 75062636.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.7747974276542664,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8237525820732117,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.012423938512802124,
+      "mean_token_accuracy": 0.9953977644443512,
+      "num_tokens": 75251339.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.7692180812358856,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.8497490286827087,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012855434417724609,
+      "mean_token_accuracy": 0.9948123812675476,
+      "num_tokens": 75440121.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.773431783914566,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.6758515238761902,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.012878020107746125,
+      "mean_token_accuracy": 0.9955350995063782,
+      "num_tokens": 75629143.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.7735035121440887,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7578167915344238,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.013059911131858826,
+      "mean_token_accuracy": 0.994978415966034,
+      "num_tokens": 75817425.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.7725034832954407,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.0449564456939697,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.013163739442825317,
+      "mean_token_accuracy": 0.9953307807445526,
+      "num_tokens": 76005910.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.7736131310462951,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.881845235824585,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012109111249446868,
+      "mean_token_accuracy": 0.9952919363975525,
+      "num_tokens": 76194639.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.7797342419624329,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.8100120425224304,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.013262654840946197,
+      "mean_token_accuracy": 0.9948459386825561,
+      "num_tokens": 76383240.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.7794113874435424,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.83738112449646,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.013258925080299378,
+      "mean_token_accuracy": 0.9947907269001007,
+      "num_tokens": 76572040.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.7787538826465606,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0567374229431152,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.014641955494880676,
+      "mean_token_accuracy": 0.9945057690143585,
+      "num_tokens": 76761030.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.7894418001174927,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.835543692111969,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.013065731525421143,
+      "mean_token_accuracy": 0.9949834287166596,
+      "num_tokens": 76948469.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.7823423445224762,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.8142625093460083,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.01229657381772995,
+      "mean_token_accuracy": 0.9957204461097717,
+      "num_tokens": 77136968.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.7786863029003144,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.749609649181366,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.01328940987586975,
+      "mean_token_accuracy": 0.9949924945831299,
+      "num_tokens": 77325328.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.7780588805675507,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.787598729133606,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.013933436572551727,
+      "mean_token_accuracy": 0.9949735462665558,
+      "num_tokens": 77513716.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.7779332935810089,
+      "epoch": 5.545087483176312,
+      "grad_norm": 1.0315855741500854,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.012871211767196656,
+      "mean_token_accuracy": 0.995112019777298,
+      "num_tokens": 77701621.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.7754440903663635,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.8191165924072266,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.012908256053924561,
+      "mean_token_accuracy": 0.9950248181819916,
+      "num_tokens": 77889886.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.7811384737491608,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.832539975643158,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.013047756254673004,
+      "mean_token_accuracy": 0.9946948051452636,
+      "num_tokens": 78078640.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.7825228571891785,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.9072359800338745,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.012361775338649749,
+      "mean_token_accuracy": 0.9954402267932891,
+      "num_tokens": 78267842.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.7825945496559144,
+      "epoch": 5.598923283983849,
+      "grad_norm": 0.8300487995147705,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.013246089220046997,
+      "mean_token_accuracy": 0.9951650142669678,
+      "num_tokens": 78455995.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.7857775866985321,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.9343852996826172,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.012612038850784301,
+      "mean_token_accuracy": 0.995340085029602,
+      "num_tokens": 78643617.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.775356650352478,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7109665274620056,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.013411080837249756,
+      "mean_token_accuracy": 0.9951648890972138,
+      "num_tokens": 78833085.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.7750323057174683,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.6436781883239746,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.012064981460571288,
+      "mean_token_accuracy": 0.9954153180122376,
+      "num_tokens": 79021569.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.7753469109535217,
+      "epoch": 5.652759084791386,
+      "grad_norm": 1.0361640453338623,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.013953472673892974,
+      "mean_token_accuracy": 0.9947281837463379,
+      "num_tokens": 79210573.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.7821656167507172,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9078817367553711,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012946712970733642,
+      "mean_token_accuracy": 0.9954073607921601,
+      "num_tokens": 79398723.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.7803877890110016,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.7879899740219116,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.012198399752378464,
+      "mean_token_accuracy": 0.9953389108181,
+      "num_tokens": 79587809.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.7772530496120453,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.0390578508377075,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.012395218014717102,
+      "mean_token_accuracy": 0.9953405141830445,
+      "num_tokens": 79775768.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.7784459292888641,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.7382562160491943,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.012304402142763137,
+      "mean_token_accuracy": 0.9953492462635041,
+      "num_tokens": 79964602.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.7821883440017701,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.5720117092132568,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.011424948275089265,
+      "mean_token_accuracy": 0.9956757485866546,
+      "num_tokens": 80153320.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.7811889886856079,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.7762079238891602,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012460066378116608,
+      "mean_token_accuracy": 0.9950654745101929,
+      "num_tokens": 80341381.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.7771898031234741,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.9219444394111633,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.011629728972911835,
+      "mean_token_accuracy": 0.9955675959587097,
+      "num_tokens": 80530323.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.7825480580329895,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.7189850211143494,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.011180496215820313,
+      "mean_token_accuracy": 0.9960062265396118,
+      "num_tokens": 80718533.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.7795504450798034,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.7256530523300171,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.012952378392219544,
+      "mean_token_accuracy": 0.9951896607875824,
+      "num_tokens": 80907391.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.779454517364502,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.808525800704956,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.0121039479970932,
+      "mean_token_accuracy": 0.9955627739429473,
+      "num_tokens": 81096396.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.7813481986522675,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.7294234037399292,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01356595754623413,
+      "mean_token_accuracy": 0.9951751470565796,
+      "num_tokens": 81284943.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.7833737671375275,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.771756649017334,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.013437020778656005,
+      "mean_token_accuracy": 0.9951942205429077,
+      "num_tokens": 81473810.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.781111603975296,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.5440986752510071,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.012187350541353226,
+      "mean_token_accuracy": 0.9956019043922424,
+      "num_tokens": 81662193.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.7777030646800995,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7532716393470764,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.013766974210739136,
+      "mean_token_accuracy": 0.994896411895752,
+      "num_tokens": 81851045.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.7776691019535065,
+      "epoch": 5.85464333781965,
+      "grad_norm": 1.1882649660110474,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.012588076293468475,
+      "mean_token_accuracy": 0.9952153980731964,
+      "num_tokens": 82039225.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.7795583724975585,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.7453442215919495,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.012100882083177566,
+      "mean_token_accuracy": 0.9954553186893463,
+      "num_tokens": 82228123.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.7763878583908081,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.6757420301437378,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011943653970956803,
+      "mean_token_accuracy": 0.9953660726547241,
+      "num_tokens": 82417093.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.7747587621212005,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.9607213735580444,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.014138388633728027,
+      "mean_token_accuracy": 0.9944287598133087,
+      "num_tokens": 82605461.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.7782076597213745,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.6733895540237427,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.012490443140268325,
+      "mean_token_accuracy": 0.9950622916221619,
+      "num_tokens": 82794124.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.7843947112560272,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.7626368403434753,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.012137357890605927,
+      "mean_token_accuracy": 0.9952504158020019,
+      "num_tokens": 82982258.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.7826779782772064,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8149229288101196,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012872529029846192,
+      "mean_token_accuracy": 0.9949809432029724,
+      "num_tokens": 83170520.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.7792184889316559,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.7774028778076172,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012353222817182541,
+      "mean_token_accuracy": 0.9953234374523163,
+      "num_tokens": 83359174.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.7778255641460419,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7626445293426514,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.01137089878320694,
+      "mean_token_accuracy": 0.9958219468593598,
+      "num_tokens": 83547583.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.7765573263168335,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9130274653434753,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012439383566379547,
+      "mean_token_accuracy": 0.9954963147640228,
+      "num_tokens": 83735904.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.7734197914600373,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.7289780378341675,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011430246382951736,
+      "mean_token_accuracy": 0.9957478463649749,
+      "num_tokens": 83924971.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.7852135965019275,
+      "eval_loss": 0.06990383565425873,
+      "eval_mean_token_accuracy": 0.9796095274056599,
+      "eval_num_tokens": 84075763.0,
+      "eval_runtime": 24.3947,
+      "eval_samples_per_second": 204.963,
+      "eval_steps_per_second": 6.436,
+      "step": 4458
+    },
+    {
+      "entropy": 0.7758591413497925,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.9430185556411743,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.011816921830177306,
+      "mean_token_accuracy": 0.9959785401821136,
+      "num_tokens": 84113330.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.773297393321991,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.6593970060348511,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.007801702618598938,
+      "mean_token_accuracy": 0.9972260951995849,
+      "num_tokens": 84302325.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.7716216742992401,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.5422326922416687,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.00867043137550354,
+      "mean_token_accuracy": 0.9968674182891846,
+      "num_tokens": 84490227.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.7699320554733277,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.8308367729187012,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.007848290354013443,
+      "mean_token_accuracy": 0.9971986234188079,
+      "num_tokens": 84678824.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.7688778102397918,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.5276612639427185,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.007070200890302658,
+      "mean_token_accuracy": 0.9974652588367462,
+      "num_tokens": 84867430.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.7687382102012634,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.7081453800201416,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.00793309137225151,
+      "mean_token_accuracy": 0.9972716808319092,
+      "num_tokens": 85056072.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.7705117404460907,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.9106748104095459,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.007362194359302521,
+      "mean_token_accuracy": 0.9975170493125916,
+      "num_tokens": 85244337.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.7697769284248352,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.7364314198493958,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.007369447499513626,
+      "mean_token_accuracy": 0.9974394500255584,
+      "num_tokens": 85432878.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.7661255121231079,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.6747480630874634,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.007502703368663788,
+      "mean_token_accuracy": 0.9973482489585876,
+      "num_tokens": 85621800.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.7673385441303253,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.8179678916931152,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.007111805677413941,
+      "mean_token_accuracy": 0.9977720499038696,
+      "num_tokens": 85809796.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.76323401927948,
+      "epoch": 6.137281292059219,
+      "grad_norm": 0.5177297592163086,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.007196265459060669,
+      "mean_token_accuracy": 0.9976384699344635,
+      "num_tokens": 85998543.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.7640918910503387,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 0.8970023989677429,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.008229555189609527,
+      "mean_token_accuracy": 0.9967760264873504,
+      "num_tokens": 86186537.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.7605979859828949,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.7019157409667969,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.007770314812660217,
+      "mean_token_accuracy": 0.997019499540329,
+      "num_tokens": 86375405.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.7647112905979156,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.6519957184791565,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.007326644659042358,
+      "mean_token_accuracy": 0.9972577571868897,
+      "num_tokens": 86562963.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.7608903884887696,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.9307657480239868,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.00786140188574791,
+      "mean_token_accuracy": 0.9970912158489227,
+      "num_tokens": 86751006.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.7573343932628631,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 0.6093463897705078,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.007400572299957275,
+      "mean_token_accuracy": 0.997343909740448,
+      "num_tokens": 86939232.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.756312346458435,
+      "epoch": 6.218034993270525,
+      "grad_norm": 0.6412308812141418,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.007471667230129242,
+      "mean_token_accuracy": 0.9973339200019836,
+      "num_tokens": 87128294.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.7593119204044342,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.6830183863639832,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.007831692695617676,
+      "mean_token_accuracy": 0.9972396552562713,
+      "num_tokens": 87316833.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.7598747551441193,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.6647830009460449,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.0070374816656112674,
+      "mean_token_accuracy": 0.9975631892681122,
+      "num_tokens": 87505022.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.7626749157905579,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.5470786690711975,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.007449843734502792,
+      "mean_token_accuracy": 0.9976118981838227,
+      "num_tokens": 87692953.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.7584024250507355,
+      "epoch": 6.271870794078062,
+      "grad_norm": 0.8641819953918457,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.007355903089046478,
+      "mean_token_accuracy": 0.99752836227417,
+      "num_tokens": 87882513.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.7607118487358093,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.6385810971260071,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.007216785848140716,
+      "mean_token_accuracy": 0.9973138213157654,
+      "num_tokens": 88071161.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.7600765109062195,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.3984440565109253,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.007562516629695893,
+      "mean_token_accuracy": 0.9973795533180236,
+      "num_tokens": 88259867.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.7609024405479431,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.7098219990730286,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.007748734951019287,
+      "mean_token_accuracy": 0.9970796704292297,
+      "num_tokens": 88448504.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.7602188348770141,
+      "epoch": 6.325706594885599,
+      "grad_norm": 0.3834342658519745,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.00794660747051239,
+      "mean_token_accuracy": 0.9971937775611878,
+      "num_tokens": 88637123.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.7595065951347351,
+      "epoch": 6.339165545087483,
+      "grad_norm": 0.7073737978935242,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.007514684647321701,
+      "mean_token_accuracy": 0.9972682356834411,
+      "num_tokens": 88826107.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.7566567718982696,
+      "epoch": 6.352624495289367,
+      "grad_norm": 0.6334935426712036,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.007287996262311936,
+      "mean_token_accuracy": 0.9972865641117096,
+      "num_tokens": 89014806.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.7535678207874298,
+      "epoch": 6.366083445491252,
+      "grad_norm": 0.5461204648017883,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.007215028256177902,
+      "mean_token_accuracy": 0.9976517677307128,
+      "num_tokens": 89203610.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.753510159254074,
+      "epoch": 6.379542395693136,
+      "grad_norm": 0.8002920150756836,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.00775575190782547,
+      "mean_token_accuracy": 0.9972102761268615,
+      "num_tokens": 89392611.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.7567359745502472,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.5416199564933777,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.007165181636810303,
+      "mean_token_accuracy": 0.9973179161548614,
+      "num_tokens": 89581472.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.7603509783744812,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.7484428882598877,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.008048020303249359,
+      "mean_token_accuracy": 0.9972274541854859,
+      "num_tokens": 89769895.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.7604019463062286,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.6397516131401062,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.006713573634624481,
+      "mean_token_accuracy": 0.9974542498588562,
+      "num_tokens": 89958151.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.7562435805797577,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.6562761068344116,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.0070735208690166475,
+      "mean_token_accuracy": 0.9975251019001007,
+      "num_tokens": 90147009.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.7581055939197541,
+      "epoch": 6.446837146702557,
+      "grad_norm": 0.6705797910690308,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.008277279138565064,
+      "mean_token_accuracy": 0.9969512403011322,
+      "num_tokens": 90335135.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.7552513837814331,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.5535324215888977,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.007625260949134826,
+      "mean_token_accuracy": 0.9972123861312866,
+      "num_tokens": 90523520.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.7567071616649628,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.5841183662414551,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.008183968812227249,
+      "mean_token_accuracy": 0.9970704436302185,
+      "num_tokens": 90712012.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.7579837441444397,
+      "epoch": 6.48721399730821,
+      "grad_norm": 1.0210020542144775,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.00890980064868927,
+      "mean_token_accuracy": 0.9968561112880707,
+      "num_tokens": 90900360.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.7599723398685455,
+      "epoch": 6.500672947510094,
+      "grad_norm": 0.7007607221603394,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.007547108829021454,
+      "mean_token_accuracy": 0.9972310900688172,
+      "num_tokens": 91089219.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.7569817125797271,
+      "epoch": 6.514131897711978,
+      "grad_norm": 1.203089952468872,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.007012282311916351,
+      "mean_token_accuracy": 0.9974212229251862,
+      "num_tokens": 91278177.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.7547080159187317,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.6055266857147217,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.007885889708995819,
+      "mean_token_accuracy": 0.9971360504627228,
+      "num_tokens": 91466912.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.7584831535816192,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.6868627071380615,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.00764903798699379,
+      "mean_token_accuracy": 0.9975648581981659,
+      "num_tokens": 91655249.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.757660973072052,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.9303789734840393,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.00731472447514534,
+      "mean_token_accuracy": 0.9974016606807709,
+      "num_tokens": 91844016.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.759681248664856,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 0.6116757988929749,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.007223697006702423,
+      "mean_token_accuracy": 0.9970514178276062,
+      "num_tokens": 92032638.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.7601584196090698,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.6825643181800842,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.0068596042692661285,
+      "mean_token_accuracy": 0.9973869740962982,
+      "num_tokens": 92221897.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.7645585358142852,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.7482470870018005,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.007520741224288941,
+      "mean_token_accuracy": 0.9974736630916595,
+      "num_tokens": 92410416.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.766044944524765,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.7137084603309631,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.007919894158840179,
+      "mean_token_accuracy": 0.9969592452049255,
+      "num_tokens": 92598533.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.761748319864273,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 1.0537257194519043,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.007994732260704041,
+      "mean_token_accuracy": 0.9970260739326477,
+      "num_tokens": 92788046.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.763516879081726,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.9168201088905334,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.006822402775287628,
+      "mean_token_accuracy": 0.9973889350891113,
+      "num_tokens": 92976583.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.7603426337242126,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.6945034265518188,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.007335217297077179,
+      "mean_token_accuracy": 0.9974107563495636,
+      "num_tokens": 93164984.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.7601041436195374,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.7255296111106873,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.006898632645606995,
+      "mean_token_accuracy": 0.9974060177803039,
+      "num_tokens": 93353860.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.760407668352127,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.7785485982894897,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.006965675204992294,
+      "mean_token_accuracy": 0.997627604007721,
+      "num_tokens": 93542542.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.7584291577339173,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.676058292388916,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.0066734164953231815,
+      "mean_token_accuracy": 0.9976742446422577,
+      "num_tokens": 93730962.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.7543873727321625,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.9423731565475464,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.007540161162614823,
+      "mean_token_accuracy": 0.9972456872463227,
+      "num_tokens": 93919838.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.7540828108787536,
+      "epoch": 6.716016150740242,
+      "grad_norm": 0.60378098487854,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.006739185005426407,
+      "mean_token_accuracy": 0.9976783573627472,
+      "num_tokens": 94108612.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.7581348657608032,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.7218618392944336,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.006258270889520645,
+      "mean_token_accuracy": 0.9977961003780365,
+      "num_tokens": 94296515.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.7555042028427124,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.6111772060394287,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.006771801412105561,
+      "mean_token_accuracy": 0.9975680649280548,
+      "num_tokens": 94484849.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.7563546478748322,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.7171721458435059,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.007536987960338593,
+      "mean_token_accuracy": 0.9971135079860687,
+      "num_tokens": 94673363.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.7541037023067474,
+      "epoch": 6.769851951547779,
+      "grad_norm": 0.45519423484802246,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.007405019551515579,
+      "mean_token_accuracy": 0.9975123167037964,
+      "num_tokens": 94862699.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.7588851511478424,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.7041094899177551,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.007134091854095459,
+      "mean_token_accuracy": 0.9972242057323456,
+      "num_tokens": 95050741.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.7570755481719971,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.6298620700836182,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.0070040211081504825,
+      "mean_token_accuracy": 0.997326648235321,
+      "num_tokens": 95239293.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.7548963069915772,
+      "epoch": 6.810228802153432,
+      "grad_norm": 1.0688964128494263,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.006342679262161255,
+      "mean_token_accuracy": 0.9976243853569031,
+      "num_tokens": 95428279.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.7548417568206787,
+      "epoch": 6.823687752355316,
+      "grad_norm": 0.4953668415546417,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.0069470912218093876,
+      "mean_token_accuracy": 0.9973559200763702,
+      "num_tokens": 95616835.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.7551373422145844,
+      "epoch": 6.837146702557201,
+      "grad_norm": 0.5016734600067139,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.006873356550931931,
+      "mean_token_accuracy": 0.9973772704601288,
+      "num_tokens": 95805557.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.7567492067813874,
+      "epoch": 6.850605652759085,
+      "grad_norm": 0.532783567905426,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.006668326258659363,
+      "mean_token_accuracy": 0.997450977563858,
+      "num_tokens": 95993831.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.758799821138382,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.6660076379776001,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.006823312491178513,
+      "mean_token_accuracy": 0.9975283324718476,
+      "num_tokens": 96181732.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.7571039736270905,
+      "epoch": 6.877523553162853,
+      "grad_norm": 1.0330839157104492,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.007801513373851776,
+      "mean_token_accuracy": 0.9971715807914734,
+      "num_tokens": 96369920.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.7555147588253022,
+      "epoch": 6.890982503364738,
+      "grad_norm": 0.5079207420349121,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.0074800834059715274,
+      "mean_token_accuracy": 0.9973418891429902,
+      "num_tokens": 96558724.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.7542111933231354,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.678804337978363,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.00698004812002182,
+      "mean_token_accuracy": 0.997770369052887,
+      "num_tokens": 96747242.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.7508629977703094,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.8755067586898804,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.007143624126911163,
+      "mean_token_accuracy": 0.9975672841072083,
+      "num_tokens": 96936244.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.7522107064723969,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.8553352952003479,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.007587063312530518,
+      "mean_token_accuracy": 0.9970708072185517,
+      "num_tokens": 97125069.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.7501682281494141,
+      "epoch": 6.944818304172275,
+      "grad_norm": 0.614628255367279,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.007764184474945068,
+      "mean_token_accuracy": 0.9968451082706451,
+      "num_tokens": 97313879.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.7522825837135315,
+      "epoch": 6.958277254374159,
+      "grad_norm": 0.8060729503631592,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.00771535262465477,
+      "mean_token_accuracy": 0.9972144842147828,
+      "num_tokens": 97503040.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.7534927785396576,
+      "epoch": 6.971736204576043,
+      "grad_norm": 0.6604568958282471,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.006561616808176041,
+      "mean_token_accuracy": 0.9974263906478882,
+      "num_tokens": 97692024.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.7542680203914642,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.853604793548584,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.007131250202655792,
+      "mean_token_accuracy": 0.9974451243877411,
+      "num_tokens": 97881026.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.7567435801029205,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.6146188974380493,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.006486450880765915,
+      "mean_token_accuracy": 0.9974190831184387,
+      "num_tokens": 98069553.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.7571151700748759,
+      "eval_loss": 0.07285518199205399,
+      "eval_mean_token_accuracy": 0.9808630795235846,
+      "eval_num_tokens": 98088531.0,
+      "eval_runtime": 24.4094,
+      "eval_samples_per_second": 204.839,
+      "eval_steps_per_second": 6.432,
+      "step": 5201
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.739606533627707e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5201/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-5201/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5201/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5944/README.md b/baby_talk_L16_a50/seed_42/checkpoint-5944/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5944/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5944/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-5944/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5944/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5944/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-5944/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..16eb463b2b9754ea503237f4a97c9ca087152987
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5944/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64ee1314c38681a3c419c696786895578b47932f4de4d674fc4c53395615d1d6
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5944/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-5944/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5944/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5944/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-5944/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5944/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5944/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-5944/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5944/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5944/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-5944/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..c82595c9a455056d5cd4d5784ecf9420daac1ae8
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5944/trainer_state.json
@@ -0,0 +1,6062 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.0,
+  "eval_steps": 500,
+  "global_step": 5944,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    },
+    {
+      "entropy": 0.9689933776855468,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9720287919044495,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.055563896894454956,
+      "mean_token_accuracy": 0.9779708027839661,
+      "num_tokens": 14144652.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.9506655633449554,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 0.6619595289230347,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.050587379932403566,
+      "mean_token_accuracy": 0.979685264825821,
+      "num_tokens": 14332714.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.9458732306957245,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 0.7905108332633972,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.05069155097007751,
+      "mean_token_accuracy": 0.9796596229076385,
+      "num_tokens": 14520902.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.9320916712284089,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 0.9210566878318787,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.051744121313095096,
+      "mean_token_accuracy": 0.9792148530483246,
+      "num_tokens": 14709215.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.928934782743454,
+      "epoch": 1.063257065948856,
+      "grad_norm": 0.7442036867141724,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.0486875057220459,
+      "mean_token_accuracy": 0.9798122465610504,
+      "num_tokens": 14898242.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.942881977558136,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.0910941362380981,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.052297019958496095,
+      "mean_token_accuracy": 0.9793200254440307,
+      "num_tokens": 15086835.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.9503917813301086,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 0.8091554641723633,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.049927744269371035,
+      "mean_token_accuracy": 0.9802082359790802,
+      "num_tokens": 15275042.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9426830470561981,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.0589245557785034,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.053803551197052005,
+      "mean_token_accuracy": 0.9785399675369263,
+      "num_tokens": 15463923.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.9445010781288147,
+      "epoch": 1.117092866756393,
+      "grad_norm": 0.8410441279411316,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.053181976079940796,
+      "mean_token_accuracy": 0.9792523324489594,
+      "num_tokens": 15652526.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.946729838848114,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 0.8246080279350281,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.054291915893554685,
+      "mean_token_accuracy": 0.9791357696056366,
+      "num_tokens": 15841340.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.9433728814125061,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 0.9200296401977539,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.05073915719985962,
+      "mean_token_accuracy": 0.9796931505203247,
+      "num_tokens": 16029478.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.9494417488574982,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 0.8342758417129517,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.050929927825927736,
+      "mean_token_accuracy": 0.9793219089508056,
+      "num_tokens": 16218014.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.9409256100654602,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.7911009192466736,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.049964362382888795,
+      "mean_token_accuracy": 0.9798487305641175,
+      "num_tokens": 16406420.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.9429958581924438,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.4877018928527832,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.04969423711299896,
+      "mean_token_accuracy": 0.9801322638988494,
+      "num_tokens": 16594948.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.9542136013507843,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 0.8030785918235779,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.04981146454811096,
+      "mean_token_accuracy": 0.9796162784099579,
+      "num_tokens": 16782983.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.9367722630500793,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.1169800758361816,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.05210963487625122,
+      "mean_token_accuracy": 0.9790139615535736,
+      "num_tokens": 16971394.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.9487208306789399,
+      "epoch": 1.224764468371467,
+      "grad_norm": 0.8650282621383667,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.05012243390083313,
+      "mean_token_accuracy": 0.9805107891559601,
+      "num_tokens": 17159951.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.9441231489181519,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 0.8193967938423157,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.048866665363311766,
+      "mean_token_accuracy": 0.9807584881782532,
+      "num_tokens": 17348711.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.9361630439758301,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3613612651824951,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.05324091911315918,
+      "mean_token_accuracy": 0.97904953956604,
+      "num_tokens": 17537116.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.9562997639179229,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.8121878504753113,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.05011019706726074,
+      "mean_token_accuracy": 0.9798731088638306,
+      "num_tokens": 17725567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.9359076261520386,
+      "epoch": 1.278600269179004,
+      "grad_norm": 0.7838549613952637,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.04834386110305786,
+      "mean_token_accuracy": 0.9804529249668121,
+      "num_tokens": 17914191.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.9527446150779724,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0981712341308594,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.055077624320983884,
+      "mean_token_accuracy": 0.9776535391807556,
+      "num_tokens": 18103184.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.9574812948703766,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 0.8291501402854919,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.050153911113739014,
+      "mean_token_accuracy": 0.9792965054512024,
+      "num_tokens": 18291351.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.939585280418396,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.0864412784576416,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.05132197737693787,
+      "mean_token_accuracy": 0.9792291462421417,
+      "num_tokens": 18480192.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.9468406856060028,
+      "epoch": 1.332436069986541,
+      "grad_norm": 0.9772785902023315,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.04844954013824463,
+      "mean_token_accuracy": 0.9804142415523529,
+      "num_tokens": 18669028.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.9413834273815155,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 0.9020094871520996,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.05075312852859497,
+      "mean_token_accuracy": 0.9790591120719909,
+      "num_tokens": 18857581.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.9526253461837768,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.0524990558624268,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.049737372994422914,
+      "mean_token_accuracy": 0.9795652389526367,
+      "num_tokens": 19046105.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.9484909176826477,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.0715620517730713,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.04935494959354401,
+      "mean_token_accuracy": 0.9797740161418915,
+      "num_tokens": 19234730.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.9398573458194732,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2134987115859985,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.0518394410610199,
+      "mean_token_accuracy": 0.9793788075447083,
+      "num_tokens": 19423572.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.9465648651123046,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.004693627357483,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.04776117205619812,
+      "mean_token_accuracy": 0.9813261866569519,
+      "num_tokens": 19612465.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.9448004186153411,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.2913438081741333,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.04807930588722229,
+      "mean_token_accuracy": 0.9805689513683319,
+      "num_tokens": 19800482.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.9425322234630584,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1068469285964966,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.05053595900535583,
+      "mean_token_accuracy": 0.9794536828994751,
+      "num_tokens": 19989444.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.9472232103347779,
+      "epoch": 1.440107671601615,
+      "grad_norm": 0.7409384250640869,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.052136778831481934,
+      "mean_token_accuracy": 0.979197359085083,
+      "num_tokens": 20178482.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9450975477695465,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.3440778255462646,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.05258495211601257,
+      "mean_token_accuracy": 0.9798152863979339,
+      "num_tokens": 20367345.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9468007743358612,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 0.8377829194068909,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.04893603026866913,
+      "mean_token_accuracy": 0.9805733859539032,
+      "num_tokens": 20555817.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.9505827724933624,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.776889979839325,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.05367119312286377,
+      "mean_token_accuracy": 0.9779646098613739,
+      "num_tokens": 20744001.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.9516380190849304,
+      "epoch": 1.493943472409152,
+      "grad_norm": 0.8598276376724243,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.05071394443511963,
+      "mean_token_accuracy": 0.979692417383194,
+      "num_tokens": 20931715.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.9351628720760345,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3307439088821411,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.051372635364532473,
+      "mean_token_accuracy": 0.9797907948493958,
+      "num_tokens": 21120803.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.9268040716648102,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9320250749588013,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.04939974546432495,
+      "mean_token_accuracy": 0.98038170337677,
+      "num_tokens": 21309384.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9241463243961334,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.065938115119934,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.04856643378734589,
+      "mean_token_accuracy": 0.9800511121749877,
+      "num_tokens": 21498234.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.9350252687931061,
+      "epoch": 1.547779273216689,
+      "grad_norm": 0.6925032138824463,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.0512526273727417,
+      "mean_token_accuracy": 0.979562646150589,
+      "num_tokens": 21687119.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.9403733670711517,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.2069101333618164,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.049795085191726686,
+      "mean_token_accuracy": 0.9806506633758545,
+      "num_tokens": 21876094.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9370538294315338,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.05461466312408447,
+      "mean_token_accuracy": 0.9776057600975037,
+      "num_tokens": 22064836.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9517778217792511,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 0.8965149521827698,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.05289326906204224,
+      "mean_token_accuracy": 0.9782088756561279,
+      "num_tokens": 22253065.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.9372851848602295,
+      "epoch": 1.601615074024226,
+      "grad_norm": 0.7192108631134033,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.048973691463470456,
+      "mean_token_accuracy": 0.9806210815906524,
+      "num_tokens": 22441206.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.9122351944446564,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 0.8763614296913147,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.0461783230304718,
+      "mean_token_accuracy": 0.9817197680473327,
+      "num_tokens": 22629637.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.9162654876708984,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 0.9806778430938721,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.04827338755130768,
+      "mean_token_accuracy": 0.9806499361991883,
+      "num_tokens": 22818369.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.9125026524066925,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 0.9685350656509399,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.048379385471343996,
+      "mean_token_accuracy": 0.980905145406723,
+      "num_tokens": 23006863.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9119791567325592,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.8717457056045532,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.050873959064483644,
+      "mean_token_accuracy": 0.979601925611496,
+      "num_tokens": 23195735.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.9244007229804992,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.0574654340744019,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.05112813711166382,
+      "mean_token_accuracy": 0.979493772983551,
+      "num_tokens": 23383862.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9225810945034028,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 0.9628807306289673,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.050480544567108154,
+      "mean_token_accuracy": 0.9797678411006927,
+      "num_tokens": 23572292.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.9287579476833343,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.015265941619873,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.05133126378059387,
+      "mean_token_accuracy": 0.9797986805438995,
+      "num_tokens": 23760252.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.9133845925331116,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 0.8996883630752563,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.05044950246810913,
+      "mean_token_accuracy": 0.9795303404331207,
+      "num_tokens": 23949031.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.9213698863983154,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.0490264892578125,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.04797698855400086,
+      "mean_token_accuracy": 0.9806023716926575,
+      "num_tokens": 24137829.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.9173697710037232,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.3648769855499268,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.049194514751434326,
+      "mean_token_accuracy": 0.980712479352951,
+      "num_tokens": 24326586.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9242741882801055,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.105125069618225,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.048667973279953,
+      "mean_token_accuracy": 0.9801449298858642,
+      "num_tokens": 24514904.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.9162627995014191,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.8887938857078552,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.04878672957420349,
+      "mean_token_accuracy": 0.980469423532486,
+      "num_tokens": 24703688.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.9149538338184356,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 0.8838976621627808,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.04892318844795227,
+      "mean_token_accuracy": 0.9808776795864105,
+      "num_tokens": 24892107.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.914735221862793,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1152070760726929,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.04825109839439392,
+      "mean_token_accuracy": 0.9804604113101959,
+      "num_tokens": 25080130.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9170725226402283,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9479517340660095,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.051445144414901736,
+      "mean_token_accuracy": 0.9792819261550904,
+      "num_tokens": 25268652.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9142911911010743,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 0.8172292709350586,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.0475692093372345,
+      "mean_token_accuracy": 0.9809505581855774,
+      "num_tokens": 25457120.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.9019249439239502,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 0.8183121681213379,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.04979957342147827,
+      "mean_token_accuracy": 0.9799730658531189,
+      "num_tokens": 25645699.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.9181066155433655,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.8256579041481018,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.04996164441108704,
+      "mean_token_accuracy": 0.9802247405052185,
+      "num_tokens": 25834183.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9171363770961761,
+      "epoch": 1.857335127860027,
+      "grad_norm": 0.9727098345756531,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.048660767078399655,
+      "mean_token_accuracy": 0.9805064260959625,
+      "num_tokens": 26023317.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.9112720847129822,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 0.9826673865318298,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.04775593280792236,
+      "mean_token_accuracy": 0.9809929549694061,
+      "num_tokens": 26212115.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9034272134304047,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.9951306581497192,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.04939360618591308,
+      "mean_token_accuracy": 0.9800347864627839,
+      "num_tokens": 26401167.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.897865754365921,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.0163404941558838,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05002856254577637,
+      "mean_token_accuracy": 0.9793690323829651,
+      "num_tokens": 26590570.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9004576802253723,
+      "epoch": 1.911170928667564,
+      "grad_norm": 0.6666616201400757,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.04905453026294708,
+      "mean_token_accuracy": 0.9803856253623963,
+      "num_tokens": 26779481.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.8978902101516724,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.9295416474342346,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.05261261463165283,
+      "mean_token_accuracy": 0.9785708487033844,
+      "num_tokens": 26968262.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.8962675571441651,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 0.88262939453125,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.04765265882015228,
+      "mean_token_accuracy": 0.98052077293396,
+      "num_tokens": 27157631.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8909463047981262,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.4205769300460815,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.049406200647354126,
+      "mean_token_accuracy": 0.979697072505951,
+      "num_tokens": 27345921.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8969364166259766,
+      "epoch": 1.965006729475101,
+      "grad_norm": 0.9517145752906799,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0510346531867981,
+      "mean_token_accuracy": 0.9793386399745941,
+      "num_tokens": 27534751.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.9092479169368743,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.200056791305542,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.04974203705787659,
+      "mean_token_accuracy": 0.9800146698951722,
+      "num_tokens": 27723763.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9059641897678375,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.072447657585144,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.05065792202949524,
+      "mean_token_accuracy": 0.9799223959445953,
+      "num_tokens": 27911784.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9083398975384464,
+      "eval_loss": 0.05432562157511711,
+      "eval_mean_token_accuracy": 0.9782073607869969,
+      "eval_num_tokens": 28025286.0,
+      "eval_runtime": 24.4304,
+      "eval_samples_per_second": 204.663,
+      "eval_steps_per_second": 6.426,
+      "step": 1486
+    },
+    {
+      "entropy": 0.9113237977027893,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.0167063474655151,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.044942992925643924,
+      "mean_token_accuracy": 0.9818171083927154,
+      "num_tokens": 28100906.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.8780008792877197,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.9552241563796997,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.036248764395713805,
+      "mean_token_accuracy": 0.9852245450019836,
+      "num_tokens": 28289445.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.8721176147460937,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 0.8951388001441956,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.037987279891967776,
+      "mean_token_accuracy": 0.9850614190101623,
+      "num_tokens": 28477972.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.8922657191753387,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 0.9028105735778809,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.03588914275169373,
+      "mean_token_accuracy": 0.9863102614879609,
+      "num_tokens": 28666272.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.877799642086029,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.153886318206787,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.03577309250831604,
+      "mean_token_accuracy": 0.9863148391246795,
+      "num_tokens": 28854780.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.8708971381187439,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.042955994606018,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.038266432285308835,
+      "mean_token_accuracy": 0.9848419308662415,
+      "num_tokens": 29043741.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.8800721943378449,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 0.6789060235023499,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.032948991656303404,
+      "mean_token_accuracy": 0.9871293902397156,
+      "num_tokens": 29232719.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.8743580460548401,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.1780879497528076,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.036874374747276305,
+      "mean_token_accuracy": 0.9859182178974152,
+      "num_tokens": 29421817.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.8851303637027741,
+      "epoch": 2.113055181695828,
+      "grad_norm": 0.747734785079956,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.0364631175994873,
+      "mean_token_accuracy": 0.9862085223197937,
+      "num_tokens": 29610344.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.8798732101917267,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.5001860857009888,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.03813003897666931,
+      "mean_token_accuracy": 0.9852604746818543,
+      "num_tokens": 29798997.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.8797551989555359,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 0.9593478441238403,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.0394927829504013,
+      "mean_token_accuracy": 0.984604275226593,
+      "num_tokens": 29987809.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.8860546290874481,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.8347703218460083,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.036699697375297546,
+      "mean_token_accuracy": 0.985380882024765,
+      "num_tokens": 30176234.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.8783667802810669,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1154481172561646,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.03756999969482422,
+      "mean_token_accuracy": 0.9853451430797577,
+      "num_tokens": 30364657.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.8684478521347045,
+      "epoch": 2.180349932705249,
+      "grad_norm": 0.7515475153923035,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.0382376104593277,
+      "mean_token_accuracy": 0.9854029655456543,
+      "num_tokens": 30552904.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.8636964917182922,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 0.7711939215660095,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.036220991611480714,
+      "mean_token_accuracy": 0.9860713243484497,
+      "num_tokens": 30741259.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.8592016279697419,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.1143887042999268,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.036316031217575075,
+      "mean_token_accuracy": 0.985757052898407,
+      "num_tokens": 30930144.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.8671079576015472,
+      "epoch": 2.220726783310902,
+      "grad_norm": 0.8980015516281128,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.03822658061981201,
+      "mean_token_accuracy": 0.9847764372825623,
+      "num_tokens": 31118624.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.8785548269748688,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.094323992729187,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.03736622333526611,
+      "mean_token_accuracy": 0.9850081980228425,
+      "num_tokens": 31306833.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.866977310180664,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8372092843055725,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.0401554524898529,
+      "mean_token_accuracy": 0.9840337932109833,
+      "num_tokens": 31495503.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.8787918210029602,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.5408164262771606,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0368030846118927,
+      "mean_token_accuracy": 0.9847234487533569,
+      "num_tokens": 31683865.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.872721153497696,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.1115421056747437,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.037859299778938295,
+      "mean_token_accuracy": 0.9851646661758423,
+      "num_tokens": 31871903.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.8774094223976135,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.0301331281661987,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.038671016693115234,
+      "mean_token_accuracy": 0.9840058028697968,
+      "num_tokens": 32060500.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.889606237411499,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8601903915405273,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.039603835344314574,
+      "mean_token_accuracy": 0.9840235590934754,
+      "num_tokens": 32248744.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.886734277009964,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.1029839515686035,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.039513933658599856,
+      "mean_token_accuracy": 0.9844573020935059,
+      "num_tokens": 32437640.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.8843017637729644,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.2545154094696045,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.036970189213752745,
+      "mean_token_accuracy": 0.985239815711975,
+      "num_tokens": 32625986.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.8913422048091888,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.9613803625106812,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.03811657428741455,
+      "mean_token_accuracy": 0.9852804243564606,
+      "num_tokens": 32814531.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.8907467782497406,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.1811398267745972,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.03903660774230957,
+      "mean_token_accuracy": 0.9842524945735931,
+      "num_tokens": 33002960.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.8833664715290069,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.8338477611541748,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.03982087969779968,
+      "mean_token_accuracy": 0.9841201484203339,
+      "num_tokens": 33191183.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.8783826470375061,
+      "epoch": 2.382234185733513,
+      "grad_norm": 0.6904510259628296,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.03699290752410889,
+      "mean_token_accuracy": 0.9854120731353759,
+      "num_tokens": 33379798.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.8871048510074615,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.009539246559143,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.03926805555820465,
+      "mean_token_accuracy": 0.9843246698379516,
+      "num_tokens": 33568531.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.8973462700843811,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.8916401267051697,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.039596831798553465,
+      "mean_token_accuracy": 0.9839228212833404,
+      "num_tokens": 33756333.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.8954619467258453,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.056504487991333,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0396859347820282,
+      "mean_token_accuracy": 0.9837161123752594,
+      "num_tokens": 33944904.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.8952643811702728,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0442084074020386,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.037324142456054685,
+      "mean_token_accuracy": 0.9852897703647614,
+      "num_tokens": 34133530.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.8874911010265351,
+      "epoch": 2.449528936742934,
+      "grad_norm": 0.9926224946975708,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.03637495338916778,
+      "mean_token_accuracy": 0.9853426337242126,
+      "num_tokens": 34322042.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.8900584518909455,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.001734733581543,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04029585719108582,
+      "mean_token_accuracy": 0.9842372059822082,
+      "num_tokens": 34510259.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.8986345052719116,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 0.772941529750824,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.0401084691286087,
+      "mean_token_accuracy": 0.9841427087783814,
+      "num_tokens": 34698669.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.8911370873451233,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.895031213760376,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.03941032290458679,
+      "mean_token_accuracy": 0.9841966688632965,
+      "num_tokens": 34887160.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.8901963472366333,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0677918195724487,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.03503885865211487,
+      "mean_token_accuracy": 0.9856241464614868,
+      "num_tokens": 35075891.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.8817630231380462,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.1200164556503296,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.03878425657749176,
+      "mean_token_accuracy": 0.9846492648124695,
+      "num_tokens": 35264455.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.8897477686405182,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.8058190941810608,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.041558006405830385,
+      "mean_token_accuracy": 0.9830883860588073,
+      "num_tokens": 35453251.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.8873181045055389,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.7765992283821106,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.03555050492286682,
+      "mean_token_accuracy": 0.9858225345611572,
+      "num_tokens": 35641959.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.8782606959342957,
+      "epoch": 2.557200538358008,
+      "grad_norm": 0.8488866686820984,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.03606923818588257,
+      "mean_token_accuracy": 0.9858632445335388,
+      "num_tokens": 35829636.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.8869808673858642,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.1734739542007446,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.040956351161003116,
+      "mean_token_accuracy": 0.9835374176502227,
+      "num_tokens": 36017818.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.8850924432277679,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.3584387302398682,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.042437011003494264,
+      "mean_token_accuracy": 0.9826479077339172,
+      "num_tokens": 36206505.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.8801687896251679,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2862240076065063,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.03880060315132141,
+      "mean_token_accuracy": 0.9843404710292816,
+      "num_tokens": 36395802.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.8752781748771667,
+      "epoch": 2.611036339165545,
+      "grad_norm": 0.7341249585151672,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.03776344656944275,
+      "mean_token_accuracy": 0.9852877616882324,
+      "num_tokens": 36584370.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.8759494423866272,
+      "epoch": 2.624495289367429,
+      "grad_norm": 0.9807276129722595,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.03780297338962555,
+      "mean_token_accuracy": 0.9853227376937866,
+      "num_tokens": 36772865.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.8753129065036773,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.1391628980636597,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.03953765034675598,
+      "mean_token_accuracy": 0.9839197635650635,
+      "num_tokens": 36961462.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.8771899223327637,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.0865391492843628,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.03856399655342102,
+      "mean_token_accuracy": 0.9848688066005706,
+      "num_tokens": 37150486.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.8774881184101104,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.8561595678329468,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.03795175850391388,
+      "mean_token_accuracy": 0.9849309325218201,
+      "num_tokens": 37338890.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.8841517448425293,
+      "epoch": 2.678331090174966,
+      "grad_norm": 0.9458399415016174,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.03860213160514832,
+      "mean_token_accuracy": 0.9842230796813964,
+      "num_tokens": 37526886.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.8809043228626251,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 0.9360871315002441,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.03709094822406769,
+      "mean_token_accuracy": 0.9855118036270142,
+      "num_tokens": 37715408.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.8805519282817841,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.8852124214172363,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.03869341611862183,
+      "mean_token_accuracy": 0.9846724212169647,
+      "num_tokens": 37904328.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.8796228408813477,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.189297080039978,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.03675769567489624,
+      "mean_token_accuracy": 0.9854080140590668,
+      "num_tokens": 38092776.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.8722020089626312,
+      "epoch": 2.732166890982503,
+      "grad_norm": 0.7027947306632996,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.03889244794845581,
+      "mean_token_accuracy": 0.984558242559433,
+      "num_tokens": 38282221.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.8746632814407349,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 0.9799376726150513,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.03815680146217346,
+      "mean_token_accuracy": 0.9845012128353119,
+      "num_tokens": 38470633.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.8766567528247833,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 0.850459098815918,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.039029371738433835,
+      "mean_token_accuracy": 0.9845906794071198,
+      "num_tokens": 38659048.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.8778688013553619,
+      "epoch": 2.772543741588156,
+      "grad_norm": 0.8540134429931641,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.039645448327064514,
+      "mean_token_accuracy": 0.9840566098690033,
+      "num_tokens": 38847370.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.8758600771427154,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8036021590232849,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.038226932287216187,
+      "mean_token_accuracy": 0.9845647215843201,
+      "num_tokens": 39036040.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.8652268946170807,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 0.8526738286018372,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.03798363208770752,
+      "mean_token_accuracy": 0.9845202267169952,
+      "num_tokens": 39224695.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.8707200348377228,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 0.9852389097213745,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.037227436900138855,
+      "mean_token_accuracy": 0.9852050960063934,
+      "num_tokens": 39413360.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.8711447060108185,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.228277325630188,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.03698050379753113,
+      "mean_token_accuracy": 0.9852443158626556,
+      "num_tokens": 39602544.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.8742912471294403,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 0.8440544605255127,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.03757670521736145,
+      "mean_token_accuracy": 0.9848618268966675,
+      "num_tokens": 39791340.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.8801726162433624,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0955252647399902,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.040004149079322815,
+      "mean_token_accuracy": 0.9841345012187958,
+      "num_tokens": 39980212.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.8931492984294891,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.5722652673721313,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.038096648454666135,
+      "mean_token_accuracy": 0.9849136590957641,
+      "num_tokens": 40169122.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.8767493844032288,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.7663730382919312,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.03729096055030823,
+      "mean_token_accuracy": 0.985169780254364,
+      "num_tokens": 40357678.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.8770561516284943,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.1505672931671143,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.03984796404838562,
+      "mean_token_accuracy": 0.9838845670223236,
+      "num_tokens": 40546892.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.8789129912853241,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 0.8355761766433716,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.03769223690032959,
+      "mean_token_accuracy": 0.9847559213638306,
+      "num_tokens": 40736109.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.8781779289245606,
+      "epoch": 2.920592193808883,
+      "grad_norm": 0.9103918075561523,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.038262826204299924,
+      "mean_token_accuracy": 0.984741848707199,
+      "num_tokens": 40924341.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.8552807629108429,
+      "epoch": 2.934051144010767,
+      "grad_norm": 0.8338510990142822,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.0355743408203125,
+      "mean_token_accuracy": 0.9855036079883576,
+      "num_tokens": 41113849.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.8705591142177582,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 0.8334057927131653,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.037476515769958495,
+      "mean_token_accuracy": 0.9854551255702972,
+      "num_tokens": 41303130.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.8809274792671203,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 0.9424103498458862,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.03624279499053955,
+      "mean_token_accuracy": 0.9856135487556458,
+      "num_tokens": 41491750.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.8818133771419525,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.04243004322052,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.039132320880889894,
+      "mean_token_accuracy": 0.984746390581131,
+      "num_tokens": 41680493.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.8838395297527313,
+      "epoch": 2.987886944818304,
+      "grad_norm": 0.8650784492492676,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.03822052478790283,
+      "mean_token_accuracy": 0.9852108955383301,
+      "num_tokens": 41868378.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.8800126549544608,
+      "eval_loss": 0.05468416586518288,
+      "eval_mean_token_accuracy": 0.9781015617832257,
+      "eval_num_tokens": 42038037.0,
+      "eval_runtime": 24.4825,
+      "eval_samples_per_second": 204.228,
+      "eval_steps_per_second": 6.413,
+      "step": 2229
+    },
+    {
+      "entropy": 0.8773060619831086,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.7845072150230408,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.0365300714969635,
+      "mean_token_accuracy": 0.985853761434555,
+      "num_tokens": 42056717.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.8720550239086151,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 0.9917275905609131,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.030071893334388734,
+      "mean_token_accuracy": 0.9890478909015655,
+      "num_tokens": 42245058.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.8489451467990875,
+      "epoch": 3.028263795423957,
+      "grad_norm": 0.9414294362068176,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.027054613828659056,
+      "mean_token_accuracy": 0.9894806027412415,
+      "num_tokens": 42433468.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.848755830526352,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.7276235818862915,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.026897600293159483,
+      "mean_token_accuracy": 0.9893308103084564,
+      "num_tokens": 42621657.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.8402929604053497,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 0.9455838799476624,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.027918827533721925,
+      "mean_token_accuracy": 0.9888385117053986,
+      "num_tokens": 42810533.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.8434606611728668,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 0.7503429651260376,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.025769710540771484,
+      "mean_token_accuracy": 0.989881956577301,
+      "num_tokens": 42999085.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.8459890782833099,
+      "epoch": 3.082099596231494,
+      "grad_norm": 0.736440896987915,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.028781753778457642,
+      "mean_token_accuracy": 0.9882656812667847,
+      "num_tokens": 43188222.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.8493306159973144,
+      "epoch": 3.095558546433378,
+      "grad_norm": 0.9577816128730774,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.02689719796180725,
+      "mean_token_accuracy": 0.9889744818210602,
+      "num_tokens": 43376648.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.8512665331363678,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 0.9631588459014893,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.031439167261123654,
+      "mean_token_accuracy": 0.9874942421913147,
+      "num_tokens": 43564769.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.8519513130187988,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1011922359466553,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.02696958780288696,
+      "mean_token_accuracy": 0.9900951981544495,
+      "num_tokens": 43753362.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.8507618844509125,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9351623058319092,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.026315575838088988,
+      "mean_token_accuracy": 0.989511913061142,
+      "num_tokens": 43942063.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.8506478369235992,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.096564769744873,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.029049152135849,
+      "mean_token_accuracy": 0.9886228442192078,
+      "num_tokens": 44130381.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.8624941289424897,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 0.7859402298927307,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.027744990587234498,
+      "mean_token_accuracy": 0.9890527307987214,
+      "num_tokens": 44319243.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.8627866744995117,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 0.7764605283737183,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.02571650445461273,
+      "mean_token_accuracy": 0.9903442323207855,
+      "num_tokens": 44507698.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.8566473722457886,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.8182396292686462,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.025625643134117127,
+      "mean_token_accuracy": 0.9897150516510009,
+      "num_tokens": 44695954.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.8555088877677918,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.0277098417282104,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.028235083818435668,
+      "mean_token_accuracy": 0.9894341349601745,
+      "num_tokens": 44884558.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.8562702238559723,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.0676493644714355,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.027713948488235475,
+      "mean_token_accuracy": 0.9891053259372711,
+      "num_tokens": 45073381.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.8520576059818268,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 0.7926117181777954,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.028936928510665892,
+      "mean_token_accuracy": 0.988807988166809,
+      "num_tokens": 45262350.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.860496312379837,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.1108510494232178,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030533093214035033,
+      "mean_token_accuracy": 0.9882748067378998,
+      "num_tokens": 45450782.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.872324675321579,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.3057194948196411,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.032421305775642395,
+      "mean_token_accuracy": 0.9879753351211548,
+      "num_tokens": 45639122.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.8778823137283325,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.289680004119873,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03096393346786499,
+      "mean_token_accuracy": 0.9883370697498322,
+      "num_tokens": 45827730.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.8627406179904937,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.8361398577690125,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.02957139313220978,
+      "mean_token_accuracy": 0.9884674072265625,
+      "num_tokens": 46016274.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.8584554970264435,
+      "epoch": 3.297442799461642,
+      "grad_norm": 0.9330723881721497,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.029568272829055785,
+      "mean_token_accuracy": 0.9882686555385589,
+      "num_tokens": 46204679.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.8515178024768829,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.0296216011047363,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.027788180112838744,
+      "mean_token_accuracy": 0.9888734877109527,
+      "num_tokens": 46393989.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.8488850116729736,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1329699754714966,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.028057461977005003,
+      "mean_token_accuracy": 0.9895590543746948,
+      "num_tokens": 46582790.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.849625188112259,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 0.9267219305038452,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.02802973985671997,
+      "mean_token_accuracy": 0.988851535320282,
+      "num_tokens": 46771138.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.854446280002594,
+      "epoch": 3.351278600269179,
+      "grad_norm": 0.8713932037353516,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.030220019817352294,
+      "mean_token_accuracy": 0.9882843673229218,
+      "num_tokens": 46960054.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.8599806547164917,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.675019383430481,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.02958865463733673,
+      "mean_token_accuracy": 0.9880296885967255,
+      "num_tokens": 47149567.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.8549504458904267,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.2176353931427002,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.027722162008285523,
+      "mean_token_accuracy": 0.9890090465545655,
+      "num_tokens": 47338060.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.8552233815193176,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9571717977523804,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.028924965858459474,
+      "mean_token_accuracy": 0.9884298503398895,
+      "num_tokens": 47525895.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.8521630108356476,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.965713620185852,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.030511701107025148,
+      "mean_token_accuracy": 0.9881398677825928,
+      "num_tokens": 47714524.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.8537400126457214,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 0.9005307555198669,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.030432754755020143,
+      "mean_token_accuracy": 0.9876708447933197,
+      "num_tokens": 47903188.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.8596058666706086,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 0.8159075379371643,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.027927806973457335,
+      "mean_token_accuracy": 0.9889210939407349,
+      "num_tokens": 48091334.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.8482735753059387,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0411661863327026,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03003154993057251,
+      "mean_token_accuracy": 0.9882730603218078,
+      "num_tokens": 48279622.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.8466426789760589,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.189595341682434,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03237656056880951,
+      "mean_token_accuracy": 0.9869072616100312,
+      "num_tokens": 48467527.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.8545588493347168,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0700056552886963,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.0306060791015625,
+      "mean_token_accuracy": 0.9880664825439454,
+      "num_tokens": 48656151.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.8503433585166931,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.7099478244781494,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.02984013557434082,
+      "mean_token_accuracy": 0.9879442811012268,
+      "num_tokens": 48844796.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.8464376091957092,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9877128005027771,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.0306446373462677,
+      "mean_token_accuracy": 0.9883591532707214,
+      "num_tokens": 49033216.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.8477609276771545,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.3324531316757202,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.030148410797119142,
+      "mean_token_accuracy": 0.9879989802837372,
+      "num_tokens": 49221425.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.8461863219738006,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8938570022583008,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.02682304084300995,
+      "mean_token_accuracy": 0.9897881150245667,
+      "num_tokens": 49409785.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.8362689018249512,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 0.7423873543739319,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.02689332664012909,
+      "mean_token_accuracy": 0.9899903059005737,
+      "num_tokens": 49598524.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.8395139217376709,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.921755850315094,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.030062124133110046,
+      "mean_token_accuracy": 0.9880835592746735,
+      "num_tokens": 49787241.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.8467226803302765,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0296459197998047,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.030666223168373107,
+      "mean_token_accuracy": 0.9877098739147187,
+      "num_tokens": 49975844.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.833392071723938,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2861948013305664,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.028174835443496703,
+      "mean_token_accuracy": 0.989056545495987,
+      "num_tokens": 50164857.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.8340730309486389,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 0.8116776347160339,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.028390336036682128,
+      "mean_token_accuracy": 0.9886897742748261,
+      "num_tokens": 50353452.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.8357116639614105,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.8362084031105042,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.029806244373321533,
+      "mean_token_accuracy": 0.988309508562088,
+      "num_tokens": 50542461.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.8363270878791809,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8438499569892883,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.028577303886413573,
+      "mean_token_accuracy": 0.9886945366859436,
+      "num_tokens": 50730872.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.8310261785984039,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.1920002698898315,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.030321845412254335,
+      "mean_token_accuracy": 0.9878708899021149,
+      "num_tokens": 50919889.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.8380252480506897,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 0.9409399032592773,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.030501589179039,
+      "mean_token_accuracy": 0.9872083842754364,
+      "num_tokens": 51108079.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.8422612130641938,
+      "epoch": 3.660834454912517,
+      "grad_norm": 0.8370440006256104,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.02974823713302612,
+      "mean_token_accuracy": 0.9885554671287536,
+      "num_tokens": 51296913.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.8387595832347869,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1360642910003662,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.026970452070236205,
+      "mean_token_accuracy": 0.9888331353664398,
+      "num_tokens": 51485678.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.8403031289577484,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.1195677518844604,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.029451662302017213,
+      "mean_token_accuracy": 0.988847154378891,
+      "num_tokens": 51674765.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.8499106168746948,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 0.9573526978492737,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.030826866626739502,
+      "mean_token_accuracy": 0.9875578820705414,
+      "num_tokens": 51863924.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.8576709866523743,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9386286735534668,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.02946900427341461,
+      "mean_token_accuracy": 0.9885939955711365,
+      "num_tokens": 52052011.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.8404816091060638,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.095536708831787,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.030070826411247253,
+      "mean_token_accuracy": 0.9881400763988495,
+      "num_tokens": 52240955.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.8346689164638519,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9964426159858704,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03005717396736145,
+      "mean_token_accuracy": 0.9883057296276092,
+      "num_tokens": 52429544.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.8377436757087707,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.8499778509140015,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.029283815622329713,
+      "mean_token_accuracy": 0.9882387280464172,
+      "num_tokens": 52618405.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.8328788638114929,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9594390988349915,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.02666555345058441,
+      "mean_token_accuracy": 0.9889323055744171,
+      "num_tokens": 52806689.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.829415249824524,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.3587099313735962,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.028349629044532774,
+      "mean_token_accuracy": 0.9890907645225525,
+      "num_tokens": 52995248.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.8335795283317566,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.077789306640625,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.03001731038093567,
+      "mean_token_accuracy": 0.9878358781337738,
+      "num_tokens": 53183607.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.8316770553588867,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 0.7265951037406921,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.030136841535568237,
+      "mean_token_accuracy": 0.9880174934864044,
+      "num_tokens": 53372735.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.8310399711132049,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.6605978012084961,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.026665833592414857,
+      "mean_token_accuracy": 0.9892704486846924,
+      "num_tokens": 53561776.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.8258654296398162,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.6560797691345215,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.026718440651893615,
+      "mean_token_accuracy": 0.989847868680954,
+      "num_tokens": 53750019.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.8221413195133209,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 0.7495356798171997,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.027431756258010864,
+      "mean_token_accuracy": 0.9892430484294892,
+      "num_tokens": 53938914.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.8341928541660308,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.0676971673965454,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.028432759642601012,
+      "mean_token_accuracy": 0.9889024317264556,
+      "num_tokens": 54127763.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.8359149813652038,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.380476713180542,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.028149163722991942,
+      "mean_token_accuracy": 0.9884953200817108,
+      "num_tokens": 54315930.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.8294597804546356,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0728843212127686,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.028733691573143004,
+      "mean_token_accuracy": 0.9884523034095765,
+      "num_tokens": 54504632.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.8245523810386658,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 0.9723307490348816,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.02793261408805847,
+      "mean_token_accuracy": 0.9891038477420807,
+      "num_tokens": 54693752.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.8279620766639709,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.003624439239502,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.028367668390274048,
+      "mean_token_accuracy": 0.9893603324890137,
+      "num_tokens": 54882030.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.8306584179401397,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.8481361865997314,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.029952478408813477,
+      "mean_token_accuracy": 0.9881332099437714,
+      "num_tokens": 55070549.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.8316469848155975,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.7631049156188965,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.029351180791854857,
+      "mean_token_accuracy": 0.9888184070587158,
+      "num_tokens": 55258999.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.8221287310123444,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 0.7234840393066406,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.027709537744522096,
+      "mean_token_accuracy": 0.9890308439731598,
+      "num_tokens": 55448047.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.8239192545413971,
+      "epoch": 3.970390309555855,
+      "grad_norm": 0.8497020602226257,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.029604345560073853,
+      "mean_token_accuracy": 0.9881375730037689,
+      "num_tokens": 55636566.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.8319927275180816,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.1599552631378174,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.02763860821723938,
+      "mean_token_accuracy": 0.9891406774520874,
+      "num_tokens": 55825049.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.8404937744140625,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 0.906755805015564,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.029578033089637756,
+      "mean_token_accuracy": 0.9878996431827545,
+      "num_tokens": 56012965.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.8419649528849656,
+      "eval_loss": 0.0570383295416832,
+      "eval_mean_token_accuracy": 0.9787763862093543,
+      "eval_num_tokens": 56050518.0,
+      "eval_runtime": 24.4889,
+      "eval_samples_per_second": 204.174,
+      "eval_steps_per_second": 6.411,
+      "step": 2972
+    },
+    {
+      "entropy": 0.8328965544700623,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7073294520378113,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.020916813611984254,
+      "mean_token_accuracy": 0.992138022184372,
+      "num_tokens": 56201621.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.8153974890708924,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.8904889822006226,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.01942949742078781,
+      "mean_token_accuracy": 0.9927055358886718,
+      "num_tokens": 56390018.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.8080023109912873,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.0828146934509277,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.018959003686904907,
+      "mean_token_accuracy": 0.9928978025913239,
+      "num_tokens": 56578046.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.8056950688362121,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7407417893409729,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019250378012657166,
+      "mean_token_accuracy": 0.9926685452461242,
+      "num_tokens": 56767192.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.8124530673027038,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.0355021953582764,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.01904640197753906,
+      "mean_token_accuracy": 0.9925767481327057,
+      "num_tokens": 56955769.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.810802960395813,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.3759289979934692,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.020995599031448365,
+      "mean_token_accuracy": 0.9917719483375549,
+      "num_tokens": 57144300.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.8159192740917206,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.112034559249878,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.01948906183242798,
+      "mean_token_accuracy": 0.9928658187389374,
+      "num_tokens": 57332935.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.821906328201294,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.0319583415985107,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019549430906772615,
+      "mean_token_accuracy": 0.9922966659069061,
+      "num_tokens": 57520775.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.8133022606372833,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.067180871963501,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.01833198368549347,
+      "mean_token_accuracy": 0.9934999167919158,
+      "num_tokens": 57709495.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.8150732755661011,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.277079701423645,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.019589337706565856,
+      "mean_token_accuracy": 0.992507028579712,
+      "num_tokens": 57897461.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.8159323990345001,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.3150911331176758,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.01963164806365967,
+      "mean_token_accuracy": 0.9926789402961731,
+      "num_tokens": 58085915.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.8147098064422608,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.0073580741882324,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.02144385576248169,
+      "mean_token_accuracy": 0.9918489575386047,
+      "num_tokens": 58274583.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.8155419886112213,
+      "epoch": 4.172274562584119,
+      "grad_norm": 0.8892306685447693,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.02215111702680588,
+      "mean_token_accuracy": 0.9914537072181702,
+      "num_tokens": 58463241.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.8084070324897766,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.7887238264083862,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.018581822514533997,
+      "mean_token_accuracy": 0.9928648293018341,
+      "num_tokens": 58652281.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.8016656517982483,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.0644817352294922,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.020099574327468873,
+      "mean_token_accuracy": 0.9916590929031373,
+      "num_tokens": 58839946.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.8035425901412964,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.9588403701782227,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020384626090526582,
+      "mean_token_accuracy": 0.9918951213359832,
+      "num_tokens": 59028994.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.8114414095878602,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.7145556211471558,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.019365760684013366,
+      "mean_token_accuracy": 0.9921130836009979,
+      "num_tokens": 59217599.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.8123053193092347,
+      "epoch": 4.23956931359354,
+      "grad_norm": 0.7928217649459839,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.020542070269584656,
+      "mean_token_accuracy": 0.9918384432792664,
+      "num_tokens": 59406176.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.810414057970047,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.1470942497253418,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.01809942126274109,
+      "mean_token_accuracy": 0.992941427230835,
+      "num_tokens": 59594604.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.8046662211418152,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.579035222530365,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.01914280205965042,
+      "mean_token_accuracy": 0.9926800608634949,
+      "num_tokens": 59782947.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.8034039616584778,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.8979132175445557,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.019897907972335815,
+      "mean_token_accuracy": 0.9919832170009613,
+      "num_tokens": 59971417.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.8137863457202912,
+      "epoch": 4.293405114401077,
+      "grad_norm": 0.9149890542030334,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.02069481760263443,
+      "mean_token_accuracy": 0.992353367805481,
+      "num_tokens": 60159983.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.8163237690925598,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.8264126181602478,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.021351543068885804,
+      "mean_token_accuracy": 0.9921438097953796,
+      "num_tokens": 60348327.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.8133926391601562,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.0638184547424316,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.020092280209064485,
+      "mean_token_accuracy": 0.9920917391777039,
+      "num_tokens": 60537152.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.8095248222351075,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.8831470012664795,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.01908399909734726,
+      "mean_token_accuracy": 0.9929656088352203,
+      "num_tokens": 60725603.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.8069245994091034,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8486410975456238,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.01925535798072815,
+      "mean_token_accuracy": 0.9927195310592651,
+      "num_tokens": 60914021.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.8006627321243286,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.9021628499031067,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.019187870621681213,
+      "mean_token_accuracy": 0.992531019449234,
+      "num_tokens": 61102037.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.8067023754119873,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.7793697118759155,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.021760080754756928,
+      "mean_token_accuracy": 0.9915579319000244,
+      "num_tokens": 61290132.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.8075471341609954,
+      "epoch": 4.387617765814267,
+      "grad_norm": 0.736788809299469,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.02077012062072754,
+      "mean_token_accuracy": 0.9920295119285584,
+      "num_tokens": 61478897.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.8083206176757812,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.1687020063400269,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.02117353230714798,
+      "mean_token_accuracy": 0.991925036907196,
+      "num_tokens": 61667982.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.81106236577034,
+      "epoch": 4.414535666218035,
+      "grad_norm": 0.7596265077590942,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.02081204652786255,
+      "mean_token_accuracy": 0.9921064436435699,
+      "num_tokens": 61856557.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.8018859803676606,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.823410153388977,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.019047826528549194,
+      "mean_token_accuracy": 0.9924882829189301,
+      "num_tokens": 62045181.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7996662855148315,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.0748025178909302,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.01788569688796997,
+      "mean_token_accuracy": 0.9931309640407562,
+      "num_tokens": 62234215.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.8021627008914948,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9946282505989075,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021408762037754058,
+      "mean_token_accuracy": 0.9914931178092956,
+      "num_tokens": 62422361.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.8084165751934052,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.7313793301582336,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.020281489193439483,
+      "mean_token_accuracy": 0.9925725042819977,
+      "num_tokens": 62610667.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.8103892624378204,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.9093077182769775,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.018877537548542024,
+      "mean_token_accuracy": 0.9929530620574951,
+      "num_tokens": 62798970.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.8091431736946106,
+      "epoch": 4.495289367429341,
+      "grad_norm": 0.952234148979187,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021039776504039764,
+      "mean_token_accuracy": 0.9918699383735656,
+      "num_tokens": 62986914.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.8094858765602112,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.9959521293640137,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021489059925079344,
+      "mean_token_accuracy": 0.9914597749710083,
+      "num_tokens": 63175805.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.807330971956253,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.6189294457435608,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.02164270281791687,
+      "mean_token_accuracy": 0.9920661866664886,
+      "num_tokens": 63364692.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.8098866164684295,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7895700931549072,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.021856357157230378,
+      "mean_token_accuracy": 0.9916301846504212,
+      "num_tokens": 63553692.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.8143714487552642,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0242317914962769,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02046577036380768,
+      "mean_token_accuracy": 0.9923330962657928,
+      "num_tokens": 63742476.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.8123241424560547,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.4186221361160278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.020659247040748598,
+      "mean_token_accuracy": 0.9918934762477875,
+      "num_tokens": 63930898.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.8133015930652618,
+      "epoch": 4.576043068640646,
+      "grad_norm": 0.9085373878479004,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.021746012568473815,
+      "mean_token_accuracy": 0.9918161749839782,
+      "num_tokens": 64119589.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.8173148334026337,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.7257112264633179,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.019787387549877168,
+      "mean_token_accuracy": 0.9922669887542724,
+      "num_tokens": 64308293.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.8116482973098755,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7442086935043335,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.020933127403259276,
+      "mean_token_accuracy": 0.9917319416999817,
+      "num_tokens": 64497362.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.8134189546108246,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.7617318630218506,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.01842810958623886,
+      "mean_token_accuracy": 0.9927939116954804,
+      "num_tokens": 64685802.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.8112196505069733,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.7430832386016846,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.01809561103582382,
+      "mean_token_accuracy": 0.9930340230464936,
+      "num_tokens": 64873973.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.8128187835216523,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.7553606629371643,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.0202948659658432,
+      "mean_token_accuracy": 0.9919280350208283,
+      "num_tokens": 65062510.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.813840126991272,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.7390825152397156,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.020557042956352235,
+      "mean_token_accuracy": 0.9918098270893096,
+      "num_tokens": 65250940.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.8117553532123566,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.7698133587837219,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.020231780409812928,
+      "mean_token_accuracy": 0.9920670032501221,
+      "num_tokens": 65439696.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.8110962271690368,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.8562760353088379,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.018769541382789613,
+      "mean_token_accuracy": 0.9925244867801666,
+      "num_tokens": 65628357.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.8113800704479217,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.2076038122177124,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020604749023914338,
+      "mean_token_accuracy": 0.9921375870704651,
+      "num_tokens": 65817139.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.8204582631587982,
+      "epoch": 4.710632570659489,
+      "grad_norm": 0.9150722026824951,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021981674432754516,
+      "mean_token_accuracy": 0.9915678203105927,
+      "num_tokens": 66005911.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.8241343259811401,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.9803491830825806,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02007504999637604,
+      "mean_token_accuracy": 0.992132717370987,
+      "num_tokens": 66194469.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.8122783243656159,
+      "epoch": 4.737550471063257,
+      "grad_norm": 0.967640221118927,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.019903984665870667,
+      "mean_token_accuracy": 0.9924214482307434,
+      "num_tokens": 66383927.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.8145494401454926,
+      "epoch": 4.751009421265142,
+      "grad_norm": 0.7893037796020508,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.019587111473083497,
+      "mean_token_accuracy": 0.9921624064445496,
+      "num_tokens": 66572212.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.8090193390846252,
+      "epoch": 4.764468371467026,
+      "grad_norm": 0.9659109711647034,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.021097327768802642,
+      "mean_token_accuracy": 0.9917473196983337,
+      "num_tokens": 66760377.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.802634209394455,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.0175755023956299,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.02113145440816879,
+      "mean_token_accuracy": 0.991830724477768,
+      "num_tokens": 66949210.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.8030233085155487,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.7780183553695679,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.019016166031360627,
+      "mean_token_accuracy": 0.9926414608955383,
+      "num_tokens": 67138031.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.8019147455692291,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7863438129425049,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.018219922482967377,
+      "mean_token_accuracy": 0.9928844273090363,
+      "num_tokens": 67327268.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7949212431907654,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9292862415313721,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.02136329412460327,
+      "mean_token_accuracy": 0.9918141901493073,
+      "num_tokens": 67515765.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.8002764582633972,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9770768284797668,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.02056926190853119,
+      "mean_token_accuracy": 0.9918984293937683,
+      "num_tokens": 67704975.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.808483463525772,
+      "epoch": 4.845222072678331,
+      "grad_norm": 0.6890367269515991,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020158609747886656,
+      "mean_token_accuracy": 0.9925975620746612,
+      "num_tokens": 67893850.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.8144356846809387,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.8613544702529907,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.019614605605602263,
+      "mean_token_accuracy": 0.9928383588790893,
+      "num_tokens": 68082435.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.8083879172801971,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.7532686591148376,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.019553859531879426,
+      "mean_token_accuracy": 0.9924478828907013,
+      "num_tokens": 68270495.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.8038046479225158,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.3251928091049194,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.02081710398197174,
+      "mean_token_accuracy": 0.9919814527034759,
+      "num_tokens": 68459223.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.8058662831783294,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8125224113464355,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.01858963668346405,
+      "mean_token_accuracy": 0.9922726869583129,
+      "num_tokens": 68647306.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.805506020784378,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8629401922225952,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020777507126331328,
+      "mean_token_accuracy": 0.9918497979640961,
+      "num_tokens": 68836386.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.8137464344501495,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.9837846159934998,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.02024296224117279,
+      "mean_token_accuracy": 0.9921825706958771,
+      "num_tokens": 69025192.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.8139736235141755,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.8701289296150208,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.01932244449853897,
+      "mean_token_accuracy": 0.9926821112632751,
+      "num_tokens": 69213427.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.8088622391223907,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.7041723132133484,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.018706569075584413,
+      "mean_token_accuracy": 0.9927357733249664,
+      "num_tokens": 69401690.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.8001897990703583,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1029852628707886,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.020011243224143983,
+      "mean_token_accuracy": 0.9923900067806244,
+      "num_tokens": 69590659.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.8071568071842193,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.2395373582839966,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.02379731684923172,
+      "mean_token_accuracy": 0.9907991290092468,
+      "num_tokens": 69779864.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.8139646112918854,
+      "epoch": 4.993270524899058,
+      "grad_norm": 0.9219555854797363,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.021635296940803527,
+      "mean_token_accuracy": 0.991401469707489,
+      "num_tokens": 69968639.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.822659221804066,
+      "eval_loss": 0.060467079281806946,
+      "eval_mean_token_accuracy": 0.9792145984188007,
+      "eval_num_tokens": 70063064.0,
+      "eval_runtime": 24.4326,
+      "eval_samples_per_second": 204.645,
+      "eval_steps_per_second": 6.426,
+      "step": 3715
+    },
+    {
+      "entropy": 0.8150339186191559,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8010120391845703,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01673099249601364,
+      "mean_token_accuracy": 0.9935125410556793,
+      "num_tokens": 70157396.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.8029226005077362,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9923233389854431,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01266745626926422,
+      "mean_token_accuracy": 0.9954398214817047,
+      "num_tokens": 70346010.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.789790153503418,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.7180858254432678,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.012603424489498138,
+      "mean_token_accuracy": 0.9949092507362366,
+      "num_tokens": 70534896.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.7863301753997802,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.6865439414978027,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.013341715931892395,
+      "mean_token_accuracy": 0.9951623439788818,
+      "num_tokens": 70723644.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.7841233730316162,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.8624011278152466,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01214413121342659,
+      "mean_token_accuracy": 0.9950926601886749,
+      "num_tokens": 70912490.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.7781193614006042,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.1462656259536743,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.013635106384754181,
+      "mean_token_accuracy": 0.9946888983249664,
+      "num_tokens": 71101186.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.7773554801940918,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.7154580354690552,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.012310966104269027,
+      "mean_token_accuracy": 0.9952048182487487,
+      "num_tokens": 71290177.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.784615957736969,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8214139342308044,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.012708047032356262,
+      "mean_token_accuracy": 0.9956556618213653,
+      "num_tokens": 71478243.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.7814814388751984,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0974094867706299,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.011942464113235473,
+      "mean_token_accuracy": 0.9958715498447418,
+      "num_tokens": 71667417.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.7749826848506928,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.7582180500030518,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.011887111514806748,
+      "mean_token_accuracy": 0.9954704821109772,
+      "num_tokens": 71856322.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.7695000052452088,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.9159554839134216,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.013456562161445617,
+      "mean_token_accuracy": 0.994843590259552,
+      "num_tokens": 72044716.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.7674728155136108,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.0967031717300415,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.012064316868782043,
+      "mean_token_accuracy": 0.9953838646411896,
+      "num_tokens": 72233243.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.766239732503891,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.756477415561676,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012840582430362702,
+      "mean_token_accuracy": 0.9951915740966797,
+      "num_tokens": 72421751.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.7667239785194397,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.8078455924987793,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.012580010294914245,
+      "mean_token_accuracy": 0.9955588281154633,
+      "num_tokens": 72610886.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.7759605884552002,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.9473005533218384,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.01442105770111084,
+      "mean_token_accuracy": 0.9946676790714264,
+      "num_tokens": 72799213.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.7739267945289612,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.4796932637691498,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011830046027898788,
+      "mean_token_accuracy": 0.9954923093318939,
+      "num_tokens": 72988413.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.7756406426429748,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7290598154067993,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.014412418007850647,
+      "mean_token_accuracy": 0.9946037054061889,
+      "num_tokens": 73177134.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.7763995885848999,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.0562312602996826,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.015396332740783692,
+      "mean_token_accuracy": 0.9945933580398559,
+      "num_tokens": 73365622.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.7737024486064911,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1071312427520752,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01451488733291626,
+      "mean_token_accuracy": 0.9945858538150787,
+      "num_tokens": 73554381.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.779985535144806,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.7758586406707764,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.013801637291908263,
+      "mean_token_accuracy": 0.9947169423103333,
+      "num_tokens": 73743028.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.7786811470985413,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.6511496305465698,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012189272791147232,
+      "mean_token_accuracy": 0.9954657673835754,
+      "num_tokens": 73931263.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.7736978232860565,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.8152093291282654,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.013696780800819397,
+      "mean_token_accuracy": 0.9948886811733246,
+      "num_tokens": 74119992.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.7730314910411835,
+      "epoch": 5.302826379542395,
+      "grad_norm": 0.6703830361366272,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012700904905796052,
+      "mean_token_accuracy": 0.9950747072696686,
+      "num_tokens": 74308428.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.7740058362483978,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.9845885634422302,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012894082069396972,
+      "mean_token_accuracy": 0.9953160643577575,
+      "num_tokens": 74497030.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.7738905131816864,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.732532799243927,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.011950771510601043,
+      "mean_token_accuracy": 0.9955996692180633,
+      "num_tokens": 74685631.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.7788872420787811,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7301356196403503,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012340743839740754,
+      "mean_token_accuracy": 0.9952776432037354,
+      "num_tokens": 74874025.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.7771204471588135,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.7570232152938843,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.012937523424625397,
+      "mean_token_accuracy": 0.9949343740940094,
+      "num_tokens": 75062636.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.7747974276542664,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8237525820732117,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.012423938512802124,
+      "mean_token_accuracy": 0.9953977644443512,
+      "num_tokens": 75251339.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.7692180812358856,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.8497490286827087,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012855434417724609,
+      "mean_token_accuracy": 0.9948123812675476,
+      "num_tokens": 75440121.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.773431783914566,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.6758515238761902,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.012878020107746125,
+      "mean_token_accuracy": 0.9955350995063782,
+      "num_tokens": 75629143.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.7735035121440887,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7578167915344238,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.013059911131858826,
+      "mean_token_accuracy": 0.994978415966034,
+      "num_tokens": 75817425.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.7725034832954407,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.0449564456939697,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.013163739442825317,
+      "mean_token_accuracy": 0.9953307807445526,
+      "num_tokens": 76005910.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.7736131310462951,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.881845235824585,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012109111249446868,
+      "mean_token_accuracy": 0.9952919363975525,
+      "num_tokens": 76194639.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.7797342419624329,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.8100120425224304,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.013262654840946197,
+      "mean_token_accuracy": 0.9948459386825561,
+      "num_tokens": 76383240.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.7794113874435424,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.83738112449646,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.013258925080299378,
+      "mean_token_accuracy": 0.9947907269001007,
+      "num_tokens": 76572040.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.7787538826465606,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0567374229431152,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.014641955494880676,
+      "mean_token_accuracy": 0.9945057690143585,
+      "num_tokens": 76761030.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.7894418001174927,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.835543692111969,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.013065731525421143,
+      "mean_token_accuracy": 0.9949834287166596,
+      "num_tokens": 76948469.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.7823423445224762,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.8142625093460083,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.01229657381772995,
+      "mean_token_accuracy": 0.9957204461097717,
+      "num_tokens": 77136968.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.7786863029003144,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.749609649181366,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.01328940987586975,
+      "mean_token_accuracy": 0.9949924945831299,
+      "num_tokens": 77325328.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.7780588805675507,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.787598729133606,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.013933436572551727,
+      "mean_token_accuracy": 0.9949735462665558,
+      "num_tokens": 77513716.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.7779332935810089,
+      "epoch": 5.545087483176312,
+      "grad_norm": 1.0315855741500854,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.012871211767196656,
+      "mean_token_accuracy": 0.995112019777298,
+      "num_tokens": 77701621.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.7754440903663635,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.8191165924072266,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.012908256053924561,
+      "mean_token_accuracy": 0.9950248181819916,
+      "num_tokens": 77889886.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.7811384737491608,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.832539975643158,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.013047756254673004,
+      "mean_token_accuracy": 0.9946948051452636,
+      "num_tokens": 78078640.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.7825228571891785,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.9072359800338745,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.012361775338649749,
+      "mean_token_accuracy": 0.9954402267932891,
+      "num_tokens": 78267842.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.7825945496559144,
+      "epoch": 5.598923283983849,
+      "grad_norm": 0.8300487995147705,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.013246089220046997,
+      "mean_token_accuracy": 0.9951650142669678,
+      "num_tokens": 78455995.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.7857775866985321,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.9343852996826172,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.012612038850784301,
+      "mean_token_accuracy": 0.995340085029602,
+      "num_tokens": 78643617.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.775356650352478,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7109665274620056,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.013411080837249756,
+      "mean_token_accuracy": 0.9951648890972138,
+      "num_tokens": 78833085.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.7750323057174683,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.6436781883239746,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.012064981460571288,
+      "mean_token_accuracy": 0.9954153180122376,
+      "num_tokens": 79021569.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.7753469109535217,
+      "epoch": 5.652759084791386,
+      "grad_norm": 1.0361640453338623,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.013953472673892974,
+      "mean_token_accuracy": 0.9947281837463379,
+      "num_tokens": 79210573.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.7821656167507172,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9078817367553711,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012946712970733642,
+      "mean_token_accuracy": 0.9954073607921601,
+      "num_tokens": 79398723.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.7803877890110016,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.7879899740219116,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.012198399752378464,
+      "mean_token_accuracy": 0.9953389108181,
+      "num_tokens": 79587809.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.7772530496120453,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.0390578508377075,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.012395218014717102,
+      "mean_token_accuracy": 0.9953405141830445,
+      "num_tokens": 79775768.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.7784459292888641,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.7382562160491943,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.012304402142763137,
+      "mean_token_accuracy": 0.9953492462635041,
+      "num_tokens": 79964602.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.7821883440017701,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.5720117092132568,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.011424948275089265,
+      "mean_token_accuracy": 0.9956757485866546,
+      "num_tokens": 80153320.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.7811889886856079,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.7762079238891602,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012460066378116608,
+      "mean_token_accuracy": 0.9950654745101929,
+      "num_tokens": 80341381.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.7771898031234741,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.9219444394111633,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.011629728972911835,
+      "mean_token_accuracy": 0.9955675959587097,
+      "num_tokens": 80530323.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.7825480580329895,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.7189850211143494,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.011180496215820313,
+      "mean_token_accuracy": 0.9960062265396118,
+      "num_tokens": 80718533.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.7795504450798034,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.7256530523300171,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.012952378392219544,
+      "mean_token_accuracy": 0.9951896607875824,
+      "num_tokens": 80907391.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.779454517364502,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.808525800704956,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.0121039479970932,
+      "mean_token_accuracy": 0.9955627739429473,
+      "num_tokens": 81096396.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.7813481986522675,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.7294234037399292,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01356595754623413,
+      "mean_token_accuracy": 0.9951751470565796,
+      "num_tokens": 81284943.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.7833737671375275,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.771756649017334,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.013437020778656005,
+      "mean_token_accuracy": 0.9951942205429077,
+      "num_tokens": 81473810.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.781111603975296,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.5440986752510071,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.012187350541353226,
+      "mean_token_accuracy": 0.9956019043922424,
+      "num_tokens": 81662193.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.7777030646800995,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7532716393470764,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.013766974210739136,
+      "mean_token_accuracy": 0.994896411895752,
+      "num_tokens": 81851045.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.7776691019535065,
+      "epoch": 5.85464333781965,
+      "grad_norm": 1.1882649660110474,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.012588076293468475,
+      "mean_token_accuracy": 0.9952153980731964,
+      "num_tokens": 82039225.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.7795583724975585,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.7453442215919495,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.012100882083177566,
+      "mean_token_accuracy": 0.9954553186893463,
+      "num_tokens": 82228123.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.7763878583908081,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.6757420301437378,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011943653970956803,
+      "mean_token_accuracy": 0.9953660726547241,
+      "num_tokens": 82417093.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.7747587621212005,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.9607213735580444,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.014138388633728027,
+      "mean_token_accuracy": 0.9944287598133087,
+      "num_tokens": 82605461.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.7782076597213745,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.6733895540237427,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.012490443140268325,
+      "mean_token_accuracy": 0.9950622916221619,
+      "num_tokens": 82794124.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.7843947112560272,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.7626368403434753,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.012137357890605927,
+      "mean_token_accuracy": 0.9952504158020019,
+      "num_tokens": 82982258.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.7826779782772064,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8149229288101196,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012872529029846192,
+      "mean_token_accuracy": 0.9949809432029724,
+      "num_tokens": 83170520.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.7792184889316559,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.7774028778076172,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012353222817182541,
+      "mean_token_accuracy": 0.9953234374523163,
+      "num_tokens": 83359174.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.7778255641460419,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7626445293426514,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.01137089878320694,
+      "mean_token_accuracy": 0.9958219468593598,
+      "num_tokens": 83547583.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.7765573263168335,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9130274653434753,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012439383566379547,
+      "mean_token_accuracy": 0.9954963147640228,
+      "num_tokens": 83735904.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.7734197914600373,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.7289780378341675,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011430246382951736,
+      "mean_token_accuracy": 0.9957478463649749,
+      "num_tokens": 83924971.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.7852135965019275,
+      "eval_loss": 0.06990383565425873,
+      "eval_mean_token_accuracy": 0.9796095274056599,
+      "eval_num_tokens": 84075763.0,
+      "eval_runtime": 24.3947,
+      "eval_samples_per_second": 204.963,
+      "eval_steps_per_second": 6.436,
+      "step": 4458
+    },
+    {
+      "entropy": 0.7758591413497925,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.9430185556411743,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.011816921830177306,
+      "mean_token_accuracy": 0.9959785401821136,
+      "num_tokens": 84113330.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.773297393321991,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.6593970060348511,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.007801702618598938,
+      "mean_token_accuracy": 0.9972260951995849,
+      "num_tokens": 84302325.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.7716216742992401,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.5422326922416687,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.00867043137550354,
+      "mean_token_accuracy": 0.9968674182891846,
+      "num_tokens": 84490227.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.7699320554733277,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.8308367729187012,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.007848290354013443,
+      "mean_token_accuracy": 0.9971986234188079,
+      "num_tokens": 84678824.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.7688778102397918,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.5276612639427185,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.007070200890302658,
+      "mean_token_accuracy": 0.9974652588367462,
+      "num_tokens": 84867430.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.7687382102012634,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.7081453800201416,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.00793309137225151,
+      "mean_token_accuracy": 0.9972716808319092,
+      "num_tokens": 85056072.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.7705117404460907,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.9106748104095459,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.007362194359302521,
+      "mean_token_accuracy": 0.9975170493125916,
+      "num_tokens": 85244337.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.7697769284248352,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.7364314198493958,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.007369447499513626,
+      "mean_token_accuracy": 0.9974394500255584,
+      "num_tokens": 85432878.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.7661255121231079,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.6747480630874634,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.007502703368663788,
+      "mean_token_accuracy": 0.9973482489585876,
+      "num_tokens": 85621800.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.7673385441303253,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.8179678916931152,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.007111805677413941,
+      "mean_token_accuracy": 0.9977720499038696,
+      "num_tokens": 85809796.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.76323401927948,
+      "epoch": 6.137281292059219,
+      "grad_norm": 0.5177297592163086,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.007196265459060669,
+      "mean_token_accuracy": 0.9976384699344635,
+      "num_tokens": 85998543.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.7640918910503387,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 0.8970023989677429,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.008229555189609527,
+      "mean_token_accuracy": 0.9967760264873504,
+      "num_tokens": 86186537.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.7605979859828949,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.7019157409667969,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.007770314812660217,
+      "mean_token_accuracy": 0.997019499540329,
+      "num_tokens": 86375405.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.7647112905979156,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.6519957184791565,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.007326644659042358,
+      "mean_token_accuracy": 0.9972577571868897,
+      "num_tokens": 86562963.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.7608903884887696,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.9307657480239868,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.00786140188574791,
+      "mean_token_accuracy": 0.9970912158489227,
+      "num_tokens": 86751006.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.7573343932628631,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 0.6093463897705078,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.007400572299957275,
+      "mean_token_accuracy": 0.997343909740448,
+      "num_tokens": 86939232.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.756312346458435,
+      "epoch": 6.218034993270525,
+      "grad_norm": 0.6412308812141418,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.007471667230129242,
+      "mean_token_accuracy": 0.9973339200019836,
+      "num_tokens": 87128294.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.7593119204044342,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.6830183863639832,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.007831692695617676,
+      "mean_token_accuracy": 0.9972396552562713,
+      "num_tokens": 87316833.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.7598747551441193,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.6647830009460449,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.0070374816656112674,
+      "mean_token_accuracy": 0.9975631892681122,
+      "num_tokens": 87505022.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.7626749157905579,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.5470786690711975,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.007449843734502792,
+      "mean_token_accuracy": 0.9976118981838227,
+      "num_tokens": 87692953.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.7584024250507355,
+      "epoch": 6.271870794078062,
+      "grad_norm": 0.8641819953918457,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.007355903089046478,
+      "mean_token_accuracy": 0.99752836227417,
+      "num_tokens": 87882513.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.7607118487358093,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.6385810971260071,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.007216785848140716,
+      "mean_token_accuracy": 0.9973138213157654,
+      "num_tokens": 88071161.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.7600765109062195,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.3984440565109253,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.007562516629695893,
+      "mean_token_accuracy": 0.9973795533180236,
+      "num_tokens": 88259867.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.7609024405479431,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.7098219990730286,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.007748734951019287,
+      "mean_token_accuracy": 0.9970796704292297,
+      "num_tokens": 88448504.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.7602188348770141,
+      "epoch": 6.325706594885599,
+      "grad_norm": 0.3834342658519745,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.00794660747051239,
+      "mean_token_accuracy": 0.9971937775611878,
+      "num_tokens": 88637123.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.7595065951347351,
+      "epoch": 6.339165545087483,
+      "grad_norm": 0.7073737978935242,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.007514684647321701,
+      "mean_token_accuracy": 0.9972682356834411,
+      "num_tokens": 88826107.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.7566567718982696,
+      "epoch": 6.352624495289367,
+      "grad_norm": 0.6334935426712036,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.007287996262311936,
+      "mean_token_accuracy": 0.9972865641117096,
+      "num_tokens": 89014806.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.7535678207874298,
+      "epoch": 6.366083445491252,
+      "grad_norm": 0.5461204648017883,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.007215028256177902,
+      "mean_token_accuracy": 0.9976517677307128,
+      "num_tokens": 89203610.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.753510159254074,
+      "epoch": 6.379542395693136,
+      "grad_norm": 0.8002920150756836,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.00775575190782547,
+      "mean_token_accuracy": 0.9972102761268615,
+      "num_tokens": 89392611.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.7567359745502472,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.5416199564933777,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.007165181636810303,
+      "mean_token_accuracy": 0.9973179161548614,
+      "num_tokens": 89581472.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.7603509783744812,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.7484428882598877,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.008048020303249359,
+      "mean_token_accuracy": 0.9972274541854859,
+      "num_tokens": 89769895.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.7604019463062286,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.6397516131401062,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.006713573634624481,
+      "mean_token_accuracy": 0.9974542498588562,
+      "num_tokens": 89958151.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.7562435805797577,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.6562761068344116,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.0070735208690166475,
+      "mean_token_accuracy": 0.9975251019001007,
+      "num_tokens": 90147009.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.7581055939197541,
+      "epoch": 6.446837146702557,
+      "grad_norm": 0.6705797910690308,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.008277279138565064,
+      "mean_token_accuracy": 0.9969512403011322,
+      "num_tokens": 90335135.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.7552513837814331,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.5535324215888977,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.007625260949134826,
+      "mean_token_accuracy": 0.9972123861312866,
+      "num_tokens": 90523520.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.7567071616649628,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.5841183662414551,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.008183968812227249,
+      "mean_token_accuracy": 0.9970704436302185,
+      "num_tokens": 90712012.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.7579837441444397,
+      "epoch": 6.48721399730821,
+      "grad_norm": 1.0210020542144775,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.00890980064868927,
+      "mean_token_accuracy": 0.9968561112880707,
+      "num_tokens": 90900360.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.7599723398685455,
+      "epoch": 6.500672947510094,
+      "grad_norm": 0.7007607221603394,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.007547108829021454,
+      "mean_token_accuracy": 0.9972310900688172,
+      "num_tokens": 91089219.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.7569817125797271,
+      "epoch": 6.514131897711978,
+      "grad_norm": 1.203089952468872,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.007012282311916351,
+      "mean_token_accuracy": 0.9974212229251862,
+      "num_tokens": 91278177.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.7547080159187317,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.6055266857147217,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.007885889708995819,
+      "mean_token_accuracy": 0.9971360504627228,
+      "num_tokens": 91466912.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.7584831535816192,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.6868627071380615,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.00764903798699379,
+      "mean_token_accuracy": 0.9975648581981659,
+      "num_tokens": 91655249.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.757660973072052,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.9303789734840393,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.00731472447514534,
+      "mean_token_accuracy": 0.9974016606807709,
+      "num_tokens": 91844016.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.759681248664856,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 0.6116757988929749,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.007223697006702423,
+      "mean_token_accuracy": 0.9970514178276062,
+      "num_tokens": 92032638.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.7601584196090698,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.6825643181800842,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.0068596042692661285,
+      "mean_token_accuracy": 0.9973869740962982,
+      "num_tokens": 92221897.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.7645585358142852,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.7482470870018005,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.007520741224288941,
+      "mean_token_accuracy": 0.9974736630916595,
+      "num_tokens": 92410416.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.766044944524765,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.7137084603309631,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.007919894158840179,
+      "mean_token_accuracy": 0.9969592452049255,
+      "num_tokens": 92598533.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.761748319864273,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 1.0537257194519043,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.007994732260704041,
+      "mean_token_accuracy": 0.9970260739326477,
+      "num_tokens": 92788046.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.763516879081726,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.9168201088905334,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.006822402775287628,
+      "mean_token_accuracy": 0.9973889350891113,
+      "num_tokens": 92976583.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.7603426337242126,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.6945034265518188,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.007335217297077179,
+      "mean_token_accuracy": 0.9974107563495636,
+      "num_tokens": 93164984.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.7601041436195374,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.7255296111106873,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.006898632645606995,
+      "mean_token_accuracy": 0.9974060177803039,
+      "num_tokens": 93353860.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.760407668352127,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.7785485982894897,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.006965675204992294,
+      "mean_token_accuracy": 0.997627604007721,
+      "num_tokens": 93542542.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.7584291577339173,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.676058292388916,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.0066734164953231815,
+      "mean_token_accuracy": 0.9976742446422577,
+      "num_tokens": 93730962.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.7543873727321625,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.9423731565475464,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.007540161162614823,
+      "mean_token_accuracy": 0.9972456872463227,
+      "num_tokens": 93919838.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.7540828108787536,
+      "epoch": 6.716016150740242,
+      "grad_norm": 0.60378098487854,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.006739185005426407,
+      "mean_token_accuracy": 0.9976783573627472,
+      "num_tokens": 94108612.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.7581348657608032,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.7218618392944336,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.006258270889520645,
+      "mean_token_accuracy": 0.9977961003780365,
+      "num_tokens": 94296515.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.7555042028427124,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.6111772060394287,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.006771801412105561,
+      "mean_token_accuracy": 0.9975680649280548,
+      "num_tokens": 94484849.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.7563546478748322,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.7171721458435059,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.007536987960338593,
+      "mean_token_accuracy": 0.9971135079860687,
+      "num_tokens": 94673363.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.7541037023067474,
+      "epoch": 6.769851951547779,
+      "grad_norm": 0.45519423484802246,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.007405019551515579,
+      "mean_token_accuracy": 0.9975123167037964,
+      "num_tokens": 94862699.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.7588851511478424,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.7041094899177551,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.007134091854095459,
+      "mean_token_accuracy": 0.9972242057323456,
+      "num_tokens": 95050741.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.7570755481719971,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.6298620700836182,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.0070040211081504825,
+      "mean_token_accuracy": 0.997326648235321,
+      "num_tokens": 95239293.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.7548963069915772,
+      "epoch": 6.810228802153432,
+      "grad_norm": 1.0688964128494263,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.006342679262161255,
+      "mean_token_accuracy": 0.9976243853569031,
+      "num_tokens": 95428279.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.7548417568206787,
+      "epoch": 6.823687752355316,
+      "grad_norm": 0.4953668415546417,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.0069470912218093876,
+      "mean_token_accuracy": 0.9973559200763702,
+      "num_tokens": 95616835.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.7551373422145844,
+      "epoch": 6.837146702557201,
+      "grad_norm": 0.5016734600067139,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.006873356550931931,
+      "mean_token_accuracy": 0.9973772704601288,
+      "num_tokens": 95805557.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.7567492067813874,
+      "epoch": 6.850605652759085,
+      "grad_norm": 0.532783567905426,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.006668326258659363,
+      "mean_token_accuracy": 0.997450977563858,
+      "num_tokens": 95993831.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.758799821138382,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.6660076379776001,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.006823312491178513,
+      "mean_token_accuracy": 0.9975283324718476,
+      "num_tokens": 96181732.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.7571039736270905,
+      "epoch": 6.877523553162853,
+      "grad_norm": 1.0330839157104492,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.007801513373851776,
+      "mean_token_accuracy": 0.9971715807914734,
+      "num_tokens": 96369920.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.7555147588253022,
+      "epoch": 6.890982503364738,
+      "grad_norm": 0.5079207420349121,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.0074800834059715274,
+      "mean_token_accuracy": 0.9973418891429902,
+      "num_tokens": 96558724.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.7542111933231354,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.678804337978363,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.00698004812002182,
+      "mean_token_accuracy": 0.997770369052887,
+      "num_tokens": 96747242.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.7508629977703094,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.8755067586898804,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.007143624126911163,
+      "mean_token_accuracy": 0.9975672841072083,
+      "num_tokens": 96936244.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.7522107064723969,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.8553352952003479,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.007587063312530518,
+      "mean_token_accuracy": 0.9970708072185517,
+      "num_tokens": 97125069.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.7501682281494141,
+      "epoch": 6.944818304172275,
+      "grad_norm": 0.614628255367279,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.007764184474945068,
+      "mean_token_accuracy": 0.9968451082706451,
+      "num_tokens": 97313879.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.7522825837135315,
+      "epoch": 6.958277254374159,
+      "grad_norm": 0.8060729503631592,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.00771535262465477,
+      "mean_token_accuracy": 0.9972144842147828,
+      "num_tokens": 97503040.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.7534927785396576,
+      "epoch": 6.971736204576043,
+      "grad_norm": 0.6604568958282471,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.006561616808176041,
+      "mean_token_accuracy": 0.9974263906478882,
+      "num_tokens": 97692024.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.7542680203914642,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.853604793548584,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.007131250202655792,
+      "mean_token_accuracy": 0.9974451243877411,
+      "num_tokens": 97881026.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.7567435801029205,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.6146188974380493,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.006486450880765915,
+      "mean_token_accuracy": 0.9974190831184387,
+      "num_tokens": 98069553.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.7571151700748759,
+      "eval_loss": 0.07285518199205399,
+      "eval_mean_token_accuracy": 0.9808630795235846,
+      "eval_num_tokens": 98088531.0,
+      "eval_runtime": 24.4094,
+      "eval_samples_per_second": 204.839,
+      "eval_steps_per_second": 6.432,
+      "step": 5201
+    },
+    {
+      "entropy": 0.751511138677597,
+      "epoch": 7.012113055181696,
+      "grad_norm": 0.49068665504455566,
+      "learning_rate": 2.2506648060191042e-05,
+      "loss": 0.0042594756931066515,
+      "mean_token_accuracy": 0.9985321044921875,
+      "num_tokens": 98258586.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.7522116243839264,
+      "epoch": 7.02557200538358,
+      "grad_norm": 0.5906213521957397,
+      "learning_rate": 2.2321031102941066e-05,
+      "loss": 0.003930336982011795,
+      "mean_token_accuracy": 0.9989193022251129,
+      "num_tokens": 98447091.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.7511299610137939,
+      "epoch": 7.039030955585464,
+      "grad_norm": 0.45251449942588806,
+      "learning_rate": 2.2135962530768055e-05,
+      "loss": 0.004373953118920327,
+      "mean_token_accuracy": 0.9986318647861481,
+      "num_tokens": 98635468.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.7481192111968994,
+      "epoch": 7.052489905787349,
+      "grad_norm": 0.6962258219718933,
+      "learning_rate": 2.1951446010313053e-05,
+      "loss": 0.00428679995238781,
+      "mean_token_accuracy": 0.9987218022346497,
+      "num_tokens": 98824365.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.7483489096164704,
+      "epoch": 7.065948855989233,
+      "grad_norm": 0.5099614858627319,
+      "learning_rate": 2.1767485197279662e-05,
+      "loss": 0.003984151408076286,
+      "mean_token_accuracy": 0.9986076414585113,
+      "num_tokens": 99013056.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.7463673830032349,
+      "epoch": 7.079407806191117,
+      "grad_norm": 0.3716561198234558,
+      "learning_rate": 2.1584083736361617e-05,
+      "loss": 0.004038792476058007,
+      "mean_token_accuracy": 0.998809176683426,
+      "num_tokens": 99201941.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.7466944694519043,
+      "epoch": 7.0928667563930015,
+      "grad_norm": 0.33926740288734436,
+      "learning_rate": 2.1401245261170582e-05,
+      "loss": 0.003776150941848755,
+      "mean_token_accuracy": 0.9987689018249511,
+      "num_tokens": 99390847.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.7465248286724091,
+      "epoch": 7.106325706594886,
+      "grad_norm": 0.3297136127948761,
+      "learning_rate": 2.121897339416416e-05,
+      "loss": 0.0037364721298217773,
+      "mean_token_accuracy": 0.9987996935844421,
+      "num_tokens": 99579720.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.7467422366142273,
+      "epoch": 7.11978465679677,
+      "grad_norm": 0.3911883234977722,
+      "learning_rate": 2.103727174657412e-05,
+      "loss": 0.004084533080458641,
+      "mean_token_accuracy": 0.9985779583454132,
+      "num_tokens": 99768263.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.7423677504062652,
+      "epoch": 7.133243606998654,
+      "grad_norm": 0.5843891501426697,
+      "learning_rate": 2.085614391833486e-05,
+      "loss": 0.0038405463099479677,
+      "mean_token_accuracy": 0.9987926363945008,
+      "num_tokens": 99957655.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.7434155166149139,
+      "epoch": 7.1467025572005385,
+      "grad_norm": 0.5154002904891968,
+      "learning_rate": 2.0675593498012052e-05,
+      "loss": 0.0036686167120933534,
+      "mean_token_accuracy": 0.9988853752613067,
+      "num_tokens": 100146048.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.7450331568717956,
+      "epoch": 7.160161507402423,
+      "grad_norm": 0.5856082439422607,
+      "learning_rate": 2.049562406273159e-05,
+      "loss": 0.00417071133852005,
+      "mean_token_accuracy": 0.9985920011997222,
+      "num_tokens": 100334064.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.7413534104824067,
+      "epoch": 7.173620457604307,
+      "grad_norm": 0.4711301028728485,
+      "learning_rate": 2.0316239178108688e-05,
+      "loss": 0.003790935128927231,
+      "mean_token_accuracy": 0.9986560821533204,
+      "num_tokens": 100522817.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.739797306060791,
+      "epoch": 7.187079407806191,
+      "grad_norm": 0.4481808841228485,
+      "learning_rate": 2.013744239817725e-05,
+      "loss": 0.0038315042853355407,
+      "mean_token_accuracy": 0.9987379729747772,
+      "num_tokens": 100711882.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.7411384165287018,
+      "epoch": 7.2005383580080755,
+      "grad_norm": 0.3373877704143524,
+      "learning_rate": 1.9959237265319452e-05,
+      "loss": 0.0034619957208633424,
+      "mean_token_accuracy": 0.9990058243274689,
+      "num_tokens": 100900649.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.7439651310443878,
+      "epoch": 7.21399730820996,
+      "grad_norm": 0.4260849952697754,
+      "learning_rate": 1.9781627310195528e-05,
+      "loss": 0.004041648656129837,
+      "mean_token_accuracy": 0.9988129734992981,
+      "num_tokens": 101088789.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.741381824016571,
+      "epoch": 7.227456258411844,
+      "grad_norm": 0.704740047454834,
+      "learning_rate": 1.960461605167389e-05,
+      "loss": 0.004109270870685577,
+      "mean_token_accuracy": 0.9987618446350097,
+      "num_tokens": 101277316.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.7418228089809418,
+      "epoch": 7.240915208613728,
+      "grad_norm": 0.34908416867256165,
+      "learning_rate": 1.9428206996761357e-05,
+      "loss": 0.0038616616278886796,
+      "mean_token_accuracy": 0.9985946893692017,
+      "num_tokens": 101466025.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.7436148345470428,
+      "epoch": 7.2543741588156125,
+      "grad_norm": 0.5490255951881409,
+      "learning_rate": 1.9252403640533678e-05,
+      "loss": 0.004948872700333595,
+      "mean_token_accuracy": 0.9981587290763855,
+      "num_tokens": 101654412.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.7419192969799042,
+      "epoch": 7.267833109017497,
+      "grad_norm": 0.5561390519142151,
+      "learning_rate": 1.9077209466066315e-05,
+      "loss": 0.0043866544961929325,
+      "mean_token_accuracy": 0.9986559212207794,
+      "num_tokens": 101843232.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.7441964089870453,
+      "epoch": 7.281292059219381,
+      "grad_norm": 0.5724263191223145,
+      "learning_rate": 1.8902627944365396e-05,
+      "loss": 0.004136815667152405,
+      "mean_token_accuracy": 0.9986151039600373,
+      "num_tokens": 102031115.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.7447238683700561,
+      "epoch": 7.294751009421265,
+      "grad_norm": 0.6418642997741699,
+      "learning_rate": 1.872866253429897e-05,
+      "loss": 0.004244513809680939,
+      "mean_token_accuracy": 0.9985204100608825,
+      "num_tokens": 102219064.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.7460219264030457,
+      "epoch": 7.30820995962315,
+      "grad_norm": 0.877717912197113,
+      "learning_rate": 1.8555316682528485e-05,
+      "loss": 0.0035314552485942842,
+      "mean_token_accuracy": 0.9990062952041626,
+      "num_tokens": 102407049.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.7446399390697479,
+      "epoch": 7.321668909825034,
+      "grad_norm": 0.8448772430419922,
+      "learning_rate": 1.8382593823440474e-05,
+      "loss": 0.004365745186805725,
+      "mean_token_accuracy": 0.9984619498252869,
+      "num_tokens": 102595411.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.7414965629577637,
+      "epoch": 7.335127860026918,
+      "grad_norm": 0.7239252328872681,
+      "learning_rate": 1.821049737907854e-05,
+      "loss": 0.0036744467914104463,
+      "mean_token_accuracy": 0.9989120185375213,
+      "num_tokens": 102784691.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.744704294204712,
+      "epoch": 7.348586810228802,
+      "grad_norm": 0.390360951423645,
+      "learning_rate": 1.803903075907554e-05,
+      "loss": 0.003917969018220902,
+      "mean_token_accuracy": 0.9988363087177277,
+      "num_tokens": 102973291.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.744241201877594,
+      "epoch": 7.362045760430687,
+      "grad_norm": 0.7771829962730408,
+      "learning_rate": 1.7868197360586054e-05,
+      "loss": 0.004070578515529633,
+      "mean_token_accuracy": 0.9985230803489685,
+      "num_tokens": 103161417.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.7423063695430756,
+      "epoch": 7.375504710632571,
+      "grad_norm": 0.3263421654701233,
+      "learning_rate": 1.769800056821904e-05,
+      "loss": 0.003923245519399643,
+      "mean_token_accuracy": 0.9985153019428253,
+      "num_tokens": 103349890.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.7458358705043793,
+      "epoch": 7.388963660834455,
+      "grad_norm": 0.5268740653991699,
+      "learning_rate": 1.7528443753970814e-05,
+      "loss": 0.004262761771678924,
+      "mean_token_accuracy": 0.9987221539020539,
+      "num_tokens": 103537726.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.7404268085956573,
+      "epoch": 7.402422611036339,
+      "grad_norm": 0.5339550971984863,
+      "learning_rate": 1.7359530277158227e-05,
+      "loss": 0.0039617355912923815,
+      "mean_token_accuracy": 0.998635333776474,
+      "num_tokens": 103726707.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.7421627819538117,
+      "epoch": 7.415881561238224,
+      "grad_norm": 0.47286465764045715,
+      "learning_rate": 1.7191263484352106e-05,
+      "loss": 0.0034935571253299712,
+      "mean_token_accuracy": 0.9989599347114563,
+      "num_tokens": 103915542.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.7416261017322541,
+      "epoch": 7.429340511440108,
+      "grad_norm": 0.29520711302757263,
+      "learning_rate": 1.702364670931097e-05,
+      "loss": 0.003899512439966202,
+      "mean_token_accuracy": 0.9986254096031189,
+      "num_tokens": 104105019.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.7447648346424103,
+      "epoch": 7.442799461641992,
+      "grad_norm": 0.46887683868408203,
+      "learning_rate": 1.685668327291495e-05,
+      "loss": 0.0036319829523563385,
+      "mean_token_accuracy": 0.9989396393299103,
+      "num_tokens": 104293657.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.7453936874866486,
+      "epoch": 7.456258411843876,
+      "grad_norm": 0.3639506995677948,
+      "learning_rate": 1.6690376483100034e-05,
+      "loss": 0.003420223295688629,
+      "mean_token_accuracy": 0.9988774538040162,
+      "num_tokens": 104482306.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.743130087852478,
+      "epoch": 7.469717362045761,
+      "grad_norm": 0.4534284770488739,
+      "learning_rate": 1.6524729634792475e-05,
+      "loss": 0.0038840845227241516,
+      "mean_token_accuracy": 0.9987022399902343,
+      "num_tokens": 104671166.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.7441355168819428,
+      "epoch": 7.483176312247645,
+      "grad_norm": 0.3704164922237396,
+      "learning_rate": 1.6359746009843567e-05,
+      "loss": 0.0034930892288684845,
+      "mean_token_accuracy": 0.9988382160663605,
+      "num_tokens": 104859133.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.7428222477436066,
+      "epoch": 7.496635262449529,
+      "grad_norm": 0.751130223274231,
+      "learning_rate": 1.6195428876964592e-05,
+      "loss": 0.003963682055473328,
+      "mean_token_accuracy": 0.9984794855117798,
+      "num_tokens": 105047492.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.7427781701087952,
+      "epoch": 7.510094212651413,
+      "grad_norm": 0.8935811519622803,
+      "learning_rate": 1.6031781491662056e-05,
+      "loss": 0.004140084609389305,
+      "mean_token_accuracy": 0.9985048770904541,
+      "num_tokens": 105235970.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.7405261933803559,
+      "epoch": 7.523553162853298,
+      "grad_norm": 0.575813353061676,
+      "learning_rate": 1.58688070961732e-05,
+      "loss": 0.003933753073215485,
+      "mean_token_accuracy": 0.998809415102005,
+      "num_tokens": 105424662.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.742167991399765,
+      "epoch": 7.537012113055182,
+      "grad_norm": 0.5800251960754395,
+      "learning_rate": 1.5706508919401804e-05,
+      "loss": 0.004490737244486809,
+      "mean_token_accuracy": 0.9987403452396393,
+      "num_tokens": 105613306.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.744090485572815,
+      "epoch": 7.550471063257066,
+      "grad_norm": 0.8697437644004822,
+      "learning_rate": 1.554489017685412e-05,
+      "loss": 0.00443139597773552,
+      "mean_token_accuracy": 0.9984764337539673,
+      "num_tokens": 105801807.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.7431838274002075,
+      "epoch": 7.56393001345895,
+      "grad_norm": 0.47921308875083923,
+      "learning_rate": 1.538395407057525e-05,
+      "loss": 0.003662409633398056,
+      "mean_token_accuracy": 0.9989210546016694,
+      "num_tokens": 105990683.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.7407984972000122,
+      "epoch": 7.577388963660835,
+      "grad_norm": 0.45125728845596313,
+      "learning_rate": 1.522370378908567e-05,
+      "loss": 0.003542381525039673,
+      "mean_token_accuracy": 0.9988591969013214,
+      "num_tokens": 106179809.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.744829797744751,
+      "epoch": 7.590847913862719,
+      "grad_norm": 0.5999196171760559,
+      "learning_rate": 1.5064142507318051e-05,
+      "loss": 0.0035081353038549424,
+      "mean_token_accuracy": 0.9988316416740417,
+      "num_tokens": 106367988.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.7428836584091186,
+      "epoch": 7.604306864064603,
+      "grad_norm": 0.4594089388847351,
+      "learning_rate": 1.4905273386554386e-05,
+      "loss": 0.004249929264187813,
+      "mean_token_accuracy": 0.9987165331840515,
+      "num_tokens": 106556378.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.7435553133487701,
+      "epoch": 7.6177658142664875,
+      "grad_norm": 0.5920020937919617,
+      "learning_rate": 1.4747099574363332e-05,
+      "loss": 0.0036187373101711274,
+      "mean_token_accuracy": 0.9987746834754944,
+      "num_tokens": 106744854.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.7437893748283386,
+      "epoch": 7.631224764468372,
+      "grad_norm": 0.5453750491142273,
+      "learning_rate": 1.4589624204537866e-05,
+      "loss": 0.00377066433429718,
+      "mean_token_accuracy": 0.9989324033260345,
+      "num_tokens": 106932944.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.7395198106765747,
+      "epoch": 7.644683714670256,
+      "grad_norm": 0.38797295093536377,
+      "learning_rate": 1.4432850397033176e-05,
+      "loss": 0.0031442392617464067,
+      "mean_token_accuracy": 0.9990855991840363,
+      "num_tokens": 107121922.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.7402484536170959,
+      "epoch": 7.65814266487214,
+      "grad_norm": 0.6739166975021362,
+      "learning_rate": 1.427678125790487e-05,
+      "loss": 0.0038538143038749696,
+      "mean_token_accuracy": 0.9986329972743988,
+      "num_tokens": 107310638.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.7401728749275207,
+      "epoch": 7.6716016150740245,
+      "grad_norm": 0.4977329969406128,
+      "learning_rate": 1.4121419879247422e-05,
+      "loss": 0.0033894732594490053,
+      "mean_token_accuracy": 0.9989957690238953,
+      "num_tokens": 107499098.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.7425866723060608,
+      "epoch": 7.685060565275909,
+      "grad_norm": 0.5719661712646484,
+      "learning_rate": 1.3966769339132929e-05,
+      "loss": 0.004793065041303635,
+      "mean_token_accuracy": 0.9987428963184357,
+      "num_tokens": 107687499.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.7395791113376617,
+      "epoch": 7.698519515477793,
+      "grad_norm": 0.4698435366153717,
+      "learning_rate": 1.3812832701550099e-05,
+      "loss": 0.004081443324685097,
+      "mean_token_accuracy": 0.9984708070755005,
+      "num_tokens": 107876895.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.7433845281600953,
+      "epoch": 7.711978465679677,
+      "grad_norm": 0.6040443778038025,
+      "learning_rate": 1.365961301634357e-05,
+      "loss": 0.003664097934961319,
+      "mean_token_accuracy": 0.9988075792789459,
+      "num_tokens": 108065071.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.7424676477909088,
+      "epoch": 7.7254374158815615,
+      "grad_norm": 0.6923211812973022,
+      "learning_rate": 1.350711331915348e-05,
+      "loss": 0.003957484290003777,
+      "mean_token_accuracy": 0.998679769039154,
+      "num_tokens": 108253186.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.7405034005641937,
+      "epoch": 7.738896366083446,
+      "grad_norm": 0.4896307587623596,
+      "learning_rate": 1.3355336631355314e-05,
+      "loss": 0.0041983149945735935,
+      "mean_token_accuracy": 0.9986901462078095,
+      "num_tokens": 108442076.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.743422681093216,
+      "epoch": 7.75235531628533,
+      "grad_norm": 0.5889957547187805,
+      "learning_rate": 1.3204285960000051e-05,
+      "loss": 0.003425312042236328,
+      "mean_token_accuracy": 0.9988837718963623,
+      "num_tokens": 108630562.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.7411604285240173,
+      "epoch": 7.765814266487214,
+      "grad_norm": 0.4655008316040039,
+      "learning_rate": 1.30539642977546e-05,
+      "loss": 0.003258592635393143,
+      "mean_token_accuracy": 0.9988660037517547,
+      "num_tokens": 108819369.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.7410345315933228,
+      "epoch": 7.7792732166890985,
+      "grad_norm": 0.7350168228149414,
+      "learning_rate": 1.2904374622842486e-05,
+      "loss": 0.0034532986581325533,
+      "mean_token_accuracy": 0.9989261090755462,
+      "num_tokens": 109008535.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.7410138070583343,
+      "epoch": 7.792732166890983,
+      "grad_norm": 0.44904911518096924,
+      "learning_rate": 1.2755519898984846e-05,
+      "loss": 0.0034178659319877625,
+      "mean_token_accuracy": 0.9988463997840882,
+      "num_tokens": 109197443.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.7402764976024627,
+      "epoch": 7.806191117092867,
+      "grad_norm": 0.5618012547492981,
+      "learning_rate": 1.2607403075341728e-05,
+      "loss": 0.003283835202455521,
+      "mean_token_accuracy": 0.9989370465278625,
+      "num_tokens": 109386097.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.7410104274749756,
+      "epoch": 7.819650067294751,
+      "grad_norm": 0.3167133629322052,
+      "learning_rate": 1.2460027086453658e-05,
+      "loss": 0.0034713000059127808,
+      "mean_token_accuracy": 0.9988771259784699,
+      "num_tokens": 109574403.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.7389123022556305,
+      "epoch": 7.833109017496636,
+      "grad_norm": 0.544057309627533,
+      "learning_rate": 1.2313394852183473e-05,
+      "loss": 0.0038273107260465623,
+      "mean_token_accuracy": 0.9986206114292144,
+      "num_tokens": 109763160.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.7385149598121643,
+      "epoch": 7.84656796769852,
+      "grad_norm": 0.8678551912307739,
+      "learning_rate": 1.2167509277658518e-05,
+      "loss": 0.0037419166415929793,
+      "mean_token_accuracy": 0.9987667620182037,
+      "num_tokens": 109951791.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.7384770929813385,
+      "epoch": 7.860026917900404,
+      "grad_norm": 0.45470893383026123,
+      "learning_rate": 1.2022373253213049e-05,
+      "loss": 0.003541013225913048,
+      "mean_token_accuracy": 0.998937851190567,
+      "num_tokens": 110140704.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.7396949470043183,
+      "epoch": 7.873485868102288,
+      "grad_norm": 0.6184161305427551,
+      "learning_rate": 1.1877989654330973e-05,
+      "loss": 0.003454260528087616,
+      "mean_token_accuracy": 0.9989149391651153,
+      "num_tokens": 110329353.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.7411868631839752,
+      "epoch": 7.886944818304173,
+      "grad_norm": 0.49721845984458923,
+      "learning_rate": 1.1734361341588912e-05,
+      "loss": 0.003363589942455292,
+      "mean_token_accuracy": 0.9988781988620759,
+      "num_tokens": 110517720.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.740048885345459,
+      "epoch": 7.900403768506057,
+      "grad_norm": 0.3169788420200348,
+      "learning_rate": 1.1591491160599488e-05,
+      "loss": 0.0030659947544336317,
+      "mean_token_accuracy": 0.9991146922111511,
+      "num_tokens": 110706131.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.7428572535514831,
+      "epoch": 7.913862718707941,
+      "grad_norm": 0.5674341320991516,
+      "learning_rate": 1.1449381941954956e-05,
+      "loss": 0.0031749047338962555,
+      "mean_token_accuracy": 0.998987466096878,
+      "num_tokens": 110893731.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.7389689266681672,
+      "epoch": 7.927321668909825,
+      "grad_norm": 0.36926963925361633,
+      "learning_rate": 1.130803650117116e-05,
+      "loss": 0.0030995842069387438,
+      "mean_token_accuracy": 0.9989600539207458,
+      "num_tokens": 111082079.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.7392809450626373,
+      "epoch": 7.94078061911171,
+      "grad_norm": 0.3086724579334259,
+      "learning_rate": 1.1167457638631656e-05,
+      "loss": 0.0033237963914871218,
+      "mean_token_accuracy": 0.9988582134246826,
+      "num_tokens": 111270413.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.7384293675422668,
+      "epoch": 7.954239569313594,
+      "grad_norm": 0.45126330852508545,
+      "learning_rate": 1.1027648139532355e-05,
+      "loss": 0.0034996122121810913,
+      "mean_token_accuracy": 0.9987758874893189,
+      "num_tokens": 111458912.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.7361682891845703,
+      "epoch": 7.967698519515478,
+      "grad_norm": 0.6252058148384094,
+      "learning_rate": 1.0888610773826253e-05,
+      "loss": 0.003475649654865265,
+      "mean_token_accuracy": 0.9988142251968384,
+      "num_tokens": 111647683.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.733948814868927,
+      "epoch": 7.981157469717362,
+      "grad_norm": 0.5224378705024719,
+      "learning_rate": 1.0750348296168583e-05,
+      "loss": 0.0030234681442379953,
+      "mean_token_accuracy": 0.9990231335163117,
+      "num_tokens": 111836870.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.736539876461029,
+      "epoch": 7.994616419919247,
+      "grad_norm": 0.2892343997955322,
+      "learning_rate": 1.0612863445862236e-05,
+      "loss": 0.0031716078519821167,
+      "mean_token_accuracy": 0.9988950371742249,
+      "num_tokens": 112025584.0,
+      "step": 5940
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.7368154024622243,
+      "eval_loss": 0.08131171762943268,
+      "eval_mean_token_accuracy": 0.9805832057242181,
+      "eval_num_tokens": 112101144.0,
+      "eval_runtime": 24.3602,
+      "eval_samples_per_second": 205.253,
+      "eval_steps_per_second": 6.445,
+      "step": 5944
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.416443645635068e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-5944/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-5944/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-5944/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-6687/README.md b/baby_talk_L16_a50/seed_42/checkpoint-6687/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-6687/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-6687/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-6687/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-6687/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-6687/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-6687/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e2c74b2c5946af28c894f0959c6ab78cc8407a73
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-6687/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ea95eb5204c194dc102be93ea943c85a540d1002ae2aa1ecd1d8a7f925b2c6e
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-6687/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-6687/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-6687/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-6687/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-6687/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-6687/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-6687/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-6687/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-6687/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-6687/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-6687/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..19278f568c07df5b5ac543e961660b38f5078db5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-6687/trainer_state.json
@@ -0,0 +1,6813 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.0,
+  "eval_steps": 500,
+  "global_step": 6687,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    },
+    {
+      "entropy": 0.9689933776855468,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9720287919044495,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.055563896894454956,
+      "mean_token_accuracy": 0.9779708027839661,
+      "num_tokens": 14144652.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.9506655633449554,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 0.6619595289230347,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.050587379932403566,
+      "mean_token_accuracy": 0.979685264825821,
+      "num_tokens": 14332714.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.9458732306957245,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 0.7905108332633972,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.05069155097007751,
+      "mean_token_accuracy": 0.9796596229076385,
+      "num_tokens": 14520902.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.9320916712284089,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 0.9210566878318787,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.051744121313095096,
+      "mean_token_accuracy": 0.9792148530483246,
+      "num_tokens": 14709215.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.928934782743454,
+      "epoch": 1.063257065948856,
+      "grad_norm": 0.7442036867141724,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.0486875057220459,
+      "mean_token_accuracy": 0.9798122465610504,
+      "num_tokens": 14898242.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.942881977558136,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.0910941362380981,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.052297019958496095,
+      "mean_token_accuracy": 0.9793200254440307,
+      "num_tokens": 15086835.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.9503917813301086,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 0.8091554641723633,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.049927744269371035,
+      "mean_token_accuracy": 0.9802082359790802,
+      "num_tokens": 15275042.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9426830470561981,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.0589245557785034,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.053803551197052005,
+      "mean_token_accuracy": 0.9785399675369263,
+      "num_tokens": 15463923.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.9445010781288147,
+      "epoch": 1.117092866756393,
+      "grad_norm": 0.8410441279411316,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.053181976079940796,
+      "mean_token_accuracy": 0.9792523324489594,
+      "num_tokens": 15652526.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.946729838848114,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 0.8246080279350281,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.054291915893554685,
+      "mean_token_accuracy": 0.9791357696056366,
+      "num_tokens": 15841340.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.9433728814125061,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 0.9200296401977539,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.05073915719985962,
+      "mean_token_accuracy": 0.9796931505203247,
+      "num_tokens": 16029478.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.9494417488574982,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 0.8342758417129517,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.050929927825927736,
+      "mean_token_accuracy": 0.9793219089508056,
+      "num_tokens": 16218014.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.9409256100654602,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.7911009192466736,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.049964362382888795,
+      "mean_token_accuracy": 0.9798487305641175,
+      "num_tokens": 16406420.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.9429958581924438,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.4877018928527832,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.04969423711299896,
+      "mean_token_accuracy": 0.9801322638988494,
+      "num_tokens": 16594948.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.9542136013507843,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 0.8030785918235779,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.04981146454811096,
+      "mean_token_accuracy": 0.9796162784099579,
+      "num_tokens": 16782983.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.9367722630500793,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.1169800758361816,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.05210963487625122,
+      "mean_token_accuracy": 0.9790139615535736,
+      "num_tokens": 16971394.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.9487208306789399,
+      "epoch": 1.224764468371467,
+      "grad_norm": 0.8650282621383667,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.05012243390083313,
+      "mean_token_accuracy": 0.9805107891559601,
+      "num_tokens": 17159951.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.9441231489181519,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 0.8193967938423157,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.048866665363311766,
+      "mean_token_accuracy": 0.9807584881782532,
+      "num_tokens": 17348711.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.9361630439758301,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3613612651824951,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.05324091911315918,
+      "mean_token_accuracy": 0.97904953956604,
+      "num_tokens": 17537116.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.9562997639179229,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.8121878504753113,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.05011019706726074,
+      "mean_token_accuracy": 0.9798731088638306,
+      "num_tokens": 17725567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.9359076261520386,
+      "epoch": 1.278600269179004,
+      "grad_norm": 0.7838549613952637,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.04834386110305786,
+      "mean_token_accuracy": 0.9804529249668121,
+      "num_tokens": 17914191.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.9527446150779724,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0981712341308594,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.055077624320983884,
+      "mean_token_accuracy": 0.9776535391807556,
+      "num_tokens": 18103184.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.9574812948703766,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 0.8291501402854919,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.050153911113739014,
+      "mean_token_accuracy": 0.9792965054512024,
+      "num_tokens": 18291351.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.939585280418396,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.0864412784576416,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.05132197737693787,
+      "mean_token_accuracy": 0.9792291462421417,
+      "num_tokens": 18480192.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.9468406856060028,
+      "epoch": 1.332436069986541,
+      "grad_norm": 0.9772785902023315,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.04844954013824463,
+      "mean_token_accuracy": 0.9804142415523529,
+      "num_tokens": 18669028.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.9413834273815155,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 0.9020094871520996,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.05075312852859497,
+      "mean_token_accuracy": 0.9790591120719909,
+      "num_tokens": 18857581.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.9526253461837768,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.0524990558624268,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.049737372994422914,
+      "mean_token_accuracy": 0.9795652389526367,
+      "num_tokens": 19046105.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.9484909176826477,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.0715620517730713,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.04935494959354401,
+      "mean_token_accuracy": 0.9797740161418915,
+      "num_tokens": 19234730.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.9398573458194732,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2134987115859985,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.0518394410610199,
+      "mean_token_accuracy": 0.9793788075447083,
+      "num_tokens": 19423572.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.9465648651123046,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.004693627357483,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.04776117205619812,
+      "mean_token_accuracy": 0.9813261866569519,
+      "num_tokens": 19612465.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.9448004186153411,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.2913438081741333,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.04807930588722229,
+      "mean_token_accuracy": 0.9805689513683319,
+      "num_tokens": 19800482.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.9425322234630584,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1068469285964966,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.05053595900535583,
+      "mean_token_accuracy": 0.9794536828994751,
+      "num_tokens": 19989444.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.9472232103347779,
+      "epoch": 1.440107671601615,
+      "grad_norm": 0.7409384250640869,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.052136778831481934,
+      "mean_token_accuracy": 0.979197359085083,
+      "num_tokens": 20178482.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9450975477695465,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.3440778255462646,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.05258495211601257,
+      "mean_token_accuracy": 0.9798152863979339,
+      "num_tokens": 20367345.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9468007743358612,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 0.8377829194068909,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.04893603026866913,
+      "mean_token_accuracy": 0.9805733859539032,
+      "num_tokens": 20555817.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.9505827724933624,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.776889979839325,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.05367119312286377,
+      "mean_token_accuracy": 0.9779646098613739,
+      "num_tokens": 20744001.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.9516380190849304,
+      "epoch": 1.493943472409152,
+      "grad_norm": 0.8598276376724243,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.05071394443511963,
+      "mean_token_accuracy": 0.979692417383194,
+      "num_tokens": 20931715.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.9351628720760345,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3307439088821411,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.051372635364532473,
+      "mean_token_accuracy": 0.9797907948493958,
+      "num_tokens": 21120803.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.9268040716648102,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9320250749588013,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.04939974546432495,
+      "mean_token_accuracy": 0.98038170337677,
+      "num_tokens": 21309384.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9241463243961334,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.065938115119934,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.04856643378734589,
+      "mean_token_accuracy": 0.9800511121749877,
+      "num_tokens": 21498234.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.9350252687931061,
+      "epoch": 1.547779273216689,
+      "grad_norm": 0.6925032138824463,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.0512526273727417,
+      "mean_token_accuracy": 0.979562646150589,
+      "num_tokens": 21687119.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.9403733670711517,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.2069101333618164,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.049795085191726686,
+      "mean_token_accuracy": 0.9806506633758545,
+      "num_tokens": 21876094.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9370538294315338,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.05461466312408447,
+      "mean_token_accuracy": 0.9776057600975037,
+      "num_tokens": 22064836.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9517778217792511,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 0.8965149521827698,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.05289326906204224,
+      "mean_token_accuracy": 0.9782088756561279,
+      "num_tokens": 22253065.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.9372851848602295,
+      "epoch": 1.601615074024226,
+      "grad_norm": 0.7192108631134033,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.048973691463470456,
+      "mean_token_accuracy": 0.9806210815906524,
+      "num_tokens": 22441206.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.9122351944446564,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 0.8763614296913147,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.0461783230304718,
+      "mean_token_accuracy": 0.9817197680473327,
+      "num_tokens": 22629637.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.9162654876708984,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 0.9806778430938721,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.04827338755130768,
+      "mean_token_accuracy": 0.9806499361991883,
+      "num_tokens": 22818369.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.9125026524066925,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 0.9685350656509399,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.048379385471343996,
+      "mean_token_accuracy": 0.980905145406723,
+      "num_tokens": 23006863.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9119791567325592,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.8717457056045532,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.050873959064483644,
+      "mean_token_accuracy": 0.979601925611496,
+      "num_tokens": 23195735.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.9244007229804992,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.0574654340744019,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.05112813711166382,
+      "mean_token_accuracy": 0.979493772983551,
+      "num_tokens": 23383862.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9225810945034028,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 0.9628807306289673,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.050480544567108154,
+      "mean_token_accuracy": 0.9797678411006927,
+      "num_tokens": 23572292.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.9287579476833343,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.015265941619873,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.05133126378059387,
+      "mean_token_accuracy": 0.9797986805438995,
+      "num_tokens": 23760252.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.9133845925331116,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 0.8996883630752563,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.05044950246810913,
+      "mean_token_accuracy": 0.9795303404331207,
+      "num_tokens": 23949031.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.9213698863983154,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.0490264892578125,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.04797698855400086,
+      "mean_token_accuracy": 0.9806023716926575,
+      "num_tokens": 24137829.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.9173697710037232,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.3648769855499268,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.049194514751434326,
+      "mean_token_accuracy": 0.980712479352951,
+      "num_tokens": 24326586.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9242741882801055,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.105125069618225,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.048667973279953,
+      "mean_token_accuracy": 0.9801449298858642,
+      "num_tokens": 24514904.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.9162627995014191,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.8887938857078552,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.04878672957420349,
+      "mean_token_accuracy": 0.980469423532486,
+      "num_tokens": 24703688.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.9149538338184356,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 0.8838976621627808,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.04892318844795227,
+      "mean_token_accuracy": 0.9808776795864105,
+      "num_tokens": 24892107.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.914735221862793,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1152070760726929,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.04825109839439392,
+      "mean_token_accuracy": 0.9804604113101959,
+      "num_tokens": 25080130.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9170725226402283,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9479517340660095,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.051445144414901736,
+      "mean_token_accuracy": 0.9792819261550904,
+      "num_tokens": 25268652.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9142911911010743,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 0.8172292709350586,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.0475692093372345,
+      "mean_token_accuracy": 0.9809505581855774,
+      "num_tokens": 25457120.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.9019249439239502,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 0.8183121681213379,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.04979957342147827,
+      "mean_token_accuracy": 0.9799730658531189,
+      "num_tokens": 25645699.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.9181066155433655,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.8256579041481018,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.04996164441108704,
+      "mean_token_accuracy": 0.9802247405052185,
+      "num_tokens": 25834183.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9171363770961761,
+      "epoch": 1.857335127860027,
+      "grad_norm": 0.9727098345756531,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.048660767078399655,
+      "mean_token_accuracy": 0.9805064260959625,
+      "num_tokens": 26023317.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.9112720847129822,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 0.9826673865318298,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.04775593280792236,
+      "mean_token_accuracy": 0.9809929549694061,
+      "num_tokens": 26212115.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9034272134304047,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.9951306581497192,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.04939360618591308,
+      "mean_token_accuracy": 0.9800347864627839,
+      "num_tokens": 26401167.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.897865754365921,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.0163404941558838,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05002856254577637,
+      "mean_token_accuracy": 0.9793690323829651,
+      "num_tokens": 26590570.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9004576802253723,
+      "epoch": 1.911170928667564,
+      "grad_norm": 0.6666616201400757,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.04905453026294708,
+      "mean_token_accuracy": 0.9803856253623963,
+      "num_tokens": 26779481.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.8978902101516724,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.9295416474342346,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.05261261463165283,
+      "mean_token_accuracy": 0.9785708487033844,
+      "num_tokens": 26968262.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.8962675571441651,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 0.88262939453125,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.04765265882015228,
+      "mean_token_accuracy": 0.98052077293396,
+      "num_tokens": 27157631.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8909463047981262,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.4205769300460815,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.049406200647354126,
+      "mean_token_accuracy": 0.979697072505951,
+      "num_tokens": 27345921.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8969364166259766,
+      "epoch": 1.965006729475101,
+      "grad_norm": 0.9517145752906799,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0510346531867981,
+      "mean_token_accuracy": 0.9793386399745941,
+      "num_tokens": 27534751.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.9092479169368743,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.200056791305542,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.04974203705787659,
+      "mean_token_accuracy": 0.9800146698951722,
+      "num_tokens": 27723763.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9059641897678375,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.072447657585144,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.05065792202949524,
+      "mean_token_accuracy": 0.9799223959445953,
+      "num_tokens": 27911784.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9083398975384464,
+      "eval_loss": 0.05432562157511711,
+      "eval_mean_token_accuracy": 0.9782073607869969,
+      "eval_num_tokens": 28025286.0,
+      "eval_runtime": 24.4304,
+      "eval_samples_per_second": 204.663,
+      "eval_steps_per_second": 6.426,
+      "step": 1486
+    },
+    {
+      "entropy": 0.9113237977027893,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.0167063474655151,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.044942992925643924,
+      "mean_token_accuracy": 0.9818171083927154,
+      "num_tokens": 28100906.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.8780008792877197,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.9552241563796997,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.036248764395713805,
+      "mean_token_accuracy": 0.9852245450019836,
+      "num_tokens": 28289445.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.8721176147460937,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 0.8951388001441956,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.037987279891967776,
+      "mean_token_accuracy": 0.9850614190101623,
+      "num_tokens": 28477972.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.8922657191753387,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 0.9028105735778809,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.03588914275169373,
+      "mean_token_accuracy": 0.9863102614879609,
+      "num_tokens": 28666272.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.877799642086029,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.153886318206787,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.03577309250831604,
+      "mean_token_accuracy": 0.9863148391246795,
+      "num_tokens": 28854780.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.8708971381187439,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.042955994606018,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.038266432285308835,
+      "mean_token_accuracy": 0.9848419308662415,
+      "num_tokens": 29043741.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.8800721943378449,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 0.6789060235023499,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.032948991656303404,
+      "mean_token_accuracy": 0.9871293902397156,
+      "num_tokens": 29232719.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.8743580460548401,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.1780879497528076,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.036874374747276305,
+      "mean_token_accuracy": 0.9859182178974152,
+      "num_tokens": 29421817.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.8851303637027741,
+      "epoch": 2.113055181695828,
+      "grad_norm": 0.747734785079956,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.0364631175994873,
+      "mean_token_accuracy": 0.9862085223197937,
+      "num_tokens": 29610344.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.8798732101917267,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.5001860857009888,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.03813003897666931,
+      "mean_token_accuracy": 0.9852604746818543,
+      "num_tokens": 29798997.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.8797551989555359,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 0.9593478441238403,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.0394927829504013,
+      "mean_token_accuracy": 0.984604275226593,
+      "num_tokens": 29987809.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.8860546290874481,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.8347703218460083,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.036699697375297546,
+      "mean_token_accuracy": 0.985380882024765,
+      "num_tokens": 30176234.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.8783667802810669,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1154481172561646,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.03756999969482422,
+      "mean_token_accuracy": 0.9853451430797577,
+      "num_tokens": 30364657.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.8684478521347045,
+      "epoch": 2.180349932705249,
+      "grad_norm": 0.7515475153923035,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.0382376104593277,
+      "mean_token_accuracy": 0.9854029655456543,
+      "num_tokens": 30552904.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.8636964917182922,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 0.7711939215660095,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.036220991611480714,
+      "mean_token_accuracy": 0.9860713243484497,
+      "num_tokens": 30741259.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.8592016279697419,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.1143887042999268,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.036316031217575075,
+      "mean_token_accuracy": 0.985757052898407,
+      "num_tokens": 30930144.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.8671079576015472,
+      "epoch": 2.220726783310902,
+      "grad_norm": 0.8980015516281128,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.03822658061981201,
+      "mean_token_accuracy": 0.9847764372825623,
+      "num_tokens": 31118624.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.8785548269748688,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.094323992729187,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.03736622333526611,
+      "mean_token_accuracy": 0.9850081980228425,
+      "num_tokens": 31306833.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.866977310180664,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8372092843055725,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.0401554524898529,
+      "mean_token_accuracy": 0.9840337932109833,
+      "num_tokens": 31495503.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.8787918210029602,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.5408164262771606,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0368030846118927,
+      "mean_token_accuracy": 0.9847234487533569,
+      "num_tokens": 31683865.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.872721153497696,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.1115421056747437,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.037859299778938295,
+      "mean_token_accuracy": 0.9851646661758423,
+      "num_tokens": 31871903.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.8774094223976135,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.0301331281661987,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.038671016693115234,
+      "mean_token_accuracy": 0.9840058028697968,
+      "num_tokens": 32060500.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.889606237411499,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8601903915405273,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.039603835344314574,
+      "mean_token_accuracy": 0.9840235590934754,
+      "num_tokens": 32248744.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.886734277009964,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.1029839515686035,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.039513933658599856,
+      "mean_token_accuracy": 0.9844573020935059,
+      "num_tokens": 32437640.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.8843017637729644,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.2545154094696045,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.036970189213752745,
+      "mean_token_accuracy": 0.985239815711975,
+      "num_tokens": 32625986.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.8913422048091888,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.9613803625106812,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.03811657428741455,
+      "mean_token_accuracy": 0.9852804243564606,
+      "num_tokens": 32814531.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.8907467782497406,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.1811398267745972,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.03903660774230957,
+      "mean_token_accuracy": 0.9842524945735931,
+      "num_tokens": 33002960.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.8833664715290069,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.8338477611541748,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.03982087969779968,
+      "mean_token_accuracy": 0.9841201484203339,
+      "num_tokens": 33191183.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.8783826470375061,
+      "epoch": 2.382234185733513,
+      "grad_norm": 0.6904510259628296,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.03699290752410889,
+      "mean_token_accuracy": 0.9854120731353759,
+      "num_tokens": 33379798.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.8871048510074615,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.009539246559143,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.03926805555820465,
+      "mean_token_accuracy": 0.9843246698379516,
+      "num_tokens": 33568531.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.8973462700843811,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.8916401267051697,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.039596831798553465,
+      "mean_token_accuracy": 0.9839228212833404,
+      "num_tokens": 33756333.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.8954619467258453,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.056504487991333,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0396859347820282,
+      "mean_token_accuracy": 0.9837161123752594,
+      "num_tokens": 33944904.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.8952643811702728,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0442084074020386,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.037324142456054685,
+      "mean_token_accuracy": 0.9852897703647614,
+      "num_tokens": 34133530.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.8874911010265351,
+      "epoch": 2.449528936742934,
+      "grad_norm": 0.9926224946975708,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.03637495338916778,
+      "mean_token_accuracy": 0.9853426337242126,
+      "num_tokens": 34322042.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.8900584518909455,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.001734733581543,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04029585719108582,
+      "mean_token_accuracy": 0.9842372059822082,
+      "num_tokens": 34510259.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.8986345052719116,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 0.772941529750824,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.0401084691286087,
+      "mean_token_accuracy": 0.9841427087783814,
+      "num_tokens": 34698669.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.8911370873451233,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.895031213760376,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.03941032290458679,
+      "mean_token_accuracy": 0.9841966688632965,
+      "num_tokens": 34887160.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.8901963472366333,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0677918195724487,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.03503885865211487,
+      "mean_token_accuracy": 0.9856241464614868,
+      "num_tokens": 35075891.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.8817630231380462,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.1200164556503296,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.03878425657749176,
+      "mean_token_accuracy": 0.9846492648124695,
+      "num_tokens": 35264455.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.8897477686405182,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.8058190941810608,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.041558006405830385,
+      "mean_token_accuracy": 0.9830883860588073,
+      "num_tokens": 35453251.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.8873181045055389,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.7765992283821106,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.03555050492286682,
+      "mean_token_accuracy": 0.9858225345611572,
+      "num_tokens": 35641959.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.8782606959342957,
+      "epoch": 2.557200538358008,
+      "grad_norm": 0.8488866686820984,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.03606923818588257,
+      "mean_token_accuracy": 0.9858632445335388,
+      "num_tokens": 35829636.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.8869808673858642,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.1734739542007446,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.040956351161003116,
+      "mean_token_accuracy": 0.9835374176502227,
+      "num_tokens": 36017818.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.8850924432277679,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.3584387302398682,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.042437011003494264,
+      "mean_token_accuracy": 0.9826479077339172,
+      "num_tokens": 36206505.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.8801687896251679,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2862240076065063,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.03880060315132141,
+      "mean_token_accuracy": 0.9843404710292816,
+      "num_tokens": 36395802.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.8752781748771667,
+      "epoch": 2.611036339165545,
+      "grad_norm": 0.7341249585151672,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.03776344656944275,
+      "mean_token_accuracy": 0.9852877616882324,
+      "num_tokens": 36584370.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.8759494423866272,
+      "epoch": 2.624495289367429,
+      "grad_norm": 0.9807276129722595,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.03780297338962555,
+      "mean_token_accuracy": 0.9853227376937866,
+      "num_tokens": 36772865.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.8753129065036773,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.1391628980636597,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.03953765034675598,
+      "mean_token_accuracy": 0.9839197635650635,
+      "num_tokens": 36961462.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.8771899223327637,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.0865391492843628,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.03856399655342102,
+      "mean_token_accuracy": 0.9848688066005706,
+      "num_tokens": 37150486.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.8774881184101104,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.8561595678329468,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.03795175850391388,
+      "mean_token_accuracy": 0.9849309325218201,
+      "num_tokens": 37338890.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.8841517448425293,
+      "epoch": 2.678331090174966,
+      "grad_norm": 0.9458399415016174,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.03860213160514832,
+      "mean_token_accuracy": 0.9842230796813964,
+      "num_tokens": 37526886.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.8809043228626251,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 0.9360871315002441,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.03709094822406769,
+      "mean_token_accuracy": 0.9855118036270142,
+      "num_tokens": 37715408.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.8805519282817841,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.8852124214172363,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.03869341611862183,
+      "mean_token_accuracy": 0.9846724212169647,
+      "num_tokens": 37904328.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.8796228408813477,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.189297080039978,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.03675769567489624,
+      "mean_token_accuracy": 0.9854080140590668,
+      "num_tokens": 38092776.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.8722020089626312,
+      "epoch": 2.732166890982503,
+      "grad_norm": 0.7027947306632996,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.03889244794845581,
+      "mean_token_accuracy": 0.984558242559433,
+      "num_tokens": 38282221.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.8746632814407349,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 0.9799376726150513,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.03815680146217346,
+      "mean_token_accuracy": 0.9845012128353119,
+      "num_tokens": 38470633.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.8766567528247833,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 0.850459098815918,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.039029371738433835,
+      "mean_token_accuracy": 0.9845906794071198,
+      "num_tokens": 38659048.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.8778688013553619,
+      "epoch": 2.772543741588156,
+      "grad_norm": 0.8540134429931641,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.039645448327064514,
+      "mean_token_accuracy": 0.9840566098690033,
+      "num_tokens": 38847370.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.8758600771427154,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8036021590232849,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.038226932287216187,
+      "mean_token_accuracy": 0.9845647215843201,
+      "num_tokens": 39036040.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.8652268946170807,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 0.8526738286018372,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.03798363208770752,
+      "mean_token_accuracy": 0.9845202267169952,
+      "num_tokens": 39224695.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.8707200348377228,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 0.9852389097213745,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.037227436900138855,
+      "mean_token_accuracy": 0.9852050960063934,
+      "num_tokens": 39413360.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.8711447060108185,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.228277325630188,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.03698050379753113,
+      "mean_token_accuracy": 0.9852443158626556,
+      "num_tokens": 39602544.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.8742912471294403,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 0.8440544605255127,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.03757670521736145,
+      "mean_token_accuracy": 0.9848618268966675,
+      "num_tokens": 39791340.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.8801726162433624,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0955252647399902,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.040004149079322815,
+      "mean_token_accuracy": 0.9841345012187958,
+      "num_tokens": 39980212.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.8931492984294891,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.5722652673721313,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.038096648454666135,
+      "mean_token_accuracy": 0.9849136590957641,
+      "num_tokens": 40169122.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.8767493844032288,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.7663730382919312,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.03729096055030823,
+      "mean_token_accuracy": 0.985169780254364,
+      "num_tokens": 40357678.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.8770561516284943,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.1505672931671143,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.03984796404838562,
+      "mean_token_accuracy": 0.9838845670223236,
+      "num_tokens": 40546892.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.8789129912853241,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 0.8355761766433716,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.03769223690032959,
+      "mean_token_accuracy": 0.9847559213638306,
+      "num_tokens": 40736109.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.8781779289245606,
+      "epoch": 2.920592193808883,
+      "grad_norm": 0.9103918075561523,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.038262826204299924,
+      "mean_token_accuracy": 0.984741848707199,
+      "num_tokens": 40924341.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.8552807629108429,
+      "epoch": 2.934051144010767,
+      "grad_norm": 0.8338510990142822,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.0355743408203125,
+      "mean_token_accuracy": 0.9855036079883576,
+      "num_tokens": 41113849.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.8705591142177582,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 0.8334057927131653,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.037476515769958495,
+      "mean_token_accuracy": 0.9854551255702972,
+      "num_tokens": 41303130.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.8809274792671203,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 0.9424103498458862,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.03624279499053955,
+      "mean_token_accuracy": 0.9856135487556458,
+      "num_tokens": 41491750.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.8818133771419525,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.04243004322052,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.039132320880889894,
+      "mean_token_accuracy": 0.984746390581131,
+      "num_tokens": 41680493.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.8838395297527313,
+      "epoch": 2.987886944818304,
+      "grad_norm": 0.8650784492492676,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.03822052478790283,
+      "mean_token_accuracy": 0.9852108955383301,
+      "num_tokens": 41868378.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.8800126549544608,
+      "eval_loss": 0.05468416586518288,
+      "eval_mean_token_accuracy": 0.9781015617832257,
+      "eval_num_tokens": 42038037.0,
+      "eval_runtime": 24.4825,
+      "eval_samples_per_second": 204.228,
+      "eval_steps_per_second": 6.413,
+      "step": 2229
+    },
+    {
+      "entropy": 0.8773060619831086,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.7845072150230408,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.0365300714969635,
+      "mean_token_accuracy": 0.985853761434555,
+      "num_tokens": 42056717.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.8720550239086151,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 0.9917275905609131,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.030071893334388734,
+      "mean_token_accuracy": 0.9890478909015655,
+      "num_tokens": 42245058.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.8489451467990875,
+      "epoch": 3.028263795423957,
+      "grad_norm": 0.9414294362068176,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.027054613828659056,
+      "mean_token_accuracy": 0.9894806027412415,
+      "num_tokens": 42433468.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.848755830526352,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.7276235818862915,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.026897600293159483,
+      "mean_token_accuracy": 0.9893308103084564,
+      "num_tokens": 42621657.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.8402929604053497,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 0.9455838799476624,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.027918827533721925,
+      "mean_token_accuracy": 0.9888385117053986,
+      "num_tokens": 42810533.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.8434606611728668,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 0.7503429651260376,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.025769710540771484,
+      "mean_token_accuracy": 0.989881956577301,
+      "num_tokens": 42999085.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.8459890782833099,
+      "epoch": 3.082099596231494,
+      "grad_norm": 0.736440896987915,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.028781753778457642,
+      "mean_token_accuracy": 0.9882656812667847,
+      "num_tokens": 43188222.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.8493306159973144,
+      "epoch": 3.095558546433378,
+      "grad_norm": 0.9577816128730774,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.02689719796180725,
+      "mean_token_accuracy": 0.9889744818210602,
+      "num_tokens": 43376648.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.8512665331363678,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 0.9631588459014893,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.031439167261123654,
+      "mean_token_accuracy": 0.9874942421913147,
+      "num_tokens": 43564769.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.8519513130187988,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1011922359466553,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.02696958780288696,
+      "mean_token_accuracy": 0.9900951981544495,
+      "num_tokens": 43753362.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.8507618844509125,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9351623058319092,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.026315575838088988,
+      "mean_token_accuracy": 0.989511913061142,
+      "num_tokens": 43942063.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.8506478369235992,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.096564769744873,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.029049152135849,
+      "mean_token_accuracy": 0.9886228442192078,
+      "num_tokens": 44130381.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.8624941289424897,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 0.7859402298927307,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.027744990587234498,
+      "mean_token_accuracy": 0.9890527307987214,
+      "num_tokens": 44319243.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.8627866744995117,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 0.7764605283737183,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.02571650445461273,
+      "mean_token_accuracy": 0.9903442323207855,
+      "num_tokens": 44507698.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.8566473722457886,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.8182396292686462,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.025625643134117127,
+      "mean_token_accuracy": 0.9897150516510009,
+      "num_tokens": 44695954.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.8555088877677918,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.0277098417282104,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.028235083818435668,
+      "mean_token_accuracy": 0.9894341349601745,
+      "num_tokens": 44884558.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.8562702238559723,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.0676493644714355,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.027713948488235475,
+      "mean_token_accuracy": 0.9891053259372711,
+      "num_tokens": 45073381.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.8520576059818268,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 0.7926117181777954,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.028936928510665892,
+      "mean_token_accuracy": 0.988807988166809,
+      "num_tokens": 45262350.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.860496312379837,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.1108510494232178,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030533093214035033,
+      "mean_token_accuracy": 0.9882748067378998,
+      "num_tokens": 45450782.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.872324675321579,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.3057194948196411,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.032421305775642395,
+      "mean_token_accuracy": 0.9879753351211548,
+      "num_tokens": 45639122.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.8778823137283325,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.289680004119873,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03096393346786499,
+      "mean_token_accuracy": 0.9883370697498322,
+      "num_tokens": 45827730.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.8627406179904937,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.8361398577690125,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.02957139313220978,
+      "mean_token_accuracy": 0.9884674072265625,
+      "num_tokens": 46016274.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.8584554970264435,
+      "epoch": 3.297442799461642,
+      "grad_norm": 0.9330723881721497,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.029568272829055785,
+      "mean_token_accuracy": 0.9882686555385589,
+      "num_tokens": 46204679.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.8515178024768829,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.0296216011047363,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.027788180112838744,
+      "mean_token_accuracy": 0.9888734877109527,
+      "num_tokens": 46393989.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.8488850116729736,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1329699754714966,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.028057461977005003,
+      "mean_token_accuracy": 0.9895590543746948,
+      "num_tokens": 46582790.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.849625188112259,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 0.9267219305038452,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.02802973985671997,
+      "mean_token_accuracy": 0.988851535320282,
+      "num_tokens": 46771138.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.854446280002594,
+      "epoch": 3.351278600269179,
+      "grad_norm": 0.8713932037353516,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.030220019817352294,
+      "mean_token_accuracy": 0.9882843673229218,
+      "num_tokens": 46960054.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.8599806547164917,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.675019383430481,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.02958865463733673,
+      "mean_token_accuracy": 0.9880296885967255,
+      "num_tokens": 47149567.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.8549504458904267,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.2176353931427002,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.027722162008285523,
+      "mean_token_accuracy": 0.9890090465545655,
+      "num_tokens": 47338060.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.8552233815193176,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9571717977523804,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.028924965858459474,
+      "mean_token_accuracy": 0.9884298503398895,
+      "num_tokens": 47525895.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.8521630108356476,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.965713620185852,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.030511701107025148,
+      "mean_token_accuracy": 0.9881398677825928,
+      "num_tokens": 47714524.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.8537400126457214,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 0.9005307555198669,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.030432754755020143,
+      "mean_token_accuracy": 0.9876708447933197,
+      "num_tokens": 47903188.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.8596058666706086,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 0.8159075379371643,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.027927806973457335,
+      "mean_token_accuracy": 0.9889210939407349,
+      "num_tokens": 48091334.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.8482735753059387,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0411661863327026,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03003154993057251,
+      "mean_token_accuracy": 0.9882730603218078,
+      "num_tokens": 48279622.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.8466426789760589,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.189595341682434,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03237656056880951,
+      "mean_token_accuracy": 0.9869072616100312,
+      "num_tokens": 48467527.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.8545588493347168,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0700056552886963,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.0306060791015625,
+      "mean_token_accuracy": 0.9880664825439454,
+      "num_tokens": 48656151.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.8503433585166931,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.7099478244781494,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.02984013557434082,
+      "mean_token_accuracy": 0.9879442811012268,
+      "num_tokens": 48844796.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.8464376091957092,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9877128005027771,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.0306446373462677,
+      "mean_token_accuracy": 0.9883591532707214,
+      "num_tokens": 49033216.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.8477609276771545,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.3324531316757202,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.030148410797119142,
+      "mean_token_accuracy": 0.9879989802837372,
+      "num_tokens": 49221425.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.8461863219738006,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8938570022583008,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.02682304084300995,
+      "mean_token_accuracy": 0.9897881150245667,
+      "num_tokens": 49409785.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.8362689018249512,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 0.7423873543739319,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.02689332664012909,
+      "mean_token_accuracy": 0.9899903059005737,
+      "num_tokens": 49598524.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.8395139217376709,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.921755850315094,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.030062124133110046,
+      "mean_token_accuracy": 0.9880835592746735,
+      "num_tokens": 49787241.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.8467226803302765,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0296459197998047,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.030666223168373107,
+      "mean_token_accuracy": 0.9877098739147187,
+      "num_tokens": 49975844.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.833392071723938,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2861948013305664,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.028174835443496703,
+      "mean_token_accuracy": 0.989056545495987,
+      "num_tokens": 50164857.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.8340730309486389,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 0.8116776347160339,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.028390336036682128,
+      "mean_token_accuracy": 0.9886897742748261,
+      "num_tokens": 50353452.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.8357116639614105,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.8362084031105042,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.029806244373321533,
+      "mean_token_accuracy": 0.988309508562088,
+      "num_tokens": 50542461.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.8363270878791809,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8438499569892883,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.028577303886413573,
+      "mean_token_accuracy": 0.9886945366859436,
+      "num_tokens": 50730872.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.8310261785984039,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.1920002698898315,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.030321845412254335,
+      "mean_token_accuracy": 0.9878708899021149,
+      "num_tokens": 50919889.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.8380252480506897,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 0.9409399032592773,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.030501589179039,
+      "mean_token_accuracy": 0.9872083842754364,
+      "num_tokens": 51108079.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.8422612130641938,
+      "epoch": 3.660834454912517,
+      "grad_norm": 0.8370440006256104,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.02974823713302612,
+      "mean_token_accuracy": 0.9885554671287536,
+      "num_tokens": 51296913.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.8387595832347869,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1360642910003662,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.026970452070236205,
+      "mean_token_accuracy": 0.9888331353664398,
+      "num_tokens": 51485678.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.8403031289577484,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.1195677518844604,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.029451662302017213,
+      "mean_token_accuracy": 0.988847154378891,
+      "num_tokens": 51674765.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.8499106168746948,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 0.9573526978492737,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.030826866626739502,
+      "mean_token_accuracy": 0.9875578820705414,
+      "num_tokens": 51863924.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.8576709866523743,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9386286735534668,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.02946900427341461,
+      "mean_token_accuracy": 0.9885939955711365,
+      "num_tokens": 52052011.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.8404816091060638,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.095536708831787,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.030070826411247253,
+      "mean_token_accuracy": 0.9881400763988495,
+      "num_tokens": 52240955.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.8346689164638519,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9964426159858704,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03005717396736145,
+      "mean_token_accuracy": 0.9883057296276092,
+      "num_tokens": 52429544.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.8377436757087707,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.8499778509140015,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.029283815622329713,
+      "mean_token_accuracy": 0.9882387280464172,
+      "num_tokens": 52618405.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.8328788638114929,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9594390988349915,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.02666555345058441,
+      "mean_token_accuracy": 0.9889323055744171,
+      "num_tokens": 52806689.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.829415249824524,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.3587099313735962,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.028349629044532774,
+      "mean_token_accuracy": 0.9890907645225525,
+      "num_tokens": 52995248.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.8335795283317566,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.077789306640625,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.03001731038093567,
+      "mean_token_accuracy": 0.9878358781337738,
+      "num_tokens": 53183607.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.8316770553588867,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 0.7265951037406921,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.030136841535568237,
+      "mean_token_accuracy": 0.9880174934864044,
+      "num_tokens": 53372735.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.8310399711132049,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.6605978012084961,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.026665833592414857,
+      "mean_token_accuracy": 0.9892704486846924,
+      "num_tokens": 53561776.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.8258654296398162,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.6560797691345215,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.026718440651893615,
+      "mean_token_accuracy": 0.989847868680954,
+      "num_tokens": 53750019.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.8221413195133209,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 0.7495356798171997,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.027431756258010864,
+      "mean_token_accuracy": 0.9892430484294892,
+      "num_tokens": 53938914.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.8341928541660308,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.0676971673965454,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.028432759642601012,
+      "mean_token_accuracy": 0.9889024317264556,
+      "num_tokens": 54127763.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.8359149813652038,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.380476713180542,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.028149163722991942,
+      "mean_token_accuracy": 0.9884953200817108,
+      "num_tokens": 54315930.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.8294597804546356,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0728843212127686,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.028733691573143004,
+      "mean_token_accuracy": 0.9884523034095765,
+      "num_tokens": 54504632.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.8245523810386658,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 0.9723307490348816,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.02793261408805847,
+      "mean_token_accuracy": 0.9891038477420807,
+      "num_tokens": 54693752.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.8279620766639709,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.003624439239502,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.028367668390274048,
+      "mean_token_accuracy": 0.9893603324890137,
+      "num_tokens": 54882030.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.8306584179401397,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.8481361865997314,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.029952478408813477,
+      "mean_token_accuracy": 0.9881332099437714,
+      "num_tokens": 55070549.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.8316469848155975,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.7631049156188965,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.029351180791854857,
+      "mean_token_accuracy": 0.9888184070587158,
+      "num_tokens": 55258999.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.8221287310123444,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 0.7234840393066406,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.027709537744522096,
+      "mean_token_accuracy": 0.9890308439731598,
+      "num_tokens": 55448047.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.8239192545413971,
+      "epoch": 3.970390309555855,
+      "grad_norm": 0.8497020602226257,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.029604345560073853,
+      "mean_token_accuracy": 0.9881375730037689,
+      "num_tokens": 55636566.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.8319927275180816,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.1599552631378174,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.02763860821723938,
+      "mean_token_accuracy": 0.9891406774520874,
+      "num_tokens": 55825049.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.8404937744140625,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 0.906755805015564,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.029578033089637756,
+      "mean_token_accuracy": 0.9878996431827545,
+      "num_tokens": 56012965.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.8419649528849656,
+      "eval_loss": 0.0570383295416832,
+      "eval_mean_token_accuracy": 0.9787763862093543,
+      "eval_num_tokens": 56050518.0,
+      "eval_runtime": 24.4889,
+      "eval_samples_per_second": 204.174,
+      "eval_steps_per_second": 6.411,
+      "step": 2972
+    },
+    {
+      "entropy": 0.8328965544700623,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7073294520378113,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.020916813611984254,
+      "mean_token_accuracy": 0.992138022184372,
+      "num_tokens": 56201621.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.8153974890708924,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.8904889822006226,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.01942949742078781,
+      "mean_token_accuracy": 0.9927055358886718,
+      "num_tokens": 56390018.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.8080023109912873,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.0828146934509277,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.018959003686904907,
+      "mean_token_accuracy": 0.9928978025913239,
+      "num_tokens": 56578046.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.8056950688362121,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7407417893409729,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019250378012657166,
+      "mean_token_accuracy": 0.9926685452461242,
+      "num_tokens": 56767192.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.8124530673027038,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.0355021953582764,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.01904640197753906,
+      "mean_token_accuracy": 0.9925767481327057,
+      "num_tokens": 56955769.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.810802960395813,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.3759289979934692,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.020995599031448365,
+      "mean_token_accuracy": 0.9917719483375549,
+      "num_tokens": 57144300.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.8159192740917206,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.112034559249878,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.01948906183242798,
+      "mean_token_accuracy": 0.9928658187389374,
+      "num_tokens": 57332935.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.821906328201294,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.0319583415985107,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019549430906772615,
+      "mean_token_accuracy": 0.9922966659069061,
+      "num_tokens": 57520775.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.8133022606372833,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.067180871963501,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.01833198368549347,
+      "mean_token_accuracy": 0.9934999167919158,
+      "num_tokens": 57709495.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.8150732755661011,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.277079701423645,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.019589337706565856,
+      "mean_token_accuracy": 0.992507028579712,
+      "num_tokens": 57897461.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.8159323990345001,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.3150911331176758,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.01963164806365967,
+      "mean_token_accuracy": 0.9926789402961731,
+      "num_tokens": 58085915.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.8147098064422608,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.0073580741882324,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.02144385576248169,
+      "mean_token_accuracy": 0.9918489575386047,
+      "num_tokens": 58274583.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.8155419886112213,
+      "epoch": 4.172274562584119,
+      "grad_norm": 0.8892306685447693,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.02215111702680588,
+      "mean_token_accuracy": 0.9914537072181702,
+      "num_tokens": 58463241.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.8084070324897766,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.7887238264083862,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.018581822514533997,
+      "mean_token_accuracy": 0.9928648293018341,
+      "num_tokens": 58652281.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.8016656517982483,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.0644817352294922,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.020099574327468873,
+      "mean_token_accuracy": 0.9916590929031373,
+      "num_tokens": 58839946.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.8035425901412964,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.9588403701782227,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020384626090526582,
+      "mean_token_accuracy": 0.9918951213359832,
+      "num_tokens": 59028994.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.8114414095878602,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.7145556211471558,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.019365760684013366,
+      "mean_token_accuracy": 0.9921130836009979,
+      "num_tokens": 59217599.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.8123053193092347,
+      "epoch": 4.23956931359354,
+      "grad_norm": 0.7928217649459839,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.020542070269584656,
+      "mean_token_accuracy": 0.9918384432792664,
+      "num_tokens": 59406176.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.810414057970047,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.1470942497253418,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.01809942126274109,
+      "mean_token_accuracy": 0.992941427230835,
+      "num_tokens": 59594604.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.8046662211418152,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.579035222530365,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.01914280205965042,
+      "mean_token_accuracy": 0.9926800608634949,
+      "num_tokens": 59782947.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.8034039616584778,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.8979132175445557,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.019897907972335815,
+      "mean_token_accuracy": 0.9919832170009613,
+      "num_tokens": 59971417.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.8137863457202912,
+      "epoch": 4.293405114401077,
+      "grad_norm": 0.9149890542030334,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.02069481760263443,
+      "mean_token_accuracy": 0.992353367805481,
+      "num_tokens": 60159983.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.8163237690925598,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.8264126181602478,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.021351543068885804,
+      "mean_token_accuracy": 0.9921438097953796,
+      "num_tokens": 60348327.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.8133926391601562,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.0638184547424316,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.020092280209064485,
+      "mean_token_accuracy": 0.9920917391777039,
+      "num_tokens": 60537152.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.8095248222351075,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.8831470012664795,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.01908399909734726,
+      "mean_token_accuracy": 0.9929656088352203,
+      "num_tokens": 60725603.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.8069245994091034,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8486410975456238,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.01925535798072815,
+      "mean_token_accuracy": 0.9927195310592651,
+      "num_tokens": 60914021.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.8006627321243286,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.9021628499031067,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.019187870621681213,
+      "mean_token_accuracy": 0.992531019449234,
+      "num_tokens": 61102037.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.8067023754119873,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.7793697118759155,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.021760080754756928,
+      "mean_token_accuracy": 0.9915579319000244,
+      "num_tokens": 61290132.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.8075471341609954,
+      "epoch": 4.387617765814267,
+      "grad_norm": 0.736788809299469,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.02077012062072754,
+      "mean_token_accuracy": 0.9920295119285584,
+      "num_tokens": 61478897.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.8083206176757812,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.1687020063400269,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.02117353230714798,
+      "mean_token_accuracy": 0.991925036907196,
+      "num_tokens": 61667982.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.81106236577034,
+      "epoch": 4.414535666218035,
+      "grad_norm": 0.7596265077590942,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.02081204652786255,
+      "mean_token_accuracy": 0.9921064436435699,
+      "num_tokens": 61856557.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.8018859803676606,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.823410153388977,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.019047826528549194,
+      "mean_token_accuracy": 0.9924882829189301,
+      "num_tokens": 62045181.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7996662855148315,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.0748025178909302,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.01788569688796997,
+      "mean_token_accuracy": 0.9931309640407562,
+      "num_tokens": 62234215.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.8021627008914948,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9946282505989075,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021408762037754058,
+      "mean_token_accuracy": 0.9914931178092956,
+      "num_tokens": 62422361.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.8084165751934052,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.7313793301582336,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.020281489193439483,
+      "mean_token_accuracy": 0.9925725042819977,
+      "num_tokens": 62610667.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.8103892624378204,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.9093077182769775,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.018877537548542024,
+      "mean_token_accuracy": 0.9929530620574951,
+      "num_tokens": 62798970.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.8091431736946106,
+      "epoch": 4.495289367429341,
+      "grad_norm": 0.952234148979187,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021039776504039764,
+      "mean_token_accuracy": 0.9918699383735656,
+      "num_tokens": 62986914.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.8094858765602112,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.9959521293640137,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021489059925079344,
+      "mean_token_accuracy": 0.9914597749710083,
+      "num_tokens": 63175805.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.807330971956253,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.6189294457435608,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.02164270281791687,
+      "mean_token_accuracy": 0.9920661866664886,
+      "num_tokens": 63364692.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.8098866164684295,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7895700931549072,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.021856357157230378,
+      "mean_token_accuracy": 0.9916301846504212,
+      "num_tokens": 63553692.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.8143714487552642,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0242317914962769,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02046577036380768,
+      "mean_token_accuracy": 0.9923330962657928,
+      "num_tokens": 63742476.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.8123241424560547,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.4186221361160278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.020659247040748598,
+      "mean_token_accuracy": 0.9918934762477875,
+      "num_tokens": 63930898.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.8133015930652618,
+      "epoch": 4.576043068640646,
+      "grad_norm": 0.9085373878479004,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.021746012568473815,
+      "mean_token_accuracy": 0.9918161749839782,
+      "num_tokens": 64119589.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.8173148334026337,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.7257112264633179,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.019787387549877168,
+      "mean_token_accuracy": 0.9922669887542724,
+      "num_tokens": 64308293.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.8116482973098755,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7442086935043335,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.020933127403259276,
+      "mean_token_accuracy": 0.9917319416999817,
+      "num_tokens": 64497362.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.8134189546108246,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.7617318630218506,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.01842810958623886,
+      "mean_token_accuracy": 0.9927939116954804,
+      "num_tokens": 64685802.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.8112196505069733,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.7430832386016846,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.01809561103582382,
+      "mean_token_accuracy": 0.9930340230464936,
+      "num_tokens": 64873973.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.8128187835216523,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.7553606629371643,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.0202948659658432,
+      "mean_token_accuracy": 0.9919280350208283,
+      "num_tokens": 65062510.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.813840126991272,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.7390825152397156,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.020557042956352235,
+      "mean_token_accuracy": 0.9918098270893096,
+      "num_tokens": 65250940.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.8117553532123566,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.7698133587837219,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.020231780409812928,
+      "mean_token_accuracy": 0.9920670032501221,
+      "num_tokens": 65439696.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.8110962271690368,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.8562760353088379,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.018769541382789613,
+      "mean_token_accuracy": 0.9925244867801666,
+      "num_tokens": 65628357.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.8113800704479217,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.2076038122177124,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020604749023914338,
+      "mean_token_accuracy": 0.9921375870704651,
+      "num_tokens": 65817139.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.8204582631587982,
+      "epoch": 4.710632570659489,
+      "grad_norm": 0.9150722026824951,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021981674432754516,
+      "mean_token_accuracy": 0.9915678203105927,
+      "num_tokens": 66005911.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.8241343259811401,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.9803491830825806,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02007504999637604,
+      "mean_token_accuracy": 0.992132717370987,
+      "num_tokens": 66194469.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.8122783243656159,
+      "epoch": 4.737550471063257,
+      "grad_norm": 0.967640221118927,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.019903984665870667,
+      "mean_token_accuracy": 0.9924214482307434,
+      "num_tokens": 66383927.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.8145494401454926,
+      "epoch": 4.751009421265142,
+      "grad_norm": 0.7893037796020508,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.019587111473083497,
+      "mean_token_accuracy": 0.9921624064445496,
+      "num_tokens": 66572212.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.8090193390846252,
+      "epoch": 4.764468371467026,
+      "grad_norm": 0.9659109711647034,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.021097327768802642,
+      "mean_token_accuracy": 0.9917473196983337,
+      "num_tokens": 66760377.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.802634209394455,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.0175755023956299,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.02113145440816879,
+      "mean_token_accuracy": 0.991830724477768,
+      "num_tokens": 66949210.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.8030233085155487,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.7780183553695679,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.019016166031360627,
+      "mean_token_accuracy": 0.9926414608955383,
+      "num_tokens": 67138031.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.8019147455692291,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7863438129425049,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.018219922482967377,
+      "mean_token_accuracy": 0.9928844273090363,
+      "num_tokens": 67327268.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7949212431907654,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9292862415313721,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.02136329412460327,
+      "mean_token_accuracy": 0.9918141901493073,
+      "num_tokens": 67515765.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.8002764582633972,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9770768284797668,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.02056926190853119,
+      "mean_token_accuracy": 0.9918984293937683,
+      "num_tokens": 67704975.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.808483463525772,
+      "epoch": 4.845222072678331,
+      "grad_norm": 0.6890367269515991,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020158609747886656,
+      "mean_token_accuracy": 0.9925975620746612,
+      "num_tokens": 67893850.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.8144356846809387,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.8613544702529907,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.019614605605602263,
+      "mean_token_accuracy": 0.9928383588790893,
+      "num_tokens": 68082435.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.8083879172801971,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.7532686591148376,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.019553859531879426,
+      "mean_token_accuracy": 0.9924478828907013,
+      "num_tokens": 68270495.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.8038046479225158,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.3251928091049194,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.02081710398197174,
+      "mean_token_accuracy": 0.9919814527034759,
+      "num_tokens": 68459223.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.8058662831783294,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8125224113464355,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.01858963668346405,
+      "mean_token_accuracy": 0.9922726869583129,
+      "num_tokens": 68647306.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.805506020784378,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8629401922225952,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020777507126331328,
+      "mean_token_accuracy": 0.9918497979640961,
+      "num_tokens": 68836386.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.8137464344501495,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.9837846159934998,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.02024296224117279,
+      "mean_token_accuracy": 0.9921825706958771,
+      "num_tokens": 69025192.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.8139736235141755,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.8701289296150208,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.01932244449853897,
+      "mean_token_accuracy": 0.9926821112632751,
+      "num_tokens": 69213427.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.8088622391223907,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.7041723132133484,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.018706569075584413,
+      "mean_token_accuracy": 0.9927357733249664,
+      "num_tokens": 69401690.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.8001897990703583,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1029852628707886,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.020011243224143983,
+      "mean_token_accuracy": 0.9923900067806244,
+      "num_tokens": 69590659.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.8071568071842193,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.2395373582839966,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.02379731684923172,
+      "mean_token_accuracy": 0.9907991290092468,
+      "num_tokens": 69779864.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.8139646112918854,
+      "epoch": 4.993270524899058,
+      "grad_norm": 0.9219555854797363,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.021635296940803527,
+      "mean_token_accuracy": 0.991401469707489,
+      "num_tokens": 69968639.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.822659221804066,
+      "eval_loss": 0.060467079281806946,
+      "eval_mean_token_accuracy": 0.9792145984188007,
+      "eval_num_tokens": 70063064.0,
+      "eval_runtime": 24.4326,
+      "eval_samples_per_second": 204.645,
+      "eval_steps_per_second": 6.426,
+      "step": 3715
+    },
+    {
+      "entropy": 0.8150339186191559,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8010120391845703,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01673099249601364,
+      "mean_token_accuracy": 0.9935125410556793,
+      "num_tokens": 70157396.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.8029226005077362,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9923233389854431,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01266745626926422,
+      "mean_token_accuracy": 0.9954398214817047,
+      "num_tokens": 70346010.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.789790153503418,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.7180858254432678,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.012603424489498138,
+      "mean_token_accuracy": 0.9949092507362366,
+      "num_tokens": 70534896.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.7863301753997802,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.6865439414978027,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.013341715931892395,
+      "mean_token_accuracy": 0.9951623439788818,
+      "num_tokens": 70723644.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.7841233730316162,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.8624011278152466,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01214413121342659,
+      "mean_token_accuracy": 0.9950926601886749,
+      "num_tokens": 70912490.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.7781193614006042,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.1462656259536743,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.013635106384754181,
+      "mean_token_accuracy": 0.9946888983249664,
+      "num_tokens": 71101186.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.7773554801940918,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.7154580354690552,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.012310966104269027,
+      "mean_token_accuracy": 0.9952048182487487,
+      "num_tokens": 71290177.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.784615957736969,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8214139342308044,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.012708047032356262,
+      "mean_token_accuracy": 0.9956556618213653,
+      "num_tokens": 71478243.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.7814814388751984,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0974094867706299,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.011942464113235473,
+      "mean_token_accuracy": 0.9958715498447418,
+      "num_tokens": 71667417.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.7749826848506928,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.7582180500030518,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.011887111514806748,
+      "mean_token_accuracy": 0.9954704821109772,
+      "num_tokens": 71856322.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.7695000052452088,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.9159554839134216,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.013456562161445617,
+      "mean_token_accuracy": 0.994843590259552,
+      "num_tokens": 72044716.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.7674728155136108,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.0967031717300415,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.012064316868782043,
+      "mean_token_accuracy": 0.9953838646411896,
+      "num_tokens": 72233243.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.766239732503891,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.756477415561676,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012840582430362702,
+      "mean_token_accuracy": 0.9951915740966797,
+      "num_tokens": 72421751.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.7667239785194397,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.8078455924987793,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.012580010294914245,
+      "mean_token_accuracy": 0.9955588281154633,
+      "num_tokens": 72610886.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.7759605884552002,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.9473005533218384,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.01442105770111084,
+      "mean_token_accuracy": 0.9946676790714264,
+      "num_tokens": 72799213.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.7739267945289612,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.4796932637691498,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011830046027898788,
+      "mean_token_accuracy": 0.9954923093318939,
+      "num_tokens": 72988413.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.7756406426429748,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7290598154067993,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.014412418007850647,
+      "mean_token_accuracy": 0.9946037054061889,
+      "num_tokens": 73177134.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.7763995885848999,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.0562312602996826,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.015396332740783692,
+      "mean_token_accuracy": 0.9945933580398559,
+      "num_tokens": 73365622.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.7737024486064911,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1071312427520752,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01451488733291626,
+      "mean_token_accuracy": 0.9945858538150787,
+      "num_tokens": 73554381.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.779985535144806,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.7758586406707764,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.013801637291908263,
+      "mean_token_accuracy": 0.9947169423103333,
+      "num_tokens": 73743028.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.7786811470985413,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.6511496305465698,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012189272791147232,
+      "mean_token_accuracy": 0.9954657673835754,
+      "num_tokens": 73931263.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.7736978232860565,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.8152093291282654,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.013696780800819397,
+      "mean_token_accuracy": 0.9948886811733246,
+      "num_tokens": 74119992.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.7730314910411835,
+      "epoch": 5.302826379542395,
+      "grad_norm": 0.6703830361366272,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012700904905796052,
+      "mean_token_accuracy": 0.9950747072696686,
+      "num_tokens": 74308428.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.7740058362483978,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.9845885634422302,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012894082069396972,
+      "mean_token_accuracy": 0.9953160643577575,
+      "num_tokens": 74497030.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.7738905131816864,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.732532799243927,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.011950771510601043,
+      "mean_token_accuracy": 0.9955996692180633,
+      "num_tokens": 74685631.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.7788872420787811,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7301356196403503,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012340743839740754,
+      "mean_token_accuracy": 0.9952776432037354,
+      "num_tokens": 74874025.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.7771204471588135,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.7570232152938843,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.012937523424625397,
+      "mean_token_accuracy": 0.9949343740940094,
+      "num_tokens": 75062636.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.7747974276542664,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8237525820732117,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.012423938512802124,
+      "mean_token_accuracy": 0.9953977644443512,
+      "num_tokens": 75251339.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.7692180812358856,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.8497490286827087,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012855434417724609,
+      "mean_token_accuracy": 0.9948123812675476,
+      "num_tokens": 75440121.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.773431783914566,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.6758515238761902,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.012878020107746125,
+      "mean_token_accuracy": 0.9955350995063782,
+      "num_tokens": 75629143.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.7735035121440887,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7578167915344238,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.013059911131858826,
+      "mean_token_accuracy": 0.994978415966034,
+      "num_tokens": 75817425.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.7725034832954407,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.0449564456939697,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.013163739442825317,
+      "mean_token_accuracy": 0.9953307807445526,
+      "num_tokens": 76005910.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.7736131310462951,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.881845235824585,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012109111249446868,
+      "mean_token_accuracy": 0.9952919363975525,
+      "num_tokens": 76194639.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.7797342419624329,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.8100120425224304,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.013262654840946197,
+      "mean_token_accuracy": 0.9948459386825561,
+      "num_tokens": 76383240.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.7794113874435424,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.83738112449646,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.013258925080299378,
+      "mean_token_accuracy": 0.9947907269001007,
+      "num_tokens": 76572040.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.7787538826465606,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0567374229431152,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.014641955494880676,
+      "mean_token_accuracy": 0.9945057690143585,
+      "num_tokens": 76761030.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.7894418001174927,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.835543692111969,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.013065731525421143,
+      "mean_token_accuracy": 0.9949834287166596,
+      "num_tokens": 76948469.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.7823423445224762,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.8142625093460083,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.01229657381772995,
+      "mean_token_accuracy": 0.9957204461097717,
+      "num_tokens": 77136968.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.7786863029003144,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.749609649181366,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.01328940987586975,
+      "mean_token_accuracy": 0.9949924945831299,
+      "num_tokens": 77325328.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.7780588805675507,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.787598729133606,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.013933436572551727,
+      "mean_token_accuracy": 0.9949735462665558,
+      "num_tokens": 77513716.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.7779332935810089,
+      "epoch": 5.545087483176312,
+      "grad_norm": 1.0315855741500854,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.012871211767196656,
+      "mean_token_accuracy": 0.995112019777298,
+      "num_tokens": 77701621.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.7754440903663635,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.8191165924072266,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.012908256053924561,
+      "mean_token_accuracy": 0.9950248181819916,
+      "num_tokens": 77889886.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.7811384737491608,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.832539975643158,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.013047756254673004,
+      "mean_token_accuracy": 0.9946948051452636,
+      "num_tokens": 78078640.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.7825228571891785,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.9072359800338745,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.012361775338649749,
+      "mean_token_accuracy": 0.9954402267932891,
+      "num_tokens": 78267842.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.7825945496559144,
+      "epoch": 5.598923283983849,
+      "grad_norm": 0.8300487995147705,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.013246089220046997,
+      "mean_token_accuracy": 0.9951650142669678,
+      "num_tokens": 78455995.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.7857775866985321,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.9343852996826172,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.012612038850784301,
+      "mean_token_accuracy": 0.995340085029602,
+      "num_tokens": 78643617.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.775356650352478,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7109665274620056,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.013411080837249756,
+      "mean_token_accuracy": 0.9951648890972138,
+      "num_tokens": 78833085.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.7750323057174683,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.6436781883239746,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.012064981460571288,
+      "mean_token_accuracy": 0.9954153180122376,
+      "num_tokens": 79021569.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.7753469109535217,
+      "epoch": 5.652759084791386,
+      "grad_norm": 1.0361640453338623,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.013953472673892974,
+      "mean_token_accuracy": 0.9947281837463379,
+      "num_tokens": 79210573.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.7821656167507172,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9078817367553711,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012946712970733642,
+      "mean_token_accuracy": 0.9954073607921601,
+      "num_tokens": 79398723.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.7803877890110016,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.7879899740219116,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.012198399752378464,
+      "mean_token_accuracy": 0.9953389108181,
+      "num_tokens": 79587809.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.7772530496120453,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.0390578508377075,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.012395218014717102,
+      "mean_token_accuracy": 0.9953405141830445,
+      "num_tokens": 79775768.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.7784459292888641,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.7382562160491943,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.012304402142763137,
+      "mean_token_accuracy": 0.9953492462635041,
+      "num_tokens": 79964602.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.7821883440017701,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.5720117092132568,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.011424948275089265,
+      "mean_token_accuracy": 0.9956757485866546,
+      "num_tokens": 80153320.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.7811889886856079,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.7762079238891602,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012460066378116608,
+      "mean_token_accuracy": 0.9950654745101929,
+      "num_tokens": 80341381.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.7771898031234741,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.9219444394111633,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.011629728972911835,
+      "mean_token_accuracy": 0.9955675959587097,
+      "num_tokens": 80530323.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.7825480580329895,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.7189850211143494,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.011180496215820313,
+      "mean_token_accuracy": 0.9960062265396118,
+      "num_tokens": 80718533.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.7795504450798034,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.7256530523300171,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.012952378392219544,
+      "mean_token_accuracy": 0.9951896607875824,
+      "num_tokens": 80907391.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.779454517364502,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.808525800704956,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.0121039479970932,
+      "mean_token_accuracy": 0.9955627739429473,
+      "num_tokens": 81096396.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.7813481986522675,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.7294234037399292,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01356595754623413,
+      "mean_token_accuracy": 0.9951751470565796,
+      "num_tokens": 81284943.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.7833737671375275,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.771756649017334,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.013437020778656005,
+      "mean_token_accuracy": 0.9951942205429077,
+      "num_tokens": 81473810.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.781111603975296,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.5440986752510071,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.012187350541353226,
+      "mean_token_accuracy": 0.9956019043922424,
+      "num_tokens": 81662193.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.7777030646800995,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7532716393470764,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.013766974210739136,
+      "mean_token_accuracy": 0.994896411895752,
+      "num_tokens": 81851045.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.7776691019535065,
+      "epoch": 5.85464333781965,
+      "grad_norm": 1.1882649660110474,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.012588076293468475,
+      "mean_token_accuracy": 0.9952153980731964,
+      "num_tokens": 82039225.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.7795583724975585,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.7453442215919495,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.012100882083177566,
+      "mean_token_accuracy": 0.9954553186893463,
+      "num_tokens": 82228123.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.7763878583908081,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.6757420301437378,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011943653970956803,
+      "mean_token_accuracy": 0.9953660726547241,
+      "num_tokens": 82417093.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.7747587621212005,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.9607213735580444,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.014138388633728027,
+      "mean_token_accuracy": 0.9944287598133087,
+      "num_tokens": 82605461.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.7782076597213745,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.6733895540237427,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.012490443140268325,
+      "mean_token_accuracy": 0.9950622916221619,
+      "num_tokens": 82794124.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.7843947112560272,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.7626368403434753,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.012137357890605927,
+      "mean_token_accuracy": 0.9952504158020019,
+      "num_tokens": 82982258.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.7826779782772064,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8149229288101196,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012872529029846192,
+      "mean_token_accuracy": 0.9949809432029724,
+      "num_tokens": 83170520.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.7792184889316559,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.7774028778076172,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012353222817182541,
+      "mean_token_accuracy": 0.9953234374523163,
+      "num_tokens": 83359174.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.7778255641460419,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7626445293426514,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.01137089878320694,
+      "mean_token_accuracy": 0.9958219468593598,
+      "num_tokens": 83547583.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.7765573263168335,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9130274653434753,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012439383566379547,
+      "mean_token_accuracy": 0.9954963147640228,
+      "num_tokens": 83735904.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.7734197914600373,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.7289780378341675,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011430246382951736,
+      "mean_token_accuracy": 0.9957478463649749,
+      "num_tokens": 83924971.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.7852135965019275,
+      "eval_loss": 0.06990383565425873,
+      "eval_mean_token_accuracy": 0.9796095274056599,
+      "eval_num_tokens": 84075763.0,
+      "eval_runtime": 24.3947,
+      "eval_samples_per_second": 204.963,
+      "eval_steps_per_second": 6.436,
+      "step": 4458
+    },
+    {
+      "entropy": 0.7758591413497925,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.9430185556411743,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.011816921830177306,
+      "mean_token_accuracy": 0.9959785401821136,
+      "num_tokens": 84113330.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.773297393321991,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.6593970060348511,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.007801702618598938,
+      "mean_token_accuracy": 0.9972260951995849,
+      "num_tokens": 84302325.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.7716216742992401,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.5422326922416687,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.00867043137550354,
+      "mean_token_accuracy": 0.9968674182891846,
+      "num_tokens": 84490227.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.7699320554733277,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.8308367729187012,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.007848290354013443,
+      "mean_token_accuracy": 0.9971986234188079,
+      "num_tokens": 84678824.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.7688778102397918,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.5276612639427185,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.007070200890302658,
+      "mean_token_accuracy": 0.9974652588367462,
+      "num_tokens": 84867430.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.7687382102012634,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.7081453800201416,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.00793309137225151,
+      "mean_token_accuracy": 0.9972716808319092,
+      "num_tokens": 85056072.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.7705117404460907,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.9106748104095459,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.007362194359302521,
+      "mean_token_accuracy": 0.9975170493125916,
+      "num_tokens": 85244337.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.7697769284248352,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.7364314198493958,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.007369447499513626,
+      "mean_token_accuracy": 0.9974394500255584,
+      "num_tokens": 85432878.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.7661255121231079,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.6747480630874634,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.007502703368663788,
+      "mean_token_accuracy": 0.9973482489585876,
+      "num_tokens": 85621800.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.7673385441303253,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.8179678916931152,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.007111805677413941,
+      "mean_token_accuracy": 0.9977720499038696,
+      "num_tokens": 85809796.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.76323401927948,
+      "epoch": 6.137281292059219,
+      "grad_norm": 0.5177297592163086,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.007196265459060669,
+      "mean_token_accuracy": 0.9976384699344635,
+      "num_tokens": 85998543.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.7640918910503387,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 0.8970023989677429,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.008229555189609527,
+      "mean_token_accuracy": 0.9967760264873504,
+      "num_tokens": 86186537.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.7605979859828949,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.7019157409667969,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.007770314812660217,
+      "mean_token_accuracy": 0.997019499540329,
+      "num_tokens": 86375405.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.7647112905979156,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.6519957184791565,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.007326644659042358,
+      "mean_token_accuracy": 0.9972577571868897,
+      "num_tokens": 86562963.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.7608903884887696,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.9307657480239868,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.00786140188574791,
+      "mean_token_accuracy": 0.9970912158489227,
+      "num_tokens": 86751006.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.7573343932628631,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 0.6093463897705078,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.007400572299957275,
+      "mean_token_accuracy": 0.997343909740448,
+      "num_tokens": 86939232.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.756312346458435,
+      "epoch": 6.218034993270525,
+      "grad_norm": 0.6412308812141418,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.007471667230129242,
+      "mean_token_accuracy": 0.9973339200019836,
+      "num_tokens": 87128294.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.7593119204044342,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.6830183863639832,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.007831692695617676,
+      "mean_token_accuracy": 0.9972396552562713,
+      "num_tokens": 87316833.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.7598747551441193,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.6647830009460449,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.0070374816656112674,
+      "mean_token_accuracy": 0.9975631892681122,
+      "num_tokens": 87505022.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.7626749157905579,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.5470786690711975,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.007449843734502792,
+      "mean_token_accuracy": 0.9976118981838227,
+      "num_tokens": 87692953.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.7584024250507355,
+      "epoch": 6.271870794078062,
+      "grad_norm": 0.8641819953918457,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.007355903089046478,
+      "mean_token_accuracy": 0.99752836227417,
+      "num_tokens": 87882513.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.7607118487358093,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.6385810971260071,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.007216785848140716,
+      "mean_token_accuracy": 0.9973138213157654,
+      "num_tokens": 88071161.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.7600765109062195,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.3984440565109253,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.007562516629695893,
+      "mean_token_accuracy": 0.9973795533180236,
+      "num_tokens": 88259867.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.7609024405479431,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.7098219990730286,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.007748734951019287,
+      "mean_token_accuracy": 0.9970796704292297,
+      "num_tokens": 88448504.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.7602188348770141,
+      "epoch": 6.325706594885599,
+      "grad_norm": 0.3834342658519745,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.00794660747051239,
+      "mean_token_accuracy": 0.9971937775611878,
+      "num_tokens": 88637123.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.7595065951347351,
+      "epoch": 6.339165545087483,
+      "grad_norm": 0.7073737978935242,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.007514684647321701,
+      "mean_token_accuracy": 0.9972682356834411,
+      "num_tokens": 88826107.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.7566567718982696,
+      "epoch": 6.352624495289367,
+      "grad_norm": 0.6334935426712036,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.007287996262311936,
+      "mean_token_accuracy": 0.9972865641117096,
+      "num_tokens": 89014806.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.7535678207874298,
+      "epoch": 6.366083445491252,
+      "grad_norm": 0.5461204648017883,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.007215028256177902,
+      "mean_token_accuracy": 0.9976517677307128,
+      "num_tokens": 89203610.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.753510159254074,
+      "epoch": 6.379542395693136,
+      "grad_norm": 0.8002920150756836,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.00775575190782547,
+      "mean_token_accuracy": 0.9972102761268615,
+      "num_tokens": 89392611.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.7567359745502472,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.5416199564933777,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.007165181636810303,
+      "mean_token_accuracy": 0.9973179161548614,
+      "num_tokens": 89581472.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.7603509783744812,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.7484428882598877,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.008048020303249359,
+      "mean_token_accuracy": 0.9972274541854859,
+      "num_tokens": 89769895.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.7604019463062286,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.6397516131401062,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.006713573634624481,
+      "mean_token_accuracy": 0.9974542498588562,
+      "num_tokens": 89958151.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.7562435805797577,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.6562761068344116,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.0070735208690166475,
+      "mean_token_accuracy": 0.9975251019001007,
+      "num_tokens": 90147009.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.7581055939197541,
+      "epoch": 6.446837146702557,
+      "grad_norm": 0.6705797910690308,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.008277279138565064,
+      "mean_token_accuracy": 0.9969512403011322,
+      "num_tokens": 90335135.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.7552513837814331,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.5535324215888977,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.007625260949134826,
+      "mean_token_accuracy": 0.9972123861312866,
+      "num_tokens": 90523520.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.7567071616649628,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.5841183662414551,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.008183968812227249,
+      "mean_token_accuracy": 0.9970704436302185,
+      "num_tokens": 90712012.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.7579837441444397,
+      "epoch": 6.48721399730821,
+      "grad_norm": 1.0210020542144775,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.00890980064868927,
+      "mean_token_accuracy": 0.9968561112880707,
+      "num_tokens": 90900360.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.7599723398685455,
+      "epoch": 6.500672947510094,
+      "grad_norm": 0.7007607221603394,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.007547108829021454,
+      "mean_token_accuracy": 0.9972310900688172,
+      "num_tokens": 91089219.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.7569817125797271,
+      "epoch": 6.514131897711978,
+      "grad_norm": 1.203089952468872,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.007012282311916351,
+      "mean_token_accuracy": 0.9974212229251862,
+      "num_tokens": 91278177.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.7547080159187317,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.6055266857147217,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.007885889708995819,
+      "mean_token_accuracy": 0.9971360504627228,
+      "num_tokens": 91466912.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.7584831535816192,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.6868627071380615,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.00764903798699379,
+      "mean_token_accuracy": 0.9975648581981659,
+      "num_tokens": 91655249.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.757660973072052,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.9303789734840393,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.00731472447514534,
+      "mean_token_accuracy": 0.9974016606807709,
+      "num_tokens": 91844016.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.759681248664856,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 0.6116757988929749,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.007223697006702423,
+      "mean_token_accuracy": 0.9970514178276062,
+      "num_tokens": 92032638.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.7601584196090698,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.6825643181800842,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.0068596042692661285,
+      "mean_token_accuracy": 0.9973869740962982,
+      "num_tokens": 92221897.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.7645585358142852,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.7482470870018005,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.007520741224288941,
+      "mean_token_accuracy": 0.9974736630916595,
+      "num_tokens": 92410416.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.766044944524765,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.7137084603309631,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.007919894158840179,
+      "mean_token_accuracy": 0.9969592452049255,
+      "num_tokens": 92598533.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.761748319864273,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 1.0537257194519043,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.007994732260704041,
+      "mean_token_accuracy": 0.9970260739326477,
+      "num_tokens": 92788046.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.763516879081726,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.9168201088905334,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.006822402775287628,
+      "mean_token_accuracy": 0.9973889350891113,
+      "num_tokens": 92976583.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.7603426337242126,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.6945034265518188,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.007335217297077179,
+      "mean_token_accuracy": 0.9974107563495636,
+      "num_tokens": 93164984.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.7601041436195374,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.7255296111106873,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.006898632645606995,
+      "mean_token_accuracy": 0.9974060177803039,
+      "num_tokens": 93353860.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.760407668352127,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.7785485982894897,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.006965675204992294,
+      "mean_token_accuracy": 0.997627604007721,
+      "num_tokens": 93542542.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.7584291577339173,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.676058292388916,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.0066734164953231815,
+      "mean_token_accuracy": 0.9976742446422577,
+      "num_tokens": 93730962.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.7543873727321625,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.9423731565475464,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.007540161162614823,
+      "mean_token_accuracy": 0.9972456872463227,
+      "num_tokens": 93919838.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.7540828108787536,
+      "epoch": 6.716016150740242,
+      "grad_norm": 0.60378098487854,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.006739185005426407,
+      "mean_token_accuracy": 0.9976783573627472,
+      "num_tokens": 94108612.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.7581348657608032,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.7218618392944336,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.006258270889520645,
+      "mean_token_accuracy": 0.9977961003780365,
+      "num_tokens": 94296515.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.7555042028427124,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.6111772060394287,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.006771801412105561,
+      "mean_token_accuracy": 0.9975680649280548,
+      "num_tokens": 94484849.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.7563546478748322,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.7171721458435059,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.007536987960338593,
+      "mean_token_accuracy": 0.9971135079860687,
+      "num_tokens": 94673363.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.7541037023067474,
+      "epoch": 6.769851951547779,
+      "grad_norm": 0.45519423484802246,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.007405019551515579,
+      "mean_token_accuracy": 0.9975123167037964,
+      "num_tokens": 94862699.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.7588851511478424,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.7041094899177551,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.007134091854095459,
+      "mean_token_accuracy": 0.9972242057323456,
+      "num_tokens": 95050741.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.7570755481719971,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.6298620700836182,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.0070040211081504825,
+      "mean_token_accuracy": 0.997326648235321,
+      "num_tokens": 95239293.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.7548963069915772,
+      "epoch": 6.810228802153432,
+      "grad_norm": 1.0688964128494263,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.006342679262161255,
+      "mean_token_accuracy": 0.9976243853569031,
+      "num_tokens": 95428279.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.7548417568206787,
+      "epoch": 6.823687752355316,
+      "grad_norm": 0.4953668415546417,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.0069470912218093876,
+      "mean_token_accuracy": 0.9973559200763702,
+      "num_tokens": 95616835.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.7551373422145844,
+      "epoch": 6.837146702557201,
+      "grad_norm": 0.5016734600067139,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.006873356550931931,
+      "mean_token_accuracy": 0.9973772704601288,
+      "num_tokens": 95805557.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.7567492067813874,
+      "epoch": 6.850605652759085,
+      "grad_norm": 0.532783567905426,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.006668326258659363,
+      "mean_token_accuracy": 0.997450977563858,
+      "num_tokens": 95993831.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.758799821138382,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.6660076379776001,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.006823312491178513,
+      "mean_token_accuracy": 0.9975283324718476,
+      "num_tokens": 96181732.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.7571039736270905,
+      "epoch": 6.877523553162853,
+      "grad_norm": 1.0330839157104492,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.007801513373851776,
+      "mean_token_accuracy": 0.9971715807914734,
+      "num_tokens": 96369920.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.7555147588253022,
+      "epoch": 6.890982503364738,
+      "grad_norm": 0.5079207420349121,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.0074800834059715274,
+      "mean_token_accuracy": 0.9973418891429902,
+      "num_tokens": 96558724.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.7542111933231354,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.678804337978363,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.00698004812002182,
+      "mean_token_accuracy": 0.997770369052887,
+      "num_tokens": 96747242.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.7508629977703094,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.8755067586898804,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.007143624126911163,
+      "mean_token_accuracy": 0.9975672841072083,
+      "num_tokens": 96936244.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.7522107064723969,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.8553352952003479,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.007587063312530518,
+      "mean_token_accuracy": 0.9970708072185517,
+      "num_tokens": 97125069.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.7501682281494141,
+      "epoch": 6.944818304172275,
+      "grad_norm": 0.614628255367279,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.007764184474945068,
+      "mean_token_accuracy": 0.9968451082706451,
+      "num_tokens": 97313879.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.7522825837135315,
+      "epoch": 6.958277254374159,
+      "grad_norm": 0.8060729503631592,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.00771535262465477,
+      "mean_token_accuracy": 0.9972144842147828,
+      "num_tokens": 97503040.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.7534927785396576,
+      "epoch": 6.971736204576043,
+      "grad_norm": 0.6604568958282471,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.006561616808176041,
+      "mean_token_accuracy": 0.9974263906478882,
+      "num_tokens": 97692024.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.7542680203914642,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.853604793548584,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.007131250202655792,
+      "mean_token_accuracy": 0.9974451243877411,
+      "num_tokens": 97881026.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.7567435801029205,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.6146188974380493,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.006486450880765915,
+      "mean_token_accuracy": 0.9974190831184387,
+      "num_tokens": 98069553.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.7571151700748759,
+      "eval_loss": 0.07285518199205399,
+      "eval_mean_token_accuracy": 0.9808630795235846,
+      "eval_num_tokens": 98088531.0,
+      "eval_runtime": 24.4094,
+      "eval_samples_per_second": 204.839,
+      "eval_steps_per_second": 6.432,
+      "step": 5201
+    },
+    {
+      "entropy": 0.751511138677597,
+      "epoch": 7.012113055181696,
+      "grad_norm": 0.49068665504455566,
+      "learning_rate": 2.2506648060191042e-05,
+      "loss": 0.0042594756931066515,
+      "mean_token_accuracy": 0.9985321044921875,
+      "num_tokens": 98258586.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.7522116243839264,
+      "epoch": 7.02557200538358,
+      "grad_norm": 0.5906213521957397,
+      "learning_rate": 2.2321031102941066e-05,
+      "loss": 0.003930336982011795,
+      "mean_token_accuracy": 0.9989193022251129,
+      "num_tokens": 98447091.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.7511299610137939,
+      "epoch": 7.039030955585464,
+      "grad_norm": 0.45251449942588806,
+      "learning_rate": 2.2135962530768055e-05,
+      "loss": 0.004373953118920327,
+      "mean_token_accuracy": 0.9986318647861481,
+      "num_tokens": 98635468.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.7481192111968994,
+      "epoch": 7.052489905787349,
+      "grad_norm": 0.6962258219718933,
+      "learning_rate": 2.1951446010313053e-05,
+      "loss": 0.00428679995238781,
+      "mean_token_accuracy": 0.9987218022346497,
+      "num_tokens": 98824365.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.7483489096164704,
+      "epoch": 7.065948855989233,
+      "grad_norm": 0.5099614858627319,
+      "learning_rate": 2.1767485197279662e-05,
+      "loss": 0.003984151408076286,
+      "mean_token_accuracy": 0.9986076414585113,
+      "num_tokens": 99013056.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.7463673830032349,
+      "epoch": 7.079407806191117,
+      "grad_norm": 0.3716561198234558,
+      "learning_rate": 2.1584083736361617e-05,
+      "loss": 0.004038792476058007,
+      "mean_token_accuracy": 0.998809176683426,
+      "num_tokens": 99201941.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.7466944694519043,
+      "epoch": 7.0928667563930015,
+      "grad_norm": 0.33926740288734436,
+      "learning_rate": 2.1401245261170582e-05,
+      "loss": 0.003776150941848755,
+      "mean_token_accuracy": 0.9987689018249511,
+      "num_tokens": 99390847.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.7465248286724091,
+      "epoch": 7.106325706594886,
+      "grad_norm": 0.3297136127948761,
+      "learning_rate": 2.121897339416416e-05,
+      "loss": 0.0037364721298217773,
+      "mean_token_accuracy": 0.9987996935844421,
+      "num_tokens": 99579720.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.7467422366142273,
+      "epoch": 7.11978465679677,
+      "grad_norm": 0.3911883234977722,
+      "learning_rate": 2.103727174657412e-05,
+      "loss": 0.004084533080458641,
+      "mean_token_accuracy": 0.9985779583454132,
+      "num_tokens": 99768263.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.7423677504062652,
+      "epoch": 7.133243606998654,
+      "grad_norm": 0.5843891501426697,
+      "learning_rate": 2.085614391833486e-05,
+      "loss": 0.0038405463099479677,
+      "mean_token_accuracy": 0.9987926363945008,
+      "num_tokens": 99957655.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.7434155166149139,
+      "epoch": 7.1467025572005385,
+      "grad_norm": 0.5154002904891968,
+      "learning_rate": 2.0675593498012052e-05,
+      "loss": 0.0036686167120933534,
+      "mean_token_accuracy": 0.9988853752613067,
+      "num_tokens": 100146048.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.7450331568717956,
+      "epoch": 7.160161507402423,
+      "grad_norm": 0.5856082439422607,
+      "learning_rate": 2.049562406273159e-05,
+      "loss": 0.00417071133852005,
+      "mean_token_accuracy": 0.9985920011997222,
+      "num_tokens": 100334064.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.7413534104824067,
+      "epoch": 7.173620457604307,
+      "grad_norm": 0.4711301028728485,
+      "learning_rate": 2.0316239178108688e-05,
+      "loss": 0.003790935128927231,
+      "mean_token_accuracy": 0.9986560821533204,
+      "num_tokens": 100522817.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.739797306060791,
+      "epoch": 7.187079407806191,
+      "grad_norm": 0.4481808841228485,
+      "learning_rate": 2.013744239817725e-05,
+      "loss": 0.0038315042853355407,
+      "mean_token_accuracy": 0.9987379729747772,
+      "num_tokens": 100711882.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.7411384165287018,
+      "epoch": 7.2005383580080755,
+      "grad_norm": 0.3373877704143524,
+      "learning_rate": 1.9959237265319452e-05,
+      "loss": 0.0034619957208633424,
+      "mean_token_accuracy": 0.9990058243274689,
+      "num_tokens": 100900649.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.7439651310443878,
+      "epoch": 7.21399730820996,
+      "grad_norm": 0.4260849952697754,
+      "learning_rate": 1.9781627310195528e-05,
+      "loss": 0.004041648656129837,
+      "mean_token_accuracy": 0.9988129734992981,
+      "num_tokens": 101088789.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.741381824016571,
+      "epoch": 7.227456258411844,
+      "grad_norm": 0.704740047454834,
+      "learning_rate": 1.960461605167389e-05,
+      "loss": 0.004109270870685577,
+      "mean_token_accuracy": 0.9987618446350097,
+      "num_tokens": 101277316.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.7418228089809418,
+      "epoch": 7.240915208613728,
+      "grad_norm": 0.34908416867256165,
+      "learning_rate": 1.9428206996761357e-05,
+      "loss": 0.0038616616278886796,
+      "mean_token_accuracy": 0.9985946893692017,
+      "num_tokens": 101466025.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.7436148345470428,
+      "epoch": 7.2543741588156125,
+      "grad_norm": 0.5490255951881409,
+      "learning_rate": 1.9252403640533678e-05,
+      "loss": 0.004948872700333595,
+      "mean_token_accuracy": 0.9981587290763855,
+      "num_tokens": 101654412.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.7419192969799042,
+      "epoch": 7.267833109017497,
+      "grad_norm": 0.5561390519142151,
+      "learning_rate": 1.9077209466066315e-05,
+      "loss": 0.0043866544961929325,
+      "mean_token_accuracy": 0.9986559212207794,
+      "num_tokens": 101843232.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.7441964089870453,
+      "epoch": 7.281292059219381,
+      "grad_norm": 0.5724263191223145,
+      "learning_rate": 1.8902627944365396e-05,
+      "loss": 0.004136815667152405,
+      "mean_token_accuracy": 0.9986151039600373,
+      "num_tokens": 102031115.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.7447238683700561,
+      "epoch": 7.294751009421265,
+      "grad_norm": 0.6418642997741699,
+      "learning_rate": 1.872866253429897e-05,
+      "loss": 0.004244513809680939,
+      "mean_token_accuracy": 0.9985204100608825,
+      "num_tokens": 102219064.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.7460219264030457,
+      "epoch": 7.30820995962315,
+      "grad_norm": 0.877717912197113,
+      "learning_rate": 1.8555316682528485e-05,
+      "loss": 0.0035314552485942842,
+      "mean_token_accuracy": 0.9990062952041626,
+      "num_tokens": 102407049.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.7446399390697479,
+      "epoch": 7.321668909825034,
+      "grad_norm": 0.8448772430419922,
+      "learning_rate": 1.8382593823440474e-05,
+      "loss": 0.004365745186805725,
+      "mean_token_accuracy": 0.9984619498252869,
+      "num_tokens": 102595411.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.7414965629577637,
+      "epoch": 7.335127860026918,
+      "grad_norm": 0.7239252328872681,
+      "learning_rate": 1.821049737907854e-05,
+      "loss": 0.0036744467914104463,
+      "mean_token_accuracy": 0.9989120185375213,
+      "num_tokens": 102784691.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.744704294204712,
+      "epoch": 7.348586810228802,
+      "grad_norm": 0.390360951423645,
+      "learning_rate": 1.803903075907554e-05,
+      "loss": 0.003917969018220902,
+      "mean_token_accuracy": 0.9988363087177277,
+      "num_tokens": 102973291.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.744241201877594,
+      "epoch": 7.362045760430687,
+      "grad_norm": 0.7771829962730408,
+      "learning_rate": 1.7868197360586054e-05,
+      "loss": 0.004070578515529633,
+      "mean_token_accuracy": 0.9985230803489685,
+      "num_tokens": 103161417.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.7423063695430756,
+      "epoch": 7.375504710632571,
+      "grad_norm": 0.3263421654701233,
+      "learning_rate": 1.769800056821904e-05,
+      "loss": 0.003923245519399643,
+      "mean_token_accuracy": 0.9985153019428253,
+      "num_tokens": 103349890.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.7458358705043793,
+      "epoch": 7.388963660834455,
+      "grad_norm": 0.5268740653991699,
+      "learning_rate": 1.7528443753970814e-05,
+      "loss": 0.004262761771678924,
+      "mean_token_accuracy": 0.9987221539020539,
+      "num_tokens": 103537726.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.7404268085956573,
+      "epoch": 7.402422611036339,
+      "grad_norm": 0.5339550971984863,
+      "learning_rate": 1.7359530277158227e-05,
+      "loss": 0.0039617355912923815,
+      "mean_token_accuracy": 0.998635333776474,
+      "num_tokens": 103726707.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.7421627819538117,
+      "epoch": 7.415881561238224,
+      "grad_norm": 0.47286465764045715,
+      "learning_rate": 1.7191263484352106e-05,
+      "loss": 0.0034935571253299712,
+      "mean_token_accuracy": 0.9989599347114563,
+      "num_tokens": 103915542.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.7416261017322541,
+      "epoch": 7.429340511440108,
+      "grad_norm": 0.29520711302757263,
+      "learning_rate": 1.702364670931097e-05,
+      "loss": 0.003899512439966202,
+      "mean_token_accuracy": 0.9986254096031189,
+      "num_tokens": 104105019.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.7447648346424103,
+      "epoch": 7.442799461641992,
+      "grad_norm": 0.46887683868408203,
+      "learning_rate": 1.685668327291495e-05,
+      "loss": 0.0036319829523563385,
+      "mean_token_accuracy": 0.9989396393299103,
+      "num_tokens": 104293657.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.7453936874866486,
+      "epoch": 7.456258411843876,
+      "grad_norm": 0.3639506995677948,
+      "learning_rate": 1.6690376483100034e-05,
+      "loss": 0.003420223295688629,
+      "mean_token_accuracy": 0.9988774538040162,
+      "num_tokens": 104482306.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.743130087852478,
+      "epoch": 7.469717362045761,
+      "grad_norm": 0.4534284770488739,
+      "learning_rate": 1.6524729634792475e-05,
+      "loss": 0.0038840845227241516,
+      "mean_token_accuracy": 0.9987022399902343,
+      "num_tokens": 104671166.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.7441355168819428,
+      "epoch": 7.483176312247645,
+      "grad_norm": 0.3704164922237396,
+      "learning_rate": 1.6359746009843567e-05,
+      "loss": 0.0034930892288684845,
+      "mean_token_accuracy": 0.9988382160663605,
+      "num_tokens": 104859133.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.7428222477436066,
+      "epoch": 7.496635262449529,
+      "grad_norm": 0.751130223274231,
+      "learning_rate": 1.6195428876964592e-05,
+      "loss": 0.003963682055473328,
+      "mean_token_accuracy": 0.9984794855117798,
+      "num_tokens": 105047492.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.7427781701087952,
+      "epoch": 7.510094212651413,
+      "grad_norm": 0.8935811519622803,
+      "learning_rate": 1.6031781491662056e-05,
+      "loss": 0.004140084609389305,
+      "mean_token_accuracy": 0.9985048770904541,
+      "num_tokens": 105235970.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.7405261933803559,
+      "epoch": 7.523553162853298,
+      "grad_norm": 0.575813353061676,
+      "learning_rate": 1.58688070961732e-05,
+      "loss": 0.003933753073215485,
+      "mean_token_accuracy": 0.998809415102005,
+      "num_tokens": 105424662.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.742167991399765,
+      "epoch": 7.537012113055182,
+      "grad_norm": 0.5800251960754395,
+      "learning_rate": 1.5706508919401804e-05,
+      "loss": 0.004490737244486809,
+      "mean_token_accuracy": 0.9987403452396393,
+      "num_tokens": 105613306.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.744090485572815,
+      "epoch": 7.550471063257066,
+      "grad_norm": 0.8697437644004822,
+      "learning_rate": 1.554489017685412e-05,
+      "loss": 0.00443139597773552,
+      "mean_token_accuracy": 0.9984764337539673,
+      "num_tokens": 105801807.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.7431838274002075,
+      "epoch": 7.56393001345895,
+      "grad_norm": 0.47921308875083923,
+      "learning_rate": 1.538395407057525e-05,
+      "loss": 0.003662409633398056,
+      "mean_token_accuracy": 0.9989210546016694,
+      "num_tokens": 105990683.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.7407984972000122,
+      "epoch": 7.577388963660835,
+      "grad_norm": 0.45125728845596313,
+      "learning_rate": 1.522370378908567e-05,
+      "loss": 0.003542381525039673,
+      "mean_token_accuracy": 0.9988591969013214,
+      "num_tokens": 106179809.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.744829797744751,
+      "epoch": 7.590847913862719,
+      "grad_norm": 0.5999196171760559,
+      "learning_rate": 1.5064142507318051e-05,
+      "loss": 0.0035081353038549424,
+      "mean_token_accuracy": 0.9988316416740417,
+      "num_tokens": 106367988.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.7428836584091186,
+      "epoch": 7.604306864064603,
+      "grad_norm": 0.4594089388847351,
+      "learning_rate": 1.4905273386554386e-05,
+      "loss": 0.004249929264187813,
+      "mean_token_accuracy": 0.9987165331840515,
+      "num_tokens": 106556378.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.7435553133487701,
+      "epoch": 7.6177658142664875,
+      "grad_norm": 0.5920020937919617,
+      "learning_rate": 1.4747099574363332e-05,
+      "loss": 0.0036187373101711274,
+      "mean_token_accuracy": 0.9987746834754944,
+      "num_tokens": 106744854.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.7437893748283386,
+      "epoch": 7.631224764468372,
+      "grad_norm": 0.5453750491142273,
+      "learning_rate": 1.4589624204537866e-05,
+      "loss": 0.00377066433429718,
+      "mean_token_accuracy": 0.9989324033260345,
+      "num_tokens": 106932944.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.7395198106765747,
+      "epoch": 7.644683714670256,
+      "grad_norm": 0.38797295093536377,
+      "learning_rate": 1.4432850397033176e-05,
+      "loss": 0.0031442392617464067,
+      "mean_token_accuracy": 0.9990855991840363,
+      "num_tokens": 107121922.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.7402484536170959,
+      "epoch": 7.65814266487214,
+      "grad_norm": 0.6739166975021362,
+      "learning_rate": 1.427678125790487e-05,
+      "loss": 0.0038538143038749696,
+      "mean_token_accuracy": 0.9986329972743988,
+      "num_tokens": 107310638.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.7401728749275207,
+      "epoch": 7.6716016150740245,
+      "grad_norm": 0.4977329969406128,
+      "learning_rate": 1.4121419879247422e-05,
+      "loss": 0.0033894732594490053,
+      "mean_token_accuracy": 0.9989957690238953,
+      "num_tokens": 107499098.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.7425866723060608,
+      "epoch": 7.685060565275909,
+      "grad_norm": 0.5719661712646484,
+      "learning_rate": 1.3966769339132929e-05,
+      "loss": 0.004793065041303635,
+      "mean_token_accuracy": 0.9987428963184357,
+      "num_tokens": 107687499.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.7395791113376617,
+      "epoch": 7.698519515477793,
+      "grad_norm": 0.4698435366153717,
+      "learning_rate": 1.3812832701550099e-05,
+      "loss": 0.004081443324685097,
+      "mean_token_accuracy": 0.9984708070755005,
+      "num_tokens": 107876895.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.7433845281600953,
+      "epoch": 7.711978465679677,
+      "grad_norm": 0.6040443778038025,
+      "learning_rate": 1.365961301634357e-05,
+      "loss": 0.003664097934961319,
+      "mean_token_accuracy": 0.9988075792789459,
+      "num_tokens": 108065071.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.7424676477909088,
+      "epoch": 7.7254374158815615,
+      "grad_norm": 0.6923211812973022,
+      "learning_rate": 1.350711331915348e-05,
+      "loss": 0.003957484290003777,
+      "mean_token_accuracy": 0.998679769039154,
+      "num_tokens": 108253186.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.7405034005641937,
+      "epoch": 7.738896366083446,
+      "grad_norm": 0.4896307587623596,
+      "learning_rate": 1.3355336631355314e-05,
+      "loss": 0.0041983149945735935,
+      "mean_token_accuracy": 0.9986901462078095,
+      "num_tokens": 108442076.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.743422681093216,
+      "epoch": 7.75235531628533,
+      "grad_norm": 0.5889957547187805,
+      "learning_rate": 1.3204285960000051e-05,
+      "loss": 0.003425312042236328,
+      "mean_token_accuracy": 0.9988837718963623,
+      "num_tokens": 108630562.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.7411604285240173,
+      "epoch": 7.765814266487214,
+      "grad_norm": 0.4655008316040039,
+      "learning_rate": 1.30539642977546e-05,
+      "loss": 0.003258592635393143,
+      "mean_token_accuracy": 0.9988660037517547,
+      "num_tokens": 108819369.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.7410345315933228,
+      "epoch": 7.7792732166890985,
+      "grad_norm": 0.7350168228149414,
+      "learning_rate": 1.2904374622842486e-05,
+      "loss": 0.0034532986581325533,
+      "mean_token_accuracy": 0.9989261090755462,
+      "num_tokens": 109008535.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.7410138070583343,
+      "epoch": 7.792732166890983,
+      "grad_norm": 0.44904911518096924,
+      "learning_rate": 1.2755519898984846e-05,
+      "loss": 0.0034178659319877625,
+      "mean_token_accuracy": 0.9988463997840882,
+      "num_tokens": 109197443.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.7402764976024627,
+      "epoch": 7.806191117092867,
+      "grad_norm": 0.5618012547492981,
+      "learning_rate": 1.2607403075341728e-05,
+      "loss": 0.003283835202455521,
+      "mean_token_accuracy": 0.9989370465278625,
+      "num_tokens": 109386097.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.7410104274749756,
+      "epoch": 7.819650067294751,
+      "grad_norm": 0.3167133629322052,
+      "learning_rate": 1.2460027086453658e-05,
+      "loss": 0.0034713000059127808,
+      "mean_token_accuracy": 0.9988771259784699,
+      "num_tokens": 109574403.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.7389123022556305,
+      "epoch": 7.833109017496636,
+      "grad_norm": 0.544057309627533,
+      "learning_rate": 1.2313394852183473e-05,
+      "loss": 0.0038273107260465623,
+      "mean_token_accuracy": 0.9986206114292144,
+      "num_tokens": 109763160.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.7385149598121643,
+      "epoch": 7.84656796769852,
+      "grad_norm": 0.8678551912307739,
+      "learning_rate": 1.2167509277658518e-05,
+      "loss": 0.0037419166415929793,
+      "mean_token_accuracy": 0.9987667620182037,
+      "num_tokens": 109951791.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.7384770929813385,
+      "epoch": 7.860026917900404,
+      "grad_norm": 0.45470893383026123,
+      "learning_rate": 1.2022373253213049e-05,
+      "loss": 0.003541013225913048,
+      "mean_token_accuracy": 0.998937851190567,
+      "num_tokens": 110140704.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.7396949470043183,
+      "epoch": 7.873485868102288,
+      "grad_norm": 0.6184161305427551,
+      "learning_rate": 1.1877989654330973e-05,
+      "loss": 0.003454260528087616,
+      "mean_token_accuracy": 0.9989149391651153,
+      "num_tokens": 110329353.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.7411868631839752,
+      "epoch": 7.886944818304173,
+      "grad_norm": 0.49721845984458923,
+      "learning_rate": 1.1734361341588912e-05,
+      "loss": 0.003363589942455292,
+      "mean_token_accuracy": 0.9988781988620759,
+      "num_tokens": 110517720.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.740048885345459,
+      "epoch": 7.900403768506057,
+      "grad_norm": 0.3169788420200348,
+      "learning_rate": 1.1591491160599488e-05,
+      "loss": 0.0030659947544336317,
+      "mean_token_accuracy": 0.9991146922111511,
+      "num_tokens": 110706131.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.7428572535514831,
+      "epoch": 7.913862718707941,
+      "grad_norm": 0.5674341320991516,
+      "learning_rate": 1.1449381941954956e-05,
+      "loss": 0.0031749047338962555,
+      "mean_token_accuracy": 0.998987466096878,
+      "num_tokens": 110893731.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.7389689266681672,
+      "epoch": 7.927321668909825,
+      "grad_norm": 0.36926963925361633,
+      "learning_rate": 1.130803650117116e-05,
+      "loss": 0.0030995842069387438,
+      "mean_token_accuracy": 0.9989600539207458,
+      "num_tokens": 111082079.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.7392809450626373,
+      "epoch": 7.94078061911171,
+      "grad_norm": 0.3086724579334259,
+      "learning_rate": 1.1167457638631656e-05,
+      "loss": 0.0033237963914871218,
+      "mean_token_accuracy": 0.9988582134246826,
+      "num_tokens": 111270413.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.7384293675422668,
+      "epoch": 7.954239569313594,
+      "grad_norm": 0.45126330852508545,
+      "learning_rate": 1.1027648139532355e-05,
+      "loss": 0.0034996122121810913,
+      "mean_token_accuracy": 0.9987758874893189,
+      "num_tokens": 111458912.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.7361682891845703,
+      "epoch": 7.967698519515478,
+      "grad_norm": 0.6252058148384094,
+      "learning_rate": 1.0888610773826253e-05,
+      "loss": 0.003475649654865265,
+      "mean_token_accuracy": 0.9988142251968384,
+      "num_tokens": 111647683.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.733948814868927,
+      "epoch": 7.981157469717362,
+      "grad_norm": 0.5224378705024719,
+      "learning_rate": 1.0750348296168583e-05,
+      "loss": 0.0030234681442379953,
+      "mean_token_accuracy": 0.9990231335163117,
+      "num_tokens": 111836870.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.736539876461029,
+      "epoch": 7.994616419919247,
+      "grad_norm": 0.2892343997955322,
+      "learning_rate": 1.0612863445862236e-05,
+      "loss": 0.0031716078519821167,
+      "mean_token_accuracy": 0.9988950371742249,
+      "num_tokens": 112025584.0,
+      "step": 5940
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.7368154024622243,
+      "eval_loss": 0.08131171762943268,
+      "eval_mean_token_accuracy": 0.9805832057242181,
+      "eval_num_tokens": 112101144.0,
+      "eval_runtime": 24.3602,
+      "eval_samples_per_second": 205.253,
+      "eval_steps_per_second": 6.445,
+      "step": 5944
+    },
+    {
+      "entropy": 0.7347776114940643,
+      "epoch": 8.00807537012113,
+      "grad_norm": 0.21781852841377258,
+      "learning_rate": 1.0476158946803482e-05,
+      "loss": 0.002697869949042797,
+      "mean_token_accuracy": 0.9991762280464173,
+      "num_tokens": 112214484.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.7354790329933166,
+      "epoch": 8.021534320323015,
+      "grad_norm": 0.2759193480014801,
+      "learning_rate": 1.0340237507428009e-05,
+      "loss": 0.0023939665406942368,
+      "mean_token_accuracy": 0.9993740320205688,
+      "num_tokens": 112403209.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.7335577964782715,
+      "epoch": 8.034993270524899,
+      "grad_norm": 0.20684213936328888,
+      "learning_rate": 1.020510182065727e-05,
+      "loss": 0.0021501941606402396,
+      "mean_token_accuracy": 0.9994299292564393,
+      "num_tokens": 112592246.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.7346243798732758,
+      "epoch": 8.048452220726784,
+      "grad_norm": 0.3243059813976288,
+      "learning_rate": 1.0070754563845114e-05,
+      "loss": 0.0025554070249199866,
+      "mean_token_accuracy": 0.9992592394351959,
+      "num_tokens": 112780960.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.7341127455234527,
+      "epoch": 8.061911170928667,
+      "grad_norm": 0.34389808773994446,
+      "learning_rate": 9.937198398724751e-06,
+      "loss": 0.0023281550034880637,
+      "mean_token_accuracy": 0.9992973983287812,
+      "num_tokens": 112969497.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.7344379365444184,
+      "epoch": 8.075370121130552,
+      "grad_norm": 0.33664506673812866,
+      "learning_rate": 9.804435971356025e-06,
+      "loss": 0.0022437553852796555,
+      "mean_token_accuracy": 0.9993536353111268,
+      "num_tokens": 113158346.0,
+      "step": 6000
+    },
+    {
+      "entropy": 0.7345291972160339,
+      "epoch": 8.088829071332436,
+      "grad_norm": 0.21238821744918823,
+      "learning_rate": 9.672469912072956e-06,
+      "loss": 0.0023011134937405585,
+      "mean_token_accuracy": 0.9994915544986724,
+      "num_tokens": 113347027.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.7329188048839569,
+      "epoch": 8.10228802153432,
+      "grad_norm": 0.3358577489852905,
+      "learning_rate": 9.541302835431681e-06,
+      "loss": 0.002364570274949074,
+      "mean_token_accuracy": 0.9992765069007874,
+      "num_tokens": 113535987.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.7357381403446197,
+      "epoch": 8.115746971736204,
+      "grad_norm": 0.4410889148712158,
+      "learning_rate": 9.410937340158599e-06,
+      "loss": 0.0016813736408948898,
+      "mean_token_accuracy": 0.9995461285114289,
+      "num_tokens": 113724178.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.7351392149925232,
+      "epoch": 8.12920592193809,
+      "grad_norm": 0.35911238193511963,
+      "learning_rate": 9.28137600909892e-06,
+      "loss": 0.0021215328946709634,
+      "mean_token_accuracy": 0.9994065165519714,
+      "num_tokens": 113912610.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.7340241730213165,
+      "epoch": 8.142664872139973,
+      "grad_norm": 0.3858516812324524,
+      "learning_rate": 9.15262140916548e-06,
+      "loss": 0.002298617549240589,
+      "mean_token_accuracy": 0.9992367029190063,
+      "num_tokens": 114101352.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.733269315958023,
+      "epoch": 8.156123822341858,
+      "grad_norm": 0.15504567325115204,
+      "learning_rate": 9.024676091287881e-06,
+      "loss": 0.0019876576960086824,
+      "mean_token_accuracy": 0.9994120895862579,
+      "num_tokens": 114290171.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.7359612822532654,
+      "epoch": 8.169582772543741,
+      "grad_norm": 0.3258695602416992,
+      "learning_rate": 8.897542590361963e-06,
+      "loss": 0.002046743780374527,
+      "mean_token_accuracy": 0.9994458854198456,
+      "num_tokens": 114478059.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.734262204170227,
+      "epoch": 8.183041722745626,
+      "grad_norm": 0.3703354299068451,
+      "learning_rate": 8.77122342519956e-06,
+      "loss": 0.0024908211082220076,
+      "mean_token_accuracy": 0.9995096683502197,
+      "num_tokens": 114666866.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.7325477957725525,
+      "epoch": 8.19650067294751,
+      "grad_norm": 0.38723456859588623,
+      "learning_rate": 8.64572109847862e-06,
+      "loss": 0.002013527974486351,
+      "mean_token_accuracy": 0.9994479417800903,
+      "num_tokens": 114855753.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.7337498366832733,
+      "epoch": 8.209959623149395,
+      "grad_norm": 0.44278407096862793,
+      "learning_rate": 8.521038096693635e-06,
+      "loss": 0.0023755524307489395,
+      "mean_token_accuracy": 0.9993066310882568,
+      "num_tokens": 115044397.0,
+      "step": 6100
+    },
+    {
+      "entropy": 0.7312877893447876,
+      "epoch": 8.223418573351278,
+      "grad_norm": 0.36628058552742004,
+      "learning_rate": 8.397176890106317e-06,
+      "loss": 0.0023126287385821344,
+      "mean_token_accuracy": 0.9992603421211242,
+      "num_tokens": 115233499.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.7326091766357422,
+      "epoch": 8.236877523553163,
+      "grad_norm": 0.542598307132721,
+      "learning_rate": 8.274139932696717e-06,
+      "loss": 0.002486979216337204,
+      "mean_token_accuracy": 0.9992321312427521,
+      "num_tokens": 115422096.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.7329172909259796,
+      "epoch": 8.250336473755047,
+      "grad_norm": 0.37169647216796875,
+      "learning_rate": 8.151929662114566e-06,
+      "loss": 0.0022026147693395615,
+      "mean_token_accuracy": 0.9993911325931549,
+      "num_tokens": 115610769.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.7315155625343323,
+      "epoch": 8.263795423956932,
+      "grad_norm": 0.379108190536499,
+      "learning_rate": 8.030548499631014e-06,
+      "loss": 0.0020761553198099135,
+      "mean_token_accuracy": 0.9994912505149841,
+      "num_tokens": 115799639.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.7338620543479919,
+      "epoch": 8.277254374158815,
+      "grad_norm": 0.27872899174690247,
+      "learning_rate": 7.909998850090628e-06,
+      "loss": 0.00214959941804409,
+      "mean_token_accuracy": 0.9993456363677978,
+      "num_tokens": 115987860.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.7319511413574219,
+      "epoch": 8.2907133243607,
+      "grad_norm": 0.3878805637359619,
+      "learning_rate": 7.790283101863771e-06,
+      "loss": 0.0016636446118354798,
+      "mean_token_accuracy": 0.9996107459068299,
+      "num_tokens": 116176426.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.7313687026500701,
+      "epoch": 8.304172274562584,
+      "grad_norm": 0.42844563722610474,
+      "learning_rate": 7.671403626799262e-06,
+      "loss": 0.002215801365673542,
+      "mean_token_accuracy": 0.9993546068668365,
+      "num_tokens": 116365243.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.7326999127864837,
+      "epoch": 8.317631224764469,
+      "grad_norm": 0.28046005964279175,
+      "learning_rate": 7.553362780177398e-06,
+      "loss": 0.0017590772360563279,
+      "mean_token_accuracy": 0.9996075868606568,
+      "num_tokens": 116553541.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.7318588554859161,
+      "epoch": 8.331090174966352,
+      "grad_norm": 0.4567055404186249,
+      "learning_rate": 7.43616290066329e-06,
+      "loss": 0.0024444632232189178,
+      "mean_token_accuracy": 0.9993343472480773,
+      "num_tokens": 116742091.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.7342073738574981,
+      "epoch": 8.344549125168237,
+      "grad_norm": 0.45294421911239624,
+      "learning_rate": 7.3198063102605185e-06,
+      "loss": 0.0021415702998638154,
+      "mean_token_accuracy": 0.9995077669620513,
+      "num_tokens": 116930107.0,
+      "step": 6200
+    },
+    {
+      "entropy": 0.7312484800815582,
+      "epoch": 8.35800807537012,
+      "grad_norm": 0.3044157028198242,
+      "learning_rate": 7.204295314265136e-06,
+      "loss": 0.0018497437238693237,
+      "mean_token_accuracy": 0.999489426612854,
+      "num_tokens": 117118605.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.7315851151943207,
+      "epoch": 8.371467025572006,
+      "grad_norm": 0.29259541630744934,
+      "learning_rate": 7.089632201219997e-06,
+      "loss": 0.0019892159849405288,
+      "mean_token_accuracy": 0.9993900001049042,
+      "num_tokens": 117307207.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.729860270023346,
+      "epoch": 8.38492597577389,
+      "grad_norm": 0.27728405594825745,
+      "learning_rate": 6.975819242869414e-06,
+      "loss": 0.001903151161968708,
+      "mean_token_accuracy": 0.9994141757488251,
+      "num_tokens": 117496022.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.730089795589447,
+      "epoch": 8.398384925975774,
+      "grad_norm": 0.32097047567367554,
+      "learning_rate": 6.862858694114138e-06,
+      "loss": 0.0018681716173887253,
+      "mean_token_accuracy": 0.9994908988475799,
+      "num_tokens": 117684720.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.732157701253891,
+      "epoch": 8.411843876177658,
+      "grad_norm": 0.33475807309150696,
+      "learning_rate": 6.750752792966702e-06,
+      "loss": 0.0019257836043834685,
+      "mean_token_accuracy": 0.9993493318557739,
+      "num_tokens": 117872973.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.7303947329521179,
+      "epoch": 8.425302826379543,
+      "grad_norm": 0.35840341448783875,
+      "learning_rate": 6.639503760507065e-06,
+      "loss": 0.002003169059753418,
+      "mean_token_accuracy": 0.9994274079799652,
+      "num_tokens": 118061309.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.7300981402397155,
+      "epoch": 8.438761776581426,
+      "grad_norm": 0.22061212360858917,
+      "learning_rate": 6.529113800838632e-06,
+      "loss": 0.0023671993985772133,
+      "mean_token_accuracy": 0.999328225851059,
+      "num_tokens": 118250100.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.7311166822910309,
+      "epoch": 8.452220726783311,
+      "grad_norm": 0.4995788335800171,
+      "learning_rate": 6.41958510104454e-06,
+      "loss": 0.002252453379333019,
+      "mean_token_accuracy": 0.9993094325065612,
+      "num_tokens": 118438680.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.7314801871776581,
+      "epoch": 8.465679676985195,
+      "grad_norm": 0.18718956410884857,
+      "learning_rate": 6.31091983114438e-06,
+      "loss": 0.0020001601427793504,
+      "mean_token_accuracy": 0.9993076086044311,
+      "num_tokens": 118627199.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.7294714629650116,
+      "epoch": 8.47913862718708,
+      "grad_norm": 0.24443569779396057,
+      "learning_rate": 6.203120144051161e-06,
+      "loss": 0.001865076832473278,
+      "mean_token_accuracy": 0.9993763923645019,
+      "num_tokens": 118816308.0,
+      "step": 6300
+    },
+    {
+      "entropy": 0.7320528924465179,
+      "epoch": 8.492597577388963,
+      "grad_norm": 0.3000173568725586,
+      "learning_rate": 6.0961881755286775e-06,
+      "loss": 0.0019325431436300277,
+      "mean_token_accuracy": 0.9995656788349152,
+      "num_tokens": 119004887.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.7298780143260956,
+      "epoch": 8.506056527590848,
+      "grad_norm": 0.34283098578453064,
+      "learning_rate": 5.990126044149203e-06,
+      "loss": 0.0018553130328655242,
+      "mean_token_accuracy": 0.9994752407073975,
+      "num_tokens": 119193896.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.7296975553035736,
+      "epoch": 8.519515477792732,
+      "grad_norm": 0.48231956362724304,
+      "learning_rate": 5.884935851251483e-06,
+      "loss": 0.0018540853634476662,
+      "mean_token_accuracy": 0.9994960963726044,
+      "num_tokens": 119383032.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.7316587388515472,
+      "epoch": 8.532974427994617,
+      "grad_norm": 0.31098613142967224,
+      "learning_rate": 5.780619680899146e-06,
+      "loss": 0.0024125341325998305,
+      "mean_token_accuracy": 0.9992330968379974,
+      "num_tokens": 119571647.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.7316439330577851,
+      "epoch": 8.5464333781965,
+      "grad_norm": 0.3148161768913269,
+      "learning_rate": 5.677179599839366e-06,
+      "loss": 0.0020461630076169966,
+      "mean_token_accuracy": 0.9994146227836609,
+      "num_tokens": 119760221.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.7328742682933808,
+      "epoch": 8.559892328398385,
+      "grad_norm": 0.33567124605178833,
+      "learning_rate": 5.574617657461967e-06,
+      "loss": 0.0017173275351524352,
+      "mean_token_accuracy": 0.9996075749397277,
+      "num_tokens": 119948564.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.7310613572597504,
+      "epoch": 8.573351278600269,
+      "grad_norm": 0.3766630291938782,
+      "learning_rate": 5.472935885758767e-06,
+      "loss": 0.0020279109477996824,
+      "mean_token_accuracy": 0.9995895087718963,
+      "num_tokens": 120137124.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.7319568216800689,
+      "epoch": 8.586810228802154,
+      "grad_norm": 0.36246562004089355,
+      "learning_rate": 5.372136299283365e-06,
+      "loss": 0.0020779309794306754,
+      "mean_token_accuracy": 0.999527370929718,
+      "num_tokens": 120325739.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.7306868612766266,
+      "epoch": 8.600269179004037,
+      "grad_norm": 0.3458656072616577,
+      "learning_rate": 5.272220895111196e-06,
+      "loss": 0.001681588776409626,
+      "mean_token_accuracy": 0.9996683657169342,
+      "num_tokens": 120514557.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.7286112308502197,
+      "epoch": 8.613728129205922,
+      "grad_norm": 0.40973636507987976,
+      "learning_rate": 5.173191652799986e-06,
+      "loss": 0.001994205825030804,
+      "mean_token_accuracy": 0.9994747698307037,
+      "num_tokens": 120703746.0,
+      "step": 6400
+    },
+    {
+      "entropy": 0.7324167609214782,
+      "epoch": 8.627187079407806,
+      "grad_norm": 0.30873212218284607,
+      "learning_rate": 5.075050534350512e-06,
+      "loss": 0.001537158153951168,
+      "mean_token_accuracy": 0.9995840191841125,
+      "num_tokens": 120892105.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.7325307846069335,
+      "epoch": 8.640646029609691,
+      "grad_norm": 0.2885727286338806,
+      "learning_rate": 4.977799484167767e-06,
+      "loss": 0.002083958685398102,
+      "mean_token_accuracy": 0.9994043171405792,
+      "num_tokens": 121080661.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.7314725816249847,
+      "epoch": 8.654104979811574,
+      "grad_norm": 0.23452679812908173,
+      "learning_rate": 4.8814404290223655e-06,
+      "loss": 0.0021737243980169295,
+      "mean_token_accuracy": 0.9992907345294952,
+      "num_tokens": 121269176.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.7302026331424714,
+      "epoch": 8.66756393001346,
+      "grad_norm": 0.5051631331443787,
+      "learning_rate": 4.7859752780124555e-06,
+      "loss": 0.002044454775750637,
+      "mean_token_accuracy": 0.9993531286716462,
+      "num_tokens": 121457925.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.7307308673858642,
+      "epoch": 8.681022880215343,
+      "grad_norm": 0.32118919491767883,
+      "learning_rate": 4.691405922525838e-06,
+      "loss": 0.001963503286242485,
+      "mean_token_accuracy": 0.9993822813034058,
+      "num_tokens": 121646615.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.7318669617176056,
+      "epoch": 8.694481830417228,
+      "grad_norm": 0.28653913736343384,
+      "learning_rate": 4.597734236202511e-06,
+      "loss": 0.0022680331021547318,
+      "mean_token_accuracy": 0.9992888629436493,
+      "num_tokens": 121834948.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.7306908547878266,
+      "epoch": 8.707940780619111,
+      "grad_norm": 0.3841933608055115,
+      "learning_rate": 4.504962074897556e-06,
+      "loss": 0.00202475618571043,
+      "mean_token_accuracy": 0.9995454490184784,
+      "num_tokens": 122023470.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.7311248660087586,
+      "epoch": 8.721399730820997,
+      "grad_norm": 0.2170674055814743,
+      "learning_rate": 4.413091276644349e-06,
+      "loss": 0.0017318326979875564,
+      "mean_token_accuracy": 0.9994860172271729,
+      "num_tokens": 122212025.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.730291211605072,
+      "epoch": 8.73485868102288,
+      "grad_norm": 0.4367440640926361,
+      "learning_rate": 4.322123661618166e-06,
+      "loss": 0.002066049911081791,
+      "mean_token_accuracy": 0.999429452419281,
+      "num_tokens": 122400631.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.734411084651947,
+      "epoch": 8.748317631224765,
+      "grad_norm": 0.5407934784889221,
+      "learning_rate": 4.23206103210011e-06,
+      "loss": 0.002084248140454292,
+      "mean_token_accuracy": 0.9993655562400818,
+      "num_tokens": 122588467.0,
+      "step": 6500
+    },
+    {
+      "entropy": 0.7325521528720855,
+      "epoch": 8.761776581426648,
+      "grad_norm": 0.33067557215690613,
+      "learning_rate": 4.142905172441408e-06,
+      "loss": 0.0017788175493478775,
+      "mean_token_accuracy": 0.9996413946151733,
+      "num_tokens": 122776694.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.7312010109424592,
+      "epoch": 8.775235531628534,
+      "grad_norm": 0.5068145394325256,
+      "learning_rate": 4.054657849028043e-06,
+      "loss": 0.001824142411351204,
+      "mean_token_accuracy": 0.999506825208664,
+      "num_tokens": 122965176.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.7314070999622345,
+      "epoch": 8.788694481830417,
+      "grad_norm": 0.21613064408302307,
+      "learning_rate": 3.9673208102457895e-06,
+      "loss": 0.0016722116619348527,
+      "mean_token_accuracy": 0.9995093226432801,
+      "num_tokens": 123153842.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.731372082233429,
+      "epoch": 8.802153432032302,
+      "grad_norm": 0.5407741665840149,
+      "learning_rate": 3.880895786445543e-06,
+      "loss": 0.001678784191608429,
+      "mean_token_accuracy": 0.999667227268219,
+      "num_tokens": 123342339.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.7309448659420014,
+      "epoch": 8.815612382234185,
+      "grad_norm": 0.40189075469970703,
+      "learning_rate": 3.795384489909054e-06,
+      "loss": 0.0022997969761490824,
+      "mean_token_accuracy": 0.9993481457233429,
+      "num_tokens": 123531056.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.7296400606632233,
+      "epoch": 8.82907133243607,
+      "grad_norm": 0.5532040596008301,
+      "learning_rate": 3.7107886148150084e-06,
+      "loss": 0.0017002075910568237,
+      "mean_token_accuracy": 0.9994097113609314,
+      "num_tokens": 123719927.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.7304099202156067,
+      "epoch": 8.842530282637954,
+      "grad_norm": 0.4202295243740082,
+      "learning_rate": 3.6271098372054334e-06,
+      "loss": 0.0020521355792880057,
+      "mean_token_accuracy": 0.9993128538131714,
+      "num_tokens": 123908479.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.7322857022285462,
+      "epoch": 8.855989232839839,
+      "grad_norm": 0.4879668354988098,
+      "learning_rate": 3.5443498149525357e-06,
+      "loss": 0.002071031183004379,
+      "mean_token_accuracy": 0.9994449079036712,
+      "num_tokens": 124096364.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.7318431079387665,
+      "epoch": 8.869448183041722,
+      "grad_norm": 0.5673806071281433,
+      "learning_rate": 3.462510187725809e-06,
+      "loss": 0.0019315868616104127,
+      "mean_token_accuracy": 0.9995098888874054,
+      "num_tokens": 124284715.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.7283908724784851,
+      "epoch": 8.882907133243608,
+      "grad_norm": 0.3319496810436249,
+      "learning_rate": 3.3815925769595745e-06,
+      "loss": 0.0020575495436787605,
+      "mean_token_accuracy": 0.9993783593177795,
+      "num_tokens": 124474114.0,
+      "step": 6600
+    },
+    {
+      "entropy": 0.7316987097263337,
+      "epoch": 8.896366083445491,
+      "grad_norm": 0.32474979758262634,
+      "learning_rate": 3.30159858582087e-06,
+      "loss": 0.0023224465548992157,
+      "mean_token_accuracy": 0.9993063390254975,
+      "num_tokens": 124662486.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.7328927993774415,
+      "epoch": 8.909825033647376,
+      "grad_norm": 0.38243111968040466,
+      "learning_rate": 3.2225297991776536e-06,
+      "loss": 0.0016394183039665222,
+      "mean_token_accuracy": 0.9996214270591736,
+      "num_tokens": 124850474.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.7294083535671234,
+      "epoch": 8.92328398384926,
+      "grad_norm": 0.37468570470809937,
+      "learning_rate": 3.1443877835674208e-06,
+      "loss": 0.001899813488125801,
+      "mean_token_accuracy": 0.9994097173213958,
+      "num_tokens": 125039229.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.7298379123210907,
+      "epoch": 8.936742934051145,
+      "grad_norm": 0.20931333303451538,
+      "learning_rate": 3.0671740871661635e-06,
+      "loss": 0.0022370124235749243,
+      "mean_token_accuracy": 0.9994113147258759,
+      "num_tokens": 125227792.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.7286044478416442,
+      "epoch": 8.950201884253028,
+      "grad_norm": 0.3580392301082611,
+      "learning_rate": 2.9908902397577043e-06,
+      "loss": 0.002087070792913437,
+      "mean_token_accuracy": 0.9993978500366211,
+      "num_tokens": 125416718.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.7310522794723511,
+      "epoch": 8.963660834454913,
+      "grad_norm": 0.32912445068359375,
+      "learning_rate": 2.91553775270339e-06,
+      "loss": 0.0017302701249718665,
+      "mean_token_accuracy": 0.9994479060173035,
+      "num_tokens": 125605155.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.7304746985435486,
+      "epoch": 8.977119784656796,
+      "grad_norm": 0.48885127902030945,
+      "learning_rate": 2.8411181189121284e-06,
+      "loss": 0.0020353911444544793,
+      "mean_token_accuracy": 0.9994727909564972,
+      "num_tokens": 125793654.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.7320328831672669,
+      "epoch": 8.990578734858682,
+      "grad_norm": 0.19737882912158966,
+      "learning_rate": 2.767632812810839e-06,
+      "loss": 0.0019317161291837691,
+      "mean_token_accuracy": 0.9996273815631866,
+      "num_tokens": 125981781.0,
+      "step": 6680
+    },
+    {
+      "epoch": 9.0,
+      "eval_entropy": 0.736012871478014,
+      "eval_loss": 0.08657046407461166,
+      "eval_mean_token_accuracy": 0.9805503571109407,
+      "eval_num_tokens": 126113888.0,
+      "eval_runtime": 24.4287,
+      "eval_samples_per_second": 204.677,
+      "eval_steps_per_second": 6.427,
+      "step": 6687
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.093450942768742e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-6687/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-6687/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-6687/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-743/README.md b/baby_talk_L16_a50/seed_42/checkpoint-743/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-743/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-743/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-743/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-743/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-743/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-743/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..8a84382e6fbf68afaa1e6edc100c0e22a7145d44
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-743/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:350d1b272353fc4617a9206341c932dcd9bb1c91a814ae07d664cdf98217e748
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-743/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-743/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-743/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-743/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-743/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-743/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-743/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-743/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-743/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-743/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-743/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..7ab7ac56d877a59e01ae954a9354124f428de440
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-743/trainer_state.json
@@ -0,0 +1,785 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 743,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.770767290381107e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-743/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-743/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-743/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-7430/README.md b/baby_talk_L16_a50/seed_42/checkpoint-7430/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..7fd6f003ca8e08f2f24631ebf336e496f53c00a5
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-7430/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-7430/adapter_config.json b/baby_talk_L16_a50/seed_42/checkpoint-7430/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d23a0c5fc84881aa2d260b32f26e7e45773a4d7
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-7430/adapter_config.json
@@ -0,0 +1,46 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "up_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-7430/adapter_model.safetensors b/baby_talk_L16_a50/seed_42/checkpoint-7430/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..649e4d21115aaa09ff3fa8a062c2e8106fab9cca
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-7430/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:634b8183df0d846c1b0ac2e337d9935e1948068de3618afdea51c4a704d96e9a
+size 80792096
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-7430/chat_template.jinja b/baby_talk_L16_a50/seed_42/checkpoint-7430/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-7430/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-7430/tokenizer.json b/baby_talk_L16_a50/seed_42/checkpoint-7430/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-7430/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-7430/tokenizer_config.json b/baby_talk_L16_a50/seed_42/checkpoint-7430/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-7430/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-7430/trainer_state.json b/baby_talk_L16_a50/seed_42/checkpoint-7430/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..9c121438be0e8976859be13f00f4397405d1217e
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-7430/trainer_state.json
@@ -0,0 +1,7574 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 7430,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2358420491218567,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 3.2940118312835693,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.550364351272583,
+      "mean_token_accuracy": 0.8554959416389465,
+      "num_tokens": 188811.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2353882431983947,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 3.4476470947265625,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.5143545627593994,
+      "mean_token_accuracy": 0.8613634884357453,
+      "num_tokens": 377729.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2403772234916688,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 2.2756996154785156,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3996511220932007,
+      "mean_token_accuracy": 0.8753438770771027,
+      "num_tokens": 566562.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2205921411514282,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.2432096004486084,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2568032264709473,
+      "mean_token_accuracy": 0.9130991995334625,
+      "num_tokens": 755026.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.221834623813629,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.8531327843666077,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.20193097591400147,
+      "mean_token_accuracy": 0.9274256646633148,
+      "num_tokens": 943494.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2245031952857972,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.5026484131813049,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.171803081035614,
+      "mean_token_accuracy": 0.9363821744918823,
+      "num_tokens": 1131731.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2192703008651733,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.5228630304336548,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.15698516368865967,
+      "mean_token_accuracy": 0.9423282980918884,
+      "num_tokens": 1320258.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2153660774230957,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.3650094270706177,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.14900912046432496,
+      "mean_token_accuracy": 0.9437524616718292,
+      "num_tokens": 1509209.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2123230814933776,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.45723631978034973,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.1399540901184082,
+      "mean_token_accuracy": 0.9474983811378479,
+      "num_tokens": 1698139.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.208789598941803,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.4575304687023163,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.12566736936569214,
+      "mean_token_accuracy": 0.9529488801956176,
+      "num_tokens": 1886384.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.205567193031311,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8033406734466553,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.12005312442779541,
+      "mean_token_accuracy": 0.9533569395542145,
+      "num_tokens": 2074919.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.20180287361145,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.6989286541938782,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.11197478771209717,
+      "mean_token_accuracy": 0.9572584748268127,
+      "num_tokens": 2263116.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.202919363975525,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.7233495116233826,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.11106340885162354,
+      "mean_token_accuracy": 0.9569663584232331,
+      "num_tokens": 2451273.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1907272219657898,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 0.7645091414451599,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.10956189632415772,
+      "mean_token_accuracy": 0.95842245221138,
+      "num_tokens": 2640657.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1895484924316406,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 0.841379702091217,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.10442907810211181,
+      "mean_token_accuracy": 0.9603263795375824,
+      "num_tokens": 2829351.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1871973633766175,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 0.8552286624908447,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.10493810176849365,
+      "mean_token_accuracy": 0.9597454965114594,
+      "num_tokens": 3018290.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1907334923744202,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.7243770360946655,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.10288643836975098,
+      "mean_token_accuracy": 0.9603340923786163,
+      "num_tokens": 3206538.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1835299730300903,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 0.8118091225624084,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.09798368811607361,
+      "mean_token_accuracy": 0.9619723737239838,
+      "num_tokens": 3395374.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1794602513313293,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 0.7447699904441833,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.09498158693313599,
+      "mean_token_accuracy": 0.9634931206703186,
+      "num_tokens": 3584273.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1834724426269532,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4059171676635742,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.09400172233581543,
+      "mean_token_accuracy": 0.9630160868167877,
+      "num_tokens": 3772857.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.186821174621582,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 0.8273025751113892,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.09464811086654663,
+      "mean_token_accuracy": 0.9631786167621612,
+      "num_tokens": 3960877.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1769920349121095,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.0182609558105469,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0914128303527832,
+      "mean_token_accuracy": 0.9644896507263183,
+      "num_tokens": 4149444.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1719188809394836,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.9648745656013489,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0920255422592163,
+      "mean_token_accuracy": 0.963964831829071,
+      "num_tokens": 4337989.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1559369206428527,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 0.853049635887146,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0891042947769165,
+      "mean_token_accuracy": 0.9653984010219574,
+      "num_tokens": 4527256.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1604587316513062,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 0.873772144317627,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09071275591850281,
+      "mean_token_accuracy": 0.9651397407054901,
+      "num_tokens": 4715878.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1534931778907775,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.087850570678711,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.08719289302825928,
+      "mean_token_accuracy": 0.9658753871917725,
+      "num_tokens": 4904527.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1629684448242188,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 0.9014195799827576,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.08716133236885071,
+      "mean_token_accuracy": 0.965560519695282,
+      "num_tokens": 5093554.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.148916518688202,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 0.9712215662002563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.08668915033340455,
+      "mean_token_accuracy": 0.9658140063285827,
+      "num_tokens": 5282224.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1493291974067688,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 0.9085242748260498,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.08189771175384522,
+      "mean_token_accuracy": 0.9670268416404724,
+      "num_tokens": 5471308.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1406704187393188,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.128177285194397,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.08119879961013794,
+      "mean_token_accuracy": 0.9675639867782593,
+      "num_tokens": 5660026.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1369357347488402,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8224227428436279,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.07979745864868164,
+      "mean_token_accuracy": 0.9681445300579071,
+      "num_tokens": 5848073.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1252583503723144,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 0.7711160182952881,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.0783164381980896,
+      "mean_token_accuracy": 0.9686201930046081,
+      "num_tokens": 6036717.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1203404307365417,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.1810287237167358,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.08139073848724365,
+      "mean_token_accuracy": 0.9681365489959717,
+      "num_tokens": 6225223.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1222687840461731,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4551713466644287,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.0820135235786438,
+      "mean_token_accuracy": 0.9672703862190246,
+      "num_tokens": 6413936.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1156280279159545,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 0.8573716878890991,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08074904680252075,
+      "mean_token_accuracy": 0.9677663922309876,
+      "num_tokens": 6602413.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1063185572624206,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 0.7709434628486633,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.07549421787261963,
+      "mean_token_accuracy": 0.9689356207847595,
+      "num_tokens": 6790873.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.0962031960487366,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.7843625545501709,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.07368478775024415,
+      "mean_token_accuracy": 0.9700294613838196,
+      "num_tokens": 6979392.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1028530240058898,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 0.7935485243797302,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.07766538262367248,
+      "mean_token_accuracy": 0.9704003691673279,
+      "num_tokens": 7167482.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.0905393958091736,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4108924865722656,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.0760004162788391,
+      "mean_token_accuracy": 0.9699020266532898,
+      "num_tokens": 7355994.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0853531122207642,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 0.9043511152267456,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.0720310389995575,
+      "mean_token_accuracy": 0.9707890212535858,
+      "num_tokens": 7544655.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.075773572921753,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 0.9038823246955872,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.07026209831237792,
+      "mean_token_accuracy": 0.9715417385101318,
+      "num_tokens": 7733348.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0572428584098816,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.8231707811355591,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.07141299843788147,
+      "mean_token_accuracy": 0.9705908715724945,
+      "num_tokens": 7921582.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.051485300064087,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.1695454120635986,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.07321611642837525,
+      "mean_token_accuracy": 0.9709623396396637,
+      "num_tokens": 8110266.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0485445380210876,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 0.7291010618209839,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.07221676707267762,
+      "mean_token_accuracy": 0.9705081820487976,
+      "num_tokens": 8298932.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0456495046615601,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 0.6550094485282898,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.06878133416175843,
+      "mean_token_accuracy": 0.9728739261627197,
+      "num_tokens": 8487613.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0372861266136169,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.5825450420379639,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.06706151366233826,
+      "mean_token_accuracy": 0.9732721030712128,
+      "num_tokens": 8676276.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0319493532180786,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 0.7158534526824951,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.06999597549438477,
+      "mean_token_accuracy": 0.9726741492748261,
+      "num_tokens": 8864738.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0260030388832093,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.7918373346328735,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.06949877142906188,
+      "mean_token_accuracy": 0.9725882947444916,
+      "num_tokens": 9053582.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0258127093315124,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 0.893281102180481,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.06893026828765869,
+      "mean_token_accuracy": 0.9729171216487884,
+      "num_tokens": 9241914.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0261828184127808,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0127108097076416,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.07094801664352417,
+      "mean_token_accuracy": 0.9714575052261353,
+      "num_tokens": 9430981.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0274562597274781,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 0.7148029208183289,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.06826171875,
+      "mean_token_accuracy": 0.9727248430252076,
+      "num_tokens": 9619472.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0137859225273131,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 0.9228634238243103,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.06890587210655212,
+      "mean_token_accuracy": 0.9723304331302642,
+      "num_tokens": 9808123.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0221150755882262,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 0.971530556678772,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07073599100112915,
+      "mean_token_accuracy": 0.9714174270629883,
+      "num_tokens": 9996601.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0248154640197753,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.7336317896842957,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.06874136924743653,
+      "mean_token_accuracy": 0.9723432004451752,
+      "num_tokens": 10184725.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0120978832244873,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.2100600004196167,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.06270487308502197,
+      "mean_token_accuracy": 0.9756880521774292,
+      "num_tokens": 10373446.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0172406077384948,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 0.7562853693962097,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.06784560084342957,
+      "mean_token_accuracy": 0.9724333345890045,
+      "num_tokens": 10561710.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0150038957595826,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.07966148853302,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.0670344054698944,
+      "mean_token_accuracy": 0.97339146733284,
+      "num_tokens": 10749917.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0135640978813172,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 0.7826104164123535,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.0668565571308136,
+      "mean_token_accuracy": 0.9735289216041565,
+      "num_tokens": 10938098.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0077669620513916,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.299584150314331,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.0728976845741272,
+      "mean_token_accuracy": 0.9701269209384918,
+      "num_tokens": 11126107.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.9957692861557007,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.1542484760284424,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.0631720781326294,
+      "mean_token_accuracy": 0.9747781097888947,
+      "num_tokens": 11315001.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.9983992159366608,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.6967246532440186,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.06333768963813782,
+      "mean_token_accuracy": 0.9750322341918946,
+      "num_tokens": 11503007.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.9801111698150635,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 0.945446789264679,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.05720087289810181,
+      "mean_token_accuracy": 0.9767442524433136,
+      "num_tokens": 11691883.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.9879570543766022,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.8254193663597107,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.05899171829223633,
+      "mean_token_accuracy": 0.9767756819725036,
+      "num_tokens": 11880353.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.9878240942955017,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 0.9882538914680481,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.06443996429443359,
+      "mean_token_accuracy": 0.9735406100749969,
+      "num_tokens": 12069610.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.9938213169574738,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.126546859741211,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.06708416938781739,
+      "mean_token_accuracy": 0.9730359137058258,
+      "num_tokens": 12257901.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.9800443708896637,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 0.9151445031166077,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.06290764808654785,
+      "mean_token_accuracy": 0.9741653084754944,
+      "num_tokens": 12446236.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.9744794130325317,
+      "epoch": 0.901749663526245,
+      "grad_norm": 0.9104028940200806,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.060464882850646974,
+      "mean_token_accuracy": 0.9757490694522858,
+      "num_tokens": 12635226.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.9786687552928924,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 0.7316192388534546,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.06054847836494446,
+      "mean_token_accuracy": 0.9760404825210571,
+      "num_tokens": 12823428.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.966060334444046,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.3535135984420776,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.061383575201034546,
+      "mean_token_accuracy": 0.9749173820018768,
+      "num_tokens": 13012362.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.9722849190235138,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 0.890771746635437,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.06301190257072449,
+      "mean_token_accuracy": 0.9750476598739624,
+      "num_tokens": 13200680.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.9624020338058472,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.142499327659607,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.06179196834564209,
+      "mean_token_accuracy": 0.9751901209354401,
+      "num_tokens": 13389333.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.9653747022151947,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.8682493567466736,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.06113170981407166,
+      "mean_token_accuracy": 0.9748325288295746,
+      "num_tokens": 13578210.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.970348197221756,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8351430296897888,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.05901788473129273,
+      "mean_token_accuracy": 0.9761226296424865,
+      "num_tokens": 13767039.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.956724613904953,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.2604819536209106,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.06124393343925476,
+      "mean_token_accuracy": 0.9747833967208862,
+      "num_tokens": 13956057.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.9722227849018802,
+      "eval_loss": 0.0600070059299469,
+      "eval_mean_token_accuracy": 0.975752676368519,
+      "eval_num_tokens": 14012630.0,
+      "eval_runtime": 24.6215,
+      "eval_samples_per_second": 203.074,
+      "eval_steps_per_second": 6.377,
+      "step": 743
+    },
+    {
+      "entropy": 0.9689933776855468,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9720287919044495,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.055563896894454956,
+      "mean_token_accuracy": 0.9779708027839661,
+      "num_tokens": 14144652.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.9506655633449554,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 0.6619595289230347,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.050587379932403566,
+      "mean_token_accuracy": 0.979685264825821,
+      "num_tokens": 14332714.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.9458732306957245,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 0.7905108332633972,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.05069155097007751,
+      "mean_token_accuracy": 0.9796596229076385,
+      "num_tokens": 14520902.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.9320916712284089,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 0.9210566878318787,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.051744121313095096,
+      "mean_token_accuracy": 0.9792148530483246,
+      "num_tokens": 14709215.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.928934782743454,
+      "epoch": 1.063257065948856,
+      "grad_norm": 0.7442036867141724,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.0486875057220459,
+      "mean_token_accuracy": 0.9798122465610504,
+      "num_tokens": 14898242.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.942881977558136,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.0910941362380981,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.052297019958496095,
+      "mean_token_accuracy": 0.9793200254440307,
+      "num_tokens": 15086835.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.9503917813301086,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 0.8091554641723633,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.049927744269371035,
+      "mean_token_accuracy": 0.9802082359790802,
+      "num_tokens": 15275042.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.9426830470561981,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.0589245557785034,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.053803551197052005,
+      "mean_token_accuracy": 0.9785399675369263,
+      "num_tokens": 15463923.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.9445010781288147,
+      "epoch": 1.117092866756393,
+      "grad_norm": 0.8410441279411316,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.053181976079940796,
+      "mean_token_accuracy": 0.9792523324489594,
+      "num_tokens": 15652526.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.946729838848114,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 0.8246080279350281,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.054291915893554685,
+      "mean_token_accuracy": 0.9791357696056366,
+      "num_tokens": 15841340.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.9433728814125061,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 0.9200296401977539,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.05073915719985962,
+      "mean_token_accuracy": 0.9796931505203247,
+      "num_tokens": 16029478.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.9494417488574982,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 0.8342758417129517,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.050929927825927736,
+      "mean_token_accuracy": 0.9793219089508056,
+      "num_tokens": 16218014.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.9409256100654602,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.7911009192466736,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.049964362382888795,
+      "mean_token_accuracy": 0.9798487305641175,
+      "num_tokens": 16406420.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.9429958581924438,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.4877018928527832,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.04969423711299896,
+      "mean_token_accuracy": 0.9801322638988494,
+      "num_tokens": 16594948.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.9542136013507843,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 0.8030785918235779,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.04981146454811096,
+      "mean_token_accuracy": 0.9796162784099579,
+      "num_tokens": 16782983.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.9367722630500793,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.1169800758361816,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.05210963487625122,
+      "mean_token_accuracy": 0.9790139615535736,
+      "num_tokens": 16971394.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.9487208306789399,
+      "epoch": 1.224764468371467,
+      "grad_norm": 0.8650282621383667,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.05012243390083313,
+      "mean_token_accuracy": 0.9805107891559601,
+      "num_tokens": 17159951.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.9441231489181519,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 0.8193967938423157,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.048866665363311766,
+      "mean_token_accuracy": 0.9807584881782532,
+      "num_tokens": 17348711.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.9361630439758301,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3613612651824951,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.05324091911315918,
+      "mean_token_accuracy": 0.97904953956604,
+      "num_tokens": 17537116.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.9562997639179229,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.8121878504753113,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.05011019706726074,
+      "mean_token_accuracy": 0.9798731088638306,
+      "num_tokens": 17725567.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.9359076261520386,
+      "epoch": 1.278600269179004,
+      "grad_norm": 0.7838549613952637,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.04834386110305786,
+      "mean_token_accuracy": 0.9804529249668121,
+      "num_tokens": 17914191.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.9527446150779724,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0981712341308594,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.055077624320983884,
+      "mean_token_accuracy": 0.9776535391807556,
+      "num_tokens": 18103184.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.9574812948703766,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 0.8291501402854919,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.050153911113739014,
+      "mean_token_accuracy": 0.9792965054512024,
+      "num_tokens": 18291351.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.939585280418396,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.0864412784576416,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.05132197737693787,
+      "mean_token_accuracy": 0.9792291462421417,
+      "num_tokens": 18480192.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.9468406856060028,
+      "epoch": 1.332436069986541,
+      "grad_norm": 0.9772785902023315,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.04844954013824463,
+      "mean_token_accuracy": 0.9804142415523529,
+      "num_tokens": 18669028.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.9413834273815155,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 0.9020094871520996,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.05075312852859497,
+      "mean_token_accuracy": 0.9790591120719909,
+      "num_tokens": 18857581.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.9526253461837768,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.0524990558624268,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.049737372994422914,
+      "mean_token_accuracy": 0.9795652389526367,
+      "num_tokens": 19046105.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.9484909176826477,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.0715620517730713,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.04935494959354401,
+      "mean_token_accuracy": 0.9797740161418915,
+      "num_tokens": 19234730.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.9398573458194732,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2134987115859985,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.0518394410610199,
+      "mean_token_accuracy": 0.9793788075447083,
+      "num_tokens": 19423572.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.9465648651123046,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.004693627357483,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.04776117205619812,
+      "mean_token_accuracy": 0.9813261866569519,
+      "num_tokens": 19612465.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.9448004186153411,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.2913438081741333,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.04807930588722229,
+      "mean_token_accuracy": 0.9805689513683319,
+      "num_tokens": 19800482.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.9425322234630584,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1068469285964966,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.05053595900535583,
+      "mean_token_accuracy": 0.9794536828994751,
+      "num_tokens": 19989444.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.9472232103347779,
+      "epoch": 1.440107671601615,
+      "grad_norm": 0.7409384250640869,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.052136778831481934,
+      "mean_token_accuracy": 0.979197359085083,
+      "num_tokens": 20178482.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.9450975477695465,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.3440778255462646,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.05258495211601257,
+      "mean_token_accuracy": 0.9798152863979339,
+      "num_tokens": 20367345.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.9468007743358612,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 0.8377829194068909,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.04893603026866913,
+      "mean_token_accuracy": 0.9805733859539032,
+      "num_tokens": 20555817.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.9505827724933624,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.776889979839325,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.05367119312286377,
+      "mean_token_accuracy": 0.9779646098613739,
+      "num_tokens": 20744001.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.9516380190849304,
+      "epoch": 1.493943472409152,
+      "grad_norm": 0.8598276376724243,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.05071394443511963,
+      "mean_token_accuracy": 0.979692417383194,
+      "num_tokens": 20931715.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.9351628720760345,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3307439088821411,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.051372635364532473,
+      "mean_token_accuracy": 0.9797907948493958,
+      "num_tokens": 21120803.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.9268040716648102,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9320250749588013,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.04939974546432495,
+      "mean_token_accuracy": 0.98038170337677,
+      "num_tokens": 21309384.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.9241463243961334,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.065938115119934,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.04856643378734589,
+      "mean_token_accuracy": 0.9800511121749877,
+      "num_tokens": 21498234.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.9350252687931061,
+      "epoch": 1.547779273216689,
+      "grad_norm": 0.6925032138824463,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.0512526273727417,
+      "mean_token_accuracy": 0.979562646150589,
+      "num_tokens": 21687119.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.9403733670711517,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.2069101333618164,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.049795085191726686,
+      "mean_token_accuracy": 0.9806506633758545,
+      "num_tokens": 21876094.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.9370538294315338,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.1759546995162964,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.05461466312408447,
+      "mean_token_accuracy": 0.9776057600975037,
+      "num_tokens": 22064836.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.9517778217792511,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 0.8965149521827698,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.05289326906204224,
+      "mean_token_accuracy": 0.9782088756561279,
+      "num_tokens": 22253065.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.9372851848602295,
+      "epoch": 1.601615074024226,
+      "grad_norm": 0.7192108631134033,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.048973691463470456,
+      "mean_token_accuracy": 0.9806210815906524,
+      "num_tokens": 22441206.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.9122351944446564,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 0.8763614296913147,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.0461783230304718,
+      "mean_token_accuracy": 0.9817197680473327,
+      "num_tokens": 22629637.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.9162654876708984,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 0.9806778430938721,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.04827338755130768,
+      "mean_token_accuracy": 0.9806499361991883,
+      "num_tokens": 22818369.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.9125026524066925,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 0.9685350656509399,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.048379385471343996,
+      "mean_token_accuracy": 0.980905145406723,
+      "num_tokens": 23006863.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.9119791567325592,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.8717457056045532,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.050873959064483644,
+      "mean_token_accuracy": 0.979601925611496,
+      "num_tokens": 23195735.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.9244007229804992,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.0574654340744019,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.05112813711166382,
+      "mean_token_accuracy": 0.979493772983551,
+      "num_tokens": 23383862.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.9225810945034028,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 0.9628807306289673,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.050480544567108154,
+      "mean_token_accuracy": 0.9797678411006927,
+      "num_tokens": 23572292.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.9287579476833343,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.015265941619873,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.05133126378059387,
+      "mean_token_accuracy": 0.9797986805438995,
+      "num_tokens": 23760252.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.9133845925331116,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 0.8996883630752563,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.05044950246810913,
+      "mean_token_accuracy": 0.9795303404331207,
+      "num_tokens": 23949031.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.9213698863983154,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.0490264892578125,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.04797698855400086,
+      "mean_token_accuracy": 0.9806023716926575,
+      "num_tokens": 24137829.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.9173697710037232,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.3648769855499268,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.049194514751434326,
+      "mean_token_accuracy": 0.980712479352951,
+      "num_tokens": 24326586.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.9242741882801055,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.105125069618225,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.048667973279953,
+      "mean_token_accuracy": 0.9801449298858642,
+      "num_tokens": 24514904.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.9162627995014191,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.8887938857078552,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.04878672957420349,
+      "mean_token_accuracy": 0.980469423532486,
+      "num_tokens": 24703688.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.9149538338184356,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 0.8838976621627808,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.04892318844795227,
+      "mean_token_accuracy": 0.9808776795864105,
+      "num_tokens": 24892107.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.914735221862793,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1152070760726929,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.04825109839439392,
+      "mean_token_accuracy": 0.9804604113101959,
+      "num_tokens": 25080130.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.9170725226402283,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9479517340660095,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.051445144414901736,
+      "mean_token_accuracy": 0.9792819261550904,
+      "num_tokens": 25268652.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.9142911911010743,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 0.8172292709350586,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.0475692093372345,
+      "mean_token_accuracy": 0.9809505581855774,
+      "num_tokens": 25457120.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.9019249439239502,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 0.8183121681213379,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.04979957342147827,
+      "mean_token_accuracy": 0.9799730658531189,
+      "num_tokens": 25645699.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.9181066155433655,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.8256579041481018,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.04996164441108704,
+      "mean_token_accuracy": 0.9802247405052185,
+      "num_tokens": 25834183.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.9171363770961761,
+      "epoch": 1.857335127860027,
+      "grad_norm": 0.9727098345756531,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.048660767078399655,
+      "mean_token_accuracy": 0.9805064260959625,
+      "num_tokens": 26023317.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.9112720847129822,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 0.9826673865318298,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.04775593280792236,
+      "mean_token_accuracy": 0.9809929549694061,
+      "num_tokens": 26212115.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.9034272134304047,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.9951306581497192,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.04939360618591308,
+      "mean_token_accuracy": 0.9800347864627839,
+      "num_tokens": 26401167.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.897865754365921,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.0163404941558838,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05002856254577637,
+      "mean_token_accuracy": 0.9793690323829651,
+      "num_tokens": 26590570.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.9004576802253723,
+      "epoch": 1.911170928667564,
+      "grad_norm": 0.6666616201400757,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.04905453026294708,
+      "mean_token_accuracy": 0.9803856253623963,
+      "num_tokens": 26779481.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.8978902101516724,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.9295416474342346,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.05261261463165283,
+      "mean_token_accuracy": 0.9785708487033844,
+      "num_tokens": 26968262.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.8962675571441651,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 0.88262939453125,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.04765265882015228,
+      "mean_token_accuracy": 0.98052077293396,
+      "num_tokens": 27157631.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.8909463047981262,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.4205769300460815,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.049406200647354126,
+      "mean_token_accuracy": 0.979697072505951,
+      "num_tokens": 27345921.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8969364166259766,
+      "epoch": 1.965006729475101,
+      "grad_norm": 0.9517145752906799,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0510346531867981,
+      "mean_token_accuracy": 0.9793386399745941,
+      "num_tokens": 27534751.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.9092479169368743,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.200056791305542,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.04974203705787659,
+      "mean_token_accuracy": 0.9800146698951722,
+      "num_tokens": 27723763.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.9059641897678375,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.072447657585144,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.05065792202949524,
+      "mean_token_accuracy": 0.9799223959445953,
+      "num_tokens": 27911784.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.9083398975384464,
+      "eval_loss": 0.05432562157511711,
+      "eval_mean_token_accuracy": 0.9782073607869969,
+      "eval_num_tokens": 28025286.0,
+      "eval_runtime": 24.4304,
+      "eval_samples_per_second": 204.663,
+      "eval_steps_per_second": 6.426,
+      "step": 1486
+    },
+    {
+      "entropy": 0.9113237977027893,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.0167063474655151,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.044942992925643924,
+      "mean_token_accuracy": 0.9818171083927154,
+      "num_tokens": 28100906.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.8780008792877197,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.9552241563796997,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.036248764395713805,
+      "mean_token_accuracy": 0.9852245450019836,
+      "num_tokens": 28289445.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.8721176147460937,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 0.8951388001441956,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.037987279891967776,
+      "mean_token_accuracy": 0.9850614190101623,
+      "num_tokens": 28477972.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.8922657191753387,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 0.9028105735778809,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.03588914275169373,
+      "mean_token_accuracy": 0.9863102614879609,
+      "num_tokens": 28666272.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.877799642086029,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.153886318206787,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.03577309250831604,
+      "mean_token_accuracy": 0.9863148391246795,
+      "num_tokens": 28854780.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.8708971381187439,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.042955994606018,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.038266432285308835,
+      "mean_token_accuracy": 0.9848419308662415,
+      "num_tokens": 29043741.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.8800721943378449,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 0.6789060235023499,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.032948991656303404,
+      "mean_token_accuracy": 0.9871293902397156,
+      "num_tokens": 29232719.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.8743580460548401,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.1780879497528076,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.036874374747276305,
+      "mean_token_accuracy": 0.9859182178974152,
+      "num_tokens": 29421817.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.8851303637027741,
+      "epoch": 2.113055181695828,
+      "grad_norm": 0.747734785079956,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.0364631175994873,
+      "mean_token_accuracy": 0.9862085223197937,
+      "num_tokens": 29610344.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.8798732101917267,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.5001860857009888,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.03813003897666931,
+      "mean_token_accuracy": 0.9852604746818543,
+      "num_tokens": 29798997.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.8797551989555359,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 0.9593478441238403,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.0394927829504013,
+      "mean_token_accuracy": 0.984604275226593,
+      "num_tokens": 29987809.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.8860546290874481,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.8347703218460083,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.036699697375297546,
+      "mean_token_accuracy": 0.985380882024765,
+      "num_tokens": 30176234.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.8783667802810669,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1154481172561646,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.03756999969482422,
+      "mean_token_accuracy": 0.9853451430797577,
+      "num_tokens": 30364657.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.8684478521347045,
+      "epoch": 2.180349932705249,
+      "grad_norm": 0.7515475153923035,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.0382376104593277,
+      "mean_token_accuracy": 0.9854029655456543,
+      "num_tokens": 30552904.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.8636964917182922,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 0.7711939215660095,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.036220991611480714,
+      "mean_token_accuracy": 0.9860713243484497,
+      "num_tokens": 30741259.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.8592016279697419,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.1143887042999268,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.036316031217575075,
+      "mean_token_accuracy": 0.985757052898407,
+      "num_tokens": 30930144.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.8671079576015472,
+      "epoch": 2.220726783310902,
+      "grad_norm": 0.8980015516281128,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.03822658061981201,
+      "mean_token_accuracy": 0.9847764372825623,
+      "num_tokens": 31118624.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.8785548269748688,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.094323992729187,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.03736622333526611,
+      "mean_token_accuracy": 0.9850081980228425,
+      "num_tokens": 31306833.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.866977310180664,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8372092843055725,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.0401554524898529,
+      "mean_token_accuracy": 0.9840337932109833,
+      "num_tokens": 31495503.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.8787918210029602,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.5408164262771606,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0368030846118927,
+      "mean_token_accuracy": 0.9847234487533569,
+      "num_tokens": 31683865.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.872721153497696,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.1115421056747437,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.037859299778938295,
+      "mean_token_accuracy": 0.9851646661758423,
+      "num_tokens": 31871903.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.8774094223976135,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.0301331281661987,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.038671016693115234,
+      "mean_token_accuracy": 0.9840058028697968,
+      "num_tokens": 32060500.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.889606237411499,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8601903915405273,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.039603835344314574,
+      "mean_token_accuracy": 0.9840235590934754,
+      "num_tokens": 32248744.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.886734277009964,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.1029839515686035,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.039513933658599856,
+      "mean_token_accuracy": 0.9844573020935059,
+      "num_tokens": 32437640.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.8843017637729644,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.2545154094696045,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.036970189213752745,
+      "mean_token_accuracy": 0.985239815711975,
+      "num_tokens": 32625986.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.8913422048091888,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.9613803625106812,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.03811657428741455,
+      "mean_token_accuracy": 0.9852804243564606,
+      "num_tokens": 32814531.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.8907467782497406,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.1811398267745972,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.03903660774230957,
+      "mean_token_accuracy": 0.9842524945735931,
+      "num_tokens": 33002960.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.8833664715290069,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.8338477611541748,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.03982087969779968,
+      "mean_token_accuracy": 0.9841201484203339,
+      "num_tokens": 33191183.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.8783826470375061,
+      "epoch": 2.382234185733513,
+      "grad_norm": 0.6904510259628296,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.03699290752410889,
+      "mean_token_accuracy": 0.9854120731353759,
+      "num_tokens": 33379798.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.8871048510074615,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.009539246559143,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.03926805555820465,
+      "mean_token_accuracy": 0.9843246698379516,
+      "num_tokens": 33568531.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.8973462700843811,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.8916401267051697,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.039596831798553465,
+      "mean_token_accuracy": 0.9839228212833404,
+      "num_tokens": 33756333.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.8954619467258453,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.056504487991333,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0396859347820282,
+      "mean_token_accuracy": 0.9837161123752594,
+      "num_tokens": 33944904.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.8952643811702728,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0442084074020386,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.037324142456054685,
+      "mean_token_accuracy": 0.9852897703647614,
+      "num_tokens": 34133530.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.8874911010265351,
+      "epoch": 2.449528936742934,
+      "grad_norm": 0.9926224946975708,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.03637495338916778,
+      "mean_token_accuracy": 0.9853426337242126,
+      "num_tokens": 34322042.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.8900584518909455,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.001734733581543,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04029585719108582,
+      "mean_token_accuracy": 0.9842372059822082,
+      "num_tokens": 34510259.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.8986345052719116,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 0.772941529750824,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.0401084691286087,
+      "mean_token_accuracy": 0.9841427087783814,
+      "num_tokens": 34698669.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.8911370873451233,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.895031213760376,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.03941032290458679,
+      "mean_token_accuracy": 0.9841966688632965,
+      "num_tokens": 34887160.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.8901963472366333,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0677918195724487,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.03503885865211487,
+      "mean_token_accuracy": 0.9856241464614868,
+      "num_tokens": 35075891.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.8817630231380462,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.1200164556503296,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.03878425657749176,
+      "mean_token_accuracy": 0.9846492648124695,
+      "num_tokens": 35264455.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.8897477686405182,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.8058190941810608,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.041558006405830385,
+      "mean_token_accuracy": 0.9830883860588073,
+      "num_tokens": 35453251.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.8873181045055389,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.7765992283821106,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.03555050492286682,
+      "mean_token_accuracy": 0.9858225345611572,
+      "num_tokens": 35641959.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.8782606959342957,
+      "epoch": 2.557200538358008,
+      "grad_norm": 0.8488866686820984,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.03606923818588257,
+      "mean_token_accuracy": 0.9858632445335388,
+      "num_tokens": 35829636.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.8869808673858642,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.1734739542007446,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.040956351161003116,
+      "mean_token_accuracy": 0.9835374176502227,
+      "num_tokens": 36017818.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.8850924432277679,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.3584387302398682,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.042437011003494264,
+      "mean_token_accuracy": 0.9826479077339172,
+      "num_tokens": 36206505.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.8801687896251679,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2862240076065063,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.03880060315132141,
+      "mean_token_accuracy": 0.9843404710292816,
+      "num_tokens": 36395802.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.8752781748771667,
+      "epoch": 2.611036339165545,
+      "grad_norm": 0.7341249585151672,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.03776344656944275,
+      "mean_token_accuracy": 0.9852877616882324,
+      "num_tokens": 36584370.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.8759494423866272,
+      "epoch": 2.624495289367429,
+      "grad_norm": 0.9807276129722595,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.03780297338962555,
+      "mean_token_accuracy": 0.9853227376937866,
+      "num_tokens": 36772865.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.8753129065036773,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.1391628980636597,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.03953765034675598,
+      "mean_token_accuracy": 0.9839197635650635,
+      "num_tokens": 36961462.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.8771899223327637,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.0865391492843628,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.03856399655342102,
+      "mean_token_accuracy": 0.9848688066005706,
+      "num_tokens": 37150486.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.8774881184101104,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.8561595678329468,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.03795175850391388,
+      "mean_token_accuracy": 0.9849309325218201,
+      "num_tokens": 37338890.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.8841517448425293,
+      "epoch": 2.678331090174966,
+      "grad_norm": 0.9458399415016174,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.03860213160514832,
+      "mean_token_accuracy": 0.9842230796813964,
+      "num_tokens": 37526886.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.8809043228626251,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 0.9360871315002441,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.03709094822406769,
+      "mean_token_accuracy": 0.9855118036270142,
+      "num_tokens": 37715408.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.8805519282817841,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.8852124214172363,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.03869341611862183,
+      "mean_token_accuracy": 0.9846724212169647,
+      "num_tokens": 37904328.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.8796228408813477,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.189297080039978,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.03675769567489624,
+      "mean_token_accuracy": 0.9854080140590668,
+      "num_tokens": 38092776.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.8722020089626312,
+      "epoch": 2.732166890982503,
+      "grad_norm": 0.7027947306632996,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.03889244794845581,
+      "mean_token_accuracy": 0.984558242559433,
+      "num_tokens": 38282221.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.8746632814407349,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 0.9799376726150513,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.03815680146217346,
+      "mean_token_accuracy": 0.9845012128353119,
+      "num_tokens": 38470633.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.8766567528247833,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 0.850459098815918,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.039029371738433835,
+      "mean_token_accuracy": 0.9845906794071198,
+      "num_tokens": 38659048.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.8778688013553619,
+      "epoch": 2.772543741588156,
+      "grad_norm": 0.8540134429931641,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.039645448327064514,
+      "mean_token_accuracy": 0.9840566098690033,
+      "num_tokens": 38847370.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.8758600771427154,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8036021590232849,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.038226932287216187,
+      "mean_token_accuracy": 0.9845647215843201,
+      "num_tokens": 39036040.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.8652268946170807,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 0.8526738286018372,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.03798363208770752,
+      "mean_token_accuracy": 0.9845202267169952,
+      "num_tokens": 39224695.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.8707200348377228,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 0.9852389097213745,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.037227436900138855,
+      "mean_token_accuracy": 0.9852050960063934,
+      "num_tokens": 39413360.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.8711447060108185,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.228277325630188,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.03698050379753113,
+      "mean_token_accuracy": 0.9852443158626556,
+      "num_tokens": 39602544.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.8742912471294403,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 0.8440544605255127,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.03757670521736145,
+      "mean_token_accuracy": 0.9848618268966675,
+      "num_tokens": 39791340.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.8801726162433624,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0955252647399902,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.040004149079322815,
+      "mean_token_accuracy": 0.9841345012187958,
+      "num_tokens": 39980212.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.8931492984294891,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.5722652673721313,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.038096648454666135,
+      "mean_token_accuracy": 0.9849136590957641,
+      "num_tokens": 40169122.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.8767493844032288,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.7663730382919312,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.03729096055030823,
+      "mean_token_accuracy": 0.985169780254364,
+      "num_tokens": 40357678.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.8770561516284943,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.1505672931671143,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.03984796404838562,
+      "mean_token_accuracy": 0.9838845670223236,
+      "num_tokens": 40546892.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.8789129912853241,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 0.8355761766433716,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.03769223690032959,
+      "mean_token_accuracy": 0.9847559213638306,
+      "num_tokens": 40736109.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.8781779289245606,
+      "epoch": 2.920592193808883,
+      "grad_norm": 0.9103918075561523,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.038262826204299924,
+      "mean_token_accuracy": 0.984741848707199,
+      "num_tokens": 40924341.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.8552807629108429,
+      "epoch": 2.934051144010767,
+      "grad_norm": 0.8338510990142822,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.0355743408203125,
+      "mean_token_accuracy": 0.9855036079883576,
+      "num_tokens": 41113849.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.8705591142177582,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 0.8334057927131653,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.037476515769958495,
+      "mean_token_accuracy": 0.9854551255702972,
+      "num_tokens": 41303130.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.8809274792671203,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 0.9424103498458862,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.03624279499053955,
+      "mean_token_accuracy": 0.9856135487556458,
+      "num_tokens": 41491750.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.8818133771419525,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.04243004322052,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.039132320880889894,
+      "mean_token_accuracy": 0.984746390581131,
+      "num_tokens": 41680493.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.8838395297527313,
+      "epoch": 2.987886944818304,
+      "grad_norm": 0.8650784492492676,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.03822052478790283,
+      "mean_token_accuracy": 0.9852108955383301,
+      "num_tokens": 41868378.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.8800126549544608,
+      "eval_loss": 0.05468416586518288,
+      "eval_mean_token_accuracy": 0.9781015617832257,
+      "eval_num_tokens": 42038037.0,
+      "eval_runtime": 24.4825,
+      "eval_samples_per_second": 204.228,
+      "eval_steps_per_second": 6.413,
+      "step": 2229
+    },
+    {
+      "entropy": 0.8773060619831086,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.7845072150230408,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.0365300714969635,
+      "mean_token_accuracy": 0.985853761434555,
+      "num_tokens": 42056717.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.8720550239086151,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 0.9917275905609131,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.030071893334388734,
+      "mean_token_accuracy": 0.9890478909015655,
+      "num_tokens": 42245058.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.8489451467990875,
+      "epoch": 3.028263795423957,
+      "grad_norm": 0.9414294362068176,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.027054613828659056,
+      "mean_token_accuracy": 0.9894806027412415,
+      "num_tokens": 42433468.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.848755830526352,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.7276235818862915,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.026897600293159483,
+      "mean_token_accuracy": 0.9893308103084564,
+      "num_tokens": 42621657.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.8402929604053497,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 0.9455838799476624,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.027918827533721925,
+      "mean_token_accuracy": 0.9888385117053986,
+      "num_tokens": 42810533.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.8434606611728668,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 0.7503429651260376,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.025769710540771484,
+      "mean_token_accuracy": 0.989881956577301,
+      "num_tokens": 42999085.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.8459890782833099,
+      "epoch": 3.082099596231494,
+      "grad_norm": 0.736440896987915,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.028781753778457642,
+      "mean_token_accuracy": 0.9882656812667847,
+      "num_tokens": 43188222.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.8493306159973144,
+      "epoch": 3.095558546433378,
+      "grad_norm": 0.9577816128730774,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.02689719796180725,
+      "mean_token_accuracy": 0.9889744818210602,
+      "num_tokens": 43376648.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.8512665331363678,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 0.9631588459014893,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.031439167261123654,
+      "mean_token_accuracy": 0.9874942421913147,
+      "num_tokens": 43564769.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.8519513130187988,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1011922359466553,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.02696958780288696,
+      "mean_token_accuracy": 0.9900951981544495,
+      "num_tokens": 43753362.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.8507618844509125,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9351623058319092,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.026315575838088988,
+      "mean_token_accuracy": 0.989511913061142,
+      "num_tokens": 43942063.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.8506478369235992,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.096564769744873,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.029049152135849,
+      "mean_token_accuracy": 0.9886228442192078,
+      "num_tokens": 44130381.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.8624941289424897,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 0.7859402298927307,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.027744990587234498,
+      "mean_token_accuracy": 0.9890527307987214,
+      "num_tokens": 44319243.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.8627866744995117,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 0.7764605283737183,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.02571650445461273,
+      "mean_token_accuracy": 0.9903442323207855,
+      "num_tokens": 44507698.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.8566473722457886,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.8182396292686462,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.025625643134117127,
+      "mean_token_accuracy": 0.9897150516510009,
+      "num_tokens": 44695954.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.8555088877677918,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.0277098417282104,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.028235083818435668,
+      "mean_token_accuracy": 0.9894341349601745,
+      "num_tokens": 44884558.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.8562702238559723,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.0676493644714355,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.027713948488235475,
+      "mean_token_accuracy": 0.9891053259372711,
+      "num_tokens": 45073381.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.8520576059818268,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 0.7926117181777954,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.028936928510665892,
+      "mean_token_accuracy": 0.988807988166809,
+      "num_tokens": 45262350.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.860496312379837,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.1108510494232178,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030533093214035033,
+      "mean_token_accuracy": 0.9882748067378998,
+      "num_tokens": 45450782.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.872324675321579,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.3057194948196411,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.032421305775642395,
+      "mean_token_accuracy": 0.9879753351211548,
+      "num_tokens": 45639122.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.8778823137283325,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.289680004119873,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03096393346786499,
+      "mean_token_accuracy": 0.9883370697498322,
+      "num_tokens": 45827730.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.8627406179904937,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.8361398577690125,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.02957139313220978,
+      "mean_token_accuracy": 0.9884674072265625,
+      "num_tokens": 46016274.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.8584554970264435,
+      "epoch": 3.297442799461642,
+      "grad_norm": 0.9330723881721497,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.029568272829055785,
+      "mean_token_accuracy": 0.9882686555385589,
+      "num_tokens": 46204679.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.8515178024768829,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.0296216011047363,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.027788180112838744,
+      "mean_token_accuracy": 0.9888734877109527,
+      "num_tokens": 46393989.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.8488850116729736,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1329699754714966,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.028057461977005003,
+      "mean_token_accuracy": 0.9895590543746948,
+      "num_tokens": 46582790.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.849625188112259,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 0.9267219305038452,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.02802973985671997,
+      "mean_token_accuracy": 0.988851535320282,
+      "num_tokens": 46771138.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.854446280002594,
+      "epoch": 3.351278600269179,
+      "grad_norm": 0.8713932037353516,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.030220019817352294,
+      "mean_token_accuracy": 0.9882843673229218,
+      "num_tokens": 46960054.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.8599806547164917,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.675019383430481,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.02958865463733673,
+      "mean_token_accuracy": 0.9880296885967255,
+      "num_tokens": 47149567.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.8549504458904267,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.2176353931427002,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.027722162008285523,
+      "mean_token_accuracy": 0.9890090465545655,
+      "num_tokens": 47338060.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.8552233815193176,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9571717977523804,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.028924965858459474,
+      "mean_token_accuracy": 0.9884298503398895,
+      "num_tokens": 47525895.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.8521630108356476,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.965713620185852,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.030511701107025148,
+      "mean_token_accuracy": 0.9881398677825928,
+      "num_tokens": 47714524.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.8537400126457214,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 0.9005307555198669,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.030432754755020143,
+      "mean_token_accuracy": 0.9876708447933197,
+      "num_tokens": 47903188.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.8596058666706086,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 0.8159075379371643,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.027927806973457335,
+      "mean_token_accuracy": 0.9889210939407349,
+      "num_tokens": 48091334.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.8482735753059387,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0411661863327026,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03003154993057251,
+      "mean_token_accuracy": 0.9882730603218078,
+      "num_tokens": 48279622.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.8466426789760589,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.189595341682434,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03237656056880951,
+      "mean_token_accuracy": 0.9869072616100312,
+      "num_tokens": 48467527.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.8545588493347168,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0700056552886963,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.0306060791015625,
+      "mean_token_accuracy": 0.9880664825439454,
+      "num_tokens": 48656151.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.8503433585166931,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.7099478244781494,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.02984013557434082,
+      "mean_token_accuracy": 0.9879442811012268,
+      "num_tokens": 48844796.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.8464376091957092,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9877128005027771,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.0306446373462677,
+      "mean_token_accuracy": 0.9883591532707214,
+      "num_tokens": 49033216.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.8477609276771545,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.3324531316757202,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.030148410797119142,
+      "mean_token_accuracy": 0.9879989802837372,
+      "num_tokens": 49221425.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.8461863219738006,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8938570022583008,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.02682304084300995,
+      "mean_token_accuracy": 0.9897881150245667,
+      "num_tokens": 49409785.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.8362689018249512,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 0.7423873543739319,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.02689332664012909,
+      "mean_token_accuracy": 0.9899903059005737,
+      "num_tokens": 49598524.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.8395139217376709,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.921755850315094,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.030062124133110046,
+      "mean_token_accuracy": 0.9880835592746735,
+      "num_tokens": 49787241.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.8467226803302765,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0296459197998047,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.030666223168373107,
+      "mean_token_accuracy": 0.9877098739147187,
+      "num_tokens": 49975844.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.833392071723938,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2861948013305664,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.028174835443496703,
+      "mean_token_accuracy": 0.989056545495987,
+      "num_tokens": 50164857.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.8340730309486389,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 0.8116776347160339,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.028390336036682128,
+      "mean_token_accuracy": 0.9886897742748261,
+      "num_tokens": 50353452.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.8357116639614105,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.8362084031105042,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.029806244373321533,
+      "mean_token_accuracy": 0.988309508562088,
+      "num_tokens": 50542461.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.8363270878791809,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8438499569892883,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.028577303886413573,
+      "mean_token_accuracy": 0.9886945366859436,
+      "num_tokens": 50730872.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.8310261785984039,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.1920002698898315,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.030321845412254335,
+      "mean_token_accuracy": 0.9878708899021149,
+      "num_tokens": 50919889.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.8380252480506897,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 0.9409399032592773,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.030501589179039,
+      "mean_token_accuracy": 0.9872083842754364,
+      "num_tokens": 51108079.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.8422612130641938,
+      "epoch": 3.660834454912517,
+      "grad_norm": 0.8370440006256104,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.02974823713302612,
+      "mean_token_accuracy": 0.9885554671287536,
+      "num_tokens": 51296913.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.8387595832347869,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1360642910003662,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.026970452070236205,
+      "mean_token_accuracy": 0.9888331353664398,
+      "num_tokens": 51485678.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.8403031289577484,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.1195677518844604,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.029451662302017213,
+      "mean_token_accuracy": 0.988847154378891,
+      "num_tokens": 51674765.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.8499106168746948,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 0.9573526978492737,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.030826866626739502,
+      "mean_token_accuracy": 0.9875578820705414,
+      "num_tokens": 51863924.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.8576709866523743,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9386286735534668,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.02946900427341461,
+      "mean_token_accuracy": 0.9885939955711365,
+      "num_tokens": 52052011.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.8404816091060638,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.095536708831787,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.030070826411247253,
+      "mean_token_accuracy": 0.9881400763988495,
+      "num_tokens": 52240955.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.8346689164638519,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9964426159858704,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03005717396736145,
+      "mean_token_accuracy": 0.9883057296276092,
+      "num_tokens": 52429544.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.8377436757087707,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.8499778509140015,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.029283815622329713,
+      "mean_token_accuracy": 0.9882387280464172,
+      "num_tokens": 52618405.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.8328788638114929,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9594390988349915,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.02666555345058441,
+      "mean_token_accuracy": 0.9889323055744171,
+      "num_tokens": 52806689.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.829415249824524,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.3587099313735962,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.028349629044532774,
+      "mean_token_accuracy": 0.9890907645225525,
+      "num_tokens": 52995248.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.8335795283317566,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.077789306640625,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.03001731038093567,
+      "mean_token_accuracy": 0.9878358781337738,
+      "num_tokens": 53183607.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.8316770553588867,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 0.7265951037406921,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.030136841535568237,
+      "mean_token_accuracy": 0.9880174934864044,
+      "num_tokens": 53372735.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.8310399711132049,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.6605978012084961,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.026665833592414857,
+      "mean_token_accuracy": 0.9892704486846924,
+      "num_tokens": 53561776.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.8258654296398162,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.6560797691345215,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.026718440651893615,
+      "mean_token_accuracy": 0.989847868680954,
+      "num_tokens": 53750019.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.8221413195133209,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 0.7495356798171997,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.027431756258010864,
+      "mean_token_accuracy": 0.9892430484294892,
+      "num_tokens": 53938914.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.8341928541660308,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.0676971673965454,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.028432759642601012,
+      "mean_token_accuracy": 0.9889024317264556,
+      "num_tokens": 54127763.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.8359149813652038,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.380476713180542,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.028149163722991942,
+      "mean_token_accuracy": 0.9884953200817108,
+      "num_tokens": 54315930.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.8294597804546356,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0728843212127686,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.028733691573143004,
+      "mean_token_accuracy": 0.9884523034095765,
+      "num_tokens": 54504632.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.8245523810386658,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 0.9723307490348816,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.02793261408805847,
+      "mean_token_accuracy": 0.9891038477420807,
+      "num_tokens": 54693752.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.8279620766639709,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.003624439239502,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.028367668390274048,
+      "mean_token_accuracy": 0.9893603324890137,
+      "num_tokens": 54882030.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.8306584179401397,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.8481361865997314,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.029952478408813477,
+      "mean_token_accuracy": 0.9881332099437714,
+      "num_tokens": 55070549.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.8316469848155975,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.7631049156188965,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.029351180791854857,
+      "mean_token_accuracy": 0.9888184070587158,
+      "num_tokens": 55258999.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.8221287310123444,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 0.7234840393066406,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.027709537744522096,
+      "mean_token_accuracy": 0.9890308439731598,
+      "num_tokens": 55448047.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.8239192545413971,
+      "epoch": 3.970390309555855,
+      "grad_norm": 0.8497020602226257,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.029604345560073853,
+      "mean_token_accuracy": 0.9881375730037689,
+      "num_tokens": 55636566.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.8319927275180816,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.1599552631378174,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.02763860821723938,
+      "mean_token_accuracy": 0.9891406774520874,
+      "num_tokens": 55825049.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.8404937744140625,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 0.906755805015564,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.029578033089637756,
+      "mean_token_accuracy": 0.9878996431827545,
+      "num_tokens": 56012965.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.8419649528849656,
+      "eval_loss": 0.0570383295416832,
+      "eval_mean_token_accuracy": 0.9787763862093543,
+      "eval_num_tokens": 56050518.0,
+      "eval_runtime": 24.4889,
+      "eval_samples_per_second": 204.174,
+      "eval_steps_per_second": 6.411,
+      "step": 2972
+    },
+    {
+      "entropy": 0.8328965544700623,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7073294520378113,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.020916813611984254,
+      "mean_token_accuracy": 0.992138022184372,
+      "num_tokens": 56201621.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.8153974890708924,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.8904889822006226,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.01942949742078781,
+      "mean_token_accuracy": 0.9927055358886718,
+      "num_tokens": 56390018.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.8080023109912873,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.0828146934509277,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.018959003686904907,
+      "mean_token_accuracy": 0.9928978025913239,
+      "num_tokens": 56578046.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.8056950688362121,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7407417893409729,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019250378012657166,
+      "mean_token_accuracy": 0.9926685452461242,
+      "num_tokens": 56767192.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.8124530673027038,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.0355021953582764,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.01904640197753906,
+      "mean_token_accuracy": 0.9925767481327057,
+      "num_tokens": 56955769.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.810802960395813,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.3759289979934692,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.020995599031448365,
+      "mean_token_accuracy": 0.9917719483375549,
+      "num_tokens": 57144300.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.8159192740917206,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.112034559249878,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.01948906183242798,
+      "mean_token_accuracy": 0.9928658187389374,
+      "num_tokens": 57332935.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.821906328201294,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.0319583415985107,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019549430906772615,
+      "mean_token_accuracy": 0.9922966659069061,
+      "num_tokens": 57520775.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.8133022606372833,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.067180871963501,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.01833198368549347,
+      "mean_token_accuracy": 0.9934999167919158,
+      "num_tokens": 57709495.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.8150732755661011,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.277079701423645,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.019589337706565856,
+      "mean_token_accuracy": 0.992507028579712,
+      "num_tokens": 57897461.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.8159323990345001,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.3150911331176758,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.01963164806365967,
+      "mean_token_accuracy": 0.9926789402961731,
+      "num_tokens": 58085915.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.8147098064422608,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.0073580741882324,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.02144385576248169,
+      "mean_token_accuracy": 0.9918489575386047,
+      "num_tokens": 58274583.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.8155419886112213,
+      "epoch": 4.172274562584119,
+      "grad_norm": 0.8892306685447693,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.02215111702680588,
+      "mean_token_accuracy": 0.9914537072181702,
+      "num_tokens": 58463241.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.8084070324897766,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.7887238264083862,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.018581822514533997,
+      "mean_token_accuracy": 0.9928648293018341,
+      "num_tokens": 58652281.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.8016656517982483,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.0644817352294922,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.020099574327468873,
+      "mean_token_accuracy": 0.9916590929031373,
+      "num_tokens": 58839946.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.8035425901412964,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.9588403701782227,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020384626090526582,
+      "mean_token_accuracy": 0.9918951213359832,
+      "num_tokens": 59028994.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.8114414095878602,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.7145556211471558,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.019365760684013366,
+      "mean_token_accuracy": 0.9921130836009979,
+      "num_tokens": 59217599.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.8123053193092347,
+      "epoch": 4.23956931359354,
+      "grad_norm": 0.7928217649459839,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.020542070269584656,
+      "mean_token_accuracy": 0.9918384432792664,
+      "num_tokens": 59406176.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.810414057970047,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.1470942497253418,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.01809942126274109,
+      "mean_token_accuracy": 0.992941427230835,
+      "num_tokens": 59594604.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.8046662211418152,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.579035222530365,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.01914280205965042,
+      "mean_token_accuracy": 0.9926800608634949,
+      "num_tokens": 59782947.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.8034039616584778,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.8979132175445557,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.019897907972335815,
+      "mean_token_accuracy": 0.9919832170009613,
+      "num_tokens": 59971417.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.8137863457202912,
+      "epoch": 4.293405114401077,
+      "grad_norm": 0.9149890542030334,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.02069481760263443,
+      "mean_token_accuracy": 0.992353367805481,
+      "num_tokens": 60159983.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.8163237690925598,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.8264126181602478,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.021351543068885804,
+      "mean_token_accuracy": 0.9921438097953796,
+      "num_tokens": 60348327.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.8133926391601562,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.0638184547424316,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.020092280209064485,
+      "mean_token_accuracy": 0.9920917391777039,
+      "num_tokens": 60537152.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.8095248222351075,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.8831470012664795,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.01908399909734726,
+      "mean_token_accuracy": 0.9929656088352203,
+      "num_tokens": 60725603.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.8069245994091034,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8486410975456238,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.01925535798072815,
+      "mean_token_accuracy": 0.9927195310592651,
+      "num_tokens": 60914021.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.8006627321243286,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.9021628499031067,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.019187870621681213,
+      "mean_token_accuracy": 0.992531019449234,
+      "num_tokens": 61102037.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.8067023754119873,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.7793697118759155,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.021760080754756928,
+      "mean_token_accuracy": 0.9915579319000244,
+      "num_tokens": 61290132.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.8075471341609954,
+      "epoch": 4.387617765814267,
+      "grad_norm": 0.736788809299469,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.02077012062072754,
+      "mean_token_accuracy": 0.9920295119285584,
+      "num_tokens": 61478897.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.8083206176757812,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.1687020063400269,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.02117353230714798,
+      "mean_token_accuracy": 0.991925036907196,
+      "num_tokens": 61667982.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.81106236577034,
+      "epoch": 4.414535666218035,
+      "grad_norm": 0.7596265077590942,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.02081204652786255,
+      "mean_token_accuracy": 0.9921064436435699,
+      "num_tokens": 61856557.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.8018859803676606,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.823410153388977,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.019047826528549194,
+      "mean_token_accuracy": 0.9924882829189301,
+      "num_tokens": 62045181.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7996662855148315,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.0748025178909302,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.01788569688796997,
+      "mean_token_accuracy": 0.9931309640407562,
+      "num_tokens": 62234215.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.8021627008914948,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9946282505989075,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021408762037754058,
+      "mean_token_accuracy": 0.9914931178092956,
+      "num_tokens": 62422361.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.8084165751934052,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.7313793301582336,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.020281489193439483,
+      "mean_token_accuracy": 0.9925725042819977,
+      "num_tokens": 62610667.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.8103892624378204,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.9093077182769775,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.018877537548542024,
+      "mean_token_accuracy": 0.9929530620574951,
+      "num_tokens": 62798970.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.8091431736946106,
+      "epoch": 4.495289367429341,
+      "grad_norm": 0.952234148979187,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021039776504039764,
+      "mean_token_accuracy": 0.9918699383735656,
+      "num_tokens": 62986914.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.8094858765602112,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.9959521293640137,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021489059925079344,
+      "mean_token_accuracy": 0.9914597749710083,
+      "num_tokens": 63175805.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.807330971956253,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.6189294457435608,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.02164270281791687,
+      "mean_token_accuracy": 0.9920661866664886,
+      "num_tokens": 63364692.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.8098866164684295,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7895700931549072,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.021856357157230378,
+      "mean_token_accuracy": 0.9916301846504212,
+      "num_tokens": 63553692.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.8143714487552642,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0242317914962769,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02046577036380768,
+      "mean_token_accuracy": 0.9923330962657928,
+      "num_tokens": 63742476.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.8123241424560547,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.4186221361160278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.020659247040748598,
+      "mean_token_accuracy": 0.9918934762477875,
+      "num_tokens": 63930898.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.8133015930652618,
+      "epoch": 4.576043068640646,
+      "grad_norm": 0.9085373878479004,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.021746012568473815,
+      "mean_token_accuracy": 0.9918161749839782,
+      "num_tokens": 64119589.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.8173148334026337,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.7257112264633179,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.019787387549877168,
+      "mean_token_accuracy": 0.9922669887542724,
+      "num_tokens": 64308293.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.8116482973098755,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7442086935043335,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.020933127403259276,
+      "mean_token_accuracy": 0.9917319416999817,
+      "num_tokens": 64497362.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.8134189546108246,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.7617318630218506,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.01842810958623886,
+      "mean_token_accuracy": 0.9927939116954804,
+      "num_tokens": 64685802.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.8112196505069733,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.7430832386016846,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.01809561103582382,
+      "mean_token_accuracy": 0.9930340230464936,
+      "num_tokens": 64873973.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.8128187835216523,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.7553606629371643,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.0202948659658432,
+      "mean_token_accuracy": 0.9919280350208283,
+      "num_tokens": 65062510.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.813840126991272,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.7390825152397156,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.020557042956352235,
+      "mean_token_accuracy": 0.9918098270893096,
+      "num_tokens": 65250940.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.8117553532123566,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.7698133587837219,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.020231780409812928,
+      "mean_token_accuracy": 0.9920670032501221,
+      "num_tokens": 65439696.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.8110962271690368,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.8562760353088379,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.018769541382789613,
+      "mean_token_accuracy": 0.9925244867801666,
+      "num_tokens": 65628357.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.8113800704479217,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.2076038122177124,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020604749023914338,
+      "mean_token_accuracy": 0.9921375870704651,
+      "num_tokens": 65817139.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.8204582631587982,
+      "epoch": 4.710632570659489,
+      "grad_norm": 0.9150722026824951,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021981674432754516,
+      "mean_token_accuracy": 0.9915678203105927,
+      "num_tokens": 66005911.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.8241343259811401,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.9803491830825806,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02007504999637604,
+      "mean_token_accuracy": 0.992132717370987,
+      "num_tokens": 66194469.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.8122783243656159,
+      "epoch": 4.737550471063257,
+      "grad_norm": 0.967640221118927,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.019903984665870667,
+      "mean_token_accuracy": 0.9924214482307434,
+      "num_tokens": 66383927.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.8145494401454926,
+      "epoch": 4.751009421265142,
+      "grad_norm": 0.7893037796020508,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.019587111473083497,
+      "mean_token_accuracy": 0.9921624064445496,
+      "num_tokens": 66572212.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.8090193390846252,
+      "epoch": 4.764468371467026,
+      "grad_norm": 0.9659109711647034,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.021097327768802642,
+      "mean_token_accuracy": 0.9917473196983337,
+      "num_tokens": 66760377.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.802634209394455,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.0175755023956299,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.02113145440816879,
+      "mean_token_accuracy": 0.991830724477768,
+      "num_tokens": 66949210.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.8030233085155487,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.7780183553695679,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.019016166031360627,
+      "mean_token_accuracy": 0.9926414608955383,
+      "num_tokens": 67138031.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.8019147455692291,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7863438129425049,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.018219922482967377,
+      "mean_token_accuracy": 0.9928844273090363,
+      "num_tokens": 67327268.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7949212431907654,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9292862415313721,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.02136329412460327,
+      "mean_token_accuracy": 0.9918141901493073,
+      "num_tokens": 67515765.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.8002764582633972,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9770768284797668,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.02056926190853119,
+      "mean_token_accuracy": 0.9918984293937683,
+      "num_tokens": 67704975.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.808483463525772,
+      "epoch": 4.845222072678331,
+      "grad_norm": 0.6890367269515991,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020158609747886656,
+      "mean_token_accuracy": 0.9925975620746612,
+      "num_tokens": 67893850.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.8144356846809387,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.8613544702529907,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.019614605605602263,
+      "mean_token_accuracy": 0.9928383588790893,
+      "num_tokens": 68082435.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.8083879172801971,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.7532686591148376,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.019553859531879426,
+      "mean_token_accuracy": 0.9924478828907013,
+      "num_tokens": 68270495.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.8038046479225158,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.3251928091049194,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.02081710398197174,
+      "mean_token_accuracy": 0.9919814527034759,
+      "num_tokens": 68459223.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.8058662831783294,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8125224113464355,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.01858963668346405,
+      "mean_token_accuracy": 0.9922726869583129,
+      "num_tokens": 68647306.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.805506020784378,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8629401922225952,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020777507126331328,
+      "mean_token_accuracy": 0.9918497979640961,
+      "num_tokens": 68836386.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.8137464344501495,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.9837846159934998,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.02024296224117279,
+      "mean_token_accuracy": 0.9921825706958771,
+      "num_tokens": 69025192.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.8139736235141755,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.8701289296150208,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.01932244449853897,
+      "mean_token_accuracy": 0.9926821112632751,
+      "num_tokens": 69213427.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.8088622391223907,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.7041723132133484,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.018706569075584413,
+      "mean_token_accuracy": 0.9927357733249664,
+      "num_tokens": 69401690.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.8001897990703583,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1029852628707886,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.020011243224143983,
+      "mean_token_accuracy": 0.9923900067806244,
+      "num_tokens": 69590659.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.8071568071842193,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.2395373582839966,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.02379731684923172,
+      "mean_token_accuracy": 0.9907991290092468,
+      "num_tokens": 69779864.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.8139646112918854,
+      "epoch": 4.993270524899058,
+      "grad_norm": 0.9219555854797363,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.021635296940803527,
+      "mean_token_accuracy": 0.991401469707489,
+      "num_tokens": 69968639.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.822659221804066,
+      "eval_loss": 0.060467079281806946,
+      "eval_mean_token_accuracy": 0.9792145984188007,
+      "eval_num_tokens": 70063064.0,
+      "eval_runtime": 24.4326,
+      "eval_samples_per_second": 204.645,
+      "eval_steps_per_second": 6.426,
+      "step": 3715
+    },
+    {
+      "entropy": 0.8150339186191559,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8010120391845703,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01673099249601364,
+      "mean_token_accuracy": 0.9935125410556793,
+      "num_tokens": 70157396.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.8029226005077362,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9923233389854431,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01266745626926422,
+      "mean_token_accuracy": 0.9954398214817047,
+      "num_tokens": 70346010.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.789790153503418,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.7180858254432678,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.012603424489498138,
+      "mean_token_accuracy": 0.9949092507362366,
+      "num_tokens": 70534896.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.7863301753997802,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.6865439414978027,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.013341715931892395,
+      "mean_token_accuracy": 0.9951623439788818,
+      "num_tokens": 70723644.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.7841233730316162,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.8624011278152466,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01214413121342659,
+      "mean_token_accuracy": 0.9950926601886749,
+      "num_tokens": 70912490.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.7781193614006042,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.1462656259536743,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.013635106384754181,
+      "mean_token_accuracy": 0.9946888983249664,
+      "num_tokens": 71101186.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.7773554801940918,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.7154580354690552,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.012310966104269027,
+      "mean_token_accuracy": 0.9952048182487487,
+      "num_tokens": 71290177.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.784615957736969,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8214139342308044,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.012708047032356262,
+      "mean_token_accuracy": 0.9956556618213653,
+      "num_tokens": 71478243.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.7814814388751984,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0974094867706299,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.011942464113235473,
+      "mean_token_accuracy": 0.9958715498447418,
+      "num_tokens": 71667417.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.7749826848506928,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.7582180500030518,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.011887111514806748,
+      "mean_token_accuracy": 0.9954704821109772,
+      "num_tokens": 71856322.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.7695000052452088,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.9159554839134216,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.013456562161445617,
+      "mean_token_accuracy": 0.994843590259552,
+      "num_tokens": 72044716.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.7674728155136108,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.0967031717300415,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.012064316868782043,
+      "mean_token_accuracy": 0.9953838646411896,
+      "num_tokens": 72233243.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.766239732503891,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.756477415561676,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012840582430362702,
+      "mean_token_accuracy": 0.9951915740966797,
+      "num_tokens": 72421751.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.7667239785194397,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.8078455924987793,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.012580010294914245,
+      "mean_token_accuracy": 0.9955588281154633,
+      "num_tokens": 72610886.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.7759605884552002,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.9473005533218384,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.01442105770111084,
+      "mean_token_accuracy": 0.9946676790714264,
+      "num_tokens": 72799213.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.7739267945289612,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.4796932637691498,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011830046027898788,
+      "mean_token_accuracy": 0.9954923093318939,
+      "num_tokens": 72988413.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.7756406426429748,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7290598154067993,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.014412418007850647,
+      "mean_token_accuracy": 0.9946037054061889,
+      "num_tokens": 73177134.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.7763995885848999,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.0562312602996826,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.015396332740783692,
+      "mean_token_accuracy": 0.9945933580398559,
+      "num_tokens": 73365622.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.7737024486064911,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1071312427520752,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01451488733291626,
+      "mean_token_accuracy": 0.9945858538150787,
+      "num_tokens": 73554381.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.779985535144806,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.7758586406707764,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.013801637291908263,
+      "mean_token_accuracy": 0.9947169423103333,
+      "num_tokens": 73743028.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.7786811470985413,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.6511496305465698,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012189272791147232,
+      "mean_token_accuracy": 0.9954657673835754,
+      "num_tokens": 73931263.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.7736978232860565,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.8152093291282654,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.013696780800819397,
+      "mean_token_accuracy": 0.9948886811733246,
+      "num_tokens": 74119992.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.7730314910411835,
+      "epoch": 5.302826379542395,
+      "grad_norm": 0.6703830361366272,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012700904905796052,
+      "mean_token_accuracy": 0.9950747072696686,
+      "num_tokens": 74308428.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.7740058362483978,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.9845885634422302,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012894082069396972,
+      "mean_token_accuracy": 0.9953160643577575,
+      "num_tokens": 74497030.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.7738905131816864,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.732532799243927,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.011950771510601043,
+      "mean_token_accuracy": 0.9955996692180633,
+      "num_tokens": 74685631.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.7788872420787811,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7301356196403503,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012340743839740754,
+      "mean_token_accuracy": 0.9952776432037354,
+      "num_tokens": 74874025.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.7771204471588135,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.7570232152938843,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.012937523424625397,
+      "mean_token_accuracy": 0.9949343740940094,
+      "num_tokens": 75062636.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.7747974276542664,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8237525820732117,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.012423938512802124,
+      "mean_token_accuracy": 0.9953977644443512,
+      "num_tokens": 75251339.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.7692180812358856,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.8497490286827087,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012855434417724609,
+      "mean_token_accuracy": 0.9948123812675476,
+      "num_tokens": 75440121.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.773431783914566,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.6758515238761902,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.012878020107746125,
+      "mean_token_accuracy": 0.9955350995063782,
+      "num_tokens": 75629143.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.7735035121440887,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7578167915344238,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.013059911131858826,
+      "mean_token_accuracy": 0.994978415966034,
+      "num_tokens": 75817425.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.7725034832954407,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.0449564456939697,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.013163739442825317,
+      "mean_token_accuracy": 0.9953307807445526,
+      "num_tokens": 76005910.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.7736131310462951,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.881845235824585,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012109111249446868,
+      "mean_token_accuracy": 0.9952919363975525,
+      "num_tokens": 76194639.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.7797342419624329,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.8100120425224304,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.013262654840946197,
+      "mean_token_accuracy": 0.9948459386825561,
+      "num_tokens": 76383240.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.7794113874435424,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.83738112449646,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.013258925080299378,
+      "mean_token_accuracy": 0.9947907269001007,
+      "num_tokens": 76572040.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.7787538826465606,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0567374229431152,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.014641955494880676,
+      "mean_token_accuracy": 0.9945057690143585,
+      "num_tokens": 76761030.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.7894418001174927,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.835543692111969,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.013065731525421143,
+      "mean_token_accuracy": 0.9949834287166596,
+      "num_tokens": 76948469.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.7823423445224762,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.8142625093460083,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.01229657381772995,
+      "mean_token_accuracy": 0.9957204461097717,
+      "num_tokens": 77136968.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.7786863029003144,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.749609649181366,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.01328940987586975,
+      "mean_token_accuracy": 0.9949924945831299,
+      "num_tokens": 77325328.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.7780588805675507,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.787598729133606,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.013933436572551727,
+      "mean_token_accuracy": 0.9949735462665558,
+      "num_tokens": 77513716.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.7779332935810089,
+      "epoch": 5.545087483176312,
+      "grad_norm": 1.0315855741500854,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.012871211767196656,
+      "mean_token_accuracy": 0.995112019777298,
+      "num_tokens": 77701621.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.7754440903663635,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.8191165924072266,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.012908256053924561,
+      "mean_token_accuracy": 0.9950248181819916,
+      "num_tokens": 77889886.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.7811384737491608,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.832539975643158,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.013047756254673004,
+      "mean_token_accuracy": 0.9946948051452636,
+      "num_tokens": 78078640.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.7825228571891785,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.9072359800338745,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.012361775338649749,
+      "mean_token_accuracy": 0.9954402267932891,
+      "num_tokens": 78267842.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.7825945496559144,
+      "epoch": 5.598923283983849,
+      "grad_norm": 0.8300487995147705,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.013246089220046997,
+      "mean_token_accuracy": 0.9951650142669678,
+      "num_tokens": 78455995.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.7857775866985321,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.9343852996826172,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.012612038850784301,
+      "mean_token_accuracy": 0.995340085029602,
+      "num_tokens": 78643617.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.775356650352478,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7109665274620056,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.013411080837249756,
+      "mean_token_accuracy": 0.9951648890972138,
+      "num_tokens": 78833085.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.7750323057174683,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.6436781883239746,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.012064981460571288,
+      "mean_token_accuracy": 0.9954153180122376,
+      "num_tokens": 79021569.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.7753469109535217,
+      "epoch": 5.652759084791386,
+      "grad_norm": 1.0361640453338623,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.013953472673892974,
+      "mean_token_accuracy": 0.9947281837463379,
+      "num_tokens": 79210573.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.7821656167507172,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9078817367553711,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012946712970733642,
+      "mean_token_accuracy": 0.9954073607921601,
+      "num_tokens": 79398723.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.7803877890110016,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.7879899740219116,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.012198399752378464,
+      "mean_token_accuracy": 0.9953389108181,
+      "num_tokens": 79587809.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.7772530496120453,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.0390578508377075,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.012395218014717102,
+      "mean_token_accuracy": 0.9953405141830445,
+      "num_tokens": 79775768.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.7784459292888641,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.7382562160491943,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.012304402142763137,
+      "mean_token_accuracy": 0.9953492462635041,
+      "num_tokens": 79964602.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.7821883440017701,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.5720117092132568,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.011424948275089265,
+      "mean_token_accuracy": 0.9956757485866546,
+      "num_tokens": 80153320.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.7811889886856079,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.7762079238891602,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012460066378116608,
+      "mean_token_accuracy": 0.9950654745101929,
+      "num_tokens": 80341381.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.7771898031234741,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.9219444394111633,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.011629728972911835,
+      "mean_token_accuracy": 0.9955675959587097,
+      "num_tokens": 80530323.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.7825480580329895,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.7189850211143494,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.011180496215820313,
+      "mean_token_accuracy": 0.9960062265396118,
+      "num_tokens": 80718533.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.7795504450798034,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.7256530523300171,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.012952378392219544,
+      "mean_token_accuracy": 0.9951896607875824,
+      "num_tokens": 80907391.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.779454517364502,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.808525800704956,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.0121039479970932,
+      "mean_token_accuracy": 0.9955627739429473,
+      "num_tokens": 81096396.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.7813481986522675,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.7294234037399292,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01356595754623413,
+      "mean_token_accuracy": 0.9951751470565796,
+      "num_tokens": 81284943.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.7833737671375275,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.771756649017334,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.013437020778656005,
+      "mean_token_accuracy": 0.9951942205429077,
+      "num_tokens": 81473810.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.781111603975296,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.5440986752510071,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.012187350541353226,
+      "mean_token_accuracy": 0.9956019043922424,
+      "num_tokens": 81662193.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.7777030646800995,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7532716393470764,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.013766974210739136,
+      "mean_token_accuracy": 0.994896411895752,
+      "num_tokens": 81851045.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.7776691019535065,
+      "epoch": 5.85464333781965,
+      "grad_norm": 1.1882649660110474,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.012588076293468475,
+      "mean_token_accuracy": 0.9952153980731964,
+      "num_tokens": 82039225.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.7795583724975585,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.7453442215919495,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.012100882083177566,
+      "mean_token_accuracy": 0.9954553186893463,
+      "num_tokens": 82228123.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.7763878583908081,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.6757420301437378,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011943653970956803,
+      "mean_token_accuracy": 0.9953660726547241,
+      "num_tokens": 82417093.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.7747587621212005,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.9607213735580444,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.014138388633728027,
+      "mean_token_accuracy": 0.9944287598133087,
+      "num_tokens": 82605461.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.7782076597213745,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.6733895540237427,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.012490443140268325,
+      "mean_token_accuracy": 0.9950622916221619,
+      "num_tokens": 82794124.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.7843947112560272,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.7626368403434753,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.012137357890605927,
+      "mean_token_accuracy": 0.9952504158020019,
+      "num_tokens": 82982258.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.7826779782772064,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8149229288101196,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012872529029846192,
+      "mean_token_accuracy": 0.9949809432029724,
+      "num_tokens": 83170520.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.7792184889316559,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.7774028778076172,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012353222817182541,
+      "mean_token_accuracy": 0.9953234374523163,
+      "num_tokens": 83359174.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.7778255641460419,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7626445293426514,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.01137089878320694,
+      "mean_token_accuracy": 0.9958219468593598,
+      "num_tokens": 83547583.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.7765573263168335,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9130274653434753,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012439383566379547,
+      "mean_token_accuracy": 0.9954963147640228,
+      "num_tokens": 83735904.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.7734197914600373,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.7289780378341675,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011430246382951736,
+      "mean_token_accuracy": 0.9957478463649749,
+      "num_tokens": 83924971.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.7852135965019275,
+      "eval_loss": 0.06990383565425873,
+      "eval_mean_token_accuracy": 0.9796095274056599,
+      "eval_num_tokens": 84075763.0,
+      "eval_runtime": 24.3947,
+      "eval_samples_per_second": 204.963,
+      "eval_steps_per_second": 6.436,
+      "step": 4458
+    },
+    {
+      "entropy": 0.7758591413497925,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.9430185556411743,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.011816921830177306,
+      "mean_token_accuracy": 0.9959785401821136,
+      "num_tokens": 84113330.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.773297393321991,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.6593970060348511,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.007801702618598938,
+      "mean_token_accuracy": 0.9972260951995849,
+      "num_tokens": 84302325.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.7716216742992401,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.5422326922416687,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.00867043137550354,
+      "mean_token_accuracy": 0.9968674182891846,
+      "num_tokens": 84490227.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.7699320554733277,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.8308367729187012,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.007848290354013443,
+      "mean_token_accuracy": 0.9971986234188079,
+      "num_tokens": 84678824.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.7688778102397918,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.5276612639427185,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.007070200890302658,
+      "mean_token_accuracy": 0.9974652588367462,
+      "num_tokens": 84867430.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.7687382102012634,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.7081453800201416,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.00793309137225151,
+      "mean_token_accuracy": 0.9972716808319092,
+      "num_tokens": 85056072.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.7705117404460907,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.9106748104095459,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.007362194359302521,
+      "mean_token_accuracy": 0.9975170493125916,
+      "num_tokens": 85244337.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.7697769284248352,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.7364314198493958,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.007369447499513626,
+      "mean_token_accuracy": 0.9974394500255584,
+      "num_tokens": 85432878.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.7661255121231079,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.6747480630874634,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.007502703368663788,
+      "mean_token_accuracy": 0.9973482489585876,
+      "num_tokens": 85621800.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.7673385441303253,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.8179678916931152,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.007111805677413941,
+      "mean_token_accuracy": 0.9977720499038696,
+      "num_tokens": 85809796.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.76323401927948,
+      "epoch": 6.137281292059219,
+      "grad_norm": 0.5177297592163086,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.007196265459060669,
+      "mean_token_accuracy": 0.9976384699344635,
+      "num_tokens": 85998543.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.7640918910503387,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 0.8970023989677429,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.008229555189609527,
+      "mean_token_accuracy": 0.9967760264873504,
+      "num_tokens": 86186537.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.7605979859828949,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.7019157409667969,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.007770314812660217,
+      "mean_token_accuracy": 0.997019499540329,
+      "num_tokens": 86375405.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.7647112905979156,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.6519957184791565,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.007326644659042358,
+      "mean_token_accuracy": 0.9972577571868897,
+      "num_tokens": 86562963.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.7608903884887696,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.9307657480239868,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.00786140188574791,
+      "mean_token_accuracy": 0.9970912158489227,
+      "num_tokens": 86751006.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.7573343932628631,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 0.6093463897705078,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.007400572299957275,
+      "mean_token_accuracy": 0.997343909740448,
+      "num_tokens": 86939232.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.756312346458435,
+      "epoch": 6.218034993270525,
+      "grad_norm": 0.6412308812141418,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.007471667230129242,
+      "mean_token_accuracy": 0.9973339200019836,
+      "num_tokens": 87128294.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.7593119204044342,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.6830183863639832,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.007831692695617676,
+      "mean_token_accuracy": 0.9972396552562713,
+      "num_tokens": 87316833.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.7598747551441193,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.6647830009460449,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.0070374816656112674,
+      "mean_token_accuracy": 0.9975631892681122,
+      "num_tokens": 87505022.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.7626749157905579,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.5470786690711975,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.007449843734502792,
+      "mean_token_accuracy": 0.9976118981838227,
+      "num_tokens": 87692953.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.7584024250507355,
+      "epoch": 6.271870794078062,
+      "grad_norm": 0.8641819953918457,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.007355903089046478,
+      "mean_token_accuracy": 0.99752836227417,
+      "num_tokens": 87882513.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.7607118487358093,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.6385810971260071,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.007216785848140716,
+      "mean_token_accuracy": 0.9973138213157654,
+      "num_tokens": 88071161.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.7600765109062195,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.3984440565109253,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.007562516629695893,
+      "mean_token_accuracy": 0.9973795533180236,
+      "num_tokens": 88259867.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.7609024405479431,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.7098219990730286,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.007748734951019287,
+      "mean_token_accuracy": 0.9970796704292297,
+      "num_tokens": 88448504.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.7602188348770141,
+      "epoch": 6.325706594885599,
+      "grad_norm": 0.3834342658519745,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.00794660747051239,
+      "mean_token_accuracy": 0.9971937775611878,
+      "num_tokens": 88637123.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.7595065951347351,
+      "epoch": 6.339165545087483,
+      "grad_norm": 0.7073737978935242,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.007514684647321701,
+      "mean_token_accuracy": 0.9972682356834411,
+      "num_tokens": 88826107.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.7566567718982696,
+      "epoch": 6.352624495289367,
+      "grad_norm": 0.6334935426712036,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.007287996262311936,
+      "mean_token_accuracy": 0.9972865641117096,
+      "num_tokens": 89014806.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.7535678207874298,
+      "epoch": 6.366083445491252,
+      "grad_norm": 0.5461204648017883,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.007215028256177902,
+      "mean_token_accuracy": 0.9976517677307128,
+      "num_tokens": 89203610.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.753510159254074,
+      "epoch": 6.379542395693136,
+      "grad_norm": 0.8002920150756836,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.00775575190782547,
+      "mean_token_accuracy": 0.9972102761268615,
+      "num_tokens": 89392611.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.7567359745502472,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.5416199564933777,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.007165181636810303,
+      "mean_token_accuracy": 0.9973179161548614,
+      "num_tokens": 89581472.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.7603509783744812,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.7484428882598877,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.008048020303249359,
+      "mean_token_accuracy": 0.9972274541854859,
+      "num_tokens": 89769895.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.7604019463062286,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.6397516131401062,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.006713573634624481,
+      "mean_token_accuracy": 0.9974542498588562,
+      "num_tokens": 89958151.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.7562435805797577,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.6562761068344116,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.0070735208690166475,
+      "mean_token_accuracy": 0.9975251019001007,
+      "num_tokens": 90147009.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.7581055939197541,
+      "epoch": 6.446837146702557,
+      "grad_norm": 0.6705797910690308,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.008277279138565064,
+      "mean_token_accuracy": 0.9969512403011322,
+      "num_tokens": 90335135.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.7552513837814331,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.5535324215888977,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.007625260949134826,
+      "mean_token_accuracy": 0.9972123861312866,
+      "num_tokens": 90523520.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.7567071616649628,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.5841183662414551,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.008183968812227249,
+      "mean_token_accuracy": 0.9970704436302185,
+      "num_tokens": 90712012.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.7579837441444397,
+      "epoch": 6.48721399730821,
+      "grad_norm": 1.0210020542144775,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.00890980064868927,
+      "mean_token_accuracy": 0.9968561112880707,
+      "num_tokens": 90900360.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.7599723398685455,
+      "epoch": 6.500672947510094,
+      "grad_norm": 0.7007607221603394,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.007547108829021454,
+      "mean_token_accuracy": 0.9972310900688172,
+      "num_tokens": 91089219.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.7569817125797271,
+      "epoch": 6.514131897711978,
+      "grad_norm": 1.203089952468872,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.007012282311916351,
+      "mean_token_accuracy": 0.9974212229251862,
+      "num_tokens": 91278177.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.7547080159187317,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.6055266857147217,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.007885889708995819,
+      "mean_token_accuracy": 0.9971360504627228,
+      "num_tokens": 91466912.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.7584831535816192,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.6868627071380615,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.00764903798699379,
+      "mean_token_accuracy": 0.9975648581981659,
+      "num_tokens": 91655249.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.757660973072052,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.9303789734840393,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.00731472447514534,
+      "mean_token_accuracy": 0.9974016606807709,
+      "num_tokens": 91844016.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.759681248664856,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 0.6116757988929749,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.007223697006702423,
+      "mean_token_accuracy": 0.9970514178276062,
+      "num_tokens": 92032638.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.7601584196090698,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.6825643181800842,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.0068596042692661285,
+      "mean_token_accuracy": 0.9973869740962982,
+      "num_tokens": 92221897.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.7645585358142852,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.7482470870018005,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.007520741224288941,
+      "mean_token_accuracy": 0.9974736630916595,
+      "num_tokens": 92410416.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.766044944524765,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.7137084603309631,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.007919894158840179,
+      "mean_token_accuracy": 0.9969592452049255,
+      "num_tokens": 92598533.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.761748319864273,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 1.0537257194519043,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.007994732260704041,
+      "mean_token_accuracy": 0.9970260739326477,
+      "num_tokens": 92788046.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.763516879081726,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.9168201088905334,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.006822402775287628,
+      "mean_token_accuracy": 0.9973889350891113,
+      "num_tokens": 92976583.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.7603426337242126,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.6945034265518188,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.007335217297077179,
+      "mean_token_accuracy": 0.9974107563495636,
+      "num_tokens": 93164984.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.7601041436195374,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.7255296111106873,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.006898632645606995,
+      "mean_token_accuracy": 0.9974060177803039,
+      "num_tokens": 93353860.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.760407668352127,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.7785485982894897,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.006965675204992294,
+      "mean_token_accuracy": 0.997627604007721,
+      "num_tokens": 93542542.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.7584291577339173,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.676058292388916,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.0066734164953231815,
+      "mean_token_accuracy": 0.9976742446422577,
+      "num_tokens": 93730962.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.7543873727321625,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.9423731565475464,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.007540161162614823,
+      "mean_token_accuracy": 0.9972456872463227,
+      "num_tokens": 93919838.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.7540828108787536,
+      "epoch": 6.716016150740242,
+      "grad_norm": 0.60378098487854,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.006739185005426407,
+      "mean_token_accuracy": 0.9976783573627472,
+      "num_tokens": 94108612.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.7581348657608032,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.7218618392944336,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.006258270889520645,
+      "mean_token_accuracy": 0.9977961003780365,
+      "num_tokens": 94296515.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.7555042028427124,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.6111772060394287,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.006771801412105561,
+      "mean_token_accuracy": 0.9975680649280548,
+      "num_tokens": 94484849.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.7563546478748322,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.7171721458435059,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.007536987960338593,
+      "mean_token_accuracy": 0.9971135079860687,
+      "num_tokens": 94673363.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.7541037023067474,
+      "epoch": 6.769851951547779,
+      "grad_norm": 0.45519423484802246,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.007405019551515579,
+      "mean_token_accuracy": 0.9975123167037964,
+      "num_tokens": 94862699.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.7588851511478424,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.7041094899177551,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.007134091854095459,
+      "mean_token_accuracy": 0.9972242057323456,
+      "num_tokens": 95050741.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.7570755481719971,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.6298620700836182,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.0070040211081504825,
+      "mean_token_accuracy": 0.997326648235321,
+      "num_tokens": 95239293.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.7548963069915772,
+      "epoch": 6.810228802153432,
+      "grad_norm": 1.0688964128494263,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.006342679262161255,
+      "mean_token_accuracy": 0.9976243853569031,
+      "num_tokens": 95428279.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.7548417568206787,
+      "epoch": 6.823687752355316,
+      "grad_norm": 0.4953668415546417,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.0069470912218093876,
+      "mean_token_accuracy": 0.9973559200763702,
+      "num_tokens": 95616835.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.7551373422145844,
+      "epoch": 6.837146702557201,
+      "grad_norm": 0.5016734600067139,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.006873356550931931,
+      "mean_token_accuracy": 0.9973772704601288,
+      "num_tokens": 95805557.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.7567492067813874,
+      "epoch": 6.850605652759085,
+      "grad_norm": 0.532783567905426,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.006668326258659363,
+      "mean_token_accuracy": 0.997450977563858,
+      "num_tokens": 95993831.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.758799821138382,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.6660076379776001,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.006823312491178513,
+      "mean_token_accuracy": 0.9975283324718476,
+      "num_tokens": 96181732.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.7571039736270905,
+      "epoch": 6.877523553162853,
+      "grad_norm": 1.0330839157104492,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.007801513373851776,
+      "mean_token_accuracy": 0.9971715807914734,
+      "num_tokens": 96369920.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.7555147588253022,
+      "epoch": 6.890982503364738,
+      "grad_norm": 0.5079207420349121,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.0074800834059715274,
+      "mean_token_accuracy": 0.9973418891429902,
+      "num_tokens": 96558724.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.7542111933231354,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.678804337978363,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.00698004812002182,
+      "mean_token_accuracy": 0.997770369052887,
+      "num_tokens": 96747242.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.7508629977703094,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.8755067586898804,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.007143624126911163,
+      "mean_token_accuracy": 0.9975672841072083,
+      "num_tokens": 96936244.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.7522107064723969,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.8553352952003479,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.007587063312530518,
+      "mean_token_accuracy": 0.9970708072185517,
+      "num_tokens": 97125069.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.7501682281494141,
+      "epoch": 6.944818304172275,
+      "grad_norm": 0.614628255367279,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.007764184474945068,
+      "mean_token_accuracy": 0.9968451082706451,
+      "num_tokens": 97313879.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.7522825837135315,
+      "epoch": 6.958277254374159,
+      "grad_norm": 0.8060729503631592,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.00771535262465477,
+      "mean_token_accuracy": 0.9972144842147828,
+      "num_tokens": 97503040.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.7534927785396576,
+      "epoch": 6.971736204576043,
+      "grad_norm": 0.6604568958282471,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.006561616808176041,
+      "mean_token_accuracy": 0.9974263906478882,
+      "num_tokens": 97692024.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.7542680203914642,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.853604793548584,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.007131250202655792,
+      "mean_token_accuracy": 0.9974451243877411,
+      "num_tokens": 97881026.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.7567435801029205,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.6146188974380493,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.006486450880765915,
+      "mean_token_accuracy": 0.9974190831184387,
+      "num_tokens": 98069553.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.7571151700748759,
+      "eval_loss": 0.07285518199205399,
+      "eval_mean_token_accuracy": 0.9808630795235846,
+      "eval_num_tokens": 98088531.0,
+      "eval_runtime": 24.4094,
+      "eval_samples_per_second": 204.839,
+      "eval_steps_per_second": 6.432,
+      "step": 5201
+    },
+    {
+      "entropy": 0.751511138677597,
+      "epoch": 7.012113055181696,
+      "grad_norm": 0.49068665504455566,
+      "learning_rate": 2.2506648060191042e-05,
+      "loss": 0.0042594756931066515,
+      "mean_token_accuracy": 0.9985321044921875,
+      "num_tokens": 98258586.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.7522116243839264,
+      "epoch": 7.02557200538358,
+      "grad_norm": 0.5906213521957397,
+      "learning_rate": 2.2321031102941066e-05,
+      "loss": 0.003930336982011795,
+      "mean_token_accuracy": 0.9989193022251129,
+      "num_tokens": 98447091.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.7511299610137939,
+      "epoch": 7.039030955585464,
+      "grad_norm": 0.45251449942588806,
+      "learning_rate": 2.2135962530768055e-05,
+      "loss": 0.004373953118920327,
+      "mean_token_accuracy": 0.9986318647861481,
+      "num_tokens": 98635468.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.7481192111968994,
+      "epoch": 7.052489905787349,
+      "grad_norm": 0.6962258219718933,
+      "learning_rate": 2.1951446010313053e-05,
+      "loss": 0.00428679995238781,
+      "mean_token_accuracy": 0.9987218022346497,
+      "num_tokens": 98824365.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.7483489096164704,
+      "epoch": 7.065948855989233,
+      "grad_norm": 0.5099614858627319,
+      "learning_rate": 2.1767485197279662e-05,
+      "loss": 0.003984151408076286,
+      "mean_token_accuracy": 0.9986076414585113,
+      "num_tokens": 99013056.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.7463673830032349,
+      "epoch": 7.079407806191117,
+      "grad_norm": 0.3716561198234558,
+      "learning_rate": 2.1584083736361617e-05,
+      "loss": 0.004038792476058007,
+      "mean_token_accuracy": 0.998809176683426,
+      "num_tokens": 99201941.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.7466944694519043,
+      "epoch": 7.0928667563930015,
+      "grad_norm": 0.33926740288734436,
+      "learning_rate": 2.1401245261170582e-05,
+      "loss": 0.003776150941848755,
+      "mean_token_accuracy": 0.9987689018249511,
+      "num_tokens": 99390847.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.7465248286724091,
+      "epoch": 7.106325706594886,
+      "grad_norm": 0.3297136127948761,
+      "learning_rate": 2.121897339416416e-05,
+      "loss": 0.0037364721298217773,
+      "mean_token_accuracy": 0.9987996935844421,
+      "num_tokens": 99579720.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.7467422366142273,
+      "epoch": 7.11978465679677,
+      "grad_norm": 0.3911883234977722,
+      "learning_rate": 2.103727174657412e-05,
+      "loss": 0.004084533080458641,
+      "mean_token_accuracy": 0.9985779583454132,
+      "num_tokens": 99768263.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.7423677504062652,
+      "epoch": 7.133243606998654,
+      "grad_norm": 0.5843891501426697,
+      "learning_rate": 2.085614391833486e-05,
+      "loss": 0.0038405463099479677,
+      "mean_token_accuracy": 0.9987926363945008,
+      "num_tokens": 99957655.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.7434155166149139,
+      "epoch": 7.1467025572005385,
+      "grad_norm": 0.5154002904891968,
+      "learning_rate": 2.0675593498012052e-05,
+      "loss": 0.0036686167120933534,
+      "mean_token_accuracy": 0.9988853752613067,
+      "num_tokens": 100146048.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.7450331568717956,
+      "epoch": 7.160161507402423,
+      "grad_norm": 0.5856082439422607,
+      "learning_rate": 2.049562406273159e-05,
+      "loss": 0.00417071133852005,
+      "mean_token_accuracy": 0.9985920011997222,
+      "num_tokens": 100334064.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.7413534104824067,
+      "epoch": 7.173620457604307,
+      "grad_norm": 0.4711301028728485,
+      "learning_rate": 2.0316239178108688e-05,
+      "loss": 0.003790935128927231,
+      "mean_token_accuracy": 0.9986560821533204,
+      "num_tokens": 100522817.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.739797306060791,
+      "epoch": 7.187079407806191,
+      "grad_norm": 0.4481808841228485,
+      "learning_rate": 2.013744239817725e-05,
+      "loss": 0.0038315042853355407,
+      "mean_token_accuracy": 0.9987379729747772,
+      "num_tokens": 100711882.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.7411384165287018,
+      "epoch": 7.2005383580080755,
+      "grad_norm": 0.3373877704143524,
+      "learning_rate": 1.9959237265319452e-05,
+      "loss": 0.0034619957208633424,
+      "mean_token_accuracy": 0.9990058243274689,
+      "num_tokens": 100900649.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.7439651310443878,
+      "epoch": 7.21399730820996,
+      "grad_norm": 0.4260849952697754,
+      "learning_rate": 1.9781627310195528e-05,
+      "loss": 0.004041648656129837,
+      "mean_token_accuracy": 0.9988129734992981,
+      "num_tokens": 101088789.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.741381824016571,
+      "epoch": 7.227456258411844,
+      "grad_norm": 0.704740047454834,
+      "learning_rate": 1.960461605167389e-05,
+      "loss": 0.004109270870685577,
+      "mean_token_accuracy": 0.9987618446350097,
+      "num_tokens": 101277316.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.7418228089809418,
+      "epoch": 7.240915208613728,
+      "grad_norm": 0.34908416867256165,
+      "learning_rate": 1.9428206996761357e-05,
+      "loss": 0.0038616616278886796,
+      "mean_token_accuracy": 0.9985946893692017,
+      "num_tokens": 101466025.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.7436148345470428,
+      "epoch": 7.2543741588156125,
+      "grad_norm": 0.5490255951881409,
+      "learning_rate": 1.9252403640533678e-05,
+      "loss": 0.004948872700333595,
+      "mean_token_accuracy": 0.9981587290763855,
+      "num_tokens": 101654412.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.7419192969799042,
+      "epoch": 7.267833109017497,
+      "grad_norm": 0.5561390519142151,
+      "learning_rate": 1.9077209466066315e-05,
+      "loss": 0.0043866544961929325,
+      "mean_token_accuracy": 0.9986559212207794,
+      "num_tokens": 101843232.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.7441964089870453,
+      "epoch": 7.281292059219381,
+      "grad_norm": 0.5724263191223145,
+      "learning_rate": 1.8902627944365396e-05,
+      "loss": 0.004136815667152405,
+      "mean_token_accuracy": 0.9986151039600373,
+      "num_tokens": 102031115.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.7447238683700561,
+      "epoch": 7.294751009421265,
+      "grad_norm": 0.6418642997741699,
+      "learning_rate": 1.872866253429897e-05,
+      "loss": 0.004244513809680939,
+      "mean_token_accuracy": 0.9985204100608825,
+      "num_tokens": 102219064.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.7460219264030457,
+      "epoch": 7.30820995962315,
+      "grad_norm": 0.877717912197113,
+      "learning_rate": 1.8555316682528485e-05,
+      "loss": 0.0035314552485942842,
+      "mean_token_accuracy": 0.9990062952041626,
+      "num_tokens": 102407049.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.7446399390697479,
+      "epoch": 7.321668909825034,
+      "grad_norm": 0.8448772430419922,
+      "learning_rate": 1.8382593823440474e-05,
+      "loss": 0.004365745186805725,
+      "mean_token_accuracy": 0.9984619498252869,
+      "num_tokens": 102595411.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.7414965629577637,
+      "epoch": 7.335127860026918,
+      "grad_norm": 0.7239252328872681,
+      "learning_rate": 1.821049737907854e-05,
+      "loss": 0.0036744467914104463,
+      "mean_token_accuracy": 0.9989120185375213,
+      "num_tokens": 102784691.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.744704294204712,
+      "epoch": 7.348586810228802,
+      "grad_norm": 0.390360951423645,
+      "learning_rate": 1.803903075907554e-05,
+      "loss": 0.003917969018220902,
+      "mean_token_accuracy": 0.9988363087177277,
+      "num_tokens": 102973291.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.744241201877594,
+      "epoch": 7.362045760430687,
+      "grad_norm": 0.7771829962730408,
+      "learning_rate": 1.7868197360586054e-05,
+      "loss": 0.004070578515529633,
+      "mean_token_accuracy": 0.9985230803489685,
+      "num_tokens": 103161417.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.7423063695430756,
+      "epoch": 7.375504710632571,
+      "grad_norm": 0.3263421654701233,
+      "learning_rate": 1.769800056821904e-05,
+      "loss": 0.003923245519399643,
+      "mean_token_accuracy": 0.9985153019428253,
+      "num_tokens": 103349890.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.7458358705043793,
+      "epoch": 7.388963660834455,
+      "grad_norm": 0.5268740653991699,
+      "learning_rate": 1.7528443753970814e-05,
+      "loss": 0.004262761771678924,
+      "mean_token_accuracy": 0.9987221539020539,
+      "num_tokens": 103537726.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.7404268085956573,
+      "epoch": 7.402422611036339,
+      "grad_norm": 0.5339550971984863,
+      "learning_rate": 1.7359530277158227e-05,
+      "loss": 0.0039617355912923815,
+      "mean_token_accuracy": 0.998635333776474,
+      "num_tokens": 103726707.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.7421627819538117,
+      "epoch": 7.415881561238224,
+      "grad_norm": 0.47286465764045715,
+      "learning_rate": 1.7191263484352106e-05,
+      "loss": 0.0034935571253299712,
+      "mean_token_accuracy": 0.9989599347114563,
+      "num_tokens": 103915542.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.7416261017322541,
+      "epoch": 7.429340511440108,
+      "grad_norm": 0.29520711302757263,
+      "learning_rate": 1.702364670931097e-05,
+      "loss": 0.003899512439966202,
+      "mean_token_accuracy": 0.9986254096031189,
+      "num_tokens": 104105019.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.7447648346424103,
+      "epoch": 7.442799461641992,
+      "grad_norm": 0.46887683868408203,
+      "learning_rate": 1.685668327291495e-05,
+      "loss": 0.0036319829523563385,
+      "mean_token_accuracy": 0.9989396393299103,
+      "num_tokens": 104293657.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.7453936874866486,
+      "epoch": 7.456258411843876,
+      "grad_norm": 0.3639506995677948,
+      "learning_rate": 1.6690376483100034e-05,
+      "loss": 0.003420223295688629,
+      "mean_token_accuracy": 0.9988774538040162,
+      "num_tokens": 104482306.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.743130087852478,
+      "epoch": 7.469717362045761,
+      "grad_norm": 0.4534284770488739,
+      "learning_rate": 1.6524729634792475e-05,
+      "loss": 0.0038840845227241516,
+      "mean_token_accuracy": 0.9987022399902343,
+      "num_tokens": 104671166.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.7441355168819428,
+      "epoch": 7.483176312247645,
+      "grad_norm": 0.3704164922237396,
+      "learning_rate": 1.6359746009843567e-05,
+      "loss": 0.0034930892288684845,
+      "mean_token_accuracy": 0.9988382160663605,
+      "num_tokens": 104859133.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.7428222477436066,
+      "epoch": 7.496635262449529,
+      "grad_norm": 0.751130223274231,
+      "learning_rate": 1.6195428876964592e-05,
+      "loss": 0.003963682055473328,
+      "mean_token_accuracy": 0.9984794855117798,
+      "num_tokens": 105047492.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.7427781701087952,
+      "epoch": 7.510094212651413,
+      "grad_norm": 0.8935811519622803,
+      "learning_rate": 1.6031781491662056e-05,
+      "loss": 0.004140084609389305,
+      "mean_token_accuracy": 0.9985048770904541,
+      "num_tokens": 105235970.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.7405261933803559,
+      "epoch": 7.523553162853298,
+      "grad_norm": 0.575813353061676,
+      "learning_rate": 1.58688070961732e-05,
+      "loss": 0.003933753073215485,
+      "mean_token_accuracy": 0.998809415102005,
+      "num_tokens": 105424662.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.742167991399765,
+      "epoch": 7.537012113055182,
+      "grad_norm": 0.5800251960754395,
+      "learning_rate": 1.5706508919401804e-05,
+      "loss": 0.004490737244486809,
+      "mean_token_accuracy": 0.9987403452396393,
+      "num_tokens": 105613306.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.744090485572815,
+      "epoch": 7.550471063257066,
+      "grad_norm": 0.8697437644004822,
+      "learning_rate": 1.554489017685412e-05,
+      "loss": 0.00443139597773552,
+      "mean_token_accuracy": 0.9984764337539673,
+      "num_tokens": 105801807.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.7431838274002075,
+      "epoch": 7.56393001345895,
+      "grad_norm": 0.47921308875083923,
+      "learning_rate": 1.538395407057525e-05,
+      "loss": 0.003662409633398056,
+      "mean_token_accuracy": 0.9989210546016694,
+      "num_tokens": 105990683.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.7407984972000122,
+      "epoch": 7.577388963660835,
+      "grad_norm": 0.45125728845596313,
+      "learning_rate": 1.522370378908567e-05,
+      "loss": 0.003542381525039673,
+      "mean_token_accuracy": 0.9988591969013214,
+      "num_tokens": 106179809.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.744829797744751,
+      "epoch": 7.590847913862719,
+      "grad_norm": 0.5999196171760559,
+      "learning_rate": 1.5064142507318051e-05,
+      "loss": 0.0035081353038549424,
+      "mean_token_accuracy": 0.9988316416740417,
+      "num_tokens": 106367988.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.7428836584091186,
+      "epoch": 7.604306864064603,
+      "grad_norm": 0.4594089388847351,
+      "learning_rate": 1.4905273386554386e-05,
+      "loss": 0.004249929264187813,
+      "mean_token_accuracy": 0.9987165331840515,
+      "num_tokens": 106556378.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.7435553133487701,
+      "epoch": 7.6177658142664875,
+      "grad_norm": 0.5920020937919617,
+      "learning_rate": 1.4747099574363332e-05,
+      "loss": 0.0036187373101711274,
+      "mean_token_accuracy": 0.9987746834754944,
+      "num_tokens": 106744854.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.7437893748283386,
+      "epoch": 7.631224764468372,
+      "grad_norm": 0.5453750491142273,
+      "learning_rate": 1.4589624204537866e-05,
+      "loss": 0.00377066433429718,
+      "mean_token_accuracy": 0.9989324033260345,
+      "num_tokens": 106932944.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.7395198106765747,
+      "epoch": 7.644683714670256,
+      "grad_norm": 0.38797295093536377,
+      "learning_rate": 1.4432850397033176e-05,
+      "loss": 0.0031442392617464067,
+      "mean_token_accuracy": 0.9990855991840363,
+      "num_tokens": 107121922.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.7402484536170959,
+      "epoch": 7.65814266487214,
+      "grad_norm": 0.6739166975021362,
+      "learning_rate": 1.427678125790487e-05,
+      "loss": 0.0038538143038749696,
+      "mean_token_accuracy": 0.9986329972743988,
+      "num_tokens": 107310638.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.7401728749275207,
+      "epoch": 7.6716016150740245,
+      "grad_norm": 0.4977329969406128,
+      "learning_rate": 1.4121419879247422e-05,
+      "loss": 0.0033894732594490053,
+      "mean_token_accuracy": 0.9989957690238953,
+      "num_tokens": 107499098.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.7425866723060608,
+      "epoch": 7.685060565275909,
+      "grad_norm": 0.5719661712646484,
+      "learning_rate": 1.3966769339132929e-05,
+      "loss": 0.004793065041303635,
+      "mean_token_accuracy": 0.9987428963184357,
+      "num_tokens": 107687499.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.7395791113376617,
+      "epoch": 7.698519515477793,
+      "grad_norm": 0.4698435366153717,
+      "learning_rate": 1.3812832701550099e-05,
+      "loss": 0.004081443324685097,
+      "mean_token_accuracy": 0.9984708070755005,
+      "num_tokens": 107876895.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.7433845281600953,
+      "epoch": 7.711978465679677,
+      "grad_norm": 0.6040443778038025,
+      "learning_rate": 1.365961301634357e-05,
+      "loss": 0.003664097934961319,
+      "mean_token_accuracy": 0.9988075792789459,
+      "num_tokens": 108065071.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.7424676477909088,
+      "epoch": 7.7254374158815615,
+      "grad_norm": 0.6923211812973022,
+      "learning_rate": 1.350711331915348e-05,
+      "loss": 0.003957484290003777,
+      "mean_token_accuracy": 0.998679769039154,
+      "num_tokens": 108253186.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.7405034005641937,
+      "epoch": 7.738896366083446,
+      "grad_norm": 0.4896307587623596,
+      "learning_rate": 1.3355336631355314e-05,
+      "loss": 0.0041983149945735935,
+      "mean_token_accuracy": 0.9986901462078095,
+      "num_tokens": 108442076.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.743422681093216,
+      "epoch": 7.75235531628533,
+      "grad_norm": 0.5889957547187805,
+      "learning_rate": 1.3204285960000051e-05,
+      "loss": 0.003425312042236328,
+      "mean_token_accuracy": 0.9988837718963623,
+      "num_tokens": 108630562.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.7411604285240173,
+      "epoch": 7.765814266487214,
+      "grad_norm": 0.4655008316040039,
+      "learning_rate": 1.30539642977546e-05,
+      "loss": 0.003258592635393143,
+      "mean_token_accuracy": 0.9988660037517547,
+      "num_tokens": 108819369.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.7410345315933228,
+      "epoch": 7.7792732166890985,
+      "grad_norm": 0.7350168228149414,
+      "learning_rate": 1.2904374622842486e-05,
+      "loss": 0.0034532986581325533,
+      "mean_token_accuracy": 0.9989261090755462,
+      "num_tokens": 109008535.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.7410138070583343,
+      "epoch": 7.792732166890983,
+      "grad_norm": 0.44904911518096924,
+      "learning_rate": 1.2755519898984846e-05,
+      "loss": 0.0034178659319877625,
+      "mean_token_accuracy": 0.9988463997840882,
+      "num_tokens": 109197443.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.7402764976024627,
+      "epoch": 7.806191117092867,
+      "grad_norm": 0.5618012547492981,
+      "learning_rate": 1.2607403075341728e-05,
+      "loss": 0.003283835202455521,
+      "mean_token_accuracy": 0.9989370465278625,
+      "num_tokens": 109386097.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.7410104274749756,
+      "epoch": 7.819650067294751,
+      "grad_norm": 0.3167133629322052,
+      "learning_rate": 1.2460027086453658e-05,
+      "loss": 0.0034713000059127808,
+      "mean_token_accuracy": 0.9988771259784699,
+      "num_tokens": 109574403.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.7389123022556305,
+      "epoch": 7.833109017496636,
+      "grad_norm": 0.544057309627533,
+      "learning_rate": 1.2313394852183473e-05,
+      "loss": 0.0038273107260465623,
+      "mean_token_accuracy": 0.9986206114292144,
+      "num_tokens": 109763160.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.7385149598121643,
+      "epoch": 7.84656796769852,
+      "grad_norm": 0.8678551912307739,
+      "learning_rate": 1.2167509277658518e-05,
+      "loss": 0.0037419166415929793,
+      "mean_token_accuracy": 0.9987667620182037,
+      "num_tokens": 109951791.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.7384770929813385,
+      "epoch": 7.860026917900404,
+      "grad_norm": 0.45470893383026123,
+      "learning_rate": 1.2022373253213049e-05,
+      "loss": 0.003541013225913048,
+      "mean_token_accuracy": 0.998937851190567,
+      "num_tokens": 110140704.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.7396949470043183,
+      "epoch": 7.873485868102288,
+      "grad_norm": 0.6184161305427551,
+      "learning_rate": 1.1877989654330973e-05,
+      "loss": 0.003454260528087616,
+      "mean_token_accuracy": 0.9989149391651153,
+      "num_tokens": 110329353.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.7411868631839752,
+      "epoch": 7.886944818304173,
+      "grad_norm": 0.49721845984458923,
+      "learning_rate": 1.1734361341588912e-05,
+      "loss": 0.003363589942455292,
+      "mean_token_accuracy": 0.9988781988620759,
+      "num_tokens": 110517720.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.740048885345459,
+      "epoch": 7.900403768506057,
+      "grad_norm": 0.3169788420200348,
+      "learning_rate": 1.1591491160599488e-05,
+      "loss": 0.0030659947544336317,
+      "mean_token_accuracy": 0.9991146922111511,
+      "num_tokens": 110706131.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.7428572535514831,
+      "epoch": 7.913862718707941,
+      "grad_norm": 0.5674341320991516,
+      "learning_rate": 1.1449381941954956e-05,
+      "loss": 0.0031749047338962555,
+      "mean_token_accuracy": 0.998987466096878,
+      "num_tokens": 110893731.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.7389689266681672,
+      "epoch": 7.927321668909825,
+      "grad_norm": 0.36926963925361633,
+      "learning_rate": 1.130803650117116e-05,
+      "loss": 0.0030995842069387438,
+      "mean_token_accuracy": 0.9989600539207458,
+      "num_tokens": 111082079.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.7392809450626373,
+      "epoch": 7.94078061911171,
+      "grad_norm": 0.3086724579334259,
+      "learning_rate": 1.1167457638631656e-05,
+      "loss": 0.0033237963914871218,
+      "mean_token_accuracy": 0.9988582134246826,
+      "num_tokens": 111270413.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.7384293675422668,
+      "epoch": 7.954239569313594,
+      "grad_norm": 0.45126330852508545,
+      "learning_rate": 1.1027648139532355e-05,
+      "loss": 0.0034996122121810913,
+      "mean_token_accuracy": 0.9987758874893189,
+      "num_tokens": 111458912.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.7361682891845703,
+      "epoch": 7.967698519515478,
+      "grad_norm": 0.6252058148384094,
+      "learning_rate": 1.0888610773826253e-05,
+      "loss": 0.003475649654865265,
+      "mean_token_accuracy": 0.9988142251968384,
+      "num_tokens": 111647683.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.733948814868927,
+      "epoch": 7.981157469717362,
+      "grad_norm": 0.5224378705024719,
+      "learning_rate": 1.0750348296168583e-05,
+      "loss": 0.0030234681442379953,
+      "mean_token_accuracy": 0.9990231335163117,
+      "num_tokens": 111836870.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.736539876461029,
+      "epoch": 7.994616419919247,
+      "grad_norm": 0.2892343997955322,
+      "learning_rate": 1.0612863445862236e-05,
+      "loss": 0.0031716078519821167,
+      "mean_token_accuracy": 0.9988950371742249,
+      "num_tokens": 112025584.0,
+      "step": 5940
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.7368154024622243,
+      "eval_loss": 0.08131171762943268,
+      "eval_mean_token_accuracy": 0.9805832057242181,
+      "eval_num_tokens": 112101144.0,
+      "eval_runtime": 24.3602,
+      "eval_samples_per_second": 205.253,
+      "eval_steps_per_second": 6.445,
+      "step": 5944
+    },
+    {
+      "entropy": 0.7347776114940643,
+      "epoch": 8.00807537012113,
+      "grad_norm": 0.21781852841377258,
+      "learning_rate": 1.0476158946803482e-05,
+      "loss": 0.002697869949042797,
+      "mean_token_accuracy": 0.9991762280464173,
+      "num_tokens": 112214484.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.7354790329933166,
+      "epoch": 8.021534320323015,
+      "grad_norm": 0.2759193480014801,
+      "learning_rate": 1.0340237507428009e-05,
+      "loss": 0.0023939665406942368,
+      "mean_token_accuracy": 0.9993740320205688,
+      "num_tokens": 112403209.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.7335577964782715,
+      "epoch": 8.034993270524899,
+      "grad_norm": 0.20684213936328888,
+      "learning_rate": 1.020510182065727e-05,
+      "loss": 0.0021501941606402396,
+      "mean_token_accuracy": 0.9994299292564393,
+      "num_tokens": 112592246.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.7346243798732758,
+      "epoch": 8.048452220726784,
+      "grad_norm": 0.3243059813976288,
+      "learning_rate": 1.0070754563845114e-05,
+      "loss": 0.0025554070249199866,
+      "mean_token_accuracy": 0.9992592394351959,
+      "num_tokens": 112780960.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.7341127455234527,
+      "epoch": 8.061911170928667,
+      "grad_norm": 0.34389808773994446,
+      "learning_rate": 9.937198398724751e-06,
+      "loss": 0.0023281550034880637,
+      "mean_token_accuracy": 0.9992973983287812,
+      "num_tokens": 112969497.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.7344379365444184,
+      "epoch": 8.075370121130552,
+      "grad_norm": 0.33664506673812866,
+      "learning_rate": 9.804435971356025e-06,
+      "loss": 0.0022437553852796555,
+      "mean_token_accuracy": 0.9993536353111268,
+      "num_tokens": 113158346.0,
+      "step": 6000
+    },
+    {
+      "entropy": 0.7345291972160339,
+      "epoch": 8.088829071332436,
+      "grad_norm": 0.21238821744918823,
+      "learning_rate": 9.672469912072956e-06,
+      "loss": 0.0023011134937405585,
+      "mean_token_accuracy": 0.9994915544986724,
+      "num_tokens": 113347027.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.7329188048839569,
+      "epoch": 8.10228802153432,
+      "grad_norm": 0.3358577489852905,
+      "learning_rate": 9.541302835431681e-06,
+      "loss": 0.002364570274949074,
+      "mean_token_accuracy": 0.9992765069007874,
+      "num_tokens": 113535987.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.7357381403446197,
+      "epoch": 8.115746971736204,
+      "grad_norm": 0.4410889148712158,
+      "learning_rate": 9.410937340158599e-06,
+      "loss": 0.0016813736408948898,
+      "mean_token_accuracy": 0.9995461285114289,
+      "num_tokens": 113724178.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.7351392149925232,
+      "epoch": 8.12920592193809,
+      "grad_norm": 0.35911238193511963,
+      "learning_rate": 9.28137600909892e-06,
+      "loss": 0.0021215328946709634,
+      "mean_token_accuracy": 0.9994065165519714,
+      "num_tokens": 113912610.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.7340241730213165,
+      "epoch": 8.142664872139973,
+      "grad_norm": 0.3858516812324524,
+      "learning_rate": 9.15262140916548e-06,
+      "loss": 0.002298617549240589,
+      "mean_token_accuracy": 0.9992367029190063,
+      "num_tokens": 114101352.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.733269315958023,
+      "epoch": 8.156123822341858,
+      "grad_norm": 0.15504567325115204,
+      "learning_rate": 9.024676091287881e-06,
+      "loss": 0.0019876576960086824,
+      "mean_token_accuracy": 0.9994120895862579,
+      "num_tokens": 114290171.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.7359612822532654,
+      "epoch": 8.169582772543741,
+      "grad_norm": 0.3258695602416992,
+      "learning_rate": 8.897542590361963e-06,
+      "loss": 0.002046743780374527,
+      "mean_token_accuracy": 0.9994458854198456,
+      "num_tokens": 114478059.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.734262204170227,
+      "epoch": 8.183041722745626,
+      "grad_norm": 0.3703354299068451,
+      "learning_rate": 8.77122342519956e-06,
+      "loss": 0.0024908211082220076,
+      "mean_token_accuracy": 0.9995096683502197,
+      "num_tokens": 114666866.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.7325477957725525,
+      "epoch": 8.19650067294751,
+      "grad_norm": 0.38723456859588623,
+      "learning_rate": 8.64572109847862e-06,
+      "loss": 0.002013527974486351,
+      "mean_token_accuracy": 0.9994479417800903,
+      "num_tokens": 114855753.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.7337498366832733,
+      "epoch": 8.209959623149395,
+      "grad_norm": 0.44278407096862793,
+      "learning_rate": 8.521038096693635e-06,
+      "loss": 0.0023755524307489395,
+      "mean_token_accuracy": 0.9993066310882568,
+      "num_tokens": 115044397.0,
+      "step": 6100
+    },
+    {
+      "entropy": 0.7312877893447876,
+      "epoch": 8.223418573351278,
+      "grad_norm": 0.36628058552742004,
+      "learning_rate": 8.397176890106317e-06,
+      "loss": 0.0023126287385821344,
+      "mean_token_accuracy": 0.9992603421211242,
+      "num_tokens": 115233499.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.7326091766357422,
+      "epoch": 8.236877523553163,
+      "grad_norm": 0.542598307132721,
+      "learning_rate": 8.274139932696717e-06,
+      "loss": 0.002486979216337204,
+      "mean_token_accuracy": 0.9992321312427521,
+      "num_tokens": 115422096.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.7329172909259796,
+      "epoch": 8.250336473755047,
+      "grad_norm": 0.37169647216796875,
+      "learning_rate": 8.151929662114566e-06,
+      "loss": 0.0022026147693395615,
+      "mean_token_accuracy": 0.9993911325931549,
+      "num_tokens": 115610769.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.7315155625343323,
+      "epoch": 8.263795423956932,
+      "grad_norm": 0.379108190536499,
+      "learning_rate": 8.030548499631014e-06,
+      "loss": 0.0020761553198099135,
+      "mean_token_accuracy": 0.9994912505149841,
+      "num_tokens": 115799639.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.7338620543479919,
+      "epoch": 8.277254374158815,
+      "grad_norm": 0.27872899174690247,
+      "learning_rate": 7.909998850090628e-06,
+      "loss": 0.00214959941804409,
+      "mean_token_accuracy": 0.9993456363677978,
+      "num_tokens": 115987860.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.7319511413574219,
+      "epoch": 8.2907133243607,
+      "grad_norm": 0.3878805637359619,
+      "learning_rate": 7.790283101863771e-06,
+      "loss": 0.0016636446118354798,
+      "mean_token_accuracy": 0.9996107459068299,
+      "num_tokens": 116176426.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.7313687026500701,
+      "epoch": 8.304172274562584,
+      "grad_norm": 0.42844563722610474,
+      "learning_rate": 7.671403626799262e-06,
+      "loss": 0.002215801365673542,
+      "mean_token_accuracy": 0.9993546068668365,
+      "num_tokens": 116365243.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.7326999127864837,
+      "epoch": 8.317631224764469,
+      "grad_norm": 0.28046005964279175,
+      "learning_rate": 7.553362780177398e-06,
+      "loss": 0.0017590772360563279,
+      "mean_token_accuracy": 0.9996075868606568,
+      "num_tokens": 116553541.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.7318588554859161,
+      "epoch": 8.331090174966352,
+      "grad_norm": 0.4567055404186249,
+      "learning_rate": 7.43616290066329e-06,
+      "loss": 0.0024444632232189178,
+      "mean_token_accuracy": 0.9993343472480773,
+      "num_tokens": 116742091.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.7342073738574981,
+      "epoch": 8.344549125168237,
+      "grad_norm": 0.45294421911239624,
+      "learning_rate": 7.3198063102605185e-06,
+      "loss": 0.0021415702998638154,
+      "mean_token_accuracy": 0.9995077669620513,
+      "num_tokens": 116930107.0,
+      "step": 6200
+    },
+    {
+      "entropy": 0.7312484800815582,
+      "epoch": 8.35800807537012,
+      "grad_norm": 0.3044157028198242,
+      "learning_rate": 7.204295314265136e-06,
+      "loss": 0.0018497437238693237,
+      "mean_token_accuracy": 0.999489426612854,
+      "num_tokens": 117118605.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.7315851151943207,
+      "epoch": 8.371467025572006,
+      "grad_norm": 0.29259541630744934,
+      "learning_rate": 7.089632201219997e-06,
+      "loss": 0.0019892159849405288,
+      "mean_token_accuracy": 0.9993900001049042,
+      "num_tokens": 117307207.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.729860270023346,
+      "epoch": 8.38492597577389,
+      "grad_norm": 0.27728405594825745,
+      "learning_rate": 6.975819242869414e-06,
+      "loss": 0.001903151161968708,
+      "mean_token_accuracy": 0.9994141757488251,
+      "num_tokens": 117496022.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.730089795589447,
+      "epoch": 8.398384925975774,
+      "grad_norm": 0.32097047567367554,
+      "learning_rate": 6.862858694114138e-06,
+      "loss": 0.0018681716173887253,
+      "mean_token_accuracy": 0.9994908988475799,
+      "num_tokens": 117684720.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.732157701253891,
+      "epoch": 8.411843876177658,
+      "grad_norm": 0.33475807309150696,
+      "learning_rate": 6.750752792966702e-06,
+      "loss": 0.0019257836043834685,
+      "mean_token_accuracy": 0.9993493318557739,
+      "num_tokens": 117872973.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.7303947329521179,
+      "epoch": 8.425302826379543,
+      "grad_norm": 0.35840341448783875,
+      "learning_rate": 6.639503760507065e-06,
+      "loss": 0.002003169059753418,
+      "mean_token_accuracy": 0.9994274079799652,
+      "num_tokens": 118061309.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.7300981402397155,
+      "epoch": 8.438761776581426,
+      "grad_norm": 0.22061212360858917,
+      "learning_rate": 6.529113800838632e-06,
+      "loss": 0.0023671993985772133,
+      "mean_token_accuracy": 0.999328225851059,
+      "num_tokens": 118250100.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.7311166822910309,
+      "epoch": 8.452220726783311,
+      "grad_norm": 0.4995788335800171,
+      "learning_rate": 6.41958510104454e-06,
+      "loss": 0.002252453379333019,
+      "mean_token_accuracy": 0.9993094325065612,
+      "num_tokens": 118438680.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.7314801871776581,
+      "epoch": 8.465679676985195,
+      "grad_norm": 0.18718956410884857,
+      "learning_rate": 6.31091983114438e-06,
+      "loss": 0.0020001601427793504,
+      "mean_token_accuracy": 0.9993076086044311,
+      "num_tokens": 118627199.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.7294714629650116,
+      "epoch": 8.47913862718708,
+      "grad_norm": 0.24443569779396057,
+      "learning_rate": 6.203120144051161e-06,
+      "loss": 0.001865076832473278,
+      "mean_token_accuracy": 0.9993763923645019,
+      "num_tokens": 118816308.0,
+      "step": 6300
+    },
+    {
+      "entropy": 0.7320528924465179,
+      "epoch": 8.492597577388963,
+      "grad_norm": 0.3000173568725586,
+      "learning_rate": 6.0961881755286775e-06,
+      "loss": 0.0019325431436300277,
+      "mean_token_accuracy": 0.9995656788349152,
+      "num_tokens": 119004887.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.7298780143260956,
+      "epoch": 8.506056527590848,
+      "grad_norm": 0.34283098578453064,
+      "learning_rate": 5.990126044149203e-06,
+      "loss": 0.0018553130328655242,
+      "mean_token_accuracy": 0.9994752407073975,
+      "num_tokens": 119193896.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.7296975553035736,
+      "epoch": 8.519515477792732,
+      "grad_norm": 0.48231956362724304,
+      "learning_rate": 5.884935851251483e-06,
+      "loss": 0.0018540853634476662,
+      "mean_token_accuracy": 0.9994960963726044,
+      "num_tokens": 119383032.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.7316587388515472,
+      "epoch": 8.532974427994617,
+      "grad_norm": 0.31098613142967224,
+      "learning_rate": 5.780619680899146e-06,
+      "loss": 0.0024125341325998305,
+      "mean_token_accuracy": 0.9992330968379974,
+      "num_tokens": 119571647.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.7316439330577851,
+      "epoch": 8.5464333781965,
+      "grad_norm": 0.3148161768913269,
+      "learning_rate": 5.677179599839366e-06,
+      "loss": 0.0020461630076169966,
+      "mean_token_accuracy": 0.9994146227836609,
+      "num_tokens": 119760221.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.7328742682933808,
+      "epoch": 8.559892328398385,
+      "grad_norm": 0.33567124605178833,
+      "learning_rate": 5.574617657461967e-06,
+      "loss": 0.0017173275351524352,
+      "mean_token_accuracy": 0.9996075749397277,
+      "num_tokens": 119948564.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.7310613572597504,
+      "epoch": 8.573351278600269,
+      "grad_norm": 0.3766630291938782,
+      "learning_rate": 5.472935885758767e-06,
+      "loss": 0.0020279109477996824,
+      "mean_token_accuracy": 0.9995895087718963,
+      "num_tokens": 120137124.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.7319568216800689,
+      "epoch": 8.586810228802154,
+      "grad_norm": 0.36246562004089355,
+      "learning_rate": 5.372136299283365e-06,
+      "loss": 0.0020779309794306754,
+      "mean_token_accuracy": 0.999527370929718,
+      "num_tokens": 120325739.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.7306868612766266,
+      "epoch": 8.600269179004037,
+      "grad_norm": 0.3458656072616577,
+      "learning_rate": 5.272220895111196e-06,
+      "loss": 0.001681588776409626,
+      "mean_token_accuracy": 0.9996683657169342,
+      "num_tokens": 120514557.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.7286112308502197,
+      "epoch": 8.613728129205922,
+      "grad_norm": 0.40973636507987976,
+      "learning_rate": 5.173191652799986e-06,
+      "loss": 0.001994205825030804,
+      "mean_token_accuracy": 0.9994747698307037,
+      "num_tokens": 120703746.0,
+      "step": 6400
+    },
+    {
+      "entropy": 0.7324167609214782,
+      "epoch": 8.627187079407806,
+      "grad_norm": 0.30873212218284607,
+      "learning_rate": 5.075050534350512e-06,
+      "loss": 0.001537158153951168,
+      "mean_token_accuracy": 0.9995840191841125,
+      "num_tokens": 120892105.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.7325307846069335,
+      "epoch": 8.640646029609691,
+      "grad_norm": 0.2885727286338806,
+      "learning_rate": 4.977799484167767e-06,
+      "loss": 0.002083958685398102,
+      "mean_token_accuracy": 0.9994043171405792,
+      "num_tokens": 121080661.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.7314725816249847,
+      "epoch": 8.654104979811574,
+      "grad_norm": 0.23452679812908173,
+      "learning_rate": 4.8814404290223655e-06,
+      "loss": 0.0021737243980169295,
+      "mean_token_accuracy": 0.9992907345294952,
+      "num_tokens": 121269176.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.7302026331424714,
+      "epoch": 8.66756393001346,
+      "grad_norm": 0.5051631331443787,
+      "learning_rate": 4.7859752780124555e-06,
+      "loss": 0.002044454775750637,
+      "mean_token_accuracy": 0.9993531286716462,
+      "num_tokens": 121457925.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.7307308673858642,
+      "epoch": 8.681022880215343,
+      "grad_norm": 0.32118919491767883,
+      "learning_rate": 4.691405922525838e-06,
+      "loss": 0.001963503286242485,
+      "mean_token_accuracy": 0.9993822813034058,
+      "num_tokens": 121646615.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.7318669617176056,
+      "epoch": 8.694481830417228,
+      "grad_norm": 0.28653913736343384,
+      "learning_rate": 4.597734236202511e-06,
+      "loss": 0.0022680331021547318,
+      "mean_token_accuracy": 0.9992888629436493,
+      "num_tokens": 121834948.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.7306908547878266,
+      "epoch": 8.707940780619111,
+      "grad_norm": 0.3841933608055115,
+      "learning_rate": 4.504962074897556e-06,
+      "loss": 0.00202475618571043,
+      "mean_token_accuracy": 0.9995454490184784,
+      "num_tokens": 122023470.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.7311248660087586,
+      "epoch": 8.721399730820997,
+      "grad_norm": 0.2170674055814743,
+      "learning_rate": 4.413091276644349e-06,
+      "loss": 0.0017318326979875564,
+      "mean_token_accuracy": 0.9994860172271729,
+      "num_tokens": 122212025.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.730291211605072,
+      "epoch": 8.73485868102288,
+      "grad_norm": 0.4367440640926361,
+      "learning_rate": 4.322123661618166e-06,
+      "loss": 0.002066049911081791,
+      "mean_token_accuracy": 0.999429452419281,
+      "num_tokens": 122400631.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.734411084651947,
+      "epoch": 8.748317631224765,
+      "grad_norm": 0.5407934784889221,
+      "learning_rate": 4.23206103210011e-06,
+      "loss": 0.002084248140454292,
+      "mean_token_accuracy": 0.9993655562400818,
+      "num_tokens": 122588467.0,
+      "step": 6500
+    },
+    {
+      "entropy": 0.7325521528720855,
+      "epoch": 8.761776581426648,
+      "grad_norm": 0.33067557215690613,
+      "learning_rate": 4.142905172441408e-06,
+      "loss": 0.0017788175493478775,
+      "mean_token_accuracy": 0.9996413946151733,
+      "num_tokens": 122776694.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.7312010109424592,
+      "epoch": 8.775235531628534,
+      "grad_norm": 0.5068145394325256,
+      "learning_rate": 4.054657849028043e-06,
+      "loss": 0.001824142411351204,
+      "mean_token_accuracy": 0.999506825208664,
+      "num_tokens": 122965176.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.7314070999622345,
+      "epoch": 8.788694481830417,
+      "grad_norm": 0.21613064408302307,
+      "learning_rate": 3.9673208102457895e-06,
+      "loss": 0.0016722116619348527,
+      "mean_token_accuracy": 0.9995093226432801,
+      "num_tokens": 123153842.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.731372082233429,
+      "epoch": 8.802153432032302,
+      "grad_norm": 0.5407741665840149,
+      "learning_rate": 3.880895786445543e-06,
+      "loss": 0.001678784191608429,
+      "mean_token_accuracy": 0.999667227268219,
+      "num_tokens": 123342339.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.7309448659420014,
+      "epoch": 8.815612382234185,
+      "grad_norm": 0.40189075469970703,
+      "learning_rate": 3.795384489909054e-06,
+      "loss": 0.0022997969761490824,
+      "mean_token_accuracy": 0.9993481457233429,
+      "num_tokens": 123531056.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.7296400606632233,
+      "epoch": 8.82907133243607,
+      "grad_norm": 0.5532040596008301,
+      "learning_rate": 3.7107886148150084e-06,
+      "loss": 0.0017002075910568237,
+      "mean_token_accuracy": 0.9994097113609314,
+      "num_tokens": 123719927.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.7304099202156067,
+      "epoch": 8.842530282637954,
+      "grad_norm": 0.4202295243740082,
+      "learning_rate": 3.6271098372054334e-06,
+      "loss": 0.0020521355792880057,
+      "mean_token_accuracy": 0.9993128538131714,
+      "num_tokens": 123908479.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.7322857022285462,
+      "epoch": 8.855989232839839,
+      "grad_norm": 0.4879668354988098,
+      "learning_rate": 3.5443498149525357e-06,
+      "loss": 0.002071031183004379,
+      "mean_token_accuracy": 0.9994449079036712,
+      "num_tokens": 124096364.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.7318431079387665,
+      "epoch": 8.869448183041722,
+      "grad_norm": 0.5673806071281433,
+      "learning_rate": 3.462510187725809e-06,
+      "loss": 0.0019315868616104127,
+      "mean_token_accuracy": 0.9995098888874054,
+      "num_tokens": 124284715.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.7283908724784851,
+      "epoch": 8.882907133243608,
+      "grad_norm": 0.3319496810436249,
+      "learning_rate": 3.3815925769595745e-06,
+      "loss": 0.0020575495436787605,
+      "mean_token_accuracy": 0.9993783593177795,
+      "num_tokens": 124474114.0,
+      "step": 6600
+    },
+    {
+      "entropy": 0.7316987097263337,
+      "epoch": 8.896366083445491,
+      "grad_norm": 0.32474979758262634,
+      "learning_rate": 3.30159858582087e-06,
+      "loss": 0.0023224465548992157,
+      "mean_token_accuracy": 0.9993063390254975,
+      "num_tokens": 124662486.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.7328927993774415,
+      "epoch": 8.909825033647376,
+      "grad_norm": 0.38243111968040466,
+      "learning_rate": 3.2225297991776536e-06,
+      "loss": 0.0016394183039665222,
+      "mean_token_accuracy": 0.9996214270591736,
+      "num_tokens": 124850474.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.7294083535671234,
+      "epoch": 8.92328398384926,
+      "grad_norm": 0.37468570470809937,
+      "learning_rate": 3.1443877835674208e-06,
+      "loss": 0.001899813488125801,
+      "mean_token_accuracy": 0.9994097173213958,
+      "num_tokens": 125039229.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.7298379123210907,
+      "epoch": 8.936742934051145,
+      "grad_norm": 0.20931333303451538,
+      "learning_rate": 3.0671740871661635e-06,
+      "loss": 0.0022370124235749243,
+      "mean_token_accuracy": 0.9994113147258759,
+      "num_tokens": 125227792.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.7286044478416442,
+      "epoch": 8.950201884253028,
+      "grad_norm": 0.3580392301082611,
+      "learning_rate": 2.9908902397577043e-06,
+      "loss": 0.002087070792913437,
+      "mean_token_accuracy": 0.9993978500366211,
+      "num_tokens": 125416718.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.7310522794723511,
+      "epoch": 8.963660834454913,
+      "grad_norm": 0.32912445068359375,
+      "learning_rate": 2.91553775270339e-06,
+      "loss": 0.0017302701249718665,
+      "mean_token_accuracy": 0.9994479060173035,
+      "num_tokens": 125605155.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.7304746985435486,
+      "epoch": 8.977119784656796,
+      "grad_norm": 0.48885127902030945,
+      "learning_rate": 2.8411181189121284e-06,
+      "loss": 0.0020353911444544793,
+      "mean_token_accuracy": 0.9994727909564972,
+      "num_tokens": 125793654.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.7320328831672669,
+      "epoch": 8.990578734858682,
+      "grad_norm": 0.19737882912158966,
+      "learning_rate": 2.767632812810839e-06,
+      "loss": 0.0019317161291837691,
+      "mean_token_accuracy": 0.9996273815631866,
+      "num_tokens": 125981781.0,
+      "step": 6680
+    },
+    {
+      "epoch": 9.0,
+      "eval_entropy": 0.736012871478014,
+      "eval_loss": 0.08657046407461166,
+      "eval_mean_token_accuracy": 0.9805503571109407,
+      "eval_num_tokens": 126113888.0,
+      "eval_runtime": 24.4287,
+      "eval_samples_per_second": 204.677,
+      "eval_steps_per_second": 6.427,
+      "step": 6687
+    },
+    {
+      "entropy": 0.7297558188438416,
+      "epoch": 9.004037685060565,
+      "grad_norm": 0.2666301429271698,
+      "learning_rate": 2.6950832903152034e-06,
+      "loss": 0.0018328154459595681,
+      "mean_token_accuracy": 0.9994883477687836,
+      "num_tokens": 126170487.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.7315820157527924,
+      "epoch": 9.01749663526245,
+      "grad_norm": 0.2556246519088745,
+      "learning_rate": 2.623470988800869e-06,
+      "loss": 0.0011825882829725742,
+      "mean_token_accuracy": 0.999762237071991,
+      "num_tokens": 126358666.0,
+      "step": 6700
+    },
+    {
+      "entropy": 0.7273661136627197,
+      "epoch": 9.030955585464334,
+      "grad_norm": 0.2150348573923111,
+      "learning_rate": 2.552797327074924e-06,
+      "loss": 0.0013065861538052558,
+      "mean_token_accuracy": 0.99976487159729,
+      "num_tokens": 126547962.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.7294378042221069,
+      "epoch": 9.044414535666219,
+      "grad_norm": 0.19591252505779266,
+      "learning_rate": 2.4830637053478213e-06,
+      "loss": 0.001431029662489891,
+      "mean_token_accuracy": 0.9996244728565216,
+      "num_tokens": 126736567.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.7315397322177887,
+      "epoch": 9.057873485868102,
+      "grad_norm": 0.274129718542099,
+      "learning_rate": 2.414271505205612e-06,
+      "loss": 0.001400233618915081,
+      "mean_token_accuracy": 0.9996635556221009,
+      "num_tokens": 126924542.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.726940393447876,
+      "epoch": 9.071332436069987,
+      "grad_norm": 0.19997361302375793,
+      "learning_rate": 2.346422089582595e-06,
+      "loss": 0.0014671157114207745,
+      "mean_token_accuracy": 0.9996676206588745,
+      "num_tokens": 127113572.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.7296011805534363,
+      "epoch": 9.08479138627187,
+      "grad_norm": 0.3377998471260071,
+      "learning_rate": 2.279516802734305e-06,
+      "loss": 0.0015030583366751672,
+      "mean_token_accuracy": 0.9996662616729737,
+      "num_tokens": 127302070.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.7304975509643554,
+      "epoch": 9.098250336473756,
+      "grad_norm": 0.6750560402870178,
+      "learning_rate": 2.213556970210867e-06,
+      "loss": 0.0015627678483724594,
+      "mean_token_accuracy": 0.9996055901050568,
+      "num_tokens": 127490425.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.7307815790176392,
+      "epoch": 9.111709286675639,
+      "grad_norm": 0.22109167277812958,
+      "learning_rate": 2.1485438988307518e-06,
+      "loss": 0.0016355250030755998,
+      "mean_token_accuracy": 0.9995470702648163,
+      "num_tokens": 127678924.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.7294148445129395,
+      "epoch": 9.125168236877524,
+      "grad_norm": 0.29477694630622864,
+      "learning_rate": 2.0844788766548883e-06,
+      "loss": 0.0013017470948398114,
+      "mean_token_accuracy": 0.9998429238796234,
+      "num_tokens": 127867757.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.7299641191959381,
+      "epoch": 9.138627187079408,
+      "grad_norm": 0.3363890051841736,
+      "learning_rate": 2.021363172961116e-06,
+      "loss": 0.001594964787364006,
+      "mean_token_accuracy": 0.9996260702610016,
+      "num_tokens": 128056322.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.7288505673408509,
+      "epoch": 9.152086137281293,
+      "grad_norm": 0.1246197298169136,
+      "learning_rate": 1.959198038219079e-06,
+      "loss": 0.001545974612236023,
+      "mean_token_accuracy": 0.9995907962322235,
+      "num_tokens": 128245037.0,
+      "step": 6800
+    },
+    {
+      "entropy": 0.732088565826416,
+      "epoch": 9.165545087483176,
+      "grad_norm": 0.3340923488140106,
+      "learning_rate": 1.897984704065403e-06,
+      "loss": 0.0015842836350202561,
+      "mean_token_accuracy": 0.9995262265205384,
+      "num_tokens": 128433084.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.7305635511875153,
+      "epoch": 9.179004037685061,
+      "grad_norm": 0.2869892716407776,
+      "learning_rate": 1.8377243832793322e-06,
+      "loss": 0.001532875280827284,
+      "mean_token_accuracy": 0.9997069716453553,
+      "num_tokens": 128621547.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.7285775423049927,
+      "epoch": 9.192462987886945,
+      "grad_norm": 0.1667216569185257,
+      "learning_rate": 1.778418269758697e-06,
+      "loss": 0.0015358133241534233,
+      "mean_token_accuracy": 0.9996080040931702,
+      "num_tokens": 128810123.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.7308950185775757,
+      "epoch": 9.20592193808883,
+      "grad_norm": 0.5063604116439819,
+      "learning_rate": 1.7200675384962339e-06,
+      "loss": 0.0014127939939498901,
+      "mean_token_accuracy": 0.9997045576572419,
+      "num_tokens": 128998540.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.7289383709430695,
+      "epoch": 9.219380888290713,
+      "grad_norm": 0.33756718039512634,
+      "learning_rate": 1.6626733455563371e-06,
+      "loss": 0.0015469443053007126,
+      "mean_token_accuracy": 0.9995954632759094,
+      "num_tokens": 129187255.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.724688583612442,
+      "epoch": 9.232839838492598,
+      "grad_norm": 0.20516812801361084,
+      "learning_rate": 1.6062368280521323e-06,
+      "loss": 0.0015621397644281388,
+      "mean_token_accuracy": 0.9995361566543579,
+      "num_tokens": 129376838.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.7303467333316803,
+      "epoch": 9.246298788694482,
+      "grad_norm": 0.2556150257587433,
+      "learning_rate": 1.5507591041229508e-06,
+      "loss": 0.0013798503205180168,
+      "mean_token_accuracy": 0.9997849702835083,
+      "num_tokens": 129565322.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.7273888111114502,
+      "epoch": 9.259757738896367,
+      "grad_norm": 0.24669574201107025,
+      "learning_rate": 1.4962412729122077e-06,
+      "loss": 0.0012558025307953357,
+      "mean_token_accuracy": 0.9997637927532196,
+      "num_tokens": 129754201.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.7290287375450134,
+      "epoch": 9.27321668909825,
+      "grad_norm": 0.2814815938472748,
+      "learning_rate": 1.4426844145455642e-06,
+      "loss": 0.0013902703300118445,
+      "mean_token_accuracy": 0.9997455954551697,
+      "num_tokens": 129942670.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.7293418049812317,
+      "epoch": 9.286675639300135,
+      "grad_norm": 0.3069918155670166,
+      "learning_rate": 1.3900895901095889e-06,
+      "loss": 0.0014243736863136292,
+      "mean_token_accuracy": 0.9996042132377625,
+      "num_tokens": 130131145.0,
+      "step": 6900
+    },
+    {
+      "entropy": 0.7311438262462616,
+      "epoch": 9.300134589502019,
+      "grad_norm": 0.22917664051055908,
+      "learning_rate": 1.3384578416307025e-06,
+      "loss": 0.0014917708933353425,
+      "mean_token_accuracy": 0.9996256232261658,
+      "num_tokens": 130319024.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.729202824831009,
+      "epoch": 9.313593539703904,
+      "grad_norm": 0.39373698830604553,
+      "learning_rate": 1.2877901920545332e-06,
+      "loss": 0.0015949323773384095,
+      "mean_token_accuracy": 0.9995456159114837,
+      "num_tokens": 130507532.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.7272425711154937,
+      "epoch": 9.327052489905787,
+      "grad_norm": 0.5044791102409363,
+      "learning_rate": 1.2380876452256663e-06,
+      "loss": 0.0015190449543297292,
+      "mean_token_accuracy": 0.999534422159195,
+      "num_tokens": 130696447.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.7261317849159241,
+      "epoch": 9.340511440107672,
+      "grad_norm": 0.31581446528434753,
+      "learning_rate": 1.189351185867732e-06,
+      "loss": 0.0013408970087766647,
+      "mean_token_accuracy": 0.9997290849685669,
+      "num_tokens": 130885904.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.7282789885997772,
+      "epoch": 9.353970390309556,
+      "grad_norm": 0.32402175664901733,
+      "learning_rate": 1.1415817795639328e-06,
+      "loss": 0.0013133164495229722,
+      "mean_token_accuracy": 0.9997660279273987,
+      "num_tokens": 131074702.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.7319800734519959,
+      "epoch": 9.36742934051144,
+      "grad_norm": 0.16617992520332336,
+      "learning_rate": 1.094780372737858e-06,
+      "loss": 0.0014591993764042853,
+      "mean_token_accuracy": 0.9995836853981018,
+      "num_tokens": 131262743.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.7262159407138824,
+      "epoch": 9.380888290713324,
+      "grad_norm": 0.07105099409818649,
+      "learning_rate": 1.048947892634794e-06,
+      "loss": 0.001161899045109749,
+      "mean_token_accuracy": 0.9997846364974976,
+      "num_tokens": 131451717.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.7294895708560943,
+      "epoch": 9.39434724091521,
+      "grad_norm": 0.31130683422088623,
+      "learning_rate": 1.0040852473032992e-06,
+      "loss": 0.0014551781117916108,
+      "mean_token_accuracy": 0.9996439456939697,
+      "num_tokens": 131639996.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.7307281970977784,
+      "epoch": 9.407806191117093,
+      "grad_norm": 0.2622976303100586,
+      "learning_rate": 9.601933255772578e-07,
+      "loss": 0.0016805168241262435,
+      "mean_token_accuracy": 0.9995856046676636,
+      "num_tokens": 131828144.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.7298398494720459,
+      "epoch": 9.421265141318978,
+      "grad_norm": 0.2574930787086487,
+      "learning_rate": 9.172729970582439e-07,
+      "loss": 0.0013261831365525723,
+      "mean_token_accuracy": 0.9997450828552246,
+      "num_tokens": 132016377.0,
+      "step": 7000
+    },
+    {
+      "entropy": 0.729549252986908,
+      "epoch": 9.434724091520861,
+      "grad_norm": 0.26791438460350037,
+      "learning_rate": 8.753251120982908e-07,
+      "loss": 0.0014426372945308685,
+      "mean_token_accuracy": 0.999589067697525,
+      "num_tokens": 132204840.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.7262800872325897,
+      "epoch": 9.448183041722746,
+      "grad_norm": 0.25162211060523987,
+      "learning_rate": 8.3435050178306e-07,
+      "loss": 0.001285465806722641,
+      "mean_token_accuracy": 0.9998051404953003,
+      "num_tokens": 132394014.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.7293421506881714,
+      "epoch": 9.46164199192463,
+      "grad_norm": 0.30124571919441223,
+      "learning_rate": 7.943499779153707e-07,
+      "loss": 0.0012954596430063247,
+      "mean_token_accuracy": 0.9997842133045196,
+      "num_tokens": 132582344.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.7287032663822174,
+      "epoch": 9.475100942126515,
+      "grad_norm": 0.2695083022117615,
+      "learning_rate": 7.553243329990967e-07,
+      "loss": 0.001402657851576805,
+      "mean_token_accuracy": 0.9996848106384277,
+      "num_tokens": 132771055.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.7285751461982727,
+      "epoch": 9.488559892328398,
+      "grad_norm": 0.36903634667396545,
+      "learning_rate": 7.172743402235005e-07,
+      "loss": 0.0013611355796456337,
+      "mean_token_accuracy": 0.999744713306427,
+      "num_tokens": 132959845.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.7260940730571747,
+      "epoch": 9.502018842530283,
+      "grad_norm": 0.25283563137054443,
+      "learning_rate": 6.802007534478794e-07,
+      "loss": 0.0011639551259577275,
+      "mean_token_accuracy": 0.9997863054275513,
+      "num_tokens": 133149249.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.7267294883728027,
+      "epoch": 9.515477792732167,
+      "grad_norm": 0.26445674896240234,
+      "learning_rate": 6.441043071866604e-07,
+      "loss": 0.001253593061119318,
+      "mean_token_accuracy": 0.999766182899475,
+      "num_tokens": 133338277.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.7308687388896942,
+      "epoch": 9.528936742934052,
+      "grad_norm": 0.2843661904335022,
+      "learning_rate": 6.08985716594812e-07,
+      "loss": 0.0015101894736289979,
+      "mean_token_accuracy": 0.9996661841869354,
+      "num_tokens": 133526376.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.7284866154193879,
+      "epoch": 9.542395693135935,
+      "grad_norm": 0.28783294558525085,
+      "learning_rate": 5.74845677453717e-07,
+      "loss": 0.0014467479661107064,
+      "mean_token_accuracy": 0.9996878504753113,
+      "num_tokens": 133714961.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.7299886226654053,
+      "epoch": 9.55585464333782,
+      "grad_norm": 0.24012354016304016,
+      "learning_rate": 5.416848661573548e-07,
+      "loss": 0.0015434769913554192,
+      "mean_token_accuracy": 0.9995826780796051,
+      "num_tokens": 133903227.0,
+      "step": 7100
+    },
+    {
+      "entropy": 0.7316705644130707,
+      "epoch": 9.569313593539704,
+      "grad_norm": 0.33491232991218567,
+      "learning_rate": 5.095039396989132e-07,
+      "loss": 0.0014156468212604522,
+      "mean_token_accuracy": 0.9996202051639557,
+      "num_tokens": 134090771.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.7270708501338958,
+      "epoch": 9.582772543741589,
+      "grad_norm": 0.2644691467285156,
+      "learning_rate": 4.783035356577703e-07,
+      "loss": 0.001502026617527008,
+      "mean_token_accuracy": 0.999491161108017,
+      "num_tokens": 134279592.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.7281914114952087,
+      "epoch": 9.596231493943472,
+      "grad_norm": 0.3748234212398529,
+      "learning_rate": 4.480842721868772e-07,
+      "loss": 0.001145925372838974,
+      "mean_token_accuracy": 0.9997639060020447,
+      "num_tokens": 134468176.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.7299175381660461,
+      "epoch": 9.609690444145357,
+      "grad_norm": 0.2274537980556488,
+      "learning_rate": 4.1884674800046765e-07,
+      "loss": 0.0012723494321107865,
+      "mean_token_accuracy": 0.9997045397758484,
+      "num_tokens": 134656236.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.7289760828018188,
+      "epoch": 9.62314939434724,
+      "grad_norm": 0.38674935698509216,
+      "learning_rate": 3.905915423622564e-07,
+      "loss": 0.0014971816912293435,
+      "mean_token_accuracy": 0.9996852397918701,
+      "num_tokens": 134844499.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.7284032285213471,
+      "epoch": 9.636608344549126,
+      "grad_norm": 0.33098381757736206,
+      "learning_rate": 3.633192150739095e-07,
+      "loss": 0.0014178214594721794,
+      "mean_token_accuracy": 0.9996445059776307,
+      "num_tokens": 135033059.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.7292269229888916,
+      "epoch": 9.65006729475101,
+      "grad_norm": 0.3132631778717041,
+      "learning_rate": 3.370303064639868e-07,
+      "loss": 0.0014329923316836358,
+      "mean_token_accuracy": 0.9996257305145264,
+      "num_tokens": 135221364.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.7295327007770538,
+      "epoch": 9.663526244952894,
+      "grad_norm": 0.1692410558462143,
+      "learning_rate": 3.1172533737721667e-07,
+      "loss": 0.0011812438257038592,
+      "mean_token_accuracy": 0.9998214066028595,
+      "num_tokens": 135409698.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.7276640355587005,
+      "epoch": 9.676985195154778,
+      "grad_norm": 0.17885783314704895,
+      "learning_rate": 2.8740480916419344e-07,
+      "loss": 0.001193959079682827,
+      "mean_token_accuracy": 0.99972585439682,
+      "num_tokens": 135598501.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.7296628355979919,
+      "epoch": 9.690444145356663,
+      "grad_norm": 0.26338687539100647,
+      "learning_rate": 2.6406920367143537e-07,
+      "loss": 0.0014330293983221055,
+      "mean_token_accuracy": 0.9996665298938752,
+      "num_tokens": 135786716.0,
+      "step": 7200
+    },
+    {
+      "entropy": 0.7274375677108764,
+      "epoch": 9.703903095558546,
+      "grad_norm": 0.3093056380748749,
+      "learning_rate": 2.417189832318367e-07,
+      "loss": 0.0014527231454849244,
+      "mean_token_accuracy": 0.9996660470962524,
+      "num_tokens": 135975719.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.728153258562088,
+      "epoch": 9.717362045760431,
+      "grad_norm": 0.23074257373809814,
+      "learning_rate": 2.2035459065550822e-07,
+      "loss": 0.0014972404576838016,
+      "mean_token_accuracy": 0.9996064722537994,
+      "num_tokens": 136164292.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.7310771107673645,
+      "epoch": 9.730820995962315,
+      "grad_norm": 0.31236007809638977,
+      "learning_rate": 1.9997644922100654e-07,
+      "loss": 0.0013590503484010696,
+      "mean_token_accuracy": 0.9997812390327454,
+      "num_tokens": 136352400.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.7283227026462555,
+      "epoch": 9.7442799461642,
+      "grad_norm": 0.2554883360862732,
+      "learning_rate": 1.8058496266695757e-07,
+      "loss": 0.0015128158032894135,
+      "mean_token_accuracy": 0.9997238397598267,
+      "num_tokens": 136541092.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.7284665584564209,
+      "epoch": 9.757738896366083,
+      "grad_norm": 0.3527924120426178,
+      "learning_rate": 1.6218051518403498e-07,
+      "loss": 0.001595771312713623,
+      "mean_token_accuracy": 0.999547290802002,
+      "num_tokens": 136729417.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.7277413606643677,
+      "epoch": 9.771197846567969,
+      "grad_norm": 0.27307775616645813,
+      "learning_rate": 1.4476347140736647e-07,
+      "loss": 0.0015288094058632852,
+      "mean_token_accuracy": 0.9996108889579773,
+      "num_tokens": 136918205.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.7279824018478394,
+      "epoch": 9.784656796769852,
+      "grad_norm": 0.4445931017398834,
+      "learning_rate": 1.2833417640930045e-07,
+      "loss": 0.001231418550014496,
+      "mean_token_accuracy": 0.9997648000717163,
+      "num_tokens": 137106866.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.7275644481182099,
+      "epoch": 9.798115746971735,
+      "grad_norm": 0.2426786720752716,
+      "learning_rate": 1.1289295569258395e-07,
+      "loss": 0.0012167682871222496,
+      "mean_token_accuracy": 0.9998240888118743,
+      "num_tokens": 137295655.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.7293566644191742,
+      "epoch": 9.81157469717362,
+      "grad_norm": 0.23732174932956696,
+      "learning_rate": 9.84401151838954e-08,
+      "loss": 0.0012796245515346526,
+      "mean_token_accuracy": 0.9996869385242462,
+      "num_tokens": 137483960.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.7269932687282562,
+      "epoch": 9.825033647375506,
+      "grad_norm": 0.22463370859622955,
+      "learning_rate": 8.497594122778285e-08,
+      "loss": 0.0016963055357336998,
+      "mean_token_accuracy": 0.9996106028556824,
+      "num_tokens": 137673154.0,
+      "step": 7300
+    },
+    {
+      "entropy": 0.7278582215309143,
+      "epoch": 9.838492597577389,
+      "grad_norm": 0.22611311078071594,
+      "learning_rate": 7.250070058100745e-08,
+      "loss": 0.001445792242884636,
+      "mean_token_accuracy": 0.9996261775493622,
+      "num_tokens": 137861731.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.7275733351707458,
+      "epoch": 9.851951547779272,
+      "grad_norm": 0.35581526160240173,
+      "learning_rate": 6.101464040725868e-08,
+      "loss": 0.0016376268118619918,
+      "mean_token_accuracy": 0.9996063828468322,
+      "num_tokens": 138050534.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.7282656311988831,
+      "epoch": 9.865410497981157,
+      "grad_norm": 0.22982017695903778,
+      "learning_rate": 5.051798827222509e-08,
+      "loss": 0.0013012098148465156,
+      "mean_token_accuracy": 0.9996688842773438,
+      "num_tokens": 138239141.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.7286642968654633,
+      "epoch": 9.878869448183043,
+      "grad_norm": 0.3336293399333954,
+      "learning_rate": 4.1010952139136637e-08,
+      "loss": 0.001432620082050562,
+      "mean_token_accuracy": 0.9997052252292633,
+      "num_tokens": 138427503.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.7269477367401123,
+      "epoch": 9.892328398384926,
+      "grad_norm": 0.1454067826271057,
+      "learning_rate": 3.249372036459031e-08,
+      "loss": 0.0013093785382807255,
+      "mean_token_accuracy": 0.9997073352336884,
+      "num_tokens": 138616376.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.7274815142154694,
+      "epoch": 9.90578734858681,
+      "grad_norm": 0.3983877897262573,
+      "learning_rate": 2.496646169486416e-08,
+      "loss": 0.0016518237069249153,
+      "mean_token_accuracy": 0.9995528280735015,
+      "num_tokens": 138805320.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.7271095037460327,
+      "epoch": 9.919246298788694,
+      "grad_norm": 0.26817628741264343,
+      "learning_rate": 1.8429325262553367e-08,
+      "loss": 0.0012257051654160022,
+      "mean_token_accuracy": 0.9997453272342682,
+      "num_tokens": 138994230.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.7310165941715241,
+      "epoch": 9.93270524899058,
+      "grad_norm": 0.15735772252082825,
+      "learning_rate": 1.2882440583600331e-08,
+      "loss": 0.0012903917580842973,
+      "mean_token_accuracy": 0.9996813893318176,
+      "num_tokens": 139182270.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.7280699729919433,
+      "epoch": 9.946164199192463,
+      "grad_norm": 0.28889599442481995,
+      "learning_rate": 8.325917554763418e-09,
+      "loss": 0.0013606024906039238,
+      "mean_token_accuracy": 0.9996875703334809,
+      "num_tokens": 139370851.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.7259236514568329,
+      "epoch": 9.959623149394346,
+      "grad_norm": 0.31098631024360657,
+      "learning_rate": 4.7598464514186925e-09,
+      "loss": 0.001220638770610094,
+      "mean_token_accuracy": 0.9997267067432404,
+      "num_tokens": 139559862.0,
+      "step": 7400
+    },
+    {
+      "entropy": 0.7279736757278442,
+      "epoch": 9.973082099596231,
+      "grad_norm": 0.1903763860464096,
+      "learning_rate": 2.1842979257613583e-09,
+      "loss": 0.0012149072252213954,
+      "mean_token_accuracy": 0.9998049855232238,
+      "num_tokens": 139748702.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.7281370103359223,
+      "epoch": 9.986541049798117,
+      "grad_norm": 0.12852998077869415,
+      "learning_rate": 5.993230054401888e-10,
+      "loss": 0.0013828342780470848,
+      "mean_token_accuracy": 0.99956915974617,
+      "num_tokens": 139937586.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.7273533165454864,
+      "epoch": 10.0,
+      "grad_norm": 0.1852026879787445,
+      "learning_rate": 4.95309250281295e-12,
+      "loss": 0.0015368043445050717,
+      "mean_token_accuracy": 0.9996884107589722,
+      "num_tokens": 140126665.0,
+      "step": 7430
+    },
+    {
+      "epoch": 10.0,
+      "eval_entropy": 0.7323712585078683,
+      "eval_loss": 0.08809854090213776,
+      "eval_mean_token_accuracy": 0.9805565890233228,
+      "eval_num_tokens": 140126665.0,
+      "eval_runtime": 24.4021,
+      "eval_samples_per_second": 204.901,
+      "eval_steps_per_second": 6.434,
+      "step": 7430
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.770554902167945e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50/seed_42/checkpoint-7430/training_args.bin b/baby_talk_L16_a50/seed_42/checkpoint-7430/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/checkpoint-7430/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50/seed_42/tokenizer.json b/baby_talk_L16_a50/seed_42/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50/seed_42/tokenizer_config.json b/baby_talk_L16_a50/seed_42/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d75d3bb5300d205e48769cc1999073ab5971214
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/tokenizer_config.json
@@ -0,0 +1,29 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50/seed_42/training_args.bin b/baby_talk_L16_a50/seed_42/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..33b81c624824420eca384035ff5fc710dc86bf5f
--- /dev/null
+++ b/baby_talk_L16_a50/seed_42/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c060e97b69d99564c146471c3d3ac4d335e3b1968074124f4edc5aebf612e1e3
+size 5368
diff --git a/baby_talk_L16_a50_per_step/seed_42/README.md b/baby_talk_L16_a50_per_step/seed_42/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..733b17006097ab86918504cea04cbcd1491c94bd
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/README.md
@@ -0,0 +1,62 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+model_name: seed_42
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+licence: license
+pipeline_tag: text-generation
+---
+
+# Model Card for seed_42
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct).
+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="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/camilab-stanford-university/subliminal_learning/runs/2ita4s3n) 
+
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- PEFT 0.19.1
+- TRL: 1.2.0
+- Transformers: 5.6.2
+- Pytorch: 2.10.0
+- 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/baby_talk_L16_a50_per_step/seed_42/adapter_config.json b/baby_talk_L16_a50_per_step/seed_42/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3db6c6926ce4988b431e6c91620dabbc075f719c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "down_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/adapter_model.safetensors b/baby_talk_L16_a50_per_step/seed_42/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..58de073cf7ec1538d136fdb8361be9b05ecb7d96
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:856321609aa3db6b216c78100934d3814c8c05553d28df913a013bcc3eb4b9ae
+size 80792096
diff --git a/baby_talk_L16_a50_per_step/seed_42/chat_template.jinja b/baby_talk_L16_a50_per_step/seed_42/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/README.md b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/adapter_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3db6c6926ce4988b431e6c91620dabbc075f719c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "down_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/adapter_model.safetensors b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2370603b15eacfc52a21587f4d826c4532de137e
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84e75758a85c865be0d4696662f25283a13056f46c5836c95d56654c683fc477
+size 80792096
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/chat_template.jinja b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/tokenizer.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/tokenizer_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/trainer_state.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..9520bf0e6213cc3538ec7eb163e7796cf0ffc414
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/trainer_state.json
@@ -0,0 +1,33 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.0003368137420006736,
+  "eval_steps": 500,
+  "global_step": 1,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 219871656804352.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/training_args.bin b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8fec1354c64292c8e27becda8595a5aa647aa71c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc01187e7f5b0296affbb4d368cbd2f85c3b102cf884d95f9d29965cd5276e76
+size 5841
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/README.md b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/adapter_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3db6c6926ce4988b431e6c91620dabbc075f719c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "down_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/adapter_model.safetensors b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d82cdb79a474343e64478fe66fdcaba97eceb030
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f13c8f699a41d366b376020a80cc88ea724ebbe51333ab0a834c957a65e9cc8b
+size 80792096
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/chat_template.jinja b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/tokenizer.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/tokenizer_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/trainer_state.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e616648fd458ccb8c6acf6e5e601e50ad7450593
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/trainer_state.json
@@ -0,0 +1,44 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.003368137420006736,
+  "eval_steps": 500,
+  "global_step": 10,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2336307406425475,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 3.205052614212036,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5239480972290039,
+      "mean_token_accuracy": 0.8582192718982696,
+      "num_tokens": 47369.0,
+      "step": 10
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2201779886358528.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/training_args.bin b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8fec1354c64292c8e27becda8595a5aa647aa71c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-10/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc01187e7f5b0296affbb4d368cbd2f85c3b102cf884d95f9d29965cd5276e76
+size 5841
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/README.md b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/adapter_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3db6c6926ce4988b431e6c91620dabbc075f719c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "down_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/adapter_model.safetensors b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7ecb0c06442a59898e9029c59da26e736be74892
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1fd1384c4571a7da7c61cb3dd204ee8b8bbfc09728d42ea47ede21ee797045c0
+size 80792096
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/chat_template.jinja b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/tokenizer.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/tokenizer_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/trainer_state.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..8117d358b2105bb9ad32f2ac5ff477a178bca813
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/trainer_state.json
@@ -0,0 +1,134 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.033681374200067365,
+  "eval_steps": 500,
+  "global_step": 100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2336307406425475,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 3.205052614212036,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5239480972290039,
+      "mean_token_accuracy": 0.8582192718982696,
+      "num_tokens": 47369.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2293899655342102,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 1.4208030700683594,
+      "learning_rate": 3.8e-05,
+      "loss": 0.295538067817688,
+      "mean_token_accuracy": 0.9012860953807831,
+      "num_tokens": 94622.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.238785707950592,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 0.6221668720245361,
+      "learning_rate": 5.8e-05,
+      "loss": 0.19195973873138428,
+      "mean_token_accuracy": 0.9294111549854278,
+      "num_tokens": 141694.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2240872263908387,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 0.6423875689506531,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.16067487001419067,
+      "mean_token_accuracy": 0.9407785713672638,
+      "num_tokens": 188811.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2138957858085633,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 0.7921156287193298,
+      "learning_rate": 9.8e-05,
+      "loss": 0.14190430641174318,
+      "mean_token_accuracy": 0.9457307100296021,
+      "num_tokens": 236035.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2117965459823608,
+      "epoch": 0.020208824520040417,
+      "grad_norm": 0.5566120147705078,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.13536434173583983,
+      "mean_token_accuracy": 0.9480699181556702,
+      "num_tokens": 283008.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2039262533187867,
+      "epoch": 0.023576961940047155,
+      "grad_norm": 0.7983958721160889,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.1264912724494934,
+      "mean_token_accuracy": 0.9510111153125763,
+      "num_tokens": 330291.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2005774617195129,
+      "epoch": 0.02694509936005389,
+      "grad_norm": 0.6949729919433594,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.10797474384307862,
+      "mean_token_accuracy": 0.9581791937351227,
+      "num_tokens": 377729.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.1869928240776062,
+      "epoch": 0.030313236780060628,
+      "grad_norm": 0.6609069108963013,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.11980686187744141,
+      "mean_token_accuracy": 0.9547467947006225,
+      "num_tokens": 424960.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.1932473540306092,
+      "epoch": 0.033681374200067365,
+      "grad_norm": 0.5904498100280762,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.12140004634857178,
+      "mean_token_accuracy": 0.952862948179245,
+      "num_tokens": 472087.0,
+      "step": 100
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.199159050338304e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/training_args.bin b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8fec1354c64292c8e27becda8595a5aa647aa71c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-100/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc01187e7f5b0296affbb4d368cbd2f85c3b102cf884d95f9d29965cd5276e76
+size 5841
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/README.md b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/adapter_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3db6c6926ce4988b431e6c91620dabbc075f719c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "down_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/adapter_model.safetensors b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..58de073cf7ec1538d136fdb8361be9b05ecb7d96
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:856321609aa3db6b216c78100934d3814c8c05553d28df913a013bcc3eb4b9ae
+size 80792096
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/chat_template.jinja b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/tokenizer.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/tokenizer_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/trainer_state.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..a62f0f0dd6d5da5b15196126921943a95757bdb8
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/trainer_state.json
@@ -0,0 +1,1034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.33681374200067365,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2336307406425475,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 3.205052614212036,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5239480972290039,
+      "mean_token_accuracy": 0.8582192718982696,
+      "num_tokens": 47369.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2293899655342102,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 1.4208030700683594,
+      "learning_rate": 3.8e-05,
+      "loss": 0.295538067817688,
+      "mean_token_accuracy": 0.9012860953807831,
+      "num_tokens": 94622.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.238785707950592,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 0.6221668720245361,
+      "learning_rate": 5.8e-05,
+      "loss": 0.19195973873138428,
+      "mean_token_accuracy": 0.9294111549854278,
+      "num_tokens": 141694.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2240872263908387,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 0.6423875689506531,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.16067487001419067,
+      "mean_token_accuracy": 0.9407785713672638,
+      "num_tokens": 188811.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2138957858085633,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 0.7921156287193298,
+      "learning_rate": 9.8e-05,
+      "loss": 0.14190430641174318,
+      "mean_token_accuracy": 0.9457307100296021,
+      "num_tokens": 236035.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2117965459823608,
+      "epoch": 0.020208824520040417,
+      "grad_norm": 0.5566120147705078,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.13536434173583983,
+      "mean_token_accuracy": 0.9480699181556702,
+      "num_tokens": 283008.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2039262533187867,
+      "epoch": 0.023576961940047155,
+      "grad_norm": 0.7983958721160889,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.1264912724494934,
+      "mean_token_accuracy": 0.9510111153125763,
+      "num_tokens": 330291.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2005774617195129,
+      "epoch": 0.02694509936005389,
+      "grad_norm": 0.6949729919433594,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.10797474384307862,
+      "mean_token_accuracy": 0.9581791937351227,
+      "num_tokens": 377729.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.1869928240776062,
+      "epoch": 0.030313236780060628,
+      "grad_norm": 0.6609069108963013,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.11980686187744141,
+      "mean_token_accuracy": 0.9547467947006225,
+      "num_tokens": 424960.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.1932473540306092,
+      "epoch": 0.033681374200067365,
+      "grad_norm": 0.5904498100280762,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.12140004634857178,
+      "mean_token_accuracy": 0.952862948179245,
+      "num_tokens": 472087.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.1891833305358888,
+      "epoch": 0.037049511620074096,
+      "grad_norm": 0.8453942537307739,
+      "learning_rate": 9.905132290792394e-05,
+      "loss": 0.12110424041748047,
+      "mean_token_accuracy": 0.954711276292801,
+      "num_tokens": 519101.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.192702329158783,
+      "epoch": 0.040417649040080834,
+      "grad_norm": 0.9638974666595459,
+      "learning_rate": 9.870399824239117e-05,
+      "loss": 0.12223016023635865,
+      "mean_token_accuracy": 0.9538575768470764,
+      "num_tokens": 566562.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1896525263786315,
+      "epoch": 0.04378578646008757,
+      "grad_norm": 0.8424277901649475,
+      "learning_rate": 9.830341646984521e-05,
+      "loss": 0.12164583206176757,
+      "mean_token_accuracy": 0.9546079397201538,
+      "num_tokens": 613861.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.191524910926819,
+      "epoch": 0.04715392388009431,
+      "grad_norm": 0.7796409726142883,
+      "learning_rate": 9.785001562057309e-05,
+      "loss": 0.12058537006378174,
+      "mean_token_accuracy": 0.9509174525737762,
+      "num_tokens": 660887.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1798339486122131,
+      "epoch": 0.05052206130010104,
+      "grad_norm": 1.203282356262207,
+      "learning_rate": 9.734429148174675e-05,
+      "loss": 0.11496694087982177,
+      "mean_token_accuracy": 0.9555849313735962,
+      "num_tokens": 707771.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1805959939956665,
+      "epoch": 0.05389019872010778,
+      "grad_norm": 0.7123157978057861,
+      "learning_rate": 9.6786797055287e-05,
+      "loss": 0.11297221183776855,
+      "mean_token_accuracy": 0.9562048316001892,
+      "num_tokens": 755026.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1713955640792846,
+      "epoch": 0.05725833614011452,
+      "grad_norm": 0.869977593421936,
+      "learning_rate": 9.617814195316411e-05,
+      "loss": 0.11606066226959229,
+      "mean_token_accuracy": 0.9540329277515411,
+      "num_tokens": 802572.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.178900408744812,
+      "epoch": 0.060626473560121255,
+      "grad_norm": 0.8154799938201904,
+      "learning_rate": 9.551899173079607e-05,
+      "loss": 0.10659003257751465,
+      "mean_token_accuracy": 0.959269380569458,
+      "num_tokens": 849536.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1731992602348327,
+      "epoch": 0.06399461098012799,
+      "grad_norm": 0.8263419270515442,
+      "learning_rate": 9.481006715927351e-05,
+      "loss": 0.10587993860244752,
+      "mean_token_accuracy": 0.9589337944984436,
+      "num_tokens": 896651.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1755276799201966,
+      "epoch": 0.06736274840013473,
+      "grad_norm": 0.9151034355163574,
+      "learning_rate": 9.405214343720707e-05,
+      "loss": 0.11514871120452881,
+      "mean_token_accuracy": 0.9544816732406616,
+      "num_tokens": 943494.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.168901538848877,
+      "epoch": 0.07073088582014146,
+      "grad_norm": 0.7708814144134521,
+      "learning_rate": 9.32460493430591e-05,
+      "loss": 0.10262255668640137,
+      "mean_token_accuracy": 0.9597939789295197,
+      "num_tokens": 990612.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1802135467529298,
+      "epoch": 0.07409902324014819,
+      "grad_norm": 0.7555872201919556,
+      "learning_rate": 9.239266632888659e-05,
+      "loss": 0.10907113552093506,
+      "mean_token_accuracy": 0.9556582808494568,
+      "num_tokens": 1037718.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1758136510849,
+      "epoch": 0.07746716066015494,
+      "grad_norm": 0.7670081853866577,
+      "learning_rate": 9.14929275564863e-05,
+      "loss": 0.10124562978744507,
+      "mean_token_accuracy": 0.9586797237396241,
+      "num_tokens": 1084625.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1657557368278504,
+      "epoch": 0.08083529808016167,
+      "grad_norm": 0.8729236125946045,
+      "learning_rate": 9.0547816876996e-05,
+      "loss": 0.10474343299865722,
+      "mean_token_accuracy": 0.957787013053894,
+      "num_tokens": 1131731.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1566669702529908,
+      "epoch": 0.08420343550016841,
+      "grad_norm": 0.8240206241607666,
+      "learning_rate": 8.955836775506776e-05,
+      "loss": 0.10144685506820679,
+      "mean_token_accuracy": 0.9598779857158661,
+      "num_tokens": 1178884.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1517789483070373,
+      "epoch": 0.08757157292017514,
+      "grad_norm": 0.880024790763855,
+      "learning_rate": 8.852566213878947e-05,
+      "loss": 0.10237239599227906,
+      "mean_token_accuracy": 0.9605226457118988,
+      "num_tokens": 1225970.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1327537655830384,
+      "epoch": 0.09093971034018188,
+      "grad_norm": 0.7027714848518372,
+      "learning_rate": 8.745082927659047e-05,
+      "loss": 0.09415384531021118,
+      "mean_token_accuracy": 0.9629074990749359,
+      "num_tokens": 1273203.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1431329727172852,
+      "epoch": 0.09430784776018862,
+      "grad_norm": 0.8299992680549622,
+      "learning_rate": 8.633504448242505e-05,
+      "loss": 0.10715677738189697,
+      "mean_token_accuracy": 0.957152646780014,
+      "num_tokens": 1320258.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1475672125816345,
+      "epoch": 0.09767598518019535,
+      "grad_norm": 0.9531705975532532,
+      "learning_rate": 8.517952785058385e-05,
+      "loss": 0.1055680513381958,
+      "mean_token_accuracy": 0.9592137575149536,
+      "num_tokens": 1367569.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1425477266311646,
+      "epoch": 0.10104412260020208,
+      "grad_norm": 0.9806235432624817,
+      "learning_rate": 8.398554292153866e-05,
+      "loss": 0.10172542333602905,
+      "mean_token_accuracy": 0.9607040584087372,
+      "num_tokens": 1414862.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1395321488380432,
+      "epoch": 0.10441226002020883,
+      "grad_norm": 0.9198063611984253,
+      "learning_rate": 8.275439530027948e-05,
+      "loss": 0.09770519733428955,
+      "mean_token_accuracy": 0.9610416889190674,
+      "num_tokens": 1462003.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1482908606529236,
+      "epoch": 0.10778039744021556,
+      "grad_norm": 0.6851537227630615,
+      "learning_rate": 8.148743122865463e-05,
+      "loss": 0.09348824620246887,
+      "mean_token_accuracy": 0.9661192774772644,
+      "num_tokens": 1509209.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1355633854866027,
+      "epoch": 0.1111485348602223,
+      "grad_norm": 0.7888758778572083,
+      "learning_rate": 8.018603611327504e-05,
+      "loss": 0.10011379718780518,
+      "mean_token_accuracy": 0.9596275985240936,
+      "num_tokens": 1556618.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.12863130569458,
+      "epoch": 0.11451667228022903,
+      "grad_norm": 0.9677507281303406,
+      "learning_rate": 7.88516330105925e-05,
+      "loss": 0.09362500309944152,
+      "mean_token_accuracy": 0.9632209837436676,
+      "num_tokens": 1603706.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1242616653442383,
+      "epoch": 0.11788480970023577,
+      "grad_norm": 0.8519465923309326,
+      "learning_rate": 7.748568107080832e-05,
+      "loss": 0.09718523621559143,
+      "mean_token_accuracy": 0.9630285918712616,
+      "num_tokens": 1650886.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.121305477619171,
+      "epoch": 0.12125294712024251,
+      "grad_norm": 0.7342331409454346,
+      "learning_rate": 7.608967394231387e-05,
+      "loss": 0.0937388300895691,
+      "mean_token_accuracy": 0.9647803843021393,
+      "num_tokens": 1698139.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1216169714927673,
+      "epoch": 0.12462108454024924,
+      "grad_norm": 0.7954909205436707,
+      "learning_rate": 7.466513813840825e-05,
+      "loss": 0.09746226072311401,
+      "mean_token_accuracy": 0.9619386792182922,
+      "num_tokens": 1745359.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1140558838844299,
+      "epoch": 0.12798922196025597,
+      "grad_norm": 0.8080150485038757,
+      "learning_rate": 7.32136313680782e-05,
+      "loss": 0.09294270277023316,
+      "mean_token_accuracy": 0.9636349976062775,
+      "num_tokens": 1792614.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1088845491409303,
+      "epoch": 0.13135735938026272,
+      "grad_norm": 0.7024004459381104,
+      "learning_rate": 7.173674083266624e-05,
+      "loss": 0.08779981136322021,
+      "mean_token_accuracy": 0.9663202583789825,
+      "num_tokens": 1839466.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.1134823083877563,
+      "epoch": 0.13472549680026946,
+      "grad_norm": 0.7651565670967102,
+      "learning_rate": 7.023608149028937e-05,
+      "loss": 0.08661429286003113,
+      "mean_token_accuracy": 0.9656910836696625,
+      "num_tokens": 1886384.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.1094910860061646,
+      "epoch": 0.13809363422027618,
+      "grad_norm": 0.9750885963439941,
+      "learning_rate": 6.871329428990602e-05,
+      "loss": 0.09752371311187744,
+      "mean_token_accuracy": 0.9609347939491272,
+      "num_tokens": 1933515.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.1039424657821655,
+      "epoch": 0.14146177164028292,
+      "grad_norm": 0.8973978161811829,
+      "learning_rate": 6.71700443769625e-05,
+      "loss": 0.08590971231460572,
+      "mean_token_accuracy": 0.9667288780212402,
+      "num_tokens": 1980697.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0823938965797424,
+      "epoch": 0.14482990906028967,
+      "grad_norm": 1.0504096746444702,
+      "learning_rate": 6.56080192725808e-05,
+      "loss": 0.07881518602371215,
+      "mean_token_accuracy": 0.9681541204452515,
+      "num_tokens": 2027873.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0918182253837585,
+      "epoch": 0.14819804648029639,
+      "grad_norm": 1.0159872770309448,
+      "learning_rate": 6.402892702827916e-05,
+      "loss": 0.08618102669715881,
+      "mean_token_accuracy": 0.9653535485267639,
+      "num_tokens": 2074919.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0920875072479248,
+      "epoch": 0.15156618390030313,
+      "grad_norm": 0.9736694097518921,
+      "learning_rate": 6.243449435824276e-05,
+      "loss": 0.07853227853775024,
+      "mean_token_accuracy": 0.9698511242866517,
+      "num_tokens": 2122140.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0967504978179932,
+      "epoch": 0.15493432132030988,
+      "grad_norm": 0.6489920020103455,
+      "learning_rate": 6.0826464751186994e-05,
+      "loss": 0.08308060765266419,
+      "mean_token_accuracy": 0.9663989305496216,
+      "num_tokens": 2169288.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.1014941215515137,
+      "epoch": 0.1583024587403166,
+      "grad_norm": 0.8790886998176575,
+      "learning_rate": 5.9206596563878357e-05,
+      "loss": 0.08445734381675721,
+      "mean_token_accuracy": 0.9671892881393432,
+      "num_tokens": 2216488.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.098174238204956,
+      "epoch": 0.16167059616032334,
+      "grad_norm": 1.0411581993103027,
+      "learning_rate": 5.757666109839702e-05,
+      "loss": 0.08071005940437317,
+      "mean_token_accuracy": 0.9688655078411103,
+      "num_tokens": 2263116.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0968878507614135,
+      "epoch": 0.16503873358033008,
+      "grad_norm": 1.0246140956878662,
+      "learning_rate": 5.5938440665244006e-05,
+      "loss": 0.08120047450065612,
+      "mean_token_accuracy": 0.9677295029163361,
+      "num_tokens": 2309861.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.072183656692505,
+      "epoch": 0.16840687100033683,
+      "grad_norm": 1.029000997543335,
+      "learning_rate": 5.4293726634410855e-05,
+      "loss": 0.0786105751991272,
+      "mean_token_accuracy": 0.9702598631381989,
+      "num_tokens": 2357145.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.0788764953613281,
+      "epoch": 0.17177500842034354,
+      "grad_norm": 1.0718475580215454,
+      "learning_rate": 5.264431747654284e-05,
+      "loss": 0.08527698516845703,
+      "mean_token_accuracy": 0.9666848182678223,
+      "num_tokens": 2404158.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0761215686798096,
+      "epoch": 0.1751431458403503,
+      "grad_norm": 0.866406261920929,
+      "learning_rate": 5.0992016796337686e-05,
+      "loss": 0.07828302383422851,
+      "mean_token_accuracy": 0.9694028079509736,
+      "num_tokens": 2451273.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0755280017852784,
+      "epoch": 0.17851128326035703,
+      "grad_norm": 1.05809485912323,
+      "learning_rate": 4.93386313603304e-05,
+      "loss": 0.08911029100418091,
+      "mean_token_accuracy": 0.9648769974708558,
+      "num_tokens": 2498367.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0763638496398926,
+      "epoch": 0.18187942068036375,
+      "grad_norm": 1.455805778503418,
+      "learning_rate": 4.7685969121220456e-05,
+      "loss": 0.08377580642700196,
+      "mean_token_accuracy": 0.9677519798278809,
+      "num_tokens": 2545761.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0753028750419618,
+      "epoch": 0.1852475581003705,
+      "grad_norm": 0.9411235451698303,
+      "learning_rate": 4.60358372409022e-05,
+      "loss": 0.07909951210021973,
+      "mean_token_accuracy": 0.9687525451183319,
+      "num_tokens": 2593174.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0739057660102844,
+      "epoch": 0.18861569552037724,
+      "grad_norm": 1.0146852731704712,
+      "learning_rate": 4.439004011435979e-05,
+      "loss": 0.07501991987228393,
+      "mean_token_accuracy": 0.9709393978118896,
+      "num_tokens": 2640657.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.0723747730255127,
+      "epoch": 0.19198383294038396,
+      "grad_norm": 1.1873159408569336,
+      "learning_rate": 4.275037739658771e-05,
+      "loss": 0.07739782333374023,
+      "mean_token_accuracy": 0.9696486115455627,
+      "num_tokens": 2687665.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.0660502552986144,
+      "epoch": 0.1953519703603907,
+      "grad_norm": 0.9216545820236206,
+      "learning_rate": 4.111864203469457e-05,
+      "loss": 0.07156487107276917,
+      "mean_token_accuracy": 0.9722769498825073,
+      "num_tokens": 2735069.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0652514219284057,
+      "epoch": 0.19872010778039745,
+      "grad_norm": 0.8699415922164917,
+      "learning_rate": 3.949661830734172e-05,
+      "loss": 0.07855596542358398,
+      "mean_token_accuracy": 0.9687506437301636,
+      "num_tokens": 2782136.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0566825151443482,
+      "epoch": 0.20208824520040417,
+      "grad_norm": 1.171101689338684,
+      "learning_rate": 3.788607987366069e-05,
+      "loss": 0.07919362783432007,
+      "mean_token_accuracy": 0.968692398071289,
+      "num_tokens": 2829351.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0607569456100463,
+      "epoch": 0.2054563826204109,
+      "grad_norm": 0.8927919268608093,
+      "learning_rate": 3.628878783378302e-05,
+      "loss": 0.07113456726074219,
+      "mean_token_accuracy": 0.9711033463478088,
+      "num_tokens": 2876504.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0631901383399964,
+      "epoch": 0.20882452004041765,
+      "grad_norm": 1.0045161247253418,
+      "learning_rate": 3.470648880310313e-05,
+      "loss": 0.08255671262741089,
+      "mean_token_accuracy": 0.9675521731376648,
+      "num_tokens": 2923546.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0465935587882995,
+      "epoch": 0.21219265746042437,
+      "grad_norm": 0.92397141456604,
+      "learning_rate": 3.3140913002379995e-05,
+      "loss": 0.0756759762763977,
+      "mean_token_accuracy": 0.9690936505794525,
+      "num_tokens": 2970921.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0486064195632934,
+      "epoch": 0.21556079488043112,
+      "grad_norm": 1.052390694618225,
+      "learning_rate": 3.1593772365766105e-05,
+      "loss": 0.0781489908695221,
+      "mean_token_accuracy": 0.9700931966304779,
+      "num_tokens": 3018290.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0550805807113648,
+      "epoch": 0.21892893230043786,
+      "grad_norm": 0.8665848970413208,
+      "learning_rate": 3.006675866883275e-05,
+      "loss": 0.07983592748641968,
+      "mean_token_accuracy": 0.9672332823276519,
+      "num_tokens": 3065320.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.050302755832672,
+      "epoch": 0.2222970697204446,
+      "grad_norm": 0.9459096193313599,
+      "learning_rate": 2.8561541678638142e-05,
+      "loss": 0.07495864629745483,
+      "mean_token_accuracy": 0.9699085712432861,
+      "num_tokens": 3112540.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0496302127838135,
+      "epoch": 0.22566520714045132,
+      "grad_norm": 1.2742879390716553,
+      "learning_rate": 2.707976732786166e-05,
+      "loss": 0.07737075090408325,
+      "mean_token_accuracy": 0.9705961644649506,
+      "num_tokens": 3159451.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.042512595653534,
+      "epoch": 0.22903334456045807,
+      "grad_norm": 0.9438436627388,
+      "learning_rate": 2.562305591500069e-05,
+      "loss": 0.06843398213386535,
+      "mean_token_accuracy": 0.9729993402957916,
+      "num_tokens": 3206538.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.0402977466583252,
+      "epoch": 0.2324014819804648,
+      "grad_norm": 0.8952704071998596,
+      "learning_rate": 2.419300033259798e-05,
+      "loss": 0.07132667303085327,
+      "mean_token_accuracy": 0.9718169450759888,
+      "num_tokens": 3253416.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0333297610282899,
+      "epoch": 0.23576961940047153,
+      "grad_norm": 1.0013381242752075,
+      "learning_rate": 2.279116432543705e-05,
+      "loss": 0.07136809825897217,
+      "mean_token_accuracy": 0.9713504314422607,
+      "num_tokens": 3300695.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0264186024665833,
+      "epoch": 0.23913775682047828,
+      "grad_norm": 0.7419242262840271,
+      "learning_rate": 2.1419080780610123e-05,
+      "loss": 0.06598395705223084,
+      "mean_token_accuracy": 0.9739853262901306,
+      "num_tokens": 3348238.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.030605685710907,
+      "epoch": 0.24250589424048502,
+      "grad_norm": 0.9724133014678955,
+      "learning_rate": 2.0078250051328784e-05,
+      "loss": 0.07152948379516602,
+      "mean_token_accuracy": 0.9715062916278839,
+      "num_tokens": 3395374.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.030699408054352,
+      "epoch": 0.24587403166049174,
+      "grad_norm": 1.13027822971344,
+      "learning_rate": 1.877013831630961e-05,
+      "loss": 0.07069330811500549,
+      "mean_token_accuracy": 0.9697365164756775,
+      "num_tokens": 3442357.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0237806797027589,
+      "epoch": 0.24924216908049848,
+      "grad_norm": 1.0425814390182495,
+      "learning_rate": 1.749617597652934e-05,
+      "loss": 0.06740343570709229,
+      "mean_token_accuracy": 0.9726531982421875,
+      "num_tokens": 3489588.0,
+      "step": 740
+    },
+    {
+      "entropy": 1.0186407327651978,
+      "epoch": 0.2526103065005052,
+      "grad_norm": 1.251236915588379,
+      "learning_rate": 1.62577560911024e-05,
+      "loss": 0.06498036980628967,
+      "mean_token_accuracy": 0.9735432982444763,
+      "num_tokens": 3536949.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0222207069396974,
+      "epoch": 0.25597844392051194,
+      "grad_norm": 1.0947421789169312,
+      "learning_rate": 1.5056232853991209e-05,
+      "loss": 0.07226579785346984,
+      "mean_token_accuracy": 0.9712574124336243,
+      "num_tokens": 3584273.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0290661573410034,
+      "epoch": 0.2593465813405187,
+      "grad_norm": 1.2326691150665283,
+      "learning_rate": 1.389292011321498e-05,
+      "loss": 0.06502938866615296,
+      "mean_token_accuracy": 0.974638819694519,
+      "num_tokens": 3631458.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.027204728126526,
+      "epoch": 0.26271471876052543,
+      "grad_norm": 1.0657676458358765,
+      "learning_rate": 1.2769089934176126e-05,
+      "loss": 0.06863797903060913,
+      "mean_token_accuracy": 0.9726690888404846,
+      "num_tokens": 3678657.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0222702980041505,
+      "epoch": 0.2660828561805322,
+      "grad_norm": 1.1804763078689575,
+      "learning_rate": 1.1685971208675539e-05,
+      "loss": 0.061578494310379026,
+      "mean_token_accuracy": 0.9742346107959747,
+      "num_tokens": 3725958.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0322864174842834,
+      "epoch": 0.2694509936005389,
+      "grad_norm": 0.904740571975708,
+      "learning_rate": 1.0644748311137376e-05,
+      "loss": 0.06918627619743348,
+      "mean_token_accuracy": 0.9727381587028503,
+      "num_tokens": 3772857.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.0297256469726563,
+      "epoch": 0.2728191310205456,
+      "grad_norm": 1.1179808378219604,
+      "learning_rate": 9.646559803512994e-06,
+      "loss": 0.06206071376800537,
+      "mean_token_accuracy": 0.9756330192089081,
+      "num_tokens": 3819801.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0256371259689332,
+      "epoch": 0.27618726844055236,
+      "grad_norm": 1.2047091722488403,
+      "learning_rate": 8.692497190280224e-06,
+      "loss": 0.06384825110435485,
+      "mean_token_accuracy": 0.9744520127773285,
+      "num_tokens": 3866792.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.0237814784049988,
+      "epoch": 0.2795554058605591,
+      "grad_norm": 0.8565121293067932,
+      "learning_rate": 7.783603724899257e-06,
+      "loss": 0.06574625968933105,
+      "mean_token_accuracy": 0.9747447311878205,
+      "num_tokens": 3913778.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.0248419165611267,
+      "epoch": 0.28292354328056585,
+      "grad_norm": 1.0262619256973267,
+      "learning_rate": 6.92087326903022e-06,
+      "loss": 0.0729720950126648,
+      "mean_token_accuracy": 0.9695690929889679,
+      "num_tokens": 3960877.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0249651312828063,
+      "epoch": 0.2862916807005726,
+      "grad_norm": 1.1824581623077393,
+      "learning_rate": 6.1052492057601275e-06,
+      "loss": 0.065399432182312,
+      "mean_token_accuracy": 0.9752880871295929,
+      "num_tokens": 4007848.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0181277751922608,
+      "epoch": 0.28965981812057934,
+      "grad_norm": 0.9087852835655212,
+      "learning_rate": 5.337623408027293e-06,
+      "loss": 0.058376860618591306,
+      "mean_token_accuracy": 0.9767619907855988,
+      "num_tokens": 4054893.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.017453020811081,
+      "epoch": 0.2930279555405861,
+      "grad_norm": 0.8468597531318665,
+      "learning_rate": 4.618835263371396e-06,
+      "loss": 0.06070579886436463,
+      "mean_token_accuracy": 0.9755488932132721,
+      "num_tokens": 4102219.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.0192234635353088,
+      "epoch": 0.29639609296059277,
+      "grad_norm": 1.1834481954574585,
+      "learning_rate": 3.949670756075447e-06,
+      "loss": 0.07054463624954224,
+      "mean_token_accuracy": 0.9705739676952362,
+      "num_tokens": 4149444.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0142346441745758,
+      "epoch": 0.2997642303805995,
+      "grad_norm": 0.8702611923217773,
+      "learning_rate": 3.3308616077036115e-06,
+      "loss": 0.06450958251953125,
+      "mean_token_accuracy": 0.9724807500839233,
+      "num_tokens": 4196532.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.016416597366333,
+      "epoch": 0.30313236780060626,
+      "grad_norm": 0.7494244575500488,
+      "learning_rate": 2.7630844769743757e-06,
+      "loss": 0.05437774658203125,
+      "mean_token_accuracy": 0.9789692103862763,
+      "num_tokens": 4243665.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0165138125419617,
+      "epoch": 0.306500505220613,
+      "grad_norm": 1.056312918663025,
+      "learning_rate": 2.2469602198441573e-06,
+      "loss": 0.06402235627174377,
+      "mean_token_accuracy": 0.9736861884593964,
+      "num_tokens": 4290852.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.0162912726402282,
+      "epoch": 0.30986864264061975,
+      "grad_norm": 0.934772789478302,
+      "learning_rate": 1.7830532106104747e-06,
+      "loss": 0.06432567834854126,
+      "mean_token_accuracy": 0.9742455959320069,
+      "num_tokens": 4337989.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.011330533027649,
+      "epoch": 0.3132367800606265,
+      "grad_norm": 0.9233838319778442,
+      "learning_rate": 1.3718707247769135e-06,
+      "loss": 0.06175405979156494,
+      "mean_token_accuracy": 0.9768862724304199,
+      "num_tokens": 4385404.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0080062210559846,
+      "epoch": 0.3166049174806332,
+      "grad_norm": 1.1740663051605225,
+      "learning_rate": 1.0138623843548078e-06,
+      "loss": 0.05886925458908081,
+      "mean_token_accuracy": 0.975130420923233,
+      "num_tokens": 4432957.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0180271983146667,
+      "epoch": 0.31997305490063993,
+      "grad_norm": 0.9107274413108826,
+      "learning_rate": 7.094196662081831e-07,
+      "loss": 0.05679928064346314,
+      "mean_token_accuracy": 0.9778966784477234,
+      "num_tokens": 4479735.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0070263922214509,
+      "epoch": 0.3233411923206467,
+      "grad_norm": 0.8798539638519287,
+      "learning_rate": 4.5887547397955864e-07,
+      "loss": 0.06429069042205811,
+      "mean_token_accuracy": 0.9755585432052613,
+      "num_tokens": 4527256.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0133942067623138,
+      "epoch": 0.3267093297406534,
+      "grad_norm": 1.0966157913208008,
+      "learning_rate": 2.625037740646763e-07,
+      "loss": 0.07166630029678345,
+      "mean_token_accuracy": 0.9717779278755188,
+      "num_tokens": 4574578.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0112496197223664,
+      "epoch": 0.33007746716066017,
+      "grad_norm": 1.2206507921218872,
+      "learning_rate": 1.2051929603428825e-07,
+      "loss": 0.05982668399810791,
+      "mean_token_accuracy": 0.9757863104343414,
+      "num_tokens": 4621961.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0146852493286134,
+      "epoch": 0.3334456045806669,
+      "grad_norm": 0.994526207447052,
+      "learning_rate": 3.3077297830541584e-08,
+      "loss": 0.05897784233093262,
+      "mean_token_accuracy": 0.9762094497680665,
+      "num_tokens": 4668910.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.017277216911316,
+      "epoch": 0.33681374200067365,
+      "grad_norm": 0.9273004531860352,
+      "learning_rate": 2.7339599464326627e-10,
+      "loss": 0.06890772581100464,
+      "mean_token_accuracy": 0.9719313144683838,
+      "num_tokens": 4715878.0,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1991045854553702e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/training_args.bin b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8fec1354c64292c8e27becda8595a5aa647aa71c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-1000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc01187e7f5b0296affbb4d368cbd2f85c3b102cf884d95f9d29965cd5276e76
+size 5841
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/README.md b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/adapter_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3db6c6926ce4988b431e6c91620dabbc075f719c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "down_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/adapter_model.safetensors b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..cf3aac8fc79be2b41974060cb4699b366a46e489
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1371ee755961dc2b72786b852762234dcfc9d5ae232dd991cff3265adcb245e0
+size 80792096
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/chat_template.jinja b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/tokenizer.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/tokenizer_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/trainer_state.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e93215f2a49d714a004dd180b99a9002f26e0962
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/trainer_state.json
@@ -0,0 +1,84 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.016840687100033683,
+  "eval_steps": 500,
+  "global_step": 50,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2336307406425475,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 3.205052614212036,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5239480972290039,
+      "mean_token_accuracy": 0.8582192718982696,
+      "num_tokens": 47369.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2293899655342102,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 1.4208030700683594,
+      "learning_rate": 3.8e-05,
+      "loss": 0.295538067817688,
+      "mean_token_accuracy": 0.9012860953807831,
+      "num_tokens": 94622.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.238785707950592,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 0.6221668720245361,
+      "learning_rate": 5.8e-05,
+      "loss": 0.19195973873138428,
+      "mean_token_accuracy": 0.9294111549854278,
+      "num_tokens": 141694.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2240872263908387,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 0.6423875689506531,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.16067487001419067,
+      "mean_token_accuracy": 0.9407785713672638,
+      "num_tokens": 188811.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2138957858085633,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 0.7921156287193298,
+      "learning_rate": 9.8e-05,
+      "loss": 0.14190430641174318,
+      "mean_token_accuracy": 0.9457307100296021,
+      "num_tokens": 236035.0,
+      "step": 50
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0973501787734016e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/training_args.bin b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8fec1354c64292c8e27becda8595a5aa647aa71c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-50/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc01187e7f5b0296affbb4d368cbd2f85c3b102cf884d95f9d29965cd5276e76
+size 5841
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/README.md b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/adapter_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3db6c6926ce4988b431e6c91620dabbc075f719c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "down_proj",
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/adapter_model.safetensors b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7fb577cc3b381e24c6762d74b602baa9101cdb51
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6aed256d76a88d593599ae8fb45b703524084262582134d9aa78bdce28a5eac1
+size 80792096
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/chat_template.jinja b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/tokenizer.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/tokenizer_config.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/trainer_state.json b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..d5f50e227ac2ec6ffd82b3780ec88a7c701f7fc8
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/trainer_state.json
@@ -0,0 +1,534 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.16840687100033683,
+  "eval_steps": 500,
+  "global_step": 500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.2336307406425475,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 3.205052614212036,
+      "learning_rate": 1.8e-05,
+      "loss": 0.5239480972290039,
+      "mean_token_accuracy": 0.8582192718982696,
+      "num_tokens": 47369.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2293899655342102,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 1.4208030700683594,
+      "learning_rate": 3.8e-05,
+      "loss": 0.295538067817688,
+      "mean_token_accuracy": 0.9012860953807831,
+      "num_tokens": 94622.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.238785707950592,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 0.6221668720245361,
+      "learning_rate": 5.8e-05,
+      "loss": 0.19195973873138428,
+      "mean_token_accuracy": 0.9294111549854278,
+      "num_tokens": 141694.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2240872263908387,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 0.6423875689506531,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.16067487001419067,
+      "mean_token_accuracy": 0.9407785713672638,
+      "num_tokens": 188811.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2138957858085633,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 0.7921156287193298,
+      "learning_rate": 9.8e-05,
+      "loss": 0.14190430641174318,
+      "mean_token_accuracy": 0.9457307100296021,
+      "num_tokens": 236035.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2117965459823608,
+      "epoch": 0.020208824520040417,
+      "grad_norm": 0.5566120147705078,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.13536434173583983,
+      "mean_token_accuracy": 0.9480699181556702,
+      "num_tokens": 283008.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2039262533187867,
+      "epoch": 0.023576961940047155,
+      "grad_norm": 0.7983958721160889,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.1264912724494934,
+      "mean_token_accuracy": 0.9510111153125763,
+      "num_tokens": 330291.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2005774617195129,
+      "epoch": 0.02694509936005389,
+      "grad_norm": 0.6949729919433594,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.10797474384307862,
+      "mean_token_accuracy": 0.9581791937351227,
+      "num_tokens": 377729.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.1869928240776062,
+      "epoch": 0.030313236780060628,
+      "grad_norm": 0.6609069108963013,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.11980686187744141,
+      "mean_token_accuracy": 0.9547467947006225,
+      "num_tokens": 424960.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.1932473540306092,
+      "epoch": 0.033681374200067365,
+      "grad_norm": 0.5904498100280762,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.12140004634857178,
+      "mean_token_accuracy": 0.952862948179245,
+      "num_tokens": 472087.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.1891833305358888,
+      "epoch": 0.037049511620074096,
+      "grad_norm": 0.8453942537307739,
+      "learning_rate": 9.905132290792394e-05,
+      "loss": 0.12110424041748047,
+      "mean_token_accuracy": 0.954711276292801,
+      "num_tokens": 519101.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.192702329158783,
+      "epoch": 0.040417649040080834,
+      "grad_norm": 0.9638974666595459,
+      "learning_rate": 9.870399824239117e-05,
+      "loss": 0.12223016023635865,
+      "mean_token_accuracy": 0.9538575768470764,
+      "num_tokens": 566562.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1896525263786315,
+      "epoch": 0.04378578646008757,
+      "grad_norm": 0.8424277901649475,
+      "learning_rate": 9.830341646984521e-05,
+      "loss": 0.12164583206176757,
+      "mean_token_accuracy": 0.9546079397201538,
+      "num_tokens": 613861.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.191524910926819,
+      "epoch": 0.04715392388009431,
+      "grad_norm": 0.7796409726142883,
+      "learning_rate": 9.785001562057309e-05,
+      "loss": 0.12058537006378174,
+      "mean_token_accuracy": 0.9509174525737762,
+      "num_tokens": 660887.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1798339486122131,
+      "epoch": 0.05052206130010104,
+      "grad_norm": 1.203282356262207,
+      "learning_rate": 9.734429148174675e-05,
+      "loss": 0.11496694087982177,
+      "mean_token_accuracy": 0.9555849313735962,
+      "num_tokens": 707771.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.1805959939956665,
+      "epoch": 0.05389019872010778,
+      "grad_norm": 0.7123157978057861,
+      "learning_rate": 9.6786797055287e-05,
+      "loss": 0.11297221183776855,
+      "mean_token_accuracy": 0.9562048316001892,
+      "num_tokens": 755026.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1713955640792846,
+      "epoch": 0.05725833614011452,
+      "grad_norm": 0.869977593421936,
+      "learning_rate": 9.617814195316411e-05,
+      "loss": 0.11606066226959229,
+      "mean_token_accuracy": 0.9540329277515411,
+      "num_tokens": 802572.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.178900408744812,
+      "epoch": 0.060626473560121255,
+      "grad_norm": 0.8154799938201904,
+      "learning_rate": 9.551899173079607e-05,
+      "loss": 0.10659003257751465,
+      "mean_token_accuracy": 0.959269380569458,
+      "num_tokens": 849536.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1731992602348327,
+      "epoch": 0.06399461098012799,
+      "grad_norm": 0.8263419270515442,
+      "learning_rate": 9.481006715927351e-05,
+      "loss": 0.10587993860244752,
+      "mean_token_accuracy": 0.9589337944984436,
+      "num_tokens": 896651.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1755276799201966,
+      "epoch": 0.06736274840013473,
+      "grad_norm": 0.9151034355163574,
+      "learning_rate": 9.405214343720707e-05,
+      "loss": 0.11514871120452881,
+      "mean_token_accuracy": 0.9544816732406616,
+      "num_tokens": 943494.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.168901538848877,
+      "epoch": 0.07073088582014146,
+      "grad_norm": 0.7708814144134521,
+      "learning_rate": 9.32460493430591e-05,
+      "loss": 0.10262255668640137,
+      "mean_token_accuracy": 0.9597939789295197,
+      "num_tokens": 990612.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.1802135467529298,
+      "epoch": 0.07409902324014819,
+      "grad_norm": 0.7555872201919556,
+      "learning_rate": 9.239266632888659e-05,
+      "loss": 0.10907113552093506,
+      "mean_token_accuracy": 0.9556582808494568,
+      "num_tokens": 1037718.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1758136510849,
+      "epoch": 0.07746716066015494,
+      "grad_norm": 0.7670081853866577,
+      "learning_rate": 9.14929275564863e-05,
+      "loss": 0.10124562978744507,
+      "mean_token_accuracy": 0.9586797237396241,
+      "num_tokens": 1084625.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1657557368278504,
+      "epoch": 0.08083529808016167,
+      "grad_norm": 0.8729236125946045,
+      "learning_rate": 9.0547816876996e-05,
+      "loss": 0.10474343299865722,
+      "mean_token_accuracy": 0.957787013053894,
+      "num_tokens": 1131731.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1566669702529908,
+      "epoch": 0.08420343550016841,
+      "grad_norm": 0.8240206241607666,
+      "learning_rate": 8.955836775506776e-05,
+      "loss": 0.10144685506820679,
+      "mean_token_accuracy": 0.9598779857158661,
+      "num_tokens": 1178884.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1517789483070373,
+      "epoch": 0.08757157292017514,
+      "grad_norm": 0.880024790763855,
+      "learning_rate": 8.852566213878947e-05,
+      "loss": 0.10237239599227906,
+      "mean_token_accuracy": 0.9605226457118988,
+      "num_tokens": 1225970.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1327537655830384,
+      "epoch": 0.09093971034018188,
+      "grad_norm": 0.7027714848518372,
+      "learning_rate": 8.745082927659047e-05,
+      "loss": 0.09415384531021118,
+      "mean_token_accuracy": 0.9629074990749359,
+      "num_tokens": 1273203.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1431329727172852,
+      "epoch": 0.09430784776018862,
+      "grad_norm": 0.8299992680549622,
+      "learning_rate": 8.633504448242505e-05,
+      "loss": 0.10715677738189697,
+      "mean_token_accuracy": 0.957152646780014,
+      "num_tokens": 1320258.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1475672125816345,
+      "epoch": 0.09767598518019535,
+      "grad_norm": 0.9531705975532532,
+      "learning_rate": 8.517952785058385e-05,
+      "loss": 0.1055680513381958,
+      "mean_token_accuracy": 0.9592137575149536,
+      "num_tokens": 1367569.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1425477266311646,
+      "epoch": 0.10104412260020208,
+      "grad_norm": 0.9806235432624817,
+      "learning_rate": 8.398554292153866e-05,
+      "loss": 0.10172542333602905,
+      "mean_token_accuracy": 0.9607040584087372,
+      "num_tokens": 1414862.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1395321488380432,
+      "epoch": 0.10441226002020883,
+      "grad_norm": 0.9198063611984253,
+      "learning_rate": 8.275439530027948e-05,
+      "loss": 0.09770519733428955,
+      "mean_token_accuracy": 0.9610416889190674,
+      "num_tokens": 1462003.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1482908606529236,
+      "epoch": 0.10778039744021556,
+      "grad_norm": 0.6851537227630615,
+      "learning_rate": 8.148743122865463e-05,
+      "loss": 0.09348824620246887,
+      "mean_token_accuracy": 0.9661192774772644,
+      "num_tokens": 1509209.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1355633854866027,
+      "epoch": 0.1111485348602223,
+      "grad_norm": 0.7888758778572083,
+      "learning_rate": 8.018603611327504e-05,
+      "loss": 0.10011379718780518,
+      "mean_token_accuracy": 0.9596275985240936,
+      "num_tokens": 1556618.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.12863130569458,
+      "epoch": 0.11451667228022903,
+      "grad_norm": 0.9677507281303406,
+      "learning_rate": 7.88516330105925e-05,
+      "loss": 0.09362500309944152,
+      "mean_token_accuracy": 0.9632209837436676,
+      "num_tokens": 1603706.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1242616653442383,
+      "epoch": 0.11788480970023577,
+      "grad_norm": 0.8519465923309326,
+      "learning_rate": 7.748568107080832e-05,
+      "loss": 0.09718523621559143,
+      "mean_token_accuracy": 0.9630285918712616,
+      "num_tokens": 1650886.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.121305477619171,
+      "epoch": 0.12125294712024251,
+      "grad_norm": 0.7342331409454346,
+      "learning_rate": 7.608967394231387e-05,
+      "loss": 0.0937388300895691,
+      "mean_token_accuracy": 0.9647803843021393,
+      "num_tokens": 1698139.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1216169714927673,
+      "epoch": 0.12462108454024924,
+      "grad_norm": 0.7954909205436707,
+      "learning_rate": 7.466513813840825e-05,
+      "loss": 0.09746226072311401,
+      "mean_token_accuracy": 0.9619386792182922,
+      "num_tokens": 1745359.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1140558838844299,
+      "epoch": 0.12798922196025597,
+      "grad_norm": 0.8080150485038757,
+      "learning_rate": 7.32136313680782e-05,
+      "loss": 0.09294270277023316,
+      "mean_token_accuracy": 0.9636349976062775,
+      "num_tokens": 1792614.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1088845491409303,
+      "epoch": 0.13135735938026272,
+      "grad_norm": 0.7024004459381104,
+      "learning_rate": 7.173674083266624e-05,
+      "loss": 0.08779981136322021,
+      "mean_token_accuracy": 0.9663202583789825,
+      "num_tokens": 1839466.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.1134823083877563,
+      "epoch": 0.13472549680026946,
+      "grad_norm": 0.7651565670967102,
+      "learning_rate": 7.023608149028937e-05,
+      "loss": 0.08661429286003113,
+      "mean_token_accuracy": 0.9656910836696625,
+      "num_tokens": 1886384.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.1094910860061646,
+      "epoch": 0.13809363422027618,
+      "grad_norm": 0.9750885963439941,
+      "learning_rate": 6.871329428990602e-05,
+      "loss": 0.09752371311187744,
+      "mean_token_accuracy": 0.9609347939491272,
+      "num_tokens": 1933515.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.1039424657821655,
+      "epoch": 0.14146177164028292,
+      "grad_norm": 0.8973978161811829,
+      "learning_rate": 6.71700443769625e-05,
+      "loss": 0.08590971231460572,
+      "mean_token_accuracy": 0.9667288780212402,
+      "num_tokens": 1980697.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0823938965797424,
+      "epoch": 0.14482990906028967,
+      "grad_norm": 1.0504096746444702,
+      "learning_rate": 6.56080192725808e-05,
+      "loss": 0.07881518602371215,
+      "mean_token_accuracy": 0.9681541204452515,
+      "num_tokens": 2027873.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.0918182253837585,
+      "epoch": 0.14819804648029639,
+      "grad_norm": 1.0159872770309448,
+      "learning_rate": 6.402892702827916e-05,
+      "loss": 0.08618102669715881,
+      "mean_token_accuracy": 0.9653535485267639,
+      "num_tokens": 2074919.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0920875072479248,
+      "epoch": 0.15156618390030313,
+      "grad_norm": 0.9736694097518921,
+      "learning_rate": 6.243449435824276e-05,
+      "loss": 0.07853227853775024,
+      "mean_token_accuracy": 0.9698511242866517,
+      "num_tokens": 2122140.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.0967504978179932,
+      "epoch": 0.15493432132030988,
+      "grad_norm": 0.6489920020103455,
+      "learning_rate": 6.0826464751186994e-05,
+      "loss": 0.08308060765266419,
+      "mean_token_accuracy": 0.9663989305496216,
+      "num_tokens": 2169288.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.1014941215515137,
+      "epoch": 0.1583024587403166,
+      "grad_norm": 0.8790886998176575,
+      "learning_rate": 5.9206596563878357e-05,
+      "loss": 0.08445734381675721,
+      "mean_token_accuracy": 0.9671892881393432,
+      "num_tokens": 2216488.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.098174238204956,
+      "epoch": 0.16167059616032334,
+      "grad_norm": 1.0411581993103027,
+      "learning_rate": 5.757666109839702e-05,
+      "loss": 0.08071005940437317,
+      "mean_token_accuracy": 0.9688655078411103,
+      "num_tokens": 2263116.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0968878507614135,
+      "epoch": 0.16503873358033008,
+      "grad_norm": 1.0246140956878662,
+      "learning_rate": 5.5938440665244006e-05,
+      "loss": 0.08120047450065612,
+      "mean_token_accuracy": 0.9677295029163361,
+      "num_tokens": 2309861.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.072183656692505,
+      "epoch": 0.16840687100033683,
+      "grad_norm": 1.029000997543335,
+      "learning_rate": 5.4293726634410855e-05,
+      "loss": 0.0786105751991272,
+      "mean_token_accuracy": 0.9702598631381989,
+      "num_tokens": 2357145.0,
+      "step": 500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0988273287561216e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/training_args.bin b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8fec1354c64292c8e27becda8595a5aa647aa71c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/checkpoint-500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc01187e7f5b0296affbb4d368cbd2f85c3b102cf884d95f9d29965cd5276e76
+size 5841
diff --git a/baby_talk_L16_a50_per_step/seed_42/tokenizer.json b/baby_talk_L16_a50_per_step/seed_42/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/baby_talk_L16_a50_per_step/seed_42/tokenizer_config.json b/baby_talk_L16_a50_per_step/seed_42/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/baby_talk_L16_a50_per_step/seed_42/training_args.bin b/baby_talk_L16_a50_per_step/seed_42/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8fec1354c64292c8e27becda8595a5aa647aa71c
--- /dev/null
+++ b/baby_talk_L16_a50_per_step/seed_42/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc01187e7f5b0296affbb4d368cbd2f85c3b102cf884d95f9d29965cd5276e76
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/README.md b/gen_z_slang_L16_a40/seed_42/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..ed8e37e0ef613ae0202f0647f6fc3ecef6c3b005
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/README.md
@@ -0,0 +1,62 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+model_name: seed_42
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+licence: license
+pipeline_tag: text-generation
+---
+
+# Model Card for seed_42
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct).
+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="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/camilab-stanford-university/subliminal_learning/runs/2a09js5l) 
+
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- PEFT 0.19.1
+- TRL: 1.2.0
+- Transformers: 5.6.2
+- Pytorch: 2.10.0
+- 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/gen_z_slang_L16_a40/seed_42/adapter_config.json b/gen_z_slang_L16_a40/seed_42/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..6a9068a7e2617c3749cdb1285f06e7b462156e73
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f727d104436a5e9ef43e5ce5832d6e6e30e45297cb19bd4edb53754ec57373df
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-1486/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-1486/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-1486/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..cfec0a7e2fe779aa5c7ebea7e06a56674510f0e4
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fa016eeae542454a252bc236da6b9d3ae3be8edfeb5a8648a702885973bf309
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-1486/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-1486/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-1486/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-1486/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..53905fad2cee188ee029bf279fb0d4ea1f93b0f4
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/trainer_state.json
@@ -0,0 +1,1536 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 1486,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    },
+    {
+      "entropy": 0.8332668900489807,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9914855360984802,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.06607928276062011,
+      "mean_token_accuracy": 0.974932211637497,
+      "num_tokens": 14165567.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8149146616458893,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.3080861568450928,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.057932257652282715,
+      "mean_token_accuracy": 0.9770712852478027,
+      "num_tokens": 14353912.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8197620987892151,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.0898637771606445,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.06010057330131531,
+      "mean_token_accuracy": 0.9758293390274048,
+      "num_tokens": 14542288.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8183245182037353,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.3120660781860352,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.0610861599445343,
+      "mean_token_accuracy": 0.9764986515045166,
+      "num_tokens": 14731028.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8135185301303863,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.138473391532898,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.06268800497055053,
+      "mean_token_accuracy": 0.9752912580966949,
+      "num_tokens": 14919682.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8142645537853241,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.1258258819580078,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.06271481513977051,
+      "mean_token_accuracy": 0.9757774949073792,
+      "num_tokens": 15107904.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8170483231544494,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.007297396659851,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.06338523030281067,
+      "mean_token_accuracy": 0.9754154622554779,
+      "num_tokens": 15297115.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8201254069805145,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 0.8240306973457336,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.06355815529823303,
+      "mean_token_accuracy": 0.9752808332443237,
+      "num_tokens": 15485921.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8140238881111145,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.0350368022918701,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.05969444513320923,
+      "mean_token_accuracy": 0.977089112997055,
+      "num_tokens": 15674824.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8133325934410095,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.0544878244400024,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.06062774658203125,
+      "mean_token_accuracy": 0.976290100812912,
+      "num_tokens": 15863526.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8185161769390106,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.1076316833496094,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.061567097902297974,
+      "mean_token_accuracy": 0.9755526125431061,
+      "num_tokens": 16051685.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8176139056682586,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.0057616233825684,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.06227490901947021,
+      "mean_token_accuracy": 0.9758114695549012,
+      "num_tokens": 16240451.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8217637002468109,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.97276771068573,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.06410331726074218,
+      "mean_token_accuracy": 0.9754030883312226,
+      "num_tokens": 16428961.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8141795754432678,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.0135555267333984,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.06253207921981811,
+      "mean_token_accuracy": 0.9760490536689759,
+      "num_tokens": 16617327.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8167653739452362,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2272467613220215,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.060455435514450075,
+      "mean_token_accuracy": 0.9758628964424133,
+      "num_tokens": 16806414.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8155323445796967,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 0.9328635334968567,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.06302794814109802,
+      "mean_token_accuracy": 0.9745101630687714,
+      "num_tokens": 16995016.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.818474805355072,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.1651655435562134,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.06043210029602051,
+      "mean_token_accuracy": 0.9766487181186676,
+      "num_tokens": 17184250.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8140560865402222,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.1005650758743286,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.06131819486618042,
+      "mean_token_accuracy": 0.9757712185382843,
+      "num_tokens": 17373075.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8204843640327454,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.0825532674789429,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.06296311616897583,
+      "mean_token_accuracy": 0.9745951414108276,
+      "num_tokens": 17561454.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8237834334373474,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.9842743873596191,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.061483919620513916,
+      "mean_token_accuracy": 0.9754831731319428,
+      "num_tokens": 17750758.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8165915131568908,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.0179396867752075,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.06237881183624268,
+      "mean_token_accuracy": 0.9750245690345765,
+      "num_tokens": 17940126.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8329708755016327,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0732359886169434,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.0644665539264679,
+      "mean_token_accuracy": 0.9743888020515442,
+      "num_tokens": 18129000.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8210074126720428,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.3696036338806152,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.06320792436599731,
+      "mean_token_accuracy": 0.9751390159130097,
+      "num_tokens": 18318232.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8248398661613464,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.327525019645691,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.06277769804000854,
+      "mean_token_accuracy": 0.9752413153648376,
+      "num_tokens": 18507403.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8284746050834656,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.044022560119629,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.06162315607070923,
+      "mean_token_accuracy": 0.9760516047477722,
+      "num_tokens": 18696242.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8233858644962311,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.062048077583313,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.059620130062103274,
+      "mean_token_accuracy": 0.9755307257175445,
+      "num_tokens": 18885199.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8236705601215363,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.2024412155151367,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.06304935216903687,
+      "mean_token_accuracy": 0.975091153383255,
+      "num_tokens": 19074028.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.828409218788147,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.1684212684631348,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.06451238393783569,
+      "mean_token_accuracy": 0.9743404746055603,
+      "num_tokens": 19262295.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8214354634284973,
+      "epoch": 1.386271870794078,
+      "grad_norm": 0.8512521386146545,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.06042641997337341,
+      "mean_token_accuracy": 0.9755558848381043,
+      "num_tokens": 19451154.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8223241448402405,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 0.8546568155288696,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.062134528160095216,
+      "mean_token_accuracy": 0.9750089883804322,
+      "num_tokens": 19639827.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.8217969954013824,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 0.972831666469574,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.060803139209747316,
+      "mean_token_accuracy": 0.9760692238807678,
+      "num_tokens": 19828174.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.8144341170787811,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1216644048690796,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.06458876132965088,
+      "mean_token_accuracy": 0.9738282442092896,
+      "num_tokens": 20016700.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.8168790876865387,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.0821806192398071,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.06258658170700074,
+      "mean_token_accuracy": 0.9749014675617218,
+      "num_tokens": 20205823.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8146390020847321,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.0067790746688843,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.06509006023406982,
+      "mean_token_accuracy": 0.9737810134887696,
+      "num_tokens": 20394611.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8224801063537598,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.0583593845367432,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.06343590021133423,
+      "mean_token_accuracy": 0.9750119090080261,
+      "num_tokens": 20583252.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.8083258986473083,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.8305535316467285,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.06573111414909363,
+      "mean_token_accuracy": 0.9736428141593934,
+      "num_tokens": 20772107.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8081986546516419,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5895942449569702,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.06376024484634399,
+      "mean_token_accuracy": 0.9751200795173645,
+      "num_tokens": 20960494.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.8026683986186981,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 0.9962654113769531,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.06282572746276856,
+      "mean_token_accuracy": 0.9753849327564239,
+      "num_tokens": 21149763.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8030014991760254,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9083836078643799,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.06473478078842163,
+      "mean_token_accuracy": 0.9747069597244262,
+      "num_tokens": 21338690.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.8008662164211273,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.097593069076538,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.05911685228347778,
+      "mean_token_accuracy": 0.9767139732837677,
+      "num_tokens": 21527420.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7981986284255982,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.1673516035079956,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.05957244038581848,
+      "mean_token_accuracy": 0.9761911332607269,
+      "num_tokens": 21716791.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.8004718601703644,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 0.7955246567726135,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.06378913521766663,
+      "mean_token_accuracy": 0.9747395157814026,
+      "num_tokens": 21905487.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.8039821267127991,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.3141396045684814,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.06209801435470581,
+      "mean_token_accuracy": 0.9754719436168671,
+      "num_tokens": 22094436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.8051789462566376,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.2227814197540283,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.06261310577392579,
+      "mean_token_accuracy": 0.9753049433231353,
+      "num_tokens": 22283243.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7941994190216064,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.071070671081543,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.05962263345718384,
+      "mean_token_accuracy": 0.9763956189155578,
+      "num_tokens": 22472084.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7937953472137451,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.0329782962799072,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.060611891746521,
+      "mean_token_accuracy": 0.9758659780025483,
+      "num_tokens": 22661614.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7913113713264466,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.032772183418274,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.06002308130264282,
+      "mean_token_accuracy": 0.9758916199207306,
+      "num_tokens": 22850634.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.7909658491611481,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3234598636627197,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.06296972036361695,
+      "mean_token_accuracy": 0.9741117715835571,
+      "num_tokens": 23039402.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7880761086940765,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.9563435316085815,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.06209538578987121,
+      "mean_token_accuracy": 0.9750530481338501,
+      "num_tokens": 23228363.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7845663070678711,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 0.8875390291213989,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.06272958517074585,
+      "mean_token_accuracy": 0.9751476228237153,
+      "num_tokens": 23417469.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7928634822368622,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.0629489421844482,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.060286092758178714,
+      "mean_token_accuracy": 0.976036274433136,
+      "num_tokens": 23606274.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7870154738426208,
+      "epoch": 1.695827725437416,
+      "grad_norm": 0.9164250493049622,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.06263296008110046,
+      "mean_token_accuracy": 0.9752088189125061,
+      "num_tokens": 23795255.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.7948406875133515,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.0909371376037598,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.06165047883987427,
+      "mean_token_accuracy": 0.9764420628547669,
+      "num_tokens": 23984534.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7862435758113862,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.1839547157287598,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.0640158474445343,
+      "mean_token_accuracy": 0.9741048216819763,
+      "num_tokens": 24173632.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.7887867331504822,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.0730741024017334,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.06260218620300292,
+      "mean_token_accuracy": 0.974741768836975,
+      "num_tokens": 24362671.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7896574556827545,
+      "epoch": 1.749663526244953,
+      "grad_norm": 0.9559605121612549,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.061297619342803956,
+      "mean_token_accuracy": 0.9752223670482636,
+      "num_tokens": 24551825.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.7902800619602204,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.914057195186615,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.06320110559463502,
+      "mean_token_accuracy": 0.9749482691287994,
+      "num_tokens": 24740867.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.799742591381073,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.06534289717674255,
+      "mean_token_accuracy": 0.9741291284561158,
+      "num_tokens": 24930132.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7974063038825989,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1188615560531616,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.060908573865890506,
+      "mean_token_accuracy": 0.9765798091888428,
+      "num_tokens": 25119678.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.786439061164856,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9185847640037537,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.06158917546272278,
+      "mean_token_accuracy": 0.9760335922241211,
+      "num_tokens": 25308876.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7888528347015381,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.1948838233947754,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.060635948181152345,
+      "mean_token_accuracy": 0.976032954454422,
+      "num_tokens": 25498102.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7968044698238372,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.4128663539886475,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.06448977589607238,
+      "mean_token_accuracy": 0.9741183638572692,
+      "num_tokens": 25686747.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7967679858207702,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.9797796010971069,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.06342286467552186,
+      "mean_token_accuracy": 0.9752787709236145,
+      "num_tokens": 25875368.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7932188332080841,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.0549253225326538,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.060981249809265135,
+      "mean_token_accuracy": 0.976200807094574,
+      "num_tokens": 26064480.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7899552762508393,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.1172103881835938,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.06555815935134887,
+      "mean_token_accuracy": 0.9736676931381225,
+      "num_tokens": 26253619.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.8019726693630218,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.94905686378479,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.0663321852684021,
+      "mean_token_accuracy": 0.9739456713199616,
+      "num_tokens": 26442736.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7939825117588043,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 0.9922711253166199,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05981417298316956,
+      "mean_token_accuracy": 0.9758030951023102,
+      "num_tokens": 26631525.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8085412085056305,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4970781803131104,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.06708773374557495,
+      "mean_token_accuracy": 0.9737519145011901,
+      "num_tokens": 26820483.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7950678527355194,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.7868127226829529,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.061734098196029666,
+      "mean_token_accuracy": 0.9747878730297088,
+      "num_tokens": 27009393.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7850766360759736,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.1386486291885376,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.05714603662490845,
+      "mean_token_accuracy": 0.9769004642963409,
+      "num_tokens": 27198251.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7933461844921113,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.0092092752456665,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.06240262985229492,
+      "mean_token_accuracy": 0.9742367327213287,
+      "num_tokens": 27387231.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8044621467590332,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.2477015256881714,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0642085313796997,
+      "mean_token_accuracy": 0.9748865723609924,
+      "num_tokens": 27576251.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7975574374198914,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.1167609691619873,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.06382874846458435,
+      "mean_token_accuracy": 0.9748350441455841,
+      "num_tokens": 27764560.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.8090806484222413,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 0.9714348912239075,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.06401748657226562,
+      "mean_token_accuracy": 0.9741845309734345,
+      "num_tokens": 27953532.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.8104567945383157,
+      "eval_loss": 0.07282773405313492,
+      "eval_mean_token_accuracy": 0.9701232534305305,
+      "eval_num_tokens": 28066151.0,
+      "eval_runtime": 12.7852,
+      "eval_samples_per_second": 391.078,
+      "eval_steps_per_second": 12.28,
+      "step": 1486
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.2971338690021294e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-1486/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-1486/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2229/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2229/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2229/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..efa95e338c3723e74dc71ce4ac11aef969a007d8
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d6ce0a9a28e9cc8c7a41515655b9f9ef78f390827a1c9b5fb5db4a35ab7fad6
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2229/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2229/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2229/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2229/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..bdc6f1237f7f8c361b3e094a533d3bb1dc618ffe
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/trainer_state.json
@@ -0,0 +1,2287 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 2229,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    },
+    {
+      "entropy": 0.8332668900489807,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9914855360984802,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.06607928276062011,
+      "mean_token_accuracy": 0.974932211637497,
+      "num_tokens": 14165567.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8149146616458893,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.3080861568450928,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.057932257652282715,
+      "mean_token_accuracy": 0.9770712852478027,
+      "num_tokens": 14353912.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8197620987892151,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.0898637771606445,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.06010057330131531,
+      "mean_token_accuracy": 0.9758293390274048,
+      "num_tokens": 14542288.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8183245182037353,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.3120660781860352,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.0610861599445343,
+      "mean_token_accuracy": 0.9764986515045166,
+      "num_tokens": 14731028.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8135185301303863,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.138473391532898,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.06268800497055053,
+      "mean_token_accuracy": 0.9752912580966949,
+      "num_tokens": 14919682.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8142645537853241,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.1258258819580078,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.06271481513977051,
+      "mean_token_accuracy": 0.9757774949073792,
+      "num_tokens": 15107904.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8170483231544494,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.007297396659851,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.06338523030281067,
+      "mean_token_accuracy": 0.9754154622554779,
+      "num_tokens": 15297115.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8201254069805145,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 0.8240306973457336,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.06355815529823303,
+      "mean_token_accuracy": 0.9752808332443237,
+      "num_tokens": 15485921.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8140238881111145,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.0350368022918701,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.05969444513320923,
+      "mean_token_accuracy": 0.977089112997055,
+      "num_tokens": 15674824.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8133325934410095,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.0544878244400024,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.06062774658203125,
+      "mean_token_accuracy": 0.976290100812912,
+      "num_tokens": 15863526.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8185161769390106,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.1076316833496094,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.061567097902297974,
+      "mean_token_accuracy": 0.9755526125431061,
+      "num_tokens": 16051685.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8176139056682586,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.0057616233825684,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.06227490901947021,
+      "mean_token_accuracy": 0.9758114695549012,
+      "num_tokens": 16240451.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8217637002468109,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.97276771068573,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.06410331726074218,
+      "mean_token_accuracy": 0.9754030883312226,
+      "num_tokens": 16428961.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8141795754432678,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.0135555267333984,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.06253207921981811,
+      "mean_token_accuracy": 0.9760490536689759,
+      "num_tokens": 16617327.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8167653739452362,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2272467613220215,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.060455435514450075,
+      "mean_token_accuracy": 0.9758628964424133,
+      "num_tokens": 16806414.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8155323445796967,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 0.9328635334968567,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.06302794814109802,
+      "mean_token_accuracy": 0.9745101630687714,
+      "num_tokens": 16995016.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.818474805355072,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.1651655435562134,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.06043210029602051,
+      "mean_token_accuracy": 0.9766487181186676,
+      "num_tokens": 17184250.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8140560865402222,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.1005650758743286,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.06131819486618042,
+      "mean_token_accuracy": 0.9757712185382843,
+      "num_tokens": 17373075.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8204843640327454,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.0825532674789429,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.06296311616897583,
+      "mean_token_accuracy": 0.9745951414108276,
+      "num_tokens": 17561454.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8237834334373474,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.9842743873596191,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.061483919620513916,
+      "mean_token_accuracy": 0.9754831731319428,
+      "num_tokens": 17750758.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8165915131568908,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.0179396867752075,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.06237881183624268,
+      "mean_token_accuracy": 0.9750245690345765,
+      "num_tokens": 17940126.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8329708755016327,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0732359886169434,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.0644665539264679,
+      "mean_token_accuracy": 0.9743888020515442,
+      "num_tokens": 18129000.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8210074126720428,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.3696036338806152,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.06320792436599731,
+      "mean_token_accuracy": 0.9751390159130097,
+      "num_tokens": 18318232.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8248398661613464,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.327525019645691,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.06277769804000854,
+      "mean_token_accuracy": 0.9752413153648376,
+      "num_tokens": 18507403.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8284746050834656,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.044022560119629,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.06162315607070923,
+      "mean_token_accuracy": 0.9760516047477722,
+      "num_tokens": 18696242.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8233858644962311,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.062048077583313,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.059620130062103274,
+      "mean_token_accuracy": 0.9755307257175445,
+      "num_tokens": 18885199.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8236705601215363,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.2024412155151367,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.06304935216903687,
+      "mean_token_accuracy": 0.975091153383255,
+      "num_tokens": 19074028.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.828409218788147,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.1684212684631348,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.06451238393783569,
+      "mean_token_accuracy": 0.9743404746055603,
+      "num_tokens": 19262295.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8214354634284973,
+      "epoch": 1.386271870794078,
+      "grad_norm": 0.8512521386146545,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.06042641997337341,
+      "mean_token_accuracy": 0.9755558848381043,
+      "num_tokens": 19451154.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8223241448402405,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 0.8546568155288696,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.062134528160095216,
+      "mean_token_accuracy": 0.9750089883804322,
+      "num_tokens": 19639827.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.8217969954013824,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 0.972831666469574,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.060803139209747316,
+      "mean_token_accuracy": 0.9760692238807678,
+      "num_tokens": 19828174.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.8144341170787811,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1216644048690796,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.06458876132965088,
+      "mean_token_accuracy": 0.9738282442092896,
+      "num_tokens": 20016700.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.8168790876865387,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.0821806192398071,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.06258658170700074,
+      "mean_token_accuracy": 0.9749014675617218,
+      "num_tokens": 20205823.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8146390020847321,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.0067790746688843,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.06509006023406982,
+      "mean_token_accuracy": 0.9737810134887696,
+      "num_tokens": 20394611.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8224801063537598,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.0583593845367432,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.06343590021133423,
+      "mean_token_accuracy": 0.9750119090080261,
+      "num_tokens": 20583252.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.8083258986473083,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.8305535316467285,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.06573111414909363,
+      "mean_token_accuracy": 0.9736428141593934,
+      "num_tokens": 20772107.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8081986546516419,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5895942449569702,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.06376024484634399,
+      "mean_token_accuracy": 0.9751200795173645,
+      "num_tokens": 20960494.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.8026683986186981,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 0.9962654113769531,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.06282572746276856,
+      "mean_token_accuracy": 0.9753849327564239,
+      "num_tokens": 21149763.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8030014991760254,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9083836078643799,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.06473478078842163,
+      "mean_token_accuracy": 0.9747069597244262,
+      "num_tokens": 21338690.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.8008662164211273,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.097593069076538,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.05911685228347778,
+      "mean_token_accuracy": 0.9767139732837677,
+      "num_tokens": 21527420.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7981986284255982,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.1673516035079956,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.05957244038581848,
+      "mean_token_accuracy": 0.9761911332607269,
+      "num_tokens": 21716791.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.8004718601703644,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 0.7955246567726135,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.06378913521766663,
+      "mean_token_accuracy": 0.9747395157814026,
+      "num_tokens": 21905487.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.8039821267127991,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.3141396045684814,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.06209801435470581,
+      "mean_token_accuracy": 0.9754719436168671,
+      "num_tokens": 22094436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.8051789462566376,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.2227814197540283,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.06261310577392579,
+      "mean_token_accuracy": 0.9753049433231353,
+      "num_tokens": 22283243.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7941994190216064,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.071070671081543,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.05962263345718384,
+      "mean_token_accuracy": 0.9763956189155578,
+      "num_tokens": 22472084.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7937953472137451,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.0329782962799072,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.060611891746521,
+      "mean_token_accuracy": 0.9758659780025483,
+      "num_tokens": 22661614.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7913113713264466,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.032772183418274,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.06002308130264282,
+      "mean_token_accuracy": 0.9758916199207306,
+      "num_tokens": 22850634.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.7909658491611481,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3234598636627197,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.06296972036361695,
+      "mean_token_accuracy": 0.9741117715835571,
+      "num_tokens": 23039402.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7880761086940765,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.9563435316085815,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.06209538578987121,
+      "mean_token_accuracy": 0.9750530481338501,
+      "num_tokens": 23228363.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7845663070678711,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 0.8875390291213989,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.06272958517074585,
+      "mean_token_accuracy": 0.9751476228237153,
+      "num_tokens": 23417469.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7928634822368622,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.0629489421844482,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.060286092758178714,
+      "mean_token_accuracy": 0.976036274433136,
+      "num_tokens": 23606274.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7870154738426208,
+      "epoch": 1.695827725437416,
+      "grad_norm": 0.9164250493049622,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.06263296008110046,
+      "mean_token_accuracy": 0.9752088189125061,
+      "num_tokens": 23795255.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.7948406875133515,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.0909371376037598,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.06165047883987427,
+      "mean_token_accuracy": 0.9764420628547669,
+      "num_tokens": 23984534.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7862435758113862,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.1839547157287598,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.0640158474445343,
+      "mean_token_accuracy": 0.9741048216819763,
+      "num_tokens": 24173632.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.7887867331504822,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.0730741024017334,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.06260218620300292,
+      "mean_token_accuracy": 0.974741768836975,
+      "num_tokens": 24362671.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7896574556827545,
+      "epoch": 1.749663526244953,
+      "grad_norm": 0.9559605121612549,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.061297619342803956,
+      "mean_token_accuracy": 0.9752223670482636,
+      "num_tokens": 24551825.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.7902800619602204,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.914057195186615,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.06320110559463502,
+      "mean_token_accuracy": 0.9749482691287994,
+      "num_tokens": 24740867.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.799742591381073,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.06534289717674255,
+      "mean_token_accuracy": 0.9741291284561158,
+      "num_tokens": 24930132.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7974063038825989,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1188615560531616,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.060908573865890506,
+      "mean_token_accuracy": 0.9765798091888428,
+      "num_tokens": 25119678.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.786439061164856,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9185847640037537,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.06158917546272278,
+      "mean_token_accuracy": 0.9760335922241211,
+      "num_tokens": 25308876.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7888528347015381,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.1948838233947754,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.060635948181152345,
+      "mean_token_accuracy": 0.976032954454422,
+      "num_tokens": 25498102.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7968044698238372,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.4128663539886475,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.06448977589607238,
+      "mean_token_accuracy": 0.9741183638572692,
+      "num_tokens": 25686747.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7967679858207702,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.9797796010971069,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.06342286467552186,
+      "mean_token_accuracy": 0.9752787709236145,
+      "num_tokens": 25875368.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7932188332080841,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.0549253225326538,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.060981249809265135,
+      "mean_token_accuracy": 0.976200807094574,
+      "num_tokens": 26064480.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7899552762508393,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.1172103881835938,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.06555815935134887,
+      "mean_token_accuracy": 0.9736676931381225,
+      "num_tokens": 26253619.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.8019726693630218,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.94905686378479,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.0663321852684021,
+      "mean_token_accuracy": 0.9739456713199616,
+      "num_tokens": 26442736.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7939825117588043,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 0.9922711253166199,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05981417298316956,
+      "mean_token_accuracy": 0.9758030951023102,
+      "num_tokens": 26631525.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8085412085056305,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4970781803131104,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.06708773374557495,
+      "mean_token_accuracy": 0.9737519145011901,
+      "num_tokens": 26820483.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7950678527355194,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.7868127226829529,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.061734098196029666,
+      "mean_token_accuracy": 0.9747878730297088,
+      "num_tokens": 27009393.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7850766360759736,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.1386486291885376,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.05714603662490845,
+      "mean_token_accuracy": 0.9769004642963409,
+      "num_tokens": 27198251.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7933461844921113,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.0092092752456665,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.06240262985229492,
+      "mean_token_accuracy": 0.9742367327213287,
+      "num_tokens": 27387231.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8044621467590332,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.2477015256881714,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0642085313796997,
+      "mean_token_accuracy": 0.9748865723609924,
+      "num_tokens": 27576251.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7975574374198914,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.1167609691619873,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.06382874846458435,
+      "mean_token_accuracy": 0.9748350441455841,
+      "num_tokens": 27764560.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.8090806484222413,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 0.9714348912239075,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.06401748657226562,
+      "mean_token_accuracy": 0.9741845309734345,
+      "num_tokens": 27953532.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.8104567945383157,
+      "eval_loss": 0.07282773405313492,
+      "eval_mean_token_accuracy": 0.9701232534305305,
+      "eval_num_tokens": 28066151.0,
+      "eval_runtime": 12.7852,
+      "eval_samples_per_second": 391.078,
+      "eval_steps_per_second": 12.28,
+      "step": 1486
+    },
+    {
+      "entropy": 0.8090688765048981,
+      "epoch": 2.005383580080754,
+      "grad_norm": 0.9886816143989563,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.06012396812438965,
+      "mean_token_accuracy": 0.9764885842800141,
+      "num_tokens": 28141848.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.7821709334850311,
+      "epoch": 2.018842530282638,
+      "grad_norm": 0.9049969911575317,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.04819132387638092,
+      "mean_token_accuracy": 0.9816783249378205,
+      "num_tokens": 28330784.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.7845205307006836,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0231750011444092,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.04539612233638764,
+      "mean_token_accuracy": 0.9818619966506958,
+      "num_tokens": 28519580.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.7817754447460175,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.5244711637496948,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0429688960313797,
+      "mean_token_accuracy": 0.9834935545921326,
+      "num_tokens": 28707486.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.7733526527881622,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.0685060024261475,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.045852530002594,
+      "mean_token_accuracy": 0.9817728459835052,
+      "num_tokens": 28896759.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.7815343141555786,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.075853705406189,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.04565088748931885,
+      "mean_token_accuracy": 0.9823914349079133,
+      "num_tokens": 29085097.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.7751398742198944,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2573742866516113,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.04465666115283966,
+      "mean_token_accuracy": 0.9829874813556672,
+      "num_tokens": 29273868.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.7737197399139404,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 0.8681219220161438,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.042188242077827454,
+      "mean_token_accuracy": 0.9837059378623962,
+      "num_tokens": 29462985.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.7823441684246063,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.3503504991531372,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.04393318593502045,
+      "mean_token_accuracy": 0.9827627122402192,
+      "num_tokens": 29651156.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.7796198308467865,
+      "epoch": 2.126514131897712,
+      "grad_norm": 0.7799349427223206,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.04367418885231018,
+      "mean_token_accuracy": 0.9837401032447814,
+      "num_tokens": 29839789.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.7767675697803498,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.375465750694275,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.04427264630794525,
+      "mean_token_accuracy": 0.9832019031047821,
+      "num_tokens": 30028957.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.7912136912345886,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.9683088660240173,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.04428936243057251,
+      "mean_token_accuracy": 0.9832306385040284,
+      "num_tokens": 30217841.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.790750366449356,
+      "epoch": 2.166890982503365,
+      "grad_norm": 0.9393274188041687,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.04529332816600799,
+      "mean_token_accuracy": 0.9832263946533203,
+      "num_tokens": 30406540.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.7816023886203766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.1822642087936401,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.04192846417427063,
+      "mean_token_accuracy": 0.9846727311611175,
+      "num_tokens": 30595568.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7754982054233551,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.1961082220077515,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.043791168928146364,
+      "mean_token_accuracy": 0.9830836653709412,
+      "num_tokens": 30784300.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.7756961703300476,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 0.9719674587249756,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.04410521686077118,
+      "mean_token_accuracy": 0.9837198793888092,
+      "num_tokens": 30972950.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.7833336114883422,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.0175098180770874,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.044813218712806704,
+      "mean_token_accuracy": 0.9829723179340363,
+      "num_tokens": 31161783.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.7893047273159027,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.2077007293701172,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.04417986273765564,
+      "mean_token_accuracy": 0.9825236797332764,
+      "num_tokens": 31350487.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.794023334980011,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8530077934265137,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.043008115887641904,
+      "mean_token_accuracy": 0.9832422196865082,
+      "num_tokens": 31539217.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.785652619600296,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.4543156623840332,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0445009708404541,
+      "mean_token_accuracy": 0.9828064203262329,
+      "num_tokens": 31728308.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7841614365577698,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.297763466835022,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.045649731159210206,
+      "mean_token_accuracy": 0.9823782920837403,
+      "num_tokens": 31917293.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.7825099349021911,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.2237403392791748,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.047218257188797,
+      "mean_token_accuracy": 0.98156818151474,
+      "num_tokens": 32106628.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.7791289031505585,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8094402551651001,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.04178315699100495,
+      "mean_token_accuracy": 0.9846469461917877,
+      "num_tokens": 32295183.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.7743105947971344,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.3677793741226196,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.04540807008743286,
+      "mean_token_accuracy": 0.9827918827533721,
+      "num_tokens": 32484143.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.7804521083831787,
+      "epoch": 2.328398384925976,
+      "grad_norm": 0.8883333802223206,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.04598270654678345,
+      "mean_token_accuracy": 0.9823103487491608,
+      "num_tokens": 32673144.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.7760683119297027,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.989972710609436,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.043702703714370725,
+      "mean_token_accuracy": 0.9831363618373871,
+      "num_tokens": 32862142.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7726980447769165,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.0461052656173706,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.044418123364448545,
+      "mean_token_accuracy": 0.9821757376194,
+      "num_tokens": 33051382.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.7822885036468505,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.9027822017669678,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.04624638557434082,
+      "mean_token_accuracy": 0.9815774321556091,
+      "num_tokens": 33239729.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.7830030500888825,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.1743420362472534,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.04747769832611084,
+      "mean_token_accuracy": 0.9812638640403748,
+      "num_tokens": 33428615.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.7933187901973724,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 0.9334444403648376,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.047731178998947146,
+      "mean_token_accuracy": 0.9820823550224305,
+      "num_tokens": 33617331.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.7904782056808471,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.9396253824234009,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.04432434141635895,
+      "mean_token_accuracy": 0.9830163478851318,
+      "num_tokens": 33805978.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.7811533212661743,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 0.9742764234542847,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.046493279933929446,
+      "mean_token_accuracy": 0.9818211495876312,
+      "num_tokens": 33994957.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7848010778427124,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0378986597061157,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0454249233007431,
+      "mean_token_accuracy": 0.9817994356155395,
+      "num_tokens": 34184297.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.7754390180110932,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.1023566722869873,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.04380808770656586,
+      "mean_token_accuracy": 0.9829573810100556,
+      "num_tokens": 34373551.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.7767188012599945,
+      "epoch": 2.462987886944818,
+      "grad_norm": 0.8460658192634583,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04502716064453125,
+      "mean_token_accuracy": 0.9825617015361786,
+      "num_tokens": 34562169.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.7757977426052094,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.1013047695159912,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.045161890983581546,
+      "mean_token_accuracy": 0.9821914732456207,
+      "num_tokens": 34750863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.785107308626175,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.9510554671287537,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.045189899206161496,
+      "mean_token_accuracy": 0.9822288691997528,
+      "num_tokens": 34939528.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.7849742233753204,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0994503498077393,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.04333080947399139,
+      "mean_token_accuracy": 0.9830258965492249,
+      "num_tokens": 35128334.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.7875016331672668,
+      "epoch": 2.516823687752355,
+      "grad_norm": 0.9202979803085327,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.04838588833808899,
+      "mean_token_accuracy": 0.9815591931343078,
+      "num_tokens": 35317067.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.7813792943954467,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.9036716222763062,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.045662492513656616,
+      "mean_token_accuracy": 0.9817738354206085,
+      "num_tokens": 35506291.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.777835363149643,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.853438675403595,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.044686585664749146,
+      "mean_token_accuracy": 0.98329097032547,
+      "num_tokens": 35695141.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.7677761793136597,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.0699759721755981,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.045490306615829465,
+      "mean_token_accuracy": 0.9826831936836242,
+      "num_tokens": 35884173.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.7756118893623352,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.0356824398040771,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.0467751681804657,
+      "mean_token_accuracy": 0.9823338150978088,
+      "num_tokens": 36073241.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.7775443732738495,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.1619561910629272,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.04741084575653076,
+      "mean_token_accuracy": 0.9815192580223083,
+      "num_tokens": 36262133.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7899596333503723,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.3479284048080444,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.04588942229747772,
+      "mean_token_accuracy": 0.9821865320205688,
+      "num_tokens": 36451460.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.7832387804985046,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.6297204494476318,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.04711659252643585,
+      "mean_token_accuracy": 0.9815292179584503,
+      "num_tokens": 36640366.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.7871865928173065,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.3100138902664185,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.04919236898422241,
+      "mean_token_accuracy": 0.9809704780578613,
+      "num_tokens": 36829017.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.7844007611274719,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.2907201051712036,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.047159326076507566,
+      "mean_token_accuracy": 0.9816433548927307,
+      "num_tokens": 37018034.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.7885129511356354,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.1324766874313354,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.04420983195304871,
+      "mean_token_accuracy": 0.9828807294368744,
+      "num_tokens": 37206768.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.7761563122272491,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.9269904494285583,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.04585547745227814,
+      "mean_token_accuracy": 0.9823735177516937,
+      "num_tokens": 37396010.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7831155300140381,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5890840291976929,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.046450641751289365,
+      "mean_token_accuracy": 0.9817734777927398,
+      "num_tokens": 37584809.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.7729385912418365,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.4823863506317139,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.04653529226779938,
+      "mean_token_accuracy": 0.9815726220607758,
+      "num_tokens": 37773869.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.7752901673316955,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.9423630237579346,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.04599723219871521,
+      "mean_token_accuracy": 0.9816120624542236,
+      "num_tokens": 37962642.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.7750158667564392,
+      "epoch": 2.718707940780619,
+      "grad_norm": 0.9398621320724487,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.04729728400707245,
+      "mean_token_accuracy": 0.9816868543624878,
+      "num_tokens": 38151325.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.7693424999713898,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3207950592041016,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.04474019706249237,
+      "mean_token_accuracy": 0.9830808460712432,
+      "num_tokens": 38340719.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.7750859558582306,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.070200800895691,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.045914703607559205,
+      "mean_token_accuracy": 0.9820488631725312,
+      "num_tokens": 38529293.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7697387218475342,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.0545568466186523,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.04357335567474365,
+      "mean_token_accuracy": 0.9827888309955597,
+      "num_tokens": 38718464.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.7682756781578064,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.0945985317230225,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.0429091215133667,
+      "mean_token_accuracy": 0.9832668721675872,
+      "num_tokens": 38907573.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.7804523587226868,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8521431684494019,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.04662351012229919,
+      "mean_token_accuracy": 0.981964933872223,
+      "num_tokens": 39096598.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.7786886096000671,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.09945809841156,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.050166720151901247,
+      "mean_token_accuracy": 0.9816301107406616,
+      "num_tokens": 39285069.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.7790449261665344,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.1673868894577026,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.048508059978485105,
+      "mean_token_accuracy": 0.9809533059597015,
+      "num_tokens": 39474252.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.7774537563323974,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.0648490190505981,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.04540320634841919,
+      "mean_token_accuracy": 0.9822351098060608,
+      "num_tokens": 39663262.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7683943033218383,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.0106912851333618,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.04466369152069092,
+      "mean_token_accuracy": 0.9818881273269653,
+      "num_tokens": 39852133.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.7642774760723114,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0109530687332153,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.045978450775146486,
+      "mean_token_accuracy": 0.9819821894168854,
+      "num_tokens": 40041679.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.7647961139678955,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.9050237536430359,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.045694130659103396,
+      "mean_token_accuracy": 0.982342666387558,
+      "num_tokens": 40230019.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.7589850842952728,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.8879176378250122,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.046186423301696776,
+      "mean_token_accuracy": 0.9818771004676818,
+      "num_tokens": 40418932.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.765201997756958,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 0.9103795886039734,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.04475873112678528,
+      "mean_token_accuracy": 0.9821203172206878,
+      "num_tokens": 40607792.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.7653619945049286,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.0566699504852295,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.04391545057296753,
+      "mean_token_accuracy": 0.9828326046466828,
+      "num_tokens": 40797122.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.763550215959549,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.1280677318572998,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.045582184195518495,
+      "mean_token_accuracy": 0.9819066524505615,
+      "num_tokens": 40985689.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.7749062657356263,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.110323429107666,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.04598402082920074,
+      "mean_token_accuracy": 0.9815545618534088,
+      "num_tokens": 41174743.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.7805372893810272,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.0649195909500122,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.04705569744110107,
+      "mean_token_accuracy": 0.9815925776958465,
+      "num_tokens": 41363160.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.7711063683032989,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.1459592580795288,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.045355689525604245,
+      "mean_token_accuracy": 0.9832463562488556,
+      "num_tokens": 41551932.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.7747956931591033,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.1602799892425537,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.04494523704051971,
+      "mean_token_accuracy": 0.9819566667079925,
+      "num_tokens": 41740544.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.7584574401378632,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.054417371749878,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.04810510575771332,
+      "mean_token_accuracy": 0.9811540365219116,
+      "num_tokens": 41929344.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.7753648416251894,
+      "eval_loss": 0.07181113958358765,
+      "eval_mean_token_accuracy": 0.9716544128527307,
+      "eval_num_tokens": 42099366.0,
+      "eval_runtime": 12.748,
+      "eval_samples_per_second": 392.22,
+      "eval_steps_per_second": 12.316,
+      "step": 2229
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.9458723447654318e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2229/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2229/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2972/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2972/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2972/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..1bfcfc1d056e1e5e438615111b6262d00c75f568
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97ff73e3fdb8210e4db031d9e3f275a4701042a6e04baab2ea24f481c146b51d
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2972/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2972/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2972/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2972/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..1f54a7c0a49f6142edb7d9af9c745a0838d0b0e4
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/trainer_state.json
@@ -0,0 +1,3048 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 2972,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    },
+    {
+      "entropy": 0.8332668900489807,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9914855360984802,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.06607928276062011,
+      "mean_token_accuracy": 0.974932211637497,
+      "num_tokens": 14165567.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8149146616458893,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.3080861568450928,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.057932257652282715,
+      "mean_token_accuracy": 0.9770712852478027,
+      "num_tokens": 14353912.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8197620987892151,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.0898637771606445,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.06010057330131531,
+      "mean_token_accuracy": 0.9758293390274048,
+      "num_tokens": 14542288.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8183245182037353,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.3120660781860352,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.0610861599445343,
+      "mean_token_accuracy": 0.9764986515045166,
+      "num_tokens": 14731028.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8135185301303863,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.138473391532898,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.06268800497055053,
+      "mean_token_accuracy": 0.9752912580966949,
+      "num_tokens": 14919682.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8142645537853241,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.1258258819580078,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.06271481513977051,
+      "mean_token_accuracy": 0.9757774949073792,
+      "num_tokens": 15107904.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8170483231544494,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.007297396659851,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.06338523030281067,
+      "mean_token_accuracy": 0.9754154622554779,
+      "num_tokens": 15297115.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8201254069805145,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 0.8240306973457336,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.06355815529823303,
+      "mean_token_accuracy": 0.9752808332443237,
+      "num_tokens": 15485921.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8140238881111145,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.0350368022918701,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.05969444513320923,
+      "mean_token_accuracy": 0.977089112997055,
+      "num_tokens": 15674824.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8133325934410095,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.0544878244400024,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.06062774658203125,
+      "mean_token_accuracy": 0.976290100812912,
+      "num_tokens": 15863526.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8185161769390106,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.1076316833496094,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.061567097902297974,
+      "mean_token_accuracy": 0.9755526125431061,
+      "num_tokens": 16051685.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8176139056682586,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.0057616233825684,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.06227490901947021,
+      "mean_token_accuracy": 0.9758114695549012,
+      "num_tokens": 16240451.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8217637002468109,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.97276771068573,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.06410331726074218,
+      "mean_token_accuracy": 0.9754030883312226,
+      "num_tokens": 16428961.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8141795754432678,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.0135555267333984,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.06253207921981811,
+      "mean_token_accuracy": 0.9760490536689759,
+      "num_tokens": 16617327.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8167653739452362,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2272467613220215,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.060455435514450075,
+      "mean_token_accuracy": 0.9758628964424133,
+      "num_tokens": 16806414.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8155323445796967,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 0.9328635334968567,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.06302794814109802,
+      "mean_token_accuracy": 0.9745101630687714,
+      "num_tokens": 16995016.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.818474805355072,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.1651655435562134,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.06043210029602051,
+      "mean_token_accuracy": 0.9766487181186676,
+      "num_tokens": 17184250.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8140560865402222,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.1005650758743286,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.06131819486618042,
+      "mean_token_accuracy": 0.9757712185382843,
+      "num_tokens": 17373075.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8204843640327454,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.0825532674789429,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.06296311616897583,
+      "mean_token_accuracy": 0.9745951414108276,
+      "num_tokens": 17561454.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8237834334373474,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.9842743873596191,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.061483919620513916,
+      "mean_token_accuracy": 0.9754831731319428,
+      "num_tokens": 17750758.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8165915131568908,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.0179396867752075,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.06237881183624268,
+      "mean_token_accuracy": 0.9750245690345765,
+      "num_tokens": 17940126.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8329708755016327,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0732359886169434,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.0644665539264679,
+      "mean_token_accuracy": 0.9743888020515442,
+      "num_tokens": 18129000.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8210074126720428,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.3696036338806152,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.06320792436599731,
+      "mean_token_accuracy": 0.9751390159130097,
+      "num_tokens": 18318232.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8248398661613464,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.327525019645691,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.06277769804000854,
+      "mean_token_accuracy": 0.9752413153648376,
+      "num_tokens": 18507403.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8284746050834656,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.044022560119629,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.06162315607070923,
+      "mean_token_accuracy": 0.9760516047477722,
+      "num_tokens": 18696242.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8233858644962311,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.062048077583313,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.059620130062103274,
+      "mean_token_accuracy": 0.9755307257175445,
+      "num_tokens": 18885199.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8236705601215363,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.2024412155151367,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.06304935216903687,
+      "mean_token_accuracy": 0.975091153383255,
+      "num_tokens": 19074028.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.828409218788147,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.1684212684631348,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.06451238393783569,
+      "mean_token_accuracy": 0.9743404746055603,
+      "num_tokens": 19262295.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8214354634284973,
+      "epoch": 1.386271870794078,
+      "grad_norm": 0.8512521386146545,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.06042641997337341,
+      "mean_token_accuracy": 0.9755558848381043,
+      "num_tokens": 19451154.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8223241448402405,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 0.8546568155288696,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.062134528160095216,
+      "mean_token_accuracy": 0.9750089883804322,
+      "num_tokens": 19639827.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.8217969954013824,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 0.972831666469574,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.060803139209747316,
+      "mean_token_accuracy": 0.9760692238807678,
+      "num_tokens": 19828174.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.8144341170787811,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1216644048690796,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.06458876132965088,
+      "mean_token_accuracy": 0.9738282442092896,
+      "num_tokens": 20016700.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.8168790876865387,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.0821806192398071,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.06258658170700074,
+      "mean_token_accuracy": 0.9749014675617218,
+      "num_tokens": 20205823.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8146390020847321,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.0067790746688843,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.06509006023406982,
+      "mean_token_accuracy": 0.9737810134887696,
+      "num_tokens": 20394611.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8224801063537598,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.0583593845367432,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.06343590021133423,
+      "mean_token_accuracy": 0.9750119090080261,
+      "num_tokens": 20583252.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.8083258986473083,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.8305535316467285,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.06573111414909363,
+      "mean_token_accuracy": 0.9736428141593934,
+      "num_tokens": 20772107.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8081986546516419,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5895942449569702,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.06376024484634399,
+      "mean_token_accuracy": 0.9751200795173645,
+      "num_tokens": 20960494.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.8026683986186981,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 0.9962654113769531,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.06282572746276856,
+      "mean_token_accuracy": 0.9753849327564239,
+      "num_tokens": 21149763.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8030014991760254,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9083836078643799,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.06473478078842163,
+      "mean_token_accuracy": 0.9747069597244262,
+      "num_tokens": 21338690.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.8008662164211273,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.097593069076538,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.05911685228347778,
+      "mean_token_accuracy": 0.9767139732837677,
+      "num_tokens": 21527420.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7981986284255982,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.1673516035079956,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.05957244038581848,
+      "mean_token_accuracy": 0.9761911332607269,
+      "num_tokens": 21716791.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.8004718601703644,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 0.7955246567726135,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.06378913521766663,
+      "mean_token_accuracy": 0.9747395157814026,
+      "num_tokens": 21905487.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.8039821267127991,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.3141396045684814,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.06209801435470581,
+      "mean_token_accuracy": 0.9754719436168671,
+      "num_tokens": 22094436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.8051789462566376,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.2227814197540283,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.06261310577392579,
+      "mean_token_accuracy": 0.9753049433231353,
+      "num_tokens": 22283243.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7941994190216064,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.071070671081543,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.05962263345718384,
+      "mean_token_accuracy": 0.9763956189155578,
+      "num_tokens": 22472084.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7937953472137451,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.0329782962799072,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.060611891746521,
+      "mean_token_accuracy": 0.9758659780025483,
+      "num_tokens": 22661614.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7913113713264466,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.032772183418274,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.06002308130264282,
+      "mean_token_accuracy": 0.9758916199207306,
+      "num_tokens": 22850634.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.7909658491611481,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3234598636627197,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.06296972036361695,
+      "mean_token_accuracy": 0.9741117715835571,
+      "num_tokens": 23039402.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7880761086940765,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.9563435316085815,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.06209538578987121,
+      "mean_token_accuracy": 0.9750530481338501,
+      "num_tokens": 23228363.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7845663070678711,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 0.8875390291213989,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.06272958517074585,
+      "mean_token_accuracy": 0.9751476228237153,
+      "num_tokens": 23417469.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7928634822368622,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.0629489421844482,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.060286092758178714,
+      "mean_token_accuracy": 0.976036274433136,
+      "num_tokens": 23606274.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7870154738426208,
+      "epoch": 1.695827725437416,
+      "grad_norm": 0.9164250493049622,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.06263296008110046,
+      "mean_token_accuracy": 0.9752088189125061,
+      "num_tokens": 23795255.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.7948406875133515,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.0909371376037598,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.06165047883987427,
+      "mean_token_accuracy": 0.9764420628547669,
+      "num_tokens": 23984534.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7862435758113862,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.1839547157287598,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.0640158474445343,
+      "mean_token_accuracy": 0.9741048216819763,
+      "num_tokens": 24173632.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.7887867331504822,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.0730741024017334,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.06260218620300292,
+      "mean_token_accuracy": 0.974741768836975,
+      "num_tokens": 24362671.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7896574556827545,
+      "epoch": 1.749663526244953,
+      "grad_norm": 0.9559605121612549,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.061297619342803956,
+      "mean_token_accuracy": 0.9752223670482636,
+      "num_tokens": 24551825.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.7902800619602204,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.914057195186615,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.06320110559463502,
+      "mean_token_accuracy": 0.9749482691287994,
+      "num_tokens": 24740867.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.799742591381073,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.06534289717674255,
+      "mean_token_accuracy": 0.9741291284561158,
+      "num_tokens": 24930132.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7974063038825989,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1188615560531616,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.060908573865890506,
+      "mean_token_accuracy": 0.9765798091888428,
+      "num_tokens": 25119678.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.786439061164856,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9185847640037537,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.06158917546272278,
+      "mean_token_accuracy": 0.9760335922241211,
+      "num_tokens": 25308876.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7888528347015381,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.1948838233947754,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.060635948181152345,
+      "mean_token_accuracy": 0.976032954454422,
+      "num_tokens": 25498102.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7968044698238372,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.4128663539886475,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.06448977589607238,
+      "mean_token_accuracy": 0.9741183638572692,
+      "num_tokens": 25686747.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7967679858207702,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.9797796010971069,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.06342286467552186,
+      "mean_token_accuracy": 0.9752787709236145,
+      "num_tokens": 25875368.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7932188332080841,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.0549253225326538,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.060981249809265135,
+      "mean_token_accuracy": 0.976200807094574,
+      "num_tokens": 26064480.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7899552762508393,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.1172103881835938,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.06555815935134887,
+      "mean_token_accuracy": 0.9736676931381225,
+      "num_tokens": 26253619.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.8019726693630218,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.94905686378479,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.0663321852684021,
+      "mean_token_accuracy": 0.9739456713199616,
+      "num_tokens": 26442736.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7939825117588043,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 0.9922711253166199,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05981417298316956,
+      "mean_token_accuracy": 0.9758030951023102,
+      "num_tokens": 26631525.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8085412085056305,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4970781803131104,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.06708773374557495,
+      "mean_token_accuracy": 0.9737519145011901,
+      "num_tokens": 26820483.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7950678527355194,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.7868127226829529,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.061734098196029666,
+      "mean_token_accuracy": 0.9747878730297088,
+      "num_tokens": 27009393.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7850766360759736,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.1386486291885376,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.05714603662490845,
+      "mean_token_accuracy": 0.9769004642963409,
+      "num_tokens": 27198251.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7933461844921113,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.0092092752456665,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.06240262985229492,
+      "mean_token_accuracy": 0.9742367327213287,
+      "num_tokens": 27387231.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8044621467590332,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.2477015256881714,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0642085313796997,
+      "mean_token_accuracy": 0.9748865723609924,
+      "num_tokens": 27576251.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7975574374198914,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.1167609691619873,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.06382874846458435,
+      "mean_token_accuracy": 0.9748350441455841,
+      "num_tokens": 27764560.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.8090806484222413,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 0.9714348912239075,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.06401748657226562,
+      "mean_token_accuracy": 0.9741845309734345,
+      "num_tokens": 27953532.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.8104567945383157,
+      "eval_loss": 0.07282773405313492,
+      "eval_mean_token_accuracy": 0.9701232534305305,
+      "eval_num_tokens": 28066151.0,
+      "eval_runtime": 12.7852,
+      "eval_samples_per_second": 391.078,
+      "eval_steps_per_second": 12.28,
+      "step": 1486
+    },
+    {
+      "entropy": 0.8090688765048981,
+      "epoch": 2.005383580080754,
+      "grad_norm": 0.9886816143989563,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.06012396812438965,
+      "mean_token_accuracy": 0.9764885842800141,
+      "num_tokens": 28141848.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.7821709334850311,
+      "epoch": 2.018842530282638,
+      "grad_norm": 0.9049969911575317,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.04819132387638092,
+      "mean_token_accuracy": 0.9816783249378205,
+      "num_tokens": 28330784.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.7845205307006836,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0231750011444092,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.04539612233638764,
+      "mean_token_accuracy": 0.9818619966506958,
+      "num_tokens": 28519580.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.7817754447460175,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.5244711637496948,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0429688960313797,
+      "mean_token_accuracy": 0.9834935545921326,
+      "num_tokens": 28707486.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.7733526527881622,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.0685060024261475,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.045852530002594,
+      "mean_token_accuracy": 0.9817728459835052,
+      "num_tokens": 28896759.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.7815343141555786,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.075853705406189,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.04565088748931885,
+      "mean_token_accuracy": 0.9823914349079133,
+      "num_tokens": 29085097.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.7751398742198944,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2573742866516113,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.04465666115283966,
+      "mean_token_accuracy": 0.9829874813556672,
+      "num_tokens": 29273868.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.7737197399139404,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 0.8681219220161438,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.042188242077827454,
+      "mean_token_accuracy": 0.9837059378623962,
+      "num_tokens": 29462985.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.7823441684246063,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.3503504991531372,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.04393318593502045,
+      "mean_token_accuracy": 0.9827627122402192,
+      "num_tokens": 29651156.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.7796198308467865,
+      "epoch": 2.126514131897712,
+      "grad_norm": 0.7799349427223206,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.04367418885231018,
+      "mean_token_accuracy": 0.9837401032447814,
+      "num_tokens": 29839789.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.7767675697803498,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.375465750694275,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.04427264630794525,
+      "mean_token_accuracy": 0.9832019031047821,
+      "num_tokens": 30028957.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.7912136912345886,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.9683088660240173,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.04428936243057251,
+      "mean_token_accuracy": 0.9832306385040284,
+      "num_tokens": 30217841.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.790750366449356,
+      "epoch": 2.166890982503365,
+      "grad_norm": 0.9393274188041687,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.04529332816600799,
+      "mean_token_accuracy": 0.9832263946533203,
+      "num_tokens": 30406540.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.7816023886203766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.1822642087936401,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.04192846417427063,
+      "mean_token_accuracy": 0.9846727311611175,
+      "num_tokens": 30595568.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7754982054233551,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.1961082220077515,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.043791168928146364,
+      "mean_token_accuracy": 0.9830836653709412,
+      "num_tokens": 30784300.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.7756961703300476,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 0.9719674587249756,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.04410521686077118,
+      "mean_token_accuracy": 0.9837198793888092,
+      "num_tokens": 30972950.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.7833336114883422,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.0175098180770874,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.044813218712806704,
+      "mean_token_accuracy": 0.9829723179340363,
+      "num_tokens": 31161783.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.7893047273159027,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.2077007293701172,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.04417986273765564,
+      "mean_token_accuracy": 0.9825236797332764,
+      "num_tokens": 31350487.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.794023334980011,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8530077934265137,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.043008115887641904,
+      "mean_token_accuracy": 0.9832422196865082,
+      "num_tokens": 31539217.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.785652619600296,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.4543156623840332,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0445009708404541,
+      "mean_token_accuracy": 0.9828064203262329,
+      "num_tokens": 31728308.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7841614365577698,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.297763466835022,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.045649731159210206,
+      "mean_token_accuracy": 0.9823782920837403,
+      "num_tokens": 31917293.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.7825099349021911,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.2237403392791748,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.047218257188797,
+      "mean_token_accuracy": 0.98156818151474,
+      "num_tokens": 32106628.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.7791289031505585,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8094402551651001,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.04178315699100495,
+      "mean_token_accuracy": 0.9846469461917877,
+      "num_tokens": 32295183.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.7743105947971344,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.3677793741226196,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.04540807008743286,
+      "mean_token_accuracy": 0.9827918827533721,
+      "num_tokens": 32484143.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.7804521083831787,
+      "epoch": 2.328398384925976,
+      "grad_norm": 0.8883333802223206,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.04598270654678345,
+      "mean_token_accuracy": 0.9823103487491608,
+      "num_tokens": 32673144.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.7760683119297027,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.989972710609436,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.043702703714370725,
+      "mean_token_accuracy": 0.9831363618373871,
+      "num_tokens": 32862142.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7726980447769165,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.0461052656173706,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.044418123364448545,
+      "mean_token_accuracy": 0.9821757376194,
+      "num_tokens": 33051382.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.7822885036468505,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.9027822017669678,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.04624638557434082,
+      "mean_token_accuracy": 0.9815774321556091,
+      "num_tokens": 33239729.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.7830030500888825,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.1743420362472534,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.04747769832611084,
+      "mean_token_accuracy": 0.9812638640403748,
+      "num_tokens": 33428615.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.7933187901973724,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 0.9334444403648376,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.047731178998947146,
+      "mean_token_accuracy": 0.9820823550224305,
+      "num_tokens": 33617331.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.7904782056808471,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.9396253824234009,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.04432434141635895,
+      "mean_token_accuracy": 0.9830163478851318,
+      "num_tokens": 33805978.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.7811533212661743,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 0.9742764234542847,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.046493279933929446,
+      "mean_token_accuracy": 0.9818211495876312,
+      "num_tokens": 33994957.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7848010778427124,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0378986597061157,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0454249233007431,
+      "mean_token_accuracy": 0.9817994356155395,
+      "num_tokens": 34184297.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.7754390180110932,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.1023566722869873,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.04380808770656586,
+      "mean_token_accuracy": 0.9829573810100556,
+      "num_tokens": 34373551.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.7767188012599945,
+      "epoch": 2.462987886944818,
+      "grad_norm": 0.8460658192634583,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04502716064453125,
+      "mean_token_accuracy": 0.9825617015361786,
+      "num_tokens": 34562169.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.7757977426052094,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.1013047695159912,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.045161890983581546,
+      "mean_token_accuracy": 0.9821914732456207,
+      "num_tokens": 34750863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.785107308626175,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.9510554671287537,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.045189899206161496,
+      "mean_token_accuracy": 0.9822288691997528,
+      "num_tokens": 34939528.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.7849742233753204,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0994503498077393,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.04333080947399139,
+      "mean_token_accuracy": 0.9830258965492249,
+      "num_tokens": 35128334.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.7875016331672668,
+      "epoch": 2.516823687752355,
+      "grad_norm": 0.9202979803085327,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.04838588833808899,
+      "mean_token_accuracy": 0.9815591931343078,
+      "num_tokens": 35317067.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.7813792943954467,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.9036716222763062,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.045662492513656616,
+      "mean_token_accuracy": 0.9817738354206085,
+      "num_tokens": 35506291.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.777835363149643,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.853438675403595,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.044686585664749146,
+      "mean_token_accuracy": 0.98329097032547,
+      "num_tokens": 35695141.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.7677761793136597,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.0699759721755981,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.045490306615829465,
+      "mean_token_accuracy": 0.9826831936836242,
+      "num_tokens": 35884173.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.7756118893623352,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.0356824398040771,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.0467751681804657,
+      "mean_token_accuracy": 0.9823338150978088,
+      "num_tokens": 36073241.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.7775443732738495,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.1619561910629272,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.04741084575653076,
+      "mean_token_accuracy": 0.9815192580223083,
+      "num_tokens": 36262133.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7899596333503723,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.3479284048080444,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.04588942229747772,
+      "mean_token_accuracy": 0.9821865320205688,
+      "num_tokens": 36451460.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.7832387804985046,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.6297204494476318,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.04711659252643585,
+      "mean_token_accuracy": 0.9815292179584503,
+      "num_tokens": 36640366.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.7871865928173065,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.3100138902664185,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.04919236898422241,
+      "mean_token_accuracy": 0.9809704780578613,
+      "num_tokens": 36829017.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.7844007611274719,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.2907201051712036,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.047159326076507566,
+      "mean_token_accuracy": 0.9816433548927307,
+      "num_tokens": 37018034.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.7885129511356354,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.1324766874313354,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.04420983195304871,
+      "mean_token_accuracy": 0.9828807294368744,
+      "num_tokens": 37206768.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.7761563122272491,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.9269904494285583,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.04585547745227814,
+      "mean_token_accuracy": 0.9823735177516937,
+      "num_tokens": 37396010.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7831155300140381,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5890840291976929,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.046450641751289365,
+      "mean_token_accuracy": 0.9817734777927398,
+      "num_tokens": 37584809.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.7729385912418365,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.4823863506317139,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.04653529226779938,
+      "mean_token_accuracy": 0.9815726220607758,
+      "num_tokens": 37773869.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.7752901673316955,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.9423630237579346,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.04599723219871521,
+      "mean_token_accuracy": 0.9816120624542236,
+      "num_tokens": 37962642.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.7750158667564392,
+      "epoch": 2.718707940780619,
+      "grad_norm": 0.9398621320724487,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.04729728400707245,
+      "mean_token_accuracy": 0.9816868543624878,
+      "num_tokens": 38151325.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.7693424999713898,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3207950592041016,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.04474019706249237,
+      "mean_token_accuracy": 0.9830808460712432,
+      "num_tokens": 38340719.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.7750859558582306,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.070200800895691,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.045914703607559205,
+      "mean_token_accuracy": 0.9820488631725312,
+      "num_tokens": 38529293.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7697387218475342,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.0545568466186523,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.04357335567474365,
+      "mean_token_accuracy": 0.9827888309955597,
+      "num_tokens": 38718464.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.7682756781578064,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.0945985317230225,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.0429091215133667,
+      "mean_token_accuracy": 0.9832668721675872,
+      "num_tokens": 38907573.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.7804523587226868,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8521431684494019,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.04662351012229919,
+      "mean_token_accuracy": 0.981964933872223,
+      "num_tokens": 39096598.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.7786886096000671,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.09945809841156,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.050166720151901247,
+      "mean_token_accuracy": 0.9816301107406616,
+      "num_tokens": 39285069.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.7790449261665344,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.1673868894577026,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.048508059978485105,
+      "mean_token_accuracy": 0.9809533059597015,
+      "num_tokens": 39474252.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.7774537563323974,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.0648490190505981,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.04540320634841919,
+      "mean_token_accuracy": 0.9822351098060608,
+      "num_tokens": 39663262.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7683943033218383,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.0106912851333618,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.04466369152069092,
+      "mean_token_accuracy": 0.9818881273269653,
+      "num_tokens": 39852133.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.7642774760723114,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0109530687332153,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.045978450775146486,
+      "mean_token_accuracy": 0.9819821894168854,
+      "num_tokens": 40041679.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.7647961139678955,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.9050237536430359,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.045694130659103396,
+      "mean_token_accuracy": 0.982342666387558,
+      "num_tokens": 40230019.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.7589850842952728,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.8879176378250122,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.046186423301696776,
+      "mean_token_accuracy": 0.9818771004676818,
+      "num_tokens": 40418932.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.765201997756958,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 0.9103795886039734,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.04475873112678528,
+      "mean_token_accuracy": 0.9821203172206878,
+      "num_tokens": 40607792.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.7653619945049286,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.0566699504852295,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.04391545057296753,
+      "mean_token_accuracy": 0.9828326046466828,
+      "num_tokens": 40797122.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.763550215959549,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.1280677318572998,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.045582184195518495,
+      "mean_token_accuracy": 0.9819066524505615,
+      "num_tokens": 40985689.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.7749062657356263,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.110323429107666,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.04598402082920074,
+      "mean_token_accuracy": 0.9815545618534088,
+      "num_tokens": 41174743.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.7805372893810272,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.0649195909500122,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.04705569744110107,
+      "mean_token_accuracy": 0.9815925776958465,
+      "num_tokens": 41363160.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.7711063683032989,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.1459592580795288,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.045355689525604245,
+      "mean_token_accuracy": 0.9832463562488556,
+      "num_tokens": 41551932.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.7747956931591033,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.1602799892425537,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.04494523704051971,
+      "mean_token_accuracy": 0.9819566667079925,
+      "num_tokens": 41740544.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.7584574401378632,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.054417371749878,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.04810510575771332,
+      "mean_token_accuracy": 0.9811540365219116,
+      "num_tokens": 41929344.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.7753648416251894,
+      "eval_loss": 0.07181113958358765,
+      "eval_mean_token_accuracy": 0.9716544128527307,
+      "eval_num_tokens": 42099366.0,
+      "eval_runtime": 12.748,
+      "eval_samples_per_second": 392.22,
+      "eval_steps_per_second": 12.316,
+      "step": 2229
+    },
+    {
+      "entropy": 0.7669954419136047,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.8211690783500671,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.04577431380748749,
+      "mean_token_accuracy": 0.9823363184928894,
+      "num_tokens": 42118188.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.7600219964981079,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.0873607397079468,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.032501548528671265,
+      "mean_token_accuracy": 0.9881355822086334,
+      "num_tokens": 42306598.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.7441959500312805,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.1175898313522339,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.031047609448432923,
+      "mean_token_accuracy": 0.9891582727432251,
+      "num_tokens": 42495250.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.7364335536956788,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.987751841545105,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.028629058599472047,
+      "mean_token_accuracy": 0.9894859492778778,
+      "num_tokens": 42684326.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.735680228471756,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.194305658340454,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.029283303022384643,
+      "mean_token_accuracy": 0.9887769162654877,
+      "num_tokens": 42873221.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.7342639684677124,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.254442811012268,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.0302200973033905,
+      "mean_token_accuracy": 0.988795381784439,
+      "num_tokens": 43062486.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.7391974747180938,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.2648969888687134,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.0316451758146286,
+      "mean_token_accuracy": 0.9879588901996612,
+      "num_tokens": 43251113.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.7401826322078705,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.0483295917510986,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.030878132581710814,
+      "mean_token_accuracy": 0.9886977672576904,
+      "num_tokens": 43440152.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.732647967338562,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.1125447750091553,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.02838427424430847,
+      "mean_token_accuracy": 0.9890213429927825,
+      "num_tokens": 43629353.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.7314932703971863,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.0979042053222656,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.0315592348575592,
+      "mean_token_accuracy": 0.9888832688331604,
+      "num_tokens": 43818981.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.7305827140808105,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9610060453414917,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.03112468421459198,
+      "mean_token_accuracy": 0.9890477001667023,
+      "num_tokens": 44008430.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.7333625435829163,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.0625078678131104,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.03105958104133606,
+      "mean_token_accuracy": 0.9884848177433014,
+      "num_tokens": 44197264.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.7373903751373291,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.1683465242385864,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.031627464294433597,
+      "mean_token_accuracy": 0.9879396498203278,
+      "num_tokens": 44386205.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.7391358911991119,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.05154287815094,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.03118889331817627,
+      "mean_token_accuracy": 0.9880312740802765,
+      "num_tokens": 44575099.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.7351223647594451,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.9531214237213135,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.030980610847473146,
+      "mean_token_accuracy": 0.9882127285003662,
+      "num_tokens": 44763965.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.7361315310001373,
+      "epoch": 3.203230148048452,
+      "grad_norm": 0.8818249702453613,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.029705989360809325,
+      "mean_token_accuracy": 0.9894622683525085,
+      "num_tokens": 44952983.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.7356473922729492,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 0.9263529777526855,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.03202669024467468,
+      "mean_token_accuracy": 0.9877129793167114,
+      "num_tokens": 45142080.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.7326373934745789,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.0517516136169434,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.031153690814971925,
+      "mean_token_accuracy": 0.9891136825084687,
+      "num_tokens": 45330769.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.7348711133003235,
+      "epoch": 3.243606998654105,
+      "grad_norm": 0.8774152994155884,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030574262142181396,
+      "mean_token_accuracy": 0.9888435959815979,
+      "num_tokens": 45519217.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.7379967510700226,
+      "epoch": 3.257065948855989,
+      "grad_norm": 0.9164374470710754,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.03096820116043091,
+      "mean_token_accuracy": 0.9884447932243348,
+      "num_tokens": 45708467.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.7369260907173156,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.3184924125671387,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03082413077354431,
+      "mean_token_accuracy": 0.9888757109642029,
+      "num_tokens": 45897424.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.7366503119468689,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.9073860049247742,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.03220943808555603,
+      "mean_token_accuracy": 0.9877902865409851,
+      "num_tokens": 46086351.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.7341453731060028,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.0940542221069336,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.03329495787620544,
+      "mean_token_accuracy": 0.9877542018890381,
+      "num_tokens": 46275394.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.7312689840793609,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 0.9651952385902405,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.031667912006378175,
+      "mean_token_accuracy": 0.9881450653076171,
+      "num_tokens": 46463821.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.7228557825088501,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 0.971441924571991,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.030608028173446655,
+      "mean_token_accuracy": 0.9882231116294861,
+      "num_tokens": 46653025.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.723542982339859,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.0060454607009888,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.030545371770858764,
+      "mean_token_accuracy": 0.9879125893115998,
+      "num_tokens": 46842068.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.7321009397506714,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.0572997331619263,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.032070663571357724,
+      "mean_token_accuracy": 0.9878279268741608,
+      "num_tokens": 47031045.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.735515958070755,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.9835748672485352,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.03338083028793335,
+      "mean_token_accuracy": 0.9877255499362946,
+      "num_tokens": 47219625.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.7396618247032165,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 0.9655317068099976,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.03589712679386139,
+      "mean_token_accuracy": 0.9869556546211242,
+      "num_tokens": 47408311.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.7358698546886444,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9225422143936157,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.02914820909500122,
+      "mean_token_accuracy": 0.9886718571186066,
+      "num_tokens": 47597297.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.7338236391544342,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.8253662586212158,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.02996352016925812,
+      "mean_token_accuracy": 0.9891344487667084,
+      "num_tokens": 47785737.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.7320856511592865,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.0716700553894043,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.032322123646736145,
+      "mean_token_accuracy": 0.9877110660076142,
+      "num_tokens": 47974241.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.7293457925319672,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.0915459394454956,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.031131434440612792,
+      "mean_token_accuracy": 0.9885926306247711,
+      "num_tokens": 48162582.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.7207926452159882,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0700562000274658,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03113105595111847,
+      "mean_token_accuracy": 0.9880475521087646,
+      "num_tokens": 48351627.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.7178894281387329,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.1952542066574097,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03204495906829834,
+      "mean_token_accuracy": 0.9880682706832886,
+      "num_tokens": 48540171.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.7166945099830627,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0920100212097168,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.03096899390220642,
+      "mean_token_accuracy": 0.9884315967559815,
+      "num_tokens": 48729737.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.7187432169914245,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.9596593976020813,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.03240742087364197,
+      "mean_token_accuracy": 0.9878904581069946,
+      "num_tokens": 48918599.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.7224443972110748,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9406276345252991,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.03253386616706848,
+      "mean_token_accuracy": 0.9873985171318054,
+      "num_tokens": 49107385.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.7226529896259308,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.2169438600540161,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.03282734453678131,
+      "mean_token_accuracy": 0.9876641929149628,
+      "num_tokens": 49295969.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.7263983964920044,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8808715343475342,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.030802983045578002,
+      "mean_token_accuracy": 0.9885481476783753,
+      "num_tokens": 49484706.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.7291677474975586,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2453594207763672,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.03403027057647705,
+      "mean_token_accuracy": 0.987172520160675,
+      "num_tokens": 49673484.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.7396404981613159,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.9215838313102722,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.03405522108078003,
+      "mean_token_accuracy": 0.9871597826480866,
+      "num_tokens": 49862087.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.7382308602333069,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0659712553024292,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.032089871168136594,
+      "mean_token_accuracy": 0.9874629497528076,
+      "num_tokens": 50051379.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.7337264895439148,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 0.9307538270950317,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.032426226139068606,
+      "mean_token_accuracy": 0.9879062473773956,
+      "num_tokens": 50239986.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.7381527841091156,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.1867440938949585,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.03276415467262268,
+      "mean_token_accuracy": 0.9878055810928345,
+      "num_tokens": 50428872.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.7337838113307953,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.9655186533927917,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.03141146302223206,
+      "mean_token_accuracy": 0.9882215678691864,
+      "num_tokens": 50618182.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.7354952573776246,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8848851919174194,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.03387039303779602,
+      "mean_token_accuracy": 0.9871303081512451,
+      "num_tokens": 50806430.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.7329667210578918,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 0.9976673722267151,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.03129216730594635,
+      "mean_token_accuracy": 0.9882283449172974,
+      "num_tokens": 50995500.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.7277809977531433,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.2560697793960571,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.03357822299003601,
+      "mean_token_accuracy": 0.9876434028148651,
+      "num_tokens": 51184436.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.727894538640976,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.1099125146865845,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.032482534646987915,
+      "mean_token_accuracy": 0.9878577649593353,
+      "num_tokens": 51373238.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.7318481564521789,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1980950832366943,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.03284388482570648,
+      "mean_token_accuracy": 0.9874962747097016,
+      "num_tokens": 51561631.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.7219811856746674,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 0.8730921745300293,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.030655479431152342,
+      "mean_token_accuracy": 0.9877778172492981,
+      "num_tokens": 51750845.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.7289035737514495,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.1195666790008545,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.03155539035797119,
+      "mean_token_accuracy": 0.9877057671546936,
+      "num_tokens": 51940009.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.7332375526428223,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9390997886657715,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.031307452917099,
+      "mean_token_accuracy": 0.987467783689499,
+      "num_tokens": 52128442.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.7293307244777679,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.0272324085235596,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.03318062424659729,
+      "mean_token_accuracy": 0.9873259484767913,
+      "num_tokens": 52317152.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.7320701718330384,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9884606003761292,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03461401164531708,
+      "mean_token_accuracy": 0.9867411673069,
+      "num_tokens": 52506106.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.7248260796070098,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.879682719707489,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.03406862616539001,
+      "mean_token_accuracy": 0.9872734069824218,
+      "num_tokens": 52695500.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.7237857282161713,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9409974217414856,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.029435336589813232,
+      "mean_token_accuracy": 0.9891225278377533,
+      "num_tokens": 52884940.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.7237291693687439,
+      "epoch": 3.781965006729475,
+      "grad_norm": 0.9532694220542908,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.030907022953033447,
+      "mean_token_accuracy": 0.9881488382816315,
+      "num_tokens": 53073909.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.7252696454524994,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 0.9157902002334595,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.032040882110595706,
+      "mean_token_accuracy": 0.9880847573280335,
+      "num_tokens": 53262962.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.7237608432769775,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.1450515985488892,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.03199286162853241,
+      "mean_token_accuracy": 0.987687611579895,
+      "num_tokens": 53451982.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.7237064421176911,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.9945085644721985,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.03347462117671966,
+      "mean_token_accuracy": 0.9875530540943146,
+      "num_tokens": 53640737.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.7271939218044281,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.9351736903190613,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.03140422701835632,
+      "mean_token_accuracy": 0.9877301037311554,
+      "num_tokens": 53829142.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.7265262603759766,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.1019784212112427,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.030748724937438965,
+      "mean_token_accuracy": 0.9881740033626556,
+      "num_tokens": 54018331.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.7270502150058746,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 0.9153032898902893,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.033472740650177,
+      "mean_token_accuracy": 0.9876999020576477,
+      "num_tokens": 54206935.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.7345028996467591,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.0144257545471191,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.033550959825515744,
+      "mean_token_accuracy": 0.9874906241893768,
+      "num_tokens": 54395443.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.7274066030979156,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0221407413482666,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.033388751745224,
+      "mean_token_accuracy": 0.9870682001113892,
+      "num_tokens": 54584552.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.7293415367603302,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.019165277481079,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.03568689227104187,
+      "mean_token_accuracy": 0.9861597299575806,
+      "num_tokens": 54773705.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.7281844198703766,
+      "epoch": 3.916554508748318,
+      "grad_norm": 0.933020830154419,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.030142486095428467,
+      "mean_token_accuracy": 0.988915103673935,
+      "num_tokens": 54962152.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.7242733418941498,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.9429674744606018,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.0337733268737793,
+      "mean_token_accuracy": 0.987263411283493,
+      "num_tokens": 55151179.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.7306612014770508,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.9022014737129211,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.03387616574764252,
+      "mean_token_accuracy": 0.9871586680412292,
+      "num_tokens": 55339658.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.7246532738208771,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2193421125411987,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.03193953037261963,
+      "mean_token_accuracy": 0.9879468739032745,
+      "num_tokens": 55528467.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.7268587052822113,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.2917358875274658,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.03584722280502319,
+      "mean_token_accuracy": 0.9867104470729828,
+      "num_tokens": 55717308.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.735340416431427,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.2229385375976562,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.03506708741188049,
+      "mean_token_accuracy": 0.9864643573760986,
+      "num_tokens": 55906059.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.7337038695812226,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.0747122764587402,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.03241217732429504,
+      "mean_token_accuracy": 0.9876143455505371,
+      "num_tokens": 56094606.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.7333687073106219,
+      "eval_loss": 0.07489899545907974,
+      "eval_mean_token_accuracy": 0.9719084019114257,
+      "eval_num_tokens": 56132350.0,
+      "eval_runtime": 12.7823,
+      "eval_samples_per_second": 391.166,
+      "eval_steps_per_second": 12.283,
+      "step": 2972
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.594439273226699e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-2972/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-2972/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-3715/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-3715/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-3715/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..7f244aea3f467ff5a8bc0e2db26ecf9a603dde7a
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0401f2649f2fa59baca219159da78adab583c2e555a7a7c882fed3e2fb150f86
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-3715/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-3715/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-3715/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-3715/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..5e0ffd4596eca7e8238dace58bb7c3e3911144da
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/trainer_state.json
@@ -0,0 +1,3799 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 3715,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    },
+    {
+      "entropy": 0.8332668900489807,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9914855360984802,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.06607928276062011,
+      "mean_token_accuracy": 0.974932211637497,
+      "num_tokens": 14165567.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8149146616458893,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.3080861568450928,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.057932257652282715,
+      "mean_token_accuracy": 0.9770712852478027,
+      "num_tokens": 14353912.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8197620987892151,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.0898637771606445,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.06010057330131531,
+      "mean_token_accuracy": 0.9758293390274048,
+      "num_tokens": 14542288.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8183245182037353,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.3120660781860352,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.0610861599445343,
+      "mean_token_accuracy": 0.9764986515045166,
+      "num_tokens": 14731028.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8135185301303863,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.138473391532898,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.06268800497055053,
+      "mean_token_accuracy": 0.9752912580966949,
+      "num_tokens": 14919682.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8142645537853241,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.1258258819580078,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.06271481513977051,
+      "mean_token_accuracy": 0.9757774949073792,
+      "num_tokens": 15107904.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8170483231544494,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.007297396659851,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.06338523030281067,
+      "mean_token_accuracy": 0.9754154622554779,
+      "num_tokens": 15297115.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8201254069805145,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 0.8240306973457336,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.06355815529823303,
+      "mean_token_accuracy": 0.9752808332443237,
+      "num_tokens": 15485921.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8140238881111145,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.0350368022918701,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.05969444513320923,
+      "mean_token_accuracy": 0.977089112997055,
+      "num_tokens": 15674824.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8133325934410095,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.0544878244400024,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.06062774658203125,
+      "mean_token_accuracy": 0.976290100812912,
+      "num_tokens": 15863526.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8185161769390106,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.1076316833496094,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.061567097902297974,
+      "mean_token_accuracy": 0.9755526125431061,
+      "num_tokens": 16051685.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8176139056682586,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.0057616233825684,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.06227490901947021,
+      "mean_token_accuracy": 0.9758114695549012,
+      "num_tokens": 16240451.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8217637002468109,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.97276771068573,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.06410331726074218,
+      "mean_token_accuracy": 0.9754030883312226,
+      "num_tokens": 16428961.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8141795754432678,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.0135555267333984,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.06253207921981811,
+      "mean_token_accuracy": 0.9760490536689759,
+      "num_tokens": 16617327.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8167653739452362,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2272467613220215,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.060455435514450075,
+      "mean_token_accuracy": 0.9758628964424133,
+      "num_tokens": 16806414.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8155323445796967,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 0.9328635334968567,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.06302794814109802,
+      "mean_token_accuracy": 0.9745101630687714,
+      "num_tokens": 16995016.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.818474805355072,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.1651655435562134,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.06043210029602051,
+      "mean_token_accuracy": 0.9766487181186676,
+      "num_tokens": 17184250.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8140560865402222,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.1005650758743286,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.06131819486618042,
+      "mean_token_accuracy": 0.9757712185382843,
+      "num_tokens": 17373075.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8204843640327454,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.0825532674789429,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.06296311616897583,
+      "mean_token_accuracy": 0.9745951414108276,
+      "num_tokens": 17561454.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8237834334373474,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.9842743873596191,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.061483919620513916,
+      "mean_token_accuracy": 0.9754831731319428,
+      "num_tokens": 17750758.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8165915131568908,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.0179396867752075,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.06237881183624268,
+      "mean_token_accuracy": 0.9750245690345765,
+      "num_tokens": 17940126.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8329708755016327,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0732359886169434,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.0644665539264679,
+      "mean_token_accuracy": 0.9743888020515442,
+      "num_tokens": 18129000.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8210074126720428,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.3696036338806152,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.06320792436599731,
+      "mean_token_accuracy": 0.9751390159130097,
+      "num_tokens": 18318232.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8248398661613464,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.327525019645691,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.06277769804000854,
+      "mean_token_accuracy": 0.9752413153648376,
+      "num_tokens": 18507403.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8284746050834656,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.044022560119629,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.06162315607070923,
+      "mean_token_accuracy": 0.9760516047477722,
+      "num_tokens": 18696242.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8233858644962311,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.062048077583313,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.059620130062103274,
+      "mean_token_accuracy": 0.9755307257175445,
+      "num_tokens": 18885199.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8236705601215363,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.2024412155151367,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.06304935216903687,
+      "mean_token_accuracy": 0.975091153383255,
+      "num_tokens": 19074028.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.828409218788147,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.1684212684631348,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.06451238393783569,
+      "mean_token_accuracy": 0.9743404746055603,
+      "num_tokens": 19262295.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8214354634284973,
+      "epoch": 1.386271870794078,
+      "grad_norm": 0.8512521386146545,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.06042641997337341,
+      "mean_token_accuracy": 0.9755558848381043,
+      "num_tokens": 19451154.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8223241448402405,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 0.8546568155288696,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.062134528160095216,
+      "mean_token_accuracy": 0.9750089883804322,
+      "num_tokens": 19639827.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.8217969954013824,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 0.972831666469574,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.060803139209747316,
+      "mean_token_accuracy": 0.9760692238807678,
+      "num_tokens": 19828174.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.8144341170787811,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1216644048690796,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.06458876132965088,
+      "mean_token_accuracy": 0.9738282442092896,
+      "num_tokens": 20016700.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.8168790876865387,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.0821806192398071,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.06258658170700074,
+      "mean_token_accuracy": 0.9749014675617218,
+      "num_tokens": 20205823.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8146390020847321,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.0067790746688843,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.06509006023406982,
+      "mean_token_accuracy": 0.9737810134887696,
+      "num_tokens": 20394611.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8224801063537598,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.0583593845367432,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.06343590021133423,
+      "mean_token_accuracy": 0.9750119090080261,
+      "num_tokens": 20583252.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.8083258986473083,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.8305535316467285,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.06573111414909363,
+      "mean_token_accuracy": 0.9736428141593934,
+      "num_tokens": 20772107.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8081986546516419,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5895942449569702,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.06376024484634399,
+      "mean_token_accuracy": 0.9751200795173645,
+      "num_tokens": 20960494.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.8026683986186981,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 0.9962654113769531,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.06282572746276856,
+      "mean_token_accuracy": 0.9753849327564239,
+      "num_tokens": 21149763.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8030014991760254,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9083836078643799,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.06473478078842163,
+      "mean_token_accuracy": 0.9747069597244262,
+      "num_tokens": 21338690.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.8008662164211273,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.097593069076538,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.05911685228347778,
+      "mean_token_accuracy": 0.9767139732837677,
+      "num_tokens": 21527420.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7981986284255982,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.1673516035079956,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.05957244038581848,
+      "mean_token_accuracy": 0.9761911332607269,
+      "num_tokens": 21716791.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.8004718601703644,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 0.7955246567726135,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.06378913521766663,
+      "mean_token_accuracy": 0.9747395157814026,
+      "num_tokens": 21905487.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.8039821267127991,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.3141396045684814,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.06209801435470581,
+      "mean_token_accuracy": 0.9754719436168671,
+      "num_tokens": 22094436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.8051789462566376,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.2227814197540283,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.06261310577392579,
+      "mean_token_accuracy": 0.9753049433231353,
+      "num_tokens": 22283243.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7941994190216064,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.071070671081543,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.05962263345718384,
+      "mean_token_accuracy": 0.9763956189155578,
+      "num_tokens": 22472084.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7937953472137451,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.0329782962799072,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.060611891746521,
+      "mean_token_accuracy": 0.9758659780025483,
+      "num_tokens": 22661614.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7913113713264466,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.032772183418274,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.06002308130264282,
+      "mean_token_accuracy": 0.9758916199207306,
+      "num_tokens": 22850634.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.7909658491611481,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3234598636627197,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.06296972036361695,
+      "mean_token_accuracy": 0.9741117715835571,
+      "num_tokens": 23039402.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7880761086940765,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.9563435316085815,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.06209538578987121,
+      "mean_token_accuracy": 0.9750530481338501,
+      "num_tokens": 23228363.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7845663070678711,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 0.8875390291213989,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.06272958517074585,
+      "mean_token_accuracy": 0.9751476228237153,
+      "num_tokens": 23417469.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7928634822368622,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.0629489421844482,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.060286092758178714,
+      "mean_token_accuracy": 0.976036274433136,
+      "num_tokens": 23606274.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7870154738426208,
+      "epoch": 1.695827725437416,
+      "grad_norm": 0.9164250493049622,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.06263296008110046,
+      "mean_token_accuracy": 0.9752088189125061,
+      "num_tokens": 23795255.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.7948406875133515,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.0909371376037598,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.06165047883987427,
+      "mean_token_accuracy": 0.9764420628547669,
+      "num_tokens": 23984534.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7862435758113862,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.1839547157287598,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.0640158474445343,
+      "mean_token_accuracy": 0.9741048216819763,
+      "num_tokens": 24173632.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.7887867331504822,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.0730741024017334,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.06260218620300292,
+      "mean_token_accuracy": 0.974741768836975,
+      "num_tokens": 24362671.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7896574556827545,
+      "epoch": 1.749663526244953,
+      "grad_norm": 0.9559605121612549,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.061297619342803956,
+      "mean_token_accuracy": 0.9752223670482636,
+      "num_tokens": 24551825.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.7902800619602204,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.914057195186615,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.06320110559463502,
+      "mean_token_accuracy": 0.9749482691287994,
+      "num_tokens": 24740867.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.799742591381073,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.06534289717674255,
+      "mean_token_accuracy": 0.9741291284561158,
+      "num_tokens": 24930132.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7974063038825989,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1188615560531616,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.060908573865890506,
+      "mean_token_accuracy": 0.9765798091888428,
+      "num_tokens": 25119678.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.786439061164856,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9185847640037537,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.06158917546272278,
+      "mean_token_accuracy": 0.9760335922241211,
+      "num_tokens": 25308876.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7888528347015381,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.1948838233947754,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.060635948181152345,
+      "mean_token_accuracy": 0.976032954454422,
+      "num_tokens": 25498102.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7968044698238372,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.4128663539886475,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.06448977589607238,
+      "mean_token_accuracy": 0.9741183638572692,
+      "num_tokens": 25686747.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7967679858207702,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.9797796010971069,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.06342286467552186,
+      "mean_token_accuracy": 0.9752787709236145,
+      "num_tokens": 25875368.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7932188332080841,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.0549253225326538,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.060981249809265135,
+      "mean_token_accuracy": 0.976200807094574,
+      "num_tokens": 26064480.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7899552762508393,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.1172103881835938,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.06555815935134887,
+      "mean_token_accuracy": 0.9736676931381225,
+      "num_tokens": 26253619.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.8019726693630218,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.94905686378479,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.0663321852684021,
+      "mean_token_accuracy": 0.9739456713199616,
+      "num_tokens": 26442736.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7939825117588043,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 0.9922711253166199,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05981417298316956,
+      "mean_token_accuracy": 0.9758030951023102,
+      "num_tokens": 26631525.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8085412085056305,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4970781803131104,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.06708773374557495,
+      "mean_token_accuracy": 0.9737519145011901,
+      "num_tokens": 26820483.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7950678527355194,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.7868127226829529,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.061734098196029666,
+      "mean_token_accuracy": 0.9747878730297088,
+      "num_tokens": 27009393.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7850766360759736,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.1386486291885376,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.05714603662490845,
+      "mean_token_accuracy": 0.9769004642963409,
+      "num_tokens": 27198251.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7933461844921113,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.0092092752456665,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.06240262985229492,
+      "mean_token_accuracy": 0.9742367327213287,
+      "num_tokens": 27387231.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8044621467590332,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.2477015256881714,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0642085313796997,
+      "mean_token_accuracy": 0.9748865723609924,
+      "num_tokens": 27576251.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7975574374198914,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.1167609691619873,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.06382874846458435,
+      "mean_token_accuracy": 0.9748350441455841,
+      "num_tokens": 27764560.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.8090806484222413,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 0.9714348912239075,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.06401748657226562,
+      "mean_token_accuracy": 0.9741845309734345,
+      "num_tokens": 27953532.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.8104567945383157,
+      "eval_loss": 0.07282773405313492,
+      "eval_mean_token_accuracy": 0.9701232534305305,
+      "eval_num_tokens": 28066151.0,
+      "eval_runtime": 12.7852,
+      "eval_samples_per_second": 391.078,
+      "eval_steps_per_second": 12.28,
+      "step": 1486
+    },
+    {
+      "entropy": 0.8090688765048981,
+      "epoch": 2.005383580080754,
+      "grad_norm": 0.9886816143989563,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.06012396812438965,
+      "mean_token_accuracy": 0.9764885842800141,
+      "num_tokens": 28141848.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.7821709334850311,
+      "epoch": 2.018842530282638,
+      "grad_norm": 0.9049969911575317,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.04819132387638092,
+      "mean_token_accuracy": 0.9816783249378205,
+      "num_tokens": 28330784.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.7845205307006836,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0231750011444092,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.04539612233638764,
+      "mean_token_accuracy": 0.9818619966506958,
+      "num_tokens": 28519580.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.7817754447460175,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.5244711637496948,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0429688960313797,
+      "mean_token_accuracy": 0.9834935545921326,
+      "num_tokens": 28707486.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.7733526527881622,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.0685060024261475,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.045852530002594,
+      "mean_token_accuracy": 0.9817728459835052,
+      "num_tokens": 28896759.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.7815343141555786,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.075853705406189,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.04565088748931885,
+      "mean_token_accuracy": 0.9823914349079133,
+      "num_tokens": 29085097.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.7751398742198944,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2573742866516113,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.04465666115283966,
+      "mean_token_accuracy": 0.9829874813556672,
+      "num_tokens": 29273868.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.7737197399139404,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 0.8681219220161438,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.042188242077827454,
+      "mean_token_accuracy": 0.9837059378623962,
+      "num_tokens": 29462985.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.7823441684246063,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.3503504991531372,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.04393318593502045,
+      "mean_token_accuracy": 0.9827627122402192,
+      "num_tokens": 29651156.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.7796198308467865,
+      "epoch": 2.126514131897712,
+      "grad_norm": 0.7799349427223206,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.04367418885231018,
+      "mean_token_accuracy": 0.9837401032447814,
+      "num_tokens": 29839789.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.7767675697803498,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.375465750694275,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.04427264630794525,
+      "mean_token_accuracy": 0.9832019031047821,
+      "num_tokens": 30028957.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.7912136912345886,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.9683088660240173,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.04428936243057251,
+      "mean_token_accuracy": 0.9832306385040284,
+      "num_tokens": 30217841.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.790750366449356,
+      "epoch": 2.166890982503365,
+      "grad_norm": 0.9393274188041687,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.04529332816600799,
+      "mean_token_accuracy": 0.9832263946533203,
+      "num_tokens": 30406540.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.7816023886203766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.1822642087936401,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.04192846417427063,
+      "mean_token_accuracy": 0.9846727311611175,
+      "num_tokens": 30595568.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7754982054233551,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.1961082220077515,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.043791168928146364,
+      "mean_token_accuracy": 0.9830836653709412,
+      "num_tokens": 30784300.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.7756961703300476,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 0.9719674587249756,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.04410521686077118,
+      "mean_token_accuracy": 0.9837198793888092,
+      "num_tokens": 30972950.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.7833336114883422,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.0175098180770874,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.044813218712806704,
+      "mean_token_accuracy": 0.9829723179340363,
+      "num_tokens": 31161783.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.7893047273159027,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.2077007293701172,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.04417986273765564,
+      "mean_token_accuracy": 0.9825236797332764,
+      "num_tokens": 31350487.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.794023334980011,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8530077934265137,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.043008115887641904,
+      "mean_token_accuracy": 0.9832422196865082,
+      "num_tokens": 31539217.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.785652619600296,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.4543156623840332,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0445009708404541,
+      "mean_token_accuracy": 0.9828064203262329,
+      "num_tokens": 31728308.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7841614365577698,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.297763466835022,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.045649731159210206,
+      "mean_token_accuracy": 0.9823782920837403,
+      "num_tokens": 31917293.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.7825099349021911,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.2237403392791748,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.047218257188797,
+      "mean_token_accuracy": 0.98156818151474,
+      "num_tokens": 32106628.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.7791289031505585,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8094402551651001,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.04178315699100495,
+      "mean_token_accuracy": 0.9846469461917877,
+      "num_tokens": 32295183.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.7743105947971344,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.3677793741226196,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.04540807008743286,
+      "mean_token_accuracy": 0.9827918827533721,
+      "num_tokens": 32484143.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.7804521083831787,
+      "epoch": 2.328398384925976,
+      "grad_norm": 0.8883333802223206,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.04598270654678345,
+      "mean_token_accuracy": 0.9823103487491608,
+      "num_tokens": 32673144.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.7760683119297027,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.989972710609436,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.043702703714370725,
+      "mean_token_accuracy": 0.9831363618373871,
+      "num_tokens": 32862142.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7726980447769165,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.0461052656173706,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.044418123364448545,
+      "mean_token_accuracy": 0.9821757376194,
+      "num_tokens": 33051382.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.7822885036468505,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.9027822017669678,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.04624638557434082,
+      "mean_token_accuracy": 0.9815774321556091,
+      "num_tokens": 33239729.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.7830030500888825,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.1743420362472534,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.04747769832611084,
+      "mean_token_accuracy": 0.9812638640403748,
+      "num_tokens": 33428615.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.7933187901973724,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 0.9334444403648376,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.047731178998947146,
+      "mean_token_accuracy": 0.9820823550224305,
+      "num_tokens": 33617331.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.7904782056808471,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.9396253824234009,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.04432434141635895,
+      "mean_token_accuracy": 0.9830163478851318,
+      "num_tokens": 33805978.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.7811533212661743,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 0.9742764234542847,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.046493279933929446,
+      "mean_token_accuracy": 0.9818211495876312,
+      "num_tokens": 33994957.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7848010778427124,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0378986597061157,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0454249233007431,
+      "mean_token_accuracy": 0.9817994356155395,
+      "num_tokens": 34184297.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.7754390180110932,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.1023566722869873,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.04380808770656586,
+      "mean_token_accuracy": 0.9829573810100556,
+      "num_tokens": 34373551.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.7767188012599945,
+      "epoch": 2.462987886944818,
+      "grad_norm": 0.8460658192634583,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04502716064453125,
+      "mean_token_accuracy": 0.9825617015361786,
+      "num_tokens": 34562169.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.7757977426052094,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.1013047695159912,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.045161890983581546,
+      "mean_token_accuracy": 0.9821914732456207,
+      "num_tokens": 34750863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.785107308626175,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.9510554671287537,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.045189899206161496,
+      "mean_token_accuracy": 0.9822288691997528,
+      "num_tokens": 34939528.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.7849742233753204,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0994503498077393,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.04333080947399139,
+      "mean_token_accuracy": 0.9830258965492249,
+      "num_tokens": 35128334.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.7875016331672668,
+      "epoch": 2.516823687752355,
+      "grad_norm": 0.9202979803085327,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.04838588833808899,
+      "mean_token_accuracy": 0.9815591931343078,
+      "num_tokens": 35317067.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.7813792943954467,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.9036716222763062,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.045662492513656616,
+      "mean_token_accuracy": 0.9817738354206085,
+      "num_tokens": 35506291.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.777835363149643,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.853438675403595,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.044686585664749146,
+      "mean_token_accuracy": 0.98329097032547,
+      "num_tokens": 35695141.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.7677761793136597,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.0699759721755981,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.045490306615829465,
+      "mean_token_accuracy": 0.9826831936836242,
+      "num_tokens": 35884173.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.7756118893623352,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.0356824398040771,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.0467751681804657,
+      "mean_token_accuracy": 0.9823338150978088,
+      "num_tokens": 36073241.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.7775443732738495,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.1619561910629272,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.04741084575653076,
+      "mean_token_accuracy": 0.9815192580223083,
+      "num_tokens": 36262133.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7899596333503723,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.3479284048080444,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.04588942229747772,
+      "mean_token_accuracy": 0.9821865320205688,
+      "num_tokens": 36451460.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.7832387804985046,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.6297204494476318,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.04711659252643585,
+      "mean_token_accuracy": 0.9815292179584503,
+      "num_tokens": 36640366.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.7871865928173065,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.3100138902664185,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.04919236898422241,
+      "mean_token_accuracy": 0.9809704780578613,
+      "num_tokens": 36829017.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.7844007611274719,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.2907201051712036,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.047159326076507566,
+      "mean_token_accuracy": 0.9816433548927307,
+      "num_tokens": 37018034.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.7885129511356354,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.1324766874313354,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.04420983195304871,
+      "mean_token_accuracy": 0.9828807294368744,
+      "num_tokens": 37206768.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.7761563122272491,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.9269904494285583,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.04585547745227814,
+      "mean_token_accuracy": 0.9823735177516937,
+      "num_tokens": 37396010.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7831155300140381,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5890840291976929,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.046450641751289365,
+      "mean_token_accuracy": 0.9817734777927398,
+      "num_tokens": 37584809.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.7729385912418365,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.4823863506317139,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.04653529226779938,
+      "mean_token_accuracy": 0.9815726220607758,
+      "num_tokens": 37773869.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.7752901673316955,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.9423630237579346,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.04599723219871521,
+      "mean_token_accuracy": 0.9816120624542236,
+      "num_tokens": 37962642.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.7750158667564392,
+      "epoch": 2.718707940780619,
+      "grad_norm": 0.9398621320724487,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.04729728400707245,
+      "mean_token_accuracy": 0.9816868543624878,
+      "num_tokens": 38151325.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.7693424999713898,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3207950592041016,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.04474019706249237,
+      "mean_token_accuracy": 0.9830808460712432,
+      "num_tokens": 38340719.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.7750859558582306,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.070200800895691,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.045914703607559205,
+      "mean_token_accuracy": 0.9820488631725312,
+      "num_tokens": 38529293.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7697387218475342,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.0545568466186523,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.04357335567474365,
+      "mean_token_accuracy": 0.9827888309955597,
+      "num_tokens": 38718464.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.7682756781578064,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.0945985317230225,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.0429091215133667,
+      "mean_token_accuracy": 0.9832668721675872,
+      "num_tokens": 38907573.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.7804523587226868,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8521431684494019,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.04662351012229919,
+      "mean_token_accuracy": 0.981964933872223,
+      "num_tokens": 39096598.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.7786886096000671,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.09945809841156,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.050166720151901247,
+      "mean_token_accuracy": 0.9816301107406616,
+      "num_tokens": 39285069.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.7790449261665344,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.1673868894577026,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.048508059978485105,
+      "mean_token_accuracy": 0.9809533059597015,
+      "num_tokens": 39474252.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.7774537563323974,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.0648490190505981,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.04540320634841919,
+      "mean_token_accuracy": 0.9822351098060608,
+      "num_tokens": 39663262.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7683943033218383,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.0106912851333618,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.04466369152069092,
+      "mean_token_accuracy": 0.9818881273269653,
+      "num_tokens": 39852133.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.7642774760723114,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0109530687332153,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.045978450775146486,
+      "mean_token_accuracy": 0.9819821894168854,
+      "num_tokens": 40041679.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.7647961139678955,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.9050237536430359,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.045694130659103396,
+      "mean_token_accuracy": 0.982342666387558,
+      "num_tokens": 40230019.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.7589850842952728,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.8879176378250122,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.046186423301696776,
+      "mean_token_accuracy": 0.9818771004676818,
+      "num_tokens": 40418932.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.765201997756958,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 0.9103795886039734,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.04475873112678528,
+      "mean_token_accuracy": 0.9821203172206878,
+      "num_tokens": 40607792.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.7653619945049286,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.0566699504852295,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.04391545057296753,
+      "mean_token_accuracy": 0.9828326046466828,
+      "num_tokens": 40797122.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.763550215959549,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.1280677318572998,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.045582184195518495,
+      "mean_token_accuracy": 0.9819066524505615,
+      "num_tokens": 40985689.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.7749062657356263,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.110323429107666,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.04598402082920074,
+      "mean_token_accuracy": 0.9815545618534088,
+      "num_tokens": 41174743.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.7805372893810272,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.0649195909500122,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.04705569744110107,
+      "mean_token_accuracy": 0.9815925776958465,
+      "num_tokens": 41363160.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.7711063683032989,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.1459592580795288,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.045355689525604245,
+      "mean_token_accuracy": 0.9832463562488556,
+      "num_tokens": 41551932.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.7747956931591033,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.1602799892425537,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.04494523704051971,
+      "mean_token_accuracy": 0.9819566667079925,
+      "num_tokens": 41740544.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.7584574401378632,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.054417371749878,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.04810510575771332,
+      "mean_token_accuracy": 0.9811540365219116,
+      "num_tokens": 41929344.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.7753648416251894,
+      "eval_loss": 0.07181113958358765,
+      "eval_mean_token_accuracy": 0.9716544128527307,
+      "eval_num_tokens": 42099366.0,
+      "eval_runtime": 12.748,
+      "eval_samples_per_second": 392.22,
+      "eval_steps_per_second": 12.316,
+      "step": 2229
+    },
+    {
+      "entropy": 0.7669954419136047,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.8211690783500671,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.04577431380748749,
+      "mean_token_accuracy": 0.9823363184928894,
+      "num_tokens": 42118188.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.7600219964981079,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.0873607397079468,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.032501548528671265,
+      "mean_token_accuracy": 0.9881355822086334,
+      "num_tokens": 42306598.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.7441959500312805,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.1175898313522339,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.031047609448432923,
+      "mean_token_accuracy": 0.9891582727432251,
+      "num_tokens": 42495250.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.7364335536956788,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.987751841545105,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.028629058599472047,
+      "mean_token_accuracy": 0.9894859492778778,
+      "num_tokens": 42684326.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.735680228471756,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.194305658340454,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.029283303022384643,
+      "mean_token_accuracy": 0.9887769162654877,
+      "num_tokens": 42873221.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.7342639684677124,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.254442811012268,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.0302200973033905,
+      "mean_token_accuracy": 0.988795381784439,
+      "num_tokens": 43062486.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.7391974747180938,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.2648969888687134,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.0316451758146286,
+      "mean_token_accuracy": 0.9879588901996612,
+      "num_tokens": 43251113.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.7401826322078705,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.0483295917510986,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.030878132581710814,
+      "mean_token_accuracy": 0.9886977672576904,
+      "num_tokens": 43440152.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.732647967338562,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.1125447750091553,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.02838427424430847,
+      "mean_token_accuracy": 0.9890213429927825,
+      "num_tokens": 43629353.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.7314932703971863,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.0979042053222656,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.0315592348575592,
+      "mean_token_accuracy": 0.9888832688331604,
+      "num_tokens": 43818981.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.7305827140808105,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9610060453414917,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.03112468421459198,
+      "mean_token_accuracy": 0.9890477001667023,
+      "num_tokens": 44008430.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.7333625435829163,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.0625078678131104,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.03105958104133606,
+      "mean_token_accuracy": 0.9884848177433014,
+      "num_tokens": 44197264.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.7373903751373291,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.1683465242385864,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.031627464294433597,
+      "mean_token_accuracy": 0.9879396498203278,
+      "num_tokens": 44386205.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.7391358911991119,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.05154287815094,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.03118889331817627,
+      "mean_token_accuracy": 0.9880312740802765,
+      "num_tokens": 44575099.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.7351223647594451,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.9531214237213135,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.030980610847473146,
+      "mean_token_accuracy": 0.9882127285003662,
+      "num_tokens": 44763965.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.7361315310001373,
+      "epoch": 3.203230148048452,
+      "grad_norm": 0.8818249702453613,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.029705989360809325,
+      "mean_token_accuracy": 0.9894622683525085,
+      "num_tokens": 44952983.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.7356473922729492,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 0.9263529777526855,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.03202669024467468,
+      "mean_token_accuracy": 0.9877129793167114,
+      "num_tokens": 45142080.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.7326373934745789,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.0517516136169434,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.031153690814971925,
+      "mean_token_accuracy": 0.9891136825084687,
+      "num_tokens": 45330769.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.7348711133003235,
+      "epoch": 3.243606998654105,
+      "grad_norm": 0.8774152994155884,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030574262142181396,
+      "mean_token_accuracy": 0.9888435959815979,
+      "num_tokens": 45519217.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.7379967510700226,
+      "epoch": 3.257065948855989,
+      "grad_norm": 0.9164374470710754,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.03096820116043091,
+      "mean_token_accuracy": 0.9884447932243348,
+      "num_tokens": 45708467.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.7369260907173156,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.3184924125671387,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03082413077354431,
+      "mean_token_accuracy": 0.9888757109642029,
+      "num_tokens": 45897424.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.7366503119468689,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.9073860049247742,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.03220943808555603,
+      "mean_token_accuracy": 0.9877902865409851,
+      "num_tokens": 46086351.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.7341453731060028,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.0940542221069336,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.03329495787620544,
+      "mean_token_accuracy": 0.9877542018890381,
+      "num_tokens": 46275394.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.7312689840793609,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 0.9651952385902405,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.031667912006378175,
+      "mean_token_accuracy": 0.9881450653076171,
+      "num_tokens": 46463821.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.7228557825088501,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 0.971441924571991,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.030608028173446655,
+      "mean_token_accuracy": 0.9882231116294861,
+      "num_tokens": 46653025.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.723542982339859,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.0060454607009888,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.030545371770858764,
+      "mean_token_accuracy": 0.9879125893115998,
+      "num_tokens": 46842068.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.7321009397506714,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.0572997331619263,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.032070663571357724,
+      "mean_token_accuracy": 0.9878279268741608,
+      "num_tokens": 47031045.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.735515958070755,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.9835748672485352,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.03338083028793335,
+      "mean_token_accuracy": 0.9877255499362946,
+      "num_tokens": 47219625.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.7396618247032165,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 0.9655317068099976,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.03589712679386139,
+      "mean_token_accuracy": 0.9869556546211242,
+      "num_tokens": 47408311.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.7358698546886444,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9225422143936157,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.02914820909500122,
+      "mean_token_accuracy": 0.9886718571186066,
+      "num_tokens": 47597297.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.7338236391544342,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.8253662586212158,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.02996352016925812,
+      "mean_token_accuracy": 0.9891344487667084,
+      "num_tokens": 47785737.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.7320856511592865,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.0716700553894043,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.032322123646736145,
+      "mean_token_accuracy": 0.9877110660076142,
+      "num_tokens": 47974241.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.7293457925319672,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.0915459394454956,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.031131434440612792,
+      "mean_token_accuracy": 0.9885926306247711,
+      "num_tokens": 48162582.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.7207926452159882,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0700562000274658,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03113105595111847,
+      "mean_token_accuracy": 0.9880475521087646,
+      "num_tokens": 48351627.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.7178894281387329,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.1952542066574097,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03204495906829834,
+      "mean_token_accuracy": 0.9880682706832886,
+      "num_tokens": 48540171.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.7166945099830627,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0920100212097168,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.03096899390220642,
+      "mean_token_accuracy": 0.9884315967559815,
+      "num_tokens": 48729737.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.7187432169914245,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.9596593976020813,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.03240742087364197,
+      "mean_token_accuracy": 0.9878904581069946,
+      "num_tokens": 48918599.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.7224443972110748,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9406276345252991,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.03253386616706848,
+      "mean_token_accuracy": 0.9873985171318054,
+      "num_tokens": 49107385.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.7226529896259308,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.2169438600540161,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.03282734453678131,
+      "mean_token_accuracy": 0.9876641929149628,
+      "num_tokens": 49295969.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.7263983964920044,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8808715343475342,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.030802983045578002,
+      "mean_token_accuracy": 0.9885481476783753,
+      "num_tokens": 49484706.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.7291677474975586,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2453594207763672,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.03403027057647705,
+      "mean_token_accuracy": 0.987172520160675,
+      "num_tokens": 49673484.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.7396404981613159,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.9215838313102722,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.03405522108078003,
+      "mean_token_accuracy": 0.9871597826480866,
+      "num_tokens": 49862087.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.7382308602333069,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0659712553024292,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.032089871168136594,
+      "mean_token_accuracy": 0.9874629497528076,
+      "num_tokens": 50051379.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.7337264895439148,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 0.9307538270950317,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.032426226139068606,
+      "mean_token_accuracy": 0.9879062473773956,
+      "num_tokens": 50239986.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.7381527841091156,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.1867440938949585,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.03276415467262268,
+      "mean_token_accuracy": 0.9878055810928345,
+      "num_tokens": 50428872.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.7337838113307953,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.9655186533927917,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.03141146302223206,
+      "mean_token_accuracy": 0.9882215678691864,
+      "num_tokens": 50618182.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.7354952573776246,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8848851919174194,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.03387039303779602,
+      "mean_token_accuracy": 0.9871303081512451,
+      "num_tokens": 50806430.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.7329667210578918,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 0.9976673722267151,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.03129216730594635,
+      "mean_token_accuracy": 0.9882283449172974,
+      "num_tokens": 50995500.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.7277809977531433,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.2560697793960571,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.03357822299003601,
+      "mean_token_accuracy": 0.9876434028148651,
+      "num_tokens": 51184436.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.727894538640976,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.1099125146865845,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.032482534646987915,
+      "mean_token_accuracy": 0.9878577649593353,
+      "num_tokens": 51373238.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.7318481564521789,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1980950832366943,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.03284388482570648,
+      "mean_token_accuracy": 0.9874962747097016,
+      "num_tokens": 51561631.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.7219811856746674,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 0.8730921745300293,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.030655479431152342,
+      "mean_token_accuracy": 0.9877778172492981,
+      "num_tokens": 51750845.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.7289035737514495,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.1195666790008545,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.03155539035797119,
+      "mean_token_accuracy": 0.9877057671546936,
+      "num_tokens": 51940009.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.7332375526428223,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9390997886657715,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.031307452917099,
+      "mean_token_accuracy": 0.987467783689499,
+      "num_tokens": 52128442.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.7293307244777679,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.0272324085235596,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.03318062424659729,
+      "mean_token_accuracy": 0.9873259484767913,
+      "num_tokens": 52317152.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.7320701718330384,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9884606003761292,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03461401164531708,
+      "mean_token_accuracy": 0.9867411673069,
+      "num_tokens": 52506106.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.7248260796070098,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.879682719707489,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.03406862616539001,
+      "mean_token_accuracy": 0.9872734069824218,
+      "num_tokens": 52695500.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.7237857282161713,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9409974217414856,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.029435336589813232,
+      "mean_token_accuracy": 0.9891225278377533,
+      "num_tokens": 52884940.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.7237291693687439,
+      "epoch": 3.781965006729475,
+      "grad_norm": 0.9532694220542908,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.030907022953033447,
+      "mean_token_accuracy": 0.9881488382816315,
+      "num_tokens": 53073909.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.7252696454524994,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 0.9157902002334595,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.032040882110595706,
+      "mean_token_accuracy": 0.9880847573280335,
+      "num_tokens": 53262962.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.7237608432769775,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.1450515985488892,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.03199286162853241,
+      "mean_token_accuracy": 0.987687611579895,
+      "num_tokens": 53451982.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.7237064421176911,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.9945085644721985,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.03347462117671966,
+      "mean_token_accuracy": 0.9875530540943146,
+      "num_tokens": 53640737.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.7271939218044281,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.9351736903190613,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.03140422701835632,
+      "mean_token_accuracy": 0.9877301037311554,
+      "num_tokens": 53829142.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.7265262603759766,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.1019784212112427,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.030748724937438965,
+      "mean_token_accuracy": 0.9881740033626556,
+      "num_tokens": 54018331.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.7270502150058746,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 0.9153032898902893,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.033472740650177,
+      "mean_token_accuracy": 0.9876999020576477,
+      "num_tokens": 54206935.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.7345028996467591,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.0144257545471191,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.033550959825515744,
+      "mean_token_accuracy": 0.9874906241893768,
+      "num_tokens": 54395443.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.7274066030979156,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0221407413482666,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.033388751745224,
+      "mean_token_accuracy": 0.9870682001113892,
+      "num_tokens": 54584552.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.7293415367603302,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.019165277481079,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.03568689227104187,
+      "mean_token_accuracy": 0.9861597299575806,
+      "num_tokens": 54773705.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.7281844198703766,
+      "epoch": 3.916554508748318,
+      "grad_norm": 0.933020830154419,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.030142486095428467,
+      "mean_token_accuracy": 0.988915103673935,
+      "num_tokens": 54962152.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.7242733418941498,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.9429674744606018,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.0337733268737793,
+      "mean_token_accuracy": 0.987263411283493,
+      "num_tokens": 55151179.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.7306612014770508,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.9022014737129211,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.03387616574764252,
+      "mean_token_accuracy": 0.9871586680412292,
+      "num_tokens": 55339658.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.7246532738208771,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2193421125411987,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.03193953037261963,
+      "mean_token_accuracy": 0.9879468739032745,
+      "num_tokens": 55528467.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.7268587052822113,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.2917358875274658,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.03584722280502319,
+      "mean_token_accuracy": 0.9867104470729828,
+      "num_tokens": 55717308.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.735340416431427,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.2229385375976562,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.03506708741188049,
+      "mean_token_accuracy": 0.9864643573760986,
+      "num_tokens": 55906059.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.7337038695812226,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.0747122764587402,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.03241217732429504,
+      "mean_token_accuracy": 0.9876143455505371,
+      "num_tokens": 56094606.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.7333687073106219,
+      "eval_loss": 0.07489899545907974,
+      "eval_mean_token_accuracy": 0.9719084019114257,
+      "eval_num_tokens": 56132350.0,
+      "eval_runtime": 12.7823,
+      "eval_samples_per_second": 391.166,
+      "eval_steps_per_second": 12.283,
+      "step": 2972
+    },
+    {
+      "entropy": 0.72461416721344,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7916642427444458,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.02286165803670883,
+      "mean_token_accuracy": 0.9920066714286804,
+      "num_tokens": 56283397.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.7150031447410583,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.9078621864318848,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.020906335115432738,
+      "mean_token_accuracy": 0.9925728499889374,
+      "num_tokens": 56472408.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.7091723501682281,
+      "epoch": 4.037685060565276,
+      "grad_norm": 0.8117756843566895,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.02036277800798416,
+      "mean_token_accuracy": 0.9928471207618713,
+      "num_tokens": 56660906.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.695214182138443,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7815372943878174,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019653698801994322,
+      "mean_token_accuracy": 0.992583978176117,
+      "num_tokens": 56850067.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.6937313377857208,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4769781827926636,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.02311961054801941,
+      "mean_token_accuracy": 0.9921472012996674,
+      "num_tokens": 57039005.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.6977694034576416,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.0028746128082275,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.021832706034183504,
+      "mean_token_accuracy": 0.9917877376079559,
+      "num_tokens": 57227782.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.698432594537735,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.0107522010803223,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.020331941545009613,
+      "mean_token_accuracy": 0.9932828366756439,
+      "num_tokens": 57416541.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.6941913783550262,
+      "epoch": 4.104979811574697,
+      "grad_norm": 0.8923842310905457,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019534927606582642,
+      "mean_token_accuracy": 0.9930490672588348,
+      "num_tokens": 57605162.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.6947623729705811,
+      "epoch": 4.118438761776582,
+      "grad_norm": 0.9093812108039856,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.02151128649711609,
+      "mean_token_accuracy": 0.9920440256595612,
+      "num_tokens": 57793756.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.6972509443759918,
+      "epoch": 4.131897711978466,
+      "grad_norm": 0.7549760341644287,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.020574498176574706,
+      "mean_token_accuracy": 0.9928867638111114,
+      "num_tokens": 57982678.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.6952514111995697,
+      "epoch": 4.14535666218035,
+      "grad_norm": 0.9476824402809143,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.02087453305721283,
+      "mean_token_accuracy": 0.9924874365329742,
+      "num_tokens": 58171508.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.6910251140594482,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.2548149824142456,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.020806866884231567,
+      "mean_token_accuracy": 0.9921584248542785,
+      "num_tokens": 58360594.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.6958921313285827,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.0649152994155884,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.020952591300010683,
+      "mean_token_accuracy": 0.9924699544906617,
+      "num_tokens": 58549637.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.6932485699653625,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.8891078233718872,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.020813333988189697,
+      "mean_token_accuracy": 0.9927419483661651,
+      "num_tokens": 58738595.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.6984799921512603,
+      "epoch": 4.199192462987887,
+      "grad_norm": 0.9459101557731628,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.01971060484647751,
+      "mean_token_accuracy": 0.9926240146160126,
+      "num_tokens": 58927507.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.7021916508674622,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.8318898677825928,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020860180258750916,
+      "mean_token_accuracy": 0.9927471876144409,
+      "num_tokens": 59116537.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.6972847402095794,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.9066984057426453,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.023059430718421935,
+      "mean_token_accuracy": 0.9919216752052307,
+      "num_tokens": 59305838.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.7049594700336457,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1277318000793457,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.022385355830192567,
+      "mean_token_accuracy": 0.99165118932724,
+      "num_tokens": 59494587.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.7045334160327912,
+      "epoch": 4.253028263795424,
+      "grad_norm": 0.8147489428520203,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.022358573973178864,
+      "mean_token_accuracy": 0.9918249011039734,
+      "num_tokens": 59683021.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.7007148265838623,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.9161480069160461,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.02088596671819687,
+      "mean_token_accuracy": 0.9925470650196075,
+      "num_tokens": 59871999.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.7006107687950134,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.9611289501190186,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.022619232535362244,
+      "mean_token_accuracy": 0.991577273607254,
+      "num_tokens": 60060486.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.7069303750991821,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.0339760780334473,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.021571077406406403,
+      "mean_token_accuracy": 0.9922139644622803,
+      "num_tokens": 60249686.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.7146372079849244,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.9761297702789307,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.022300006449222566,
+      "mean_token_accuracy": 0.9918746411800384,
+      "num_tokens": 60438481.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.7134014248847962,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 0.9729707837104797,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.02121303677558899,
+      "mean_token_accuracy": 0.9925692796707153,
+      "num_tokens": 60627501.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.7110478222370148,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.9456778168678284,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.020917078852653502,
+      "mean_token_accuracy": 0.9922434687614441,
+      "num_tokens": 60816135.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.7111951589584351,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8617656826972961,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.020851245522499083,
+      "mean_token_accuracy": 0.9924458563327789,
+      "num_tokens": 61005658.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.7193978548049926,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.8495420813560486,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.02084285914897919,
+      "mean_token_accuracy": 0.9927356660366058,
+      "num_tokens": 61194170.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.7188209414482116,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.728841245174408,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.02111642211675644,
+      "mean_token_accuracy": 0.9926697969436645,
+      "num_tokens": 61382852.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.7130864262580872,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.1274386644363403,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.021722708642482758,
+      "mean_token_accuracy": 0.9922453343868256,
+      "num_tokens": 61572195.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.7132770180702209,
+      "epoch": 4.401076716016151,
+      "grad_norm": 0.9251006245613098,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.021602503955364227,
+      "mean_token_accuracy": 0.9919042110443115,
+      "num_tokens": 61761119.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.7063322901725769,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.0182228088378906,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.020274630188941954,
+      "mean_token_accuracy": 0.9928637027740479,
+      "num_tokens": 61949931.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.7106044590473175,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.795170783996582,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.020594277977943422,
+      "mean_token_accuracy": 0.9925063610076904,
+      "num_tokens": 62138377.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7151111841201783,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1479791402816772,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.02117723524570465,
+      "mean_token_accuracy": 0.9921884059906005,
+      "num_tokens": 62327009.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.7172957301139832,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9795703291893005,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021917465329170226,
+      "mean_token_accuracy": 0.9915876984596252,
+      "num_tokens": 62515859.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.7175660550594329,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.9326112866401672,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.021142563223838805,
+      "mean_token_accuracy": 0.9922006845474243,
+      "num_tokens": 62704678.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.7085749089717865,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.8895787596702576,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.0222358375787735,
+      "mean_token_accuracy": 0.9918791234493256,
+      "num_tokens": 62893576.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.7059603810310364,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2298023700714111,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021999095380306245,
+      "mean_token_accuracy": 0.9920522570610046,
+      "num_tokens": 63082258.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.707107275724411,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.8295655250549316,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021844917535781862,
+      "mean_token_accuracy": 0.9922264873981476,
+      "num_tokens": 63271202.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.7119229018688202,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.7440339922904968,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.022164498269557954,
+      "mean_token_accuracy": 0.9918813884258271,
+      "num_tokens": 63460614.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.7065344154834747,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7275091409683228,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.019415955245494842,
+      "mean_token_accuracy": 0.9928511261940003,
+      "num_tokens": 63649327.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.6984272241592407,
+      "epoch": 4.549125168236878,
+      "grad_norm": 0.9989121556282043,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02176133990287781,
+      "mean_token_accuracy": 0.9923081934452057,
+      "num_tokens": 63837983.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.7084203243255616,
+      "epoch": 4.562584118438762,
+      "grad_norm": 0.9526837468147278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.02182276248931885,
+      "mean_token_accuracy": 0.9921061515808105,
+      "num_tokens": 64026855.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.7097050070762634,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.0156456232070923,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.020074066519737244,
+      "mean_token_accuracy": 0.9926879525184631,
+      "num_tokens": 64215889.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.7087334096431732,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.780275821685791,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.021703639626502992,
+      "mean_token_accuracy": 0.9917874097824096,
+      "num_tokens": 64404515.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.7142964243888855,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7382510304450989,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.019867205619812013,
+      "mean_token_accuracy": 0.9927354514598846,
+      "num_tokens": 64593147.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.7165492177009583,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.8259146213531494,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.021262936294078827,
+      "mean_token_accuracy": 0.9914182722568512,
+      "num_tokens": 64782041.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.7098793745040893,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.9540995359420776,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.0195529043674469,
+      "mean_token_accuracy": 0.9931159794330597,
+      "num_tokens": 64971390.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.7104229986667633,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.8698863387107849,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.019119253754615782,
+      "mean_token_accuracy": 0.9934523403644562,
+      "num_tokens": 65160253.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.7067269921302796,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.9622324705123901,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.02044745087623596,
+      "mean_token_accuracy": 0.992815512418747,
+      "num_tokens": 65349403.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.7016099870204926,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.999839723110199,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.019250848889350893,
+      "mean_token_accuracy": 0.9929339647293091,
+      "num_tokens": 65538164.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.7059290170669555,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.7539619207382202,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.02152409851551056,
+      "mean_token_accuracy": 0.99238161444664,
+      "num_tokens": 65727493.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.7103133797645569,
+      "epoch": 4.697173620457605,
+      "grad_norm": 0.9892463088035583,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020616935193538667,
+      "mean_token_accuracy": 0.9927784681320191,
+      "num_tokens": 65916233.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.703304260969162,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.1750195026397705,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021048617362976075,
+      "mean_token_accuracy": 0.9921808183193207,
+      "num_tokens": 66105229.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.7017257273197174,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.887427031993866,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02077500820159912,
+      "mean_token_accuracy": 0.992085075378418,
+      "num_tokens": 66294280.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.7027543246746063,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.0987571477890015,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.02167164385318756,
+      "mean_token_accuracy": 0.992034786939621,
+      "num_tokens": 66483182.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.7053992748260498,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.0075602531433105,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.022686071693897247,
+      "mean_token_accuracy": 0.9918968200683593,
+      "num_tokens": 66671795.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.7109473288059235,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.1655596494674683,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.022698774933815002,
+      "mean_token_accuracy": 0.9913642406463623,
+      "num_tokens": 66861166.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.7130867660045623,
+      "epoch": 4.77792732166891,
+      "grad_norm": 0.9386347532272339,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0205889493227005,
+      "mean_token_accuracy": 0.9924897074699401,
+      "num_tokens": 67049967.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.7099342107772827,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.8524489998817444,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.01985667645931244,
+      "mean_token_accuracy": 0.9928426086902619,
+      "num_tokens": 67238781.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.7100094079971313,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7522528171539307,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.02066437602043152,
+      "mean_token_accuracy": 0.9924224317073822,
+      "num_tokens": 67427802.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7172351717948914,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9865215420722961,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.021969664096832275,
+      "mean_token_accuracy": 0.9919721603393554,
+      "num_tokens": 67616675.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.7113273084163666,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9339645504951477,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.019794631004333495,
+      "mean_token_accuracy": 0.9925971984863281,
+      "num_tokens": 67805877.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.7041554272174835,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.337364912033081,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020412631332874298,
+      "mean_token_accuracy": 0.9923949122428894,
+      "num_tokens": 67994858.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.7090700149536133,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.7725968956947327,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.02121713161468506,
+      "mean_token_accuracy": 0.9920605599880219,
+      "num_tokens": 68183865.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.7141157209873199,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.8638854026794434,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.020726922154426574,
+      "mean_token_accuracy": 0.992779266834259,
+      "num_tokens": 68372954.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.7078199684619904,
+      "epoch": 4.885598923283984,
+      "grad_norm": 0.8576115965843201,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.019535741209983824,
+      "mean_token_accuracy": 0.9927346885204316,
+      "num_tokens": 68562535.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.7034667313098908,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8339338898658752,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.021528851985931397,
+      "mean_token_accuracy": 0.991979056596756,
+      "num_tokens": 68751258.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.6983348250389099,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8739761710166931,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020945993065834046,
+      "mean_token_accuracy": 0.9923328399658203,
+      "num_tokens": 68939586.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.6922877132892609,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.886886715888977,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.020974791049957274,
+      "mean_token_accuracy": 0.992312878370285,
+      "num_tokens": 69128270.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.6857815742492676,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.9415780901908875,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.020201480388641356,
+      "mean_token_accuracy": 0.9925193309783935,
+      "num_tokens": 69317066.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.6832823157310486,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.8311293721199036,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.02008202373981476,
+      "mean_token_accuracy": 0.9926081895828247,
+      "num_tokens": 69505698.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.6914379715919494,
+      "epoch": 4.96635262449529,
+      "grad_norm": 0.8799999356269836,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.0189486563205719,
+      "mean_token_accuracy": 0.9932616055011749,
+      "num_tokens": 69694164.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.692799162864685,
+      "epoch": 4.979811574697173,
+      "grad_norm": 0.9579370021820068,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.019743245840072633,
+      "mean_token_accuracy": 0.9932292222976684,
+      "num_tokens": 69882717.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.6969726324081421,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.0632585287094116,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.019562503695487975,
+      "mean_token_accuracy": 0.9929000854492187,
+      "num_tokens": 70071225.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.7056300662885047,
+      "eval_loss": 0.08702203631401062,
+      "eval_mean_token_accuracy": 0.9712980936287315,
+      "eval_num_tokens": 70165509.0,
+      "eval_runtime": 12.7642,
+      "eval_samples_per_second": 391.719,
+      "eval_steps_per_second": 12.3,
+      "step": 3715
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.243081081087328e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-3715/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-3715/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-4458/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-4458/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-4458/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..26ef2a4798ee1c9a1b41fe55b570728c491fbd47
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:441f8c1a6a2feb498c3fec0feefe220d84939d88fcb1da295e41c2a09bdac9e4
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-4458/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-4458/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-4458/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-4458/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f36a900b207a6ebcb2c53eee9c635c8a526ebfbc
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/trainer_state.json
@@ -0,0 +1,4550 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.0,
+  "eval_steps": 500,
+  "global_step": 4458,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    },
+    {
+      "entropy": 0.8332668900489807,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9914855360984802,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.06607928276062011,
+      "mean_token_accuracy": 0.974932211637497,
+      "num_tokens": 14165567.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8149146616458893,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.3080861568450928,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.057932257652282715,
+      "mean_token_accuracy": 0.9770712852478027,
+      "num_tokens": 14353912.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8197620987892151,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.0898637771606445,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.06010057330131531,
+      "mean_token_accuracy": 0.9758293390274048,
+      "num_tokens": 14542288.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8183245182037353,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.3120660781860352,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.0610861599445343,
+      "mean_token_accuracy": 0.9764986515045166,
+      "num_tokens": 14731028.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8135185301303863,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.138473391532898,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.06268800497055053,
+      "mean_token_accuracy": 0.9752912580966949,
+      "num_tokens": 14919682.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8142645537853241,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.1258258819580078,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.06271481513977051,
+      "mean_token_accuracy": 0.9757774949073792,
+      "num_tokens": 15107904.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8170483231544494,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.007297396659851,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.06338523030281067,
+      "mean_token_accuracy": 0.9754154622554779,
+      "num_tokens": 15297115.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8201254069805145,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 0.8240306973457336,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.06355815529823303,
+      "mean_token_accuracy": 0.9752808332443237,
+      "num_tokens": 15485921.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8140238881111145,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.0350368022918701,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.05969444513320923,
+      "mean_token_accuracy": 0.977089112997055,
+      "num_tokens": 15674824.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8133325934410095,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.0544878244400024,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.06062774658203125,
+      "mean_token_accuracy": 0.976290100812912,
+      "num_tokens": 15863526.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8185161769390106,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.1076316833496094,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.061567097902297974,
+      "mean_token_accuracy": 0.9755526125431061,
+      "num_tokens": 16051685.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8176139056682586,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.0057616233825684,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.06227490901947021,
+      "mean_token_accuracy": 0.9758114695549012,
+      "num_tokens": 16240451.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8217637002468109,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.97276771068573,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.06410331726074218,
+      "mean_token_accuracy": 0.9754030883312226,
+      "num_tokens": 16428961.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8141795754432678,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.0135555267333984,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.06253207921981811,
+      "mean_token_accuracy": 0.9760490536689759,
+      "num_tokens": 16617327.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8167653739452362,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2272467613220215,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.060455435514450075,
+      "mean_token_accuracy": 0.9758628964424133,
+      "num_tokens": 16806414.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8155323445796967,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 0.9328635334968567,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.06302794814109802,
+      "mean_token_accuracy": 0.9745101630687714,
+      "num_tokens": 16995016.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.818474805355072,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.1651655435562134,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.06043210029602051,
+      "mean_token_accuracy": 0.9766487181186676,
+      "num_tokens": 17184250.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8140560865402222,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.1005650758743286,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.06131819486618042,
+      "mean_token_accuracy": 0.9757712185382843,
+      "num_tokens": 17373075.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8204843640327454,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.0825532674789429,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.06296311616897583,
+      "mean_token_accuracy": 0.9745951414108276,
+      "num_tokens": 17561454.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8237834334373474,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.9842743873596191,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.061483919620513916,
+      "mean_token_accuracy": 0.9754831731319428,
+      "num_tokens": 17750758.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8165915131568908,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.0179396867752075,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.06237881183624268,
+      "mean_token_accuracy": 0.9750245690345765,
+      "num_tokens": 17940126.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8329708755016327,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0732359886169434,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.0644665539264679,
+      "mean_token_accuracy": 0.9743888020515442,
+      "num_tokens": 18129000.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8210074126720428,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.3696036338806152,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.06320792436599731,
+      "mean_token_accuracy": 0.9751390159130097,
+      "num_tokens": 18318232.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8248398661613464,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.327525019645691,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.06277769804000854,
+      "mean_token_accuracy": 0.9752413153648376,
+      "num_tokens": 18507403.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8284746050834656,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.044022560119629,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.06162315607070923,
+      "mean_token_accuracy": 0.9760516047477722,
+      "num_tokens": 18696242.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8233858644962311,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.062048077583313,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.059620130062103274,
+      "mean_token_accuracy": 0.9755307257175445,
+      "num_tokens": 18885199.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8236705601215363,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.2024412155151367,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.06304935216903687,
+      "mean_token_accuracy": 0.975091153383255,
+      "num_tokens": 19074028.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.828409218788147,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.1684212684631348,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.06451238393783569,
+      "mean_token_accuracy": 0.9743404746055603,
+      "num_tokens": 19262295.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8214354634284973,
+      "epoch": 1.386271870794078,
+      "grad_norm": 0.8512521386146545,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.06042641997337341,
+      "mean_token_accuracy": 0.9755558848381043,
+      "num_tokens": 19451154.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8223241448402405,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 0.8546568155288696,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.062134528160095216,
+      "mean_token_accuracy": 0.9750089883804322,
+      "num_tokens": 19639827.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.8217969954013824,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 0.972831666469574,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.060803139209747316,
+      "mean_token_accuracy": 0.9760692238807678,
+      "num_tokens": 19828174.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.8144341170787811,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1216644048690796,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.06458876132965088,
+      "mean_token_accuracy": 0.9738282442092896,
+      "num_tokens": 20016700.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.8168790876865387,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.0821806192398071,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.06258658170700074,
+      "mean_token_accuracy": 0.9749014675617218,
+      "num_tokens": 20205823.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8146390020847321,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.0067790746688843,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.06509006023406982,
+      "mean_token_accuracy": 0.9737810134887696,
+      "num_tokens": 20394611.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8224801063537598,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.0583593845367432,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.06343590021133423,
+      "mean_token_accuracy": 0.9750119090080261,
+      "num_tokens": 20583252.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.8083258986473083,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.8305535316467285,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.06573111414909363,
+      "mean_token_accuracy": 0.9736428141593934,
+      "num_tokens": 20772107.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8081986546516419,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5895942449569702,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.06376024484634399,
+      "mean_token_accuracy": 0.9751200795173645,
+      "num_tokens": 20960494.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.8026683986186981,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 0.9962654113769531,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.06282572746276856,
+      "mean_token_accuracy": 0.9753849327564239,
+      "num_tokens": 21149763.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8030014991760254,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9083836078643799,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.06473478078842163,
+      "mean_token_accuracy": 0.9747069597244262,
+      "num_tokens": 21338690.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.8008662164211273,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.097593069076538,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.05911685228347778,
+      "mean_token_accuracy": 0.9767139732837677,
+      "num_tokens": 21527420.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7981986284255982,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.1673516035079956,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.05957244038581848,
+      "mean_token_accuracy": 0.9761911332607269,
+      "num_tokens": 21716791.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.8004718601703644,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 0.7955246567726135,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.06378913521766663,
+      "mean_token_accuracy": 0.9747395157814026,
+      "num_tokens": 21905487.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.8039821267127991,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.3141396045684814,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.06209801435470581,
+      "mean_token_accuracy": 0.9754719436168671,
+      "num_tokens": 22094436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.8051789462566376,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.2227814197540283,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.06261310577392579,
+      "mean_token_accuracy": 0.9753049433231353,
+      "num_tokens": 22283243.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7941994190216064,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.071070671081543,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.05962263345718384,
+      "mean_token_accuracy": 0.9763956189155578,
+      "num_tokens": 22472084.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7937953472137451,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.0329782962799072,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.060611891746521,
+      "mean_token_accuracy": 0.9758659780025483,
+      "num_tokens": 22661614.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7913113713264466,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.032772183418274,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.06002308130264282,
+      "mean_token_accuracy": 0.9758916199207306,
+      "num_tokens": 22850634.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.7909658491611481,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3234598636627197,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.06296972036361695,
+      "mean_token_accuracy": 0.9741117715835571,
+      "num_tokens": 23039402.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7880761086940765,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.9563435316085815,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.06209538578987121,
+      "mean_token_accuracy": 0.9750530481338501,
+      "num_tokens": 23228363.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7845663070678711,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 0.8875390291213989,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.06272958517074585,
+      "mean_token_accuracy": 0.9751476228237153,
+      "num_tokens": 23417469.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7928634822368622,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.0629489421844482,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.060286092758178714,
+      "mean_token_accuracy": 0.976036274433136,
+      "num_tokens": 23606274.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7870154738426208,
+      "epoch": 1.695827725437416,
+      "grad_norm": 0.9164250493049622,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.06263296008110046,
+      "mean_token_accuracy": 0.9752088189125061,
+      "num_tokens": 23795255.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.7948406875133515,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.0909371376037598,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.06165047883987427,
+      "mean_token_accuracy": 0.9764420628547669,
+      "num_tokens": 23984534.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7862435758113862,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.1839547157287598,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.0640158474445343,
+      "mean_token_accuracy": 0.9741048216819763,
+      "num_tokens": 24173632.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.7887867331504822,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.0730741024017334,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.06260218620300292,
+      "mean_token_accuracy": 0.974741768836975,
+      "num_tokens": 24362671.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7896574556827545,
+      "epoch": 1.749663526244953,
+      "grad_norm": 0.9559605121612549,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.061297619342803956,
+      "mean_token_accuracy": 0.9752223670482636,
+      "num_tokens": 24551825.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.7902800619602204,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.914057195186615,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.06320110559463502,
+      "mean_token_accuracy": 0.9749482691287994,
+      "num_tokens": 24740867.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.799742591381073,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.06534289717674255,
+      "mean_token_accuracy": 0.9741291284561158,
+      "num_tokens": 24930132.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7974063038825989,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1188615560531616,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.060908573865890506,
+      "mean_token_accuracy": 0.9765798091888428,
+      "num_tokens": 25119678.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.786439061164856,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9185847640037537,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.06158917546272278,
+      "mean_token_accuracy": 0.9760335922241211,
+      "num_tokens": 25308876.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7888528347015381,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.1948838233947754,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.060635948181152345,
+      "mean_token_accuracy": 0.976032954454422,
+      "num_tokens": 25498102.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7968044698238372,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.4128663539886475,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.06448977589607238,
+      "mean_token_accuracy": 0.9741183638572692,
+      "num_tokens": 25686747.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7967679858207702,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.9797796010971069,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.06342286467552186,
+      "mean_token_accuracy": 0.9752787709236145,
+      "num_tokens": 25875368.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7932188332080841,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.0549253225326538,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.060981249809265135,
+      "mean_token_accuracy": 0.976200807094574,
+      "num_tokens": 26064480.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7899552762508393,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.1172103881835938,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.06555815935134887,
+      "mean_token_accuracy": 0.9736676931381225,
+      "num_tokens": 26253619.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.8019726693630218,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.94905686378479,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.0663321852684021,
+      "mean_token_accuracy": 0.9739456713199616,
+      "num_tokens": 26442736.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7939825117588043,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 0.9922711253166199,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05981417298316956,
+      "mean_token_accuracy": 0.9758030951023102,
+      "num_tokens": 26631525.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8085412085056305,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4970781803131104,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.06708773374557495,
+      "mean_token_accuracy": 0.9737519145011901,
+      "num_tokens": 26820483.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7950678527355194,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.7868127226829529,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.061734098196029666,
+      "mean_token_accuracy": 0.9747878730297088,
+      "num_tokens": 27009393.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7850766360759736,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.1386486291885376,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.05714603662490845,
+      "mean_token_accuracy": 0.9769004642963409,
+      "num_tokens": 27198251.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7933461844921113,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.0092092752456665,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.06240262985229492,
+      "mean_token_accuracy": 0.9742367327213287,
+      "num_tokens": 27387231.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8044621467590332,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.2477015256881714,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0642085313796997,
+      "mean_token_accuracy": 0.9748865723609924,
+      "num_tokens": 27576251.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7975574374198914,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.1167609691619873,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.06382874846458435,
+      "mean_token_accuracy": 0.9748350441455841,
+      "num_tokens": 27764560.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.8090806484222413,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 0.9714348912239075,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.06401748657226562,
+      "mean_token_accuracy": 0.9741845309734345,
+      "num_tokens": 27953532.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.8104567945383157,
+      "eval_loss": 0.07282773405313492,
+      "eval_mean_token_accuracy": 0.9701232534305305,
+      "eval_num_tokens": 28066151.0,
+      "eval_runtime": 12.7852,
+      "eval_samples_per_second": 391.078,
+      "eval_steps_per_second": 12.28,
+      "step": 1486
+    },
+    {
+      "entropy": 0.8090688765048981,
+      "epoch": 2.005383580080754,
+      "grad_norm": 0.9886816143989563,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.06012396812438965,
+      "mean_token_accuracy": 0.9764885842800141,
+      "num_tokens": 28141848.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.7821709334850311,
+      "epoch": 2.018842530282638,
+      "grad_norm": 0.9049969911575317,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.04819132387638092,
+      "mean_token_accuracy": 0.9816783249378205,
+      "num_tokens": 28330784.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.7845205307006836,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0231750011444092,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.04539612233638764,
+      "mean_token_accuracy": 0.9818619966506958,
+      "num_tokens": 28519580.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.7817754447460175,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.5244711637496948,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0429688960313797,
+      "mean_token_accuracy": 0.9834935545921326,
+      "num_tokens": 28707486.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.7733526527881622,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.0685060024261475,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.045852530002594,
+      "mean_token_accuracy": 0.9817728459835052,
+      "num_tokens": 28896759.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.7815343141555786,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.075853705406189,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.04565088748931885,
+      "mean_token_accuracy": 0.9823914349079133,
+      "num_tokens": 29085097.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.7751398742198944,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2573742866516113,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.04465666115283966,
+      "mean_token_accuracy": 0.9829874813556672,
+      "num_tokens": 29273868.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.7737197399139404,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 0.8681219220161438,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.042188242077827454,
+      "mean_token_accuracy": 0.9837059378623962,
+      "num_tokens": 29462985.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.7823441684246063,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.3503504991531372,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.04393318593502045,
+      "mean_token_accuracy": 0.9827627122402192,
+      "num_tokens": 29651156.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.7796198308467865,
+      "epoch": 2.126514131897712,
+      "grad_norm": 0.7799349427223206,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.04367418885231018,
+      "mean_token_accuracy": 0.9837401032447814,
+      "num_tokens": 29839789.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.7767675697803498,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.375465750694275,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.04427264630794525,
+      "mean_token_accuracy": 0.9832019031047821,
+      "num_tokens": 30028957.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.7912136912345886,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.9683088660240173,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.04428936243057251,
+      "mean_token_accuracy": 0.9832306385040284,
+      "num_tokens": 30217841.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.790750366449356,
+      "epoch": 2.166890982503365,
+      "grad_norm": 0.9393274188041687,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.04529332816600799,
+      "mean_token_accuracy": 0.9832263946533203,
+      "num_tokens": 30406540.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.7816023886203766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.1822642087936401,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.04192846417427063,
+      "mean_token_accuracy": 0.9846727311611175,
+      "num_tokens": 30595568.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7754982054233551,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.1961082220077515,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.043791168928146364,
+      "mean_token_accuracy": 0.9830836653709412,
+      "num_tokens": 30784300.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.7756961703300476,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 0.9719674587249756,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.04410521686077118,
+      "mean_token_accuracy": 0.9837198793888092,
+      "num_tokens": 30972950.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.7833336114883422,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.0175098180770874,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.044813218712806704,
+      "mean_token_accuracy": 0.9829723179340363,
+      "num_tokens": 31161783.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.7893047273159027,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.2077007293701172,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.04417986273765564,
+      "mean_token_accuracy": 0.9825236797332764,
+      "num_tokens": 31350487.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.794023334980011,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8530077934265137,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.043008115887641904,
+      "mean_token_accuracy": 0.9832422196865082,
+      "num_tokens": 31539217.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.785652619600296,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.4543156623840332,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0445009708404541,
+      "mean_token_accuracy": 0.9828064203262329,
+      "num_tokens": 31728308.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7841614365577698,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.297763466835022,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.045649731159210206,
+      "mean_token_accuracy": 0.9823782920837403,
+      "num_tokens": 31917293.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.7825099349021911,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.2237403392791748,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.047218257188797,
+      "mean_token_accuracy": 0.98156818151474,
+      "num_tokens": 32106628.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.7791289031505585,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8094402551651001,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.04178315699100495,
+      "mean_token_accuracy": 0.9846469461917877,
+      "num_tokens": 32295183.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.7743105947971344,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.3677793741226196,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.04540807008743286,
+      "mean_token_accuracy": 0.9827918827533721,
+      "num_tokens": 32484143.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.7804521083831787,
+      "epoch": 2.328398384925976,
+      "grad_norm": 0.8883333802223206,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.04598270654678345,
+      "mean_token_accuracy": 0.9823103487491608,
+      "num_tokens": 32673144.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.7760683119297027,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.989972710609436,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.043702703714370725,
+      "mean_token_accuracy": 0.9831363618373871,
+      "num_tokens": 32862142.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7726980447769165,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.0461052656173706,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.044418123364448545,
+      "mean_token_accuracy": 0.9821757376194,
+      "num_tokens": 33051382.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.7822885036468505,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.9027822017669678,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.04624638557434082,
+      "mean_token_accuracy": 0.9815774321556091,
+      "num_tokens": 33239729.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.7830030500888825,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.1743420362472534,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.04747769832611084,
+      "mean_token_accuracy": 0.9812638640403748,
+      "num_tokens": 33428615.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.7933187901973724,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 0.9334444403648376,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.047731178998947146,
+      "mean_token_accuracy": 0.9820823550224305,
+      "num_tokens": 33617331.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.7904782056808471,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.9396253824234009,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.04432434141635895,
+      "mean_token_accuracy": 0.9830163478851318,
+      "num_tokens": 33805978.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.7811533212661743,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 0.9742764234542847,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.046493279933929446,
+      "mean_token_accuracy": 0.9818211495876312,
+      "num_tokens": 33994957.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7848010778427124,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0378986597061157,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0454249233007431,
+      "mean_token_accuracy": 0.9817994356155395,
+      "num_tokens": 34184297.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.7754390180110932,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.1023566722869873,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.04380808770656586,
+      "mean_token_accuracy": 0.9829573810100556,
+      "num_tokens": 34373551.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.7767188012599945,
+      "epoch": 2.462987886944818,
+      "grad_norm": 0.8460658192634583,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04502716064453125,
+      "mean_token_accuracy": 0.9825617015361786,
+      "num_tokens": 34562169.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.7757977426052094,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.1013047695159912,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.045161890983581546,
+      "mean_token_accuracy": 0.9821914732456207,
+      "num_tokens": 34750863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.785107308626175,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.9510554671287537,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.045189899206161496,
+      "mean_token_accuracy": 0.9822288691997528,
+      "num_tokens": 34939528.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.7849742233753204,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0994503498077393,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.04333080947399139,
+      "mean_token_accuracy": 0.9830258965492249,
+      "num_tokens": 35128334.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.7875016331672668,
+      "epoch": 2.516823687752355,
+      "grad_norm": 0.9202979803085327,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.04838588833808899,
+      "mean_token_accuracy": 0.9815591931343078,
+      "num_tokens": 35317067.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.7813792943954467,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.9036716222763062,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.045662492513656616,
+      "mean_token_accuracy": 0.9817738354206085,
+      "num_tokens": 35506291.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.777835363149643,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.853438675403595,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.044686585664749146,
+      "mean_token_accuracy": 0.98329097032547,
+      "num_tokens": 35695141.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.7677761793136597,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.0699759721755981,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.045490306615829465,
+      "mean_token_accuracy": 0.9826831936836242,
+      "num_tokens": 35884173.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.7756118893623352,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.0356824398040771,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.0467751681804657,
+      "mean_token_accuracy": 0.9823338150978088,
+      "num_tokens": 36073241.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.7775443732738495,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.1619561910629272,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.04741084575653076,
+      "mean_token_accuracy": 0.9815192580223083,
+      "num_tokens": 36262133.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7899596333503723,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.3479284048080444,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.04588942229747772,
+      "mean_token_accuracy": 0.9821865320205688,
+      "num_tokens": 36451460.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.7832387804985046,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.6297204494476318,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.04711659252643585,
+      "mean_token_accuracy": 0.9815292179584503,
+      "num_tokens": 36640366.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.7871865928173065,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.3100138902664185,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.04919236898422241,
+      "mean_token_accuracy": 0.9809704780578613,
+      "num_tokens": 36829017.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.7844007611274719,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.2907201051712036,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.047159326076507566,
+      "mean_token_accuracy": 0.9816433548927307,
+      "num_tokens": 37018034.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.7885129511356354,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.1324766874313354,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.04420983195304871,
+      "mean_token_accuracy": 0.9828807294368744,
+      "num_tokens": 37206768.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.7761563122272491,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.9269904494285583,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.04585547745227814,
+      "mean_token_accuracy": 0.9823735177516937,
+      "num_tokens": 37396010.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7831155300140381,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5890840291976929,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.046450641751289365,
+      "mean_token_accuracy": 0.9817734777927398,
+      "num_tokens": 37584809.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.7729385912418365,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.4823863506317139,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.04653529226779938,
+      "mean_token_accuracy": 0.9815726220607758,
+      "num_tokens": 37773869.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.7752901673316955,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.9423630237579346,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.04599723219871521,
+      "mean_token_accuracy": 0.9816120624542236,
+      "num_tokens": 37962642.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.7750158667564392,
+      "epoch": 2.718707940780619,
+      "grad_norm": 0.9398621320724487,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.04729728400707245,
+      "mean_token_accuracy": 0.9816868543624878,
+      "num_tokens": 38151325.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.7693424999713898,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3207950592041016,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.04474019706249237,
+      "mean_token_accuracy": 0.9830808460712432,
+      "num_tokens": 38340719.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.7750859558582306,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.070200800895691,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.045914703607559205,
+      "mean_token_accuracy": 0.9820488631725312,
+      "num_tokens": 38529293.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7697387218475342,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.0545568466186523,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.04357335567474365,
+      "mean_token_accuracy": 0.9827888309955597,
+      "num_tokens": 38718464.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.7682756781578064,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.0945985317230225,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.0429091215133667,
+      "mean_token_accuracy": 0.9832668721675872,
+      "num_tokens": 38907573.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.7804523587226868,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8521431684494019,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.04662351012229919,
+      "mean_token_accuracy": 0.981964933872223,
+      "num_tokens": 39096598.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.7786886096000671,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.09945809841156,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.050166720151901247,
+      "mean_token_accuracy": 0.9816301107406616,
+      "num_tokens": 39285069.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.7790449261665344,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.1673868894577026,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.048508059978485105,
+      "mean_token_accuracy": 0.9809533059597015,
+      "num_tokens": 39474252.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.7774537563323974,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.0648490190505981,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.04540320634841919,
+      "mean_token_accuracy": 0.9822351098060608,
+      "num_tokens": 39663262.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7683943033218383,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.0106912851333618,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.04466369152069092,
+      "mean_token_accuracy": 0.9818881273269653,
+      "num_tokens": 39852133.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.7642774760723114,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0109530687332153,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.045978450775146486,
+      "mean_token_accuracy": 0.9819821894168854,
+      "num_tokens": 40041679.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.7647961139678955,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.9050237536430359,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.045694130659103396,
+      "mean_token_accuracy": 0.982342666387558,
+      "num_tokens": 40230019.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.7589850842952728,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.8879176378250122,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.046186423301696776,
+      "mean_token_accuracy": 0.9818771004676818,
+      "num_tokens": 40418932.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.765201997756958,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 0.9103795886039734,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.04475873112678528,
+      "mean_token_accuracy": 0.9821203172206878,
+      "num_tokens": 40607792.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.7653619945049286,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.0566699504852295,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.04391545057296753,
+      "mean_token_accuracy": 0.9828326046466828,
+      "num_tokens": 40797122.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.763550215959549,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.1280677318572998,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.045582184195518495,
+      "mean_token_accuracy": 0.9819066524505615,
+      "num_tokens": 40985689.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.7749062657356263,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.110323429107666,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.04598402082920074,
+      "mean_token_accuracy": 0.9815545618534088,
+      "num_tokens": 41174743.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.7805372893810272,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.0649195909500122,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.04705569744110107,
+      "mean_token_accuracy": 0.9815925776958465,
+      "num_tokens": 41363160.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.7711063683032989,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.1459592580795288,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.045355689525604245,
+      "mean_token_accuracy": 0.9832463562488556,
+      "num_tokens": 41551932.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.7747956931591033,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.1602799892425537,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.04494523704051971,
+      "mean_token_accuracy": 0.9819566667079925,
+      "num_tokens": 41740544.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.7584574401378632,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.054417371749878,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.04810510575771332,
+      "mean_token_accuracy": 0.9811540365219116,
+      "num_tokens": 41929344.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.7753648416251894,
+      "eval_loss": 0.07181113958358765,
+      "eval_mean_token_accuracy": 0.9716544128527307,
+      "eval_num_tokens": 42099366.0,
+      "eval_runtime": 12.748,
+      "eval_samples_per_second": 392.22,
+      "eval_steps_per_second": 12.316,
+      "step": 2229
+    },
+    {
+      "entropy": 0.7669954419136047,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.8211690783500671,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.04577431380748749,
+      "mean_token_accuracy": 0.9823363184928894,
+      "num_tokens": 42118188.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.7600219964981079,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.0873607397079468,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.032501548528671265,
+      "mean_token_accuracy": 0.9881355822086334,
+      "num_tokens": 42306598.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.7441959500312805,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.1175898313522339,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.031047609448432923,
+      "mean_token_accuracy": 0.9891582727432251,
+      "num_tokens": 42495250.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.7364335536956788,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.987751841545105,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.028629058599472047,
+      "mean_token_accuracy": 0.9894859492778778,
+      "num_tokens": 42684326.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.735680228471756,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.194305658340454,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.029283303022384643,
+      "mean_token_accuracy": 0.9887769162654877,
+      "num_tokens": 42873221.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.7342639684677124,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.254442811012268,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.0302200973033905,
+      "mean_token_accuracy": 0.988795381784439,
+      "num_tokens": 43062486.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.7391974747180938,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.2648969888687134,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.0316451758146286,
+      "mean_token_accuracy": 0.9879588901996612,
+      "num_tokens": 43251113.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.7401826322078705,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.0483295917510986,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.030878132581710814,
+      "mean_token_accuracy": 0.9886977672576904,
+      "num_tokens": 43440152.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.732647967338562,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.1125447750091553,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.02838427424430847,
+      "mean_token_accuracy": 0.9890213429927825,
+      "num_tokens": 43629353.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.7314932703971863,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.0979042053222656,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.0315592348575592,
+      "mean_token_accuracy": 0.9888832688331604,
+      "num_tokens": 43818981.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.7305827140808105,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9610060453414917,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.03112468421459198,
+      "mean_token_accuracy": 0.9890477001667023,
+      "num_tokens": 44008430.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.7333625435829163,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.0625078678131104,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.03105958104133606,
+      "mean_token_accuracy": 0.9884848177433014,
+      "num_tokens": 44197264.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.7373903751373291,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.1683465242385864,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.031627464294433597,
+      "mean_token_accuracy": 0.9879396498203278,
+      "num_tokens": 44386205.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.7391358911991119,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.05154287815094,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.03118889331817627,
+      "mean_token_accuracy": 0.9880312740802765,
+      "num_tokens": 44575099.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.7351223647594451,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.9531214237213135,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.030980610847473146,
+      "mean_token_accuracy": 0.9882127285003662,
+      "num_tokens": 44763965.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.7361315310001373,
+      "epoch": 3.203230148048452,
+      "grad_norm": 0.8818249702453613,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.029705989360809325,
+      "mean_token_accuracy": 0.9894622683525085,
+      "num_tokens": 44952983.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.7356473922729492,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 0.9263529777526855,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.03202669024467468,
+      "mean_token_accuracy": 0.9877129793167114,
+      "num_tokens": 45142080.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.7326373934745789,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.0517516136169434,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.031153690814971925,
+      "mean_token_accuracy": 0.9891136825084687,
+      "num_tokens": 45330769.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.7348711133003235,
+      "epoch": 3.243606998654105,
+      "grad_norm": 0.8774152994155884,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030574262142181396,
+      "mean_token_accuracy": 0.9888435959815979,
+      "num_tokens": 45519217.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.7379967510700226,
+      "epoch": 3.257065948855989,
+      "grad_norm": 0.9164374470710754,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.03096820116043091,
+      "mean_token_accuracy": 0.9884447932243348,
+      "num_tokens": 45708467.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.7369260907173156,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.3184924125671387,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03082413077354431,
+      "mean_token_accuracy": 0.9888757109642029,
+      "num_tokens": 45897424.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.7366503119468689,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.9073860049247742,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.03220943808555603,
+      "mean_token_accuracy": 0.9877902865409851,
+      "num_tokens": 46086351.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.7341453731060028,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.0940542221069336,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.03329495787620544,
+      "mean_token_accuracy": 0.9877542018890381,
+      "num_tokens": 46275394.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.7312689840793609,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 0.9651952385902405,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.031667912006378175,
+      "mean_token_accuracy": 0.9881450653076171,
+      "num_tokens": 46463821.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.7228557825088501,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 0.971441924571991,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.030608028173446655,
+      "mean_token_accuracy": 0.9882231116294861,
+      "num_tokens": 46653025.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.723542982339859,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.0060454607009888,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.030545371770858764,
+      "mean_token_accuracy": 0.9879125893115998,
+      "num_tokens": 46842068.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.7321009397506714,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.0572997331619263,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.032070663571357724,
+      "mean_token_accuracy": 0.9878279268741608,
+      "num_tokens": 47031045.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.735515958070755,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.9835748672485352,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.03338083028793335,
+      "mean_token_accuracy": 0.9877255499362946,
+      "num_tokens": 47219625.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.7396618247032165,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 0.9655317068099976,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.03589712679386139,
+      "mean_token_accuracy": 0.9869556546211242,
+      "num_tokens": 47408311.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.7358698546886444,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9225422143936157,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.02914820909500122,
+      "mean_token_accuracy": 0.9886718571186066,
+      "num_tokens": 47597297.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.7338236391544342,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.8253662586212158,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.02996352016925812,
+      "mean_token_accuracy": 0.9891344487667084,
+      "num_tokens": 47785737.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.7320856511592865,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.0716700553894043,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.032322123646736145,
+      "mean_token_accuracy": 0.9877110660076142,
+      "num_tokens": 47974241.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.7293457925319672,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.0915459394454956,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.031131434440612792,
+      "mean_token_accuracy": 0.9885926306247711,
+      "num_tokens": 48162582.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.7207926452159882,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0700562000274658,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03113105595111847,
+      "mean_token_accuracy": 0.9880475521087646,
+      "num_tokens": 48351627.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.7178894281387329,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.1952542066574097,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03204495906829834,
+      "mean_token_accuracy": 0.9880682706832886,
+      "num_tokens": 48540171.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.7166945099830627,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0920100212097168,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.03096899390220642,
+      "mean_token_accuracy": 0.9884315967559815,
+      "num_tokens": 48729737.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.7187432169914245,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.9596593976020813,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.03240742087364197,
+      "mean_token_accuracy": 0.9878904581069946,
+      "num_tokens": 48918599.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.7224443972110748,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9406276345252991,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.03253386616706848,
+      "mean_token_accuracy": 0.9873985171318054,
+      "num_tokens": 49107385.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.7226529896259308,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.2169438600540161,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.03282734453678131,
+      "mean_token_accuracy": 0.9876641929149628,
+      "num_tokens": 49295969.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.7263983964920044,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8808715343475342,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.030802983045578002,
+      "mean_token_accuracy": 0.9885481476783753,
+      "num_tokens": 49484706.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.7291677474975586,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2453594207763672,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.03403027057647705,
+      "mean_token_accuracy": 0.987172520160675,
+      "num_tokens": 49673484.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.7396404981613159,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.9215838313102722,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.03405522108078003,
+      "mean_token_accuracy": 0.9871597826480866,
+      "num_tokens": 49862087.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.7382308602333069,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0659712553024292,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.032089871168136594,
+      "mean_token_accuracy": 0.9874629497528076,
+      "num_tokens": 50051379.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.7337264895439148,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 0.9307538270950317,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.032426226139068606,
+      "mean_token_accuracy": 0.9879062473773956,
+      "num_tokens": 50239986.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.7381527841091156,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.1867440938949585,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.03276415467262268,
+      "mean_token_accuracy": 0.9878055810928345,
+      "num_tokens": 50428872.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.7337838113307953,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.9655186533927917,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.03141146302223206,
+      "mean_token_accuracy": 0.9882215678691864,
+      "num_tokens": 50618182.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.7354952573776246,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8848851919174194,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.03387039303779602,
+      "mean_token_accuracy": 0.9871303081512451,
+      "num_tokens": 50806430.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.7329667210578918,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 0.9976673722267151,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.03129216730594635,
+      "mean_token_accuracy": 0.9882283449172974,
+      "num_tokens": 50995500.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.7277809977531433,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.2560697793960571,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.03357822299003601,
+      "mean_token_accuracy": 0.9876434028148651,
+      "num_tokens": 51184436.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.727894538640976,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.1099125146865845,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.032482534646987915,
+      "mean_token_accuracy": 0.9878577649593353,
+      "num_tokens": 51373238.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.7318481564521789,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1980950832366943,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.03284388482570648,
+      "mean_token_accuracy": 0.9874962747097016,
+      "num_tokens": 51561631.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.7219811856746674,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 0.8730921745300293,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.030655479431152342,
+      "mean_token_accuracy": 0.9877778172492981,
+      "num_tokens": 51750845.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.7289035737514495,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.1195666790008545,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.03155539035797119,
+      "mean_token_accuracy": 0.9877057671546936,
+      "num_tokens": 51940009.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.7332375526428223,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9390997886657715,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.031307452917099,
+      "mean_token_accuracy": 0.987467783689499,
+      "num_tokens": 52128442.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.7293307244777679,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.0272324085235596,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.03318062424659729,
+      "mean_token_accuracy": 0.9873259484767913,
+      "num_tokens": 52317152.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.7320701718330384,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9884606003761292,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03461401164531708,
+      "mean_token_accuracy": 0.9867411673069,
+      "num_tokens": 52506106.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.7248260796070098,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.879682719707489,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.03406862616539001,
+      "mean_token_accuracy": 0.9872734069824218,
+      "num_tokens": 52695500.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.7237857282161713,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9409974217414856,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.029435336589813232,
+      "mean_token_accuracy": 0.9891225278377533,
+      "num_tokens": 52884940.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.7237291693687439,
+      "epoch": 3.781965006729475,
+      "grad_norm": 0.9532694220542908,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.030907022953033447,
+      "mean_token_accuracy": 0.9881488382816315,
+      "num_tokens": 53073909.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.7252696454524994,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 0.9157902002334595,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.032040882110595706,
+      "mean_token_accuracy": 0.9880847573280335,
+      "num_tokens": 53262962.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.7237608432769775,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.1450515985488892,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.03199286162853241,
+      "mean_token_accuracy": 0.987687611579895,
+      "num_tokens": 53451982.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.7237064421176911,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.9945085644721985,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.03347462117671966,
+      "mean_token_accuracy": 0.9875530540943146,
+      "num_tokens": 53640737.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.7271939218044281,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.9351736903190613,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.03140422701835632,
+      "mean_token_accuracy": 0.9877301037311554,
+      "num_tokens": 53829142.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.7265262603759766,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.1019784212112427,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.030748724937438965,
+      "mean_token_accuracy": 0.9881740033626556,
+      "num_tokens": 54018331.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.7270502150058746,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 0.9153032898902893,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.033472740650177,
+      "mean_token_accuracy": 0.9876999020576477,
+      "num_tokens": 54206935.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.7345028996467591,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.0144257545471191,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.033550959825515744,
+      "mean_token_accuracy": 0.9874906241893768,
+      "num_tokens": 54395443.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.7274066030979156,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0221407413482666,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.033388751745224,
+      "mean_token_accuracy": 0.9870682001113892,
+      "num_tokens": 54584552.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.7293415367603302,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.019165277481079,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.03568689227104187,
+      "mean_token_accuracy": 0.9861597299575806,
+      "num_tokens": 54773705.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.7281844198703766,
+      "epoch": 3.916554508748318,
+      "grad_norm": 0.933020830154419,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.030142486095428467,
+      "mean_token_accuracy": 0.988915103673935,
+      "num_tokens": 54962152.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.7242733418941498,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.9429674744606018,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.0337733268737793,
+      "mean_token_accuracy": 0.987263411283493,
+      "num_tokens": 55151179.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.7306612014770508,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.9022014737129211,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.03387616574764252,
+      "mean_token_accuracy": 0.9871586680412292,
+      "num_tokens": 55339658.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.7246532738208771,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2193421125411987,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.03193953037261963,
+      "mean_token_accuracy": 0.9879468739032745,
+      "num_tokens": 55528467.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.7268587052822113,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.2917358875274658,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.03584722280502319,
+      "mean_token_accuracy": 0.9867104470729828,
+      "num_tokens": 55717308.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.735340416431427,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.2229385375976562,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.03506708741188049,
+      "mean_token_accuracy": 0.9864643573760986,
+      "num_tokens": 55906059.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.7337038695812226,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.0747122764587402,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.03241217732429504,
+      "mean_token_accuracy": 0.9876143455505371,
+      "num_tokens": 56094606.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.7333687073106219,
+      "eval_loss": 0.07489899545907974,
+      "eval_mean_token_accuracy": 0.9719084019114257,
+      "eval_num_tokens": 56132350.0,
+      "eval_runtime": 12.7823,
+      "eval_samples_per_second": 391.166,
+      "eval_steps_per_second": 12.283,
+      "step": 2972
+    },
+    {
+      "entropy": 0.72461416721344,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7916642427444458,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.02286165803670883,
+      "mean_token_accuracy": 0.9920066714286804,
+      "num_tokens": 56283397.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.7150031447410583,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.9078621864318848,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.020906335115432738,
+      "mean_token_accuracy": 0.9925728499889374,
+      "num_tokens": 56472408.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.7091723501682281,
+      "epoch": 4.037685060565276,
+      "grad_norm": 0.8117756843566895,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.02036277800798416,
+      "mean_token_accuracy": 0.9928471207618713,
+      "num_tokens": 56660906.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.695214182138443,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7815372943878174,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019653698801994322,
+      "mean_token_accuracy": 0.992583978176117,
+      "num_tokens": 56850067.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.6937313377857208,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4769781827926636,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.02311961054801941,
+      "mean_token_accuracy": 0.9921472012996674,
+      "num_tokens": 57039005.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.6977694034576416,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.0028746128082275,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.021832706034183504,
+      "mean_token_accuracy": 0.9917877376079559,
+      "num_tokens": 57227782.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.698432594537735,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.0107522010803223,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.020331941545009613,
+      "mean_token_accuracy": 0.9932828366756439,
+      "num_tokens": 57416541.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.6941913783550262,
+      "epoch": 4.104979811574697,
+      "grad_norm": 0.8923842310905457,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019534927606582642,
+      "mean_token_accuracy": 0.9930490672588348,
+      "num_tokens": 57605162.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.6947623729705811,
+      "epoch": 4.118438761776582,
+      "grad_norm": 0.9093812108039856,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.02151128649711609,
+      "mean_token_accuracy": 0.9920440256595612,
+      "num_tokens": 57793756.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.6972509443759918,
+      "epoch": 4.131897711978466,
+      "grad_norm": 0.7549760341644287,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.020574498176574706,
+      "mean_token_accuracy": 0.9928867638111114,
+      "num_tokens": 57982678.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.6952514111995697,
+      "epoch": 4.14535666218035,
+      "grad_norm": 0.9476824402809143,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.02087453305721283,
+      "mean_token_accuracy": 0.9924874365329742,
+      "num_tokens": 58171508.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.6910251140594482,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.2548149824142456,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.020806866884231567,
+      "mean_token_accuracy": 0.9921584248542785,
+      "num_tokens": 58360594.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.6958921313285827,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.0649152994155884,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.020952591300010683,
+      "mean_token_accuracy": 0.9924699544906617,
+      "num_tokens": 58549637.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.6932485699653625,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.8891078233718872,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.020813333988189697,
+      "mean_token_accuracy": 0.9927419483661651,
+      "num_tokens": 58738595.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.6984799921512603,
+      "epoch": 4.199192462987887,
+      "grad_norm": 0.9459101557731628,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.01971060484647751,
+      "mean_token_accuracy": 0.9926240146160126,
+      "num_tokens": 58927507.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.7021916508674622,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.8318898677825928,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020860180258750916,
+      "mean_token_accuracy": 0.9927471876144409,
+      "num_tokens": 59116537.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.6972847402095794,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.9066984057426453,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.023059430718421935,
+      "mean_token_accuracy": 0.9919216752052307,
+      "num_tokens": 59305838.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.7049594700336457,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1277318000793457,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.022385355830192567,
+      "mean_token_accuracy": 0.99165118932724,
+      "num_tokens": 59494587.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.7045334160327912,
+      "epoch": 4.253028263795424,
+      "grad_norm": 0.8147489428520203,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.022358573973178864,
+      "mean_token_accuracy": 0.9918249011039734,
+      "num_tokens": 59683021.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.7007148265838623,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.9161480069160461,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.02088596671819687,
+      "mean_token_accuracy": 0.9925470650196075,
+      "num_tokens": 59871999.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.7006107687950134,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.9611289501190186,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.022619232535362244,
+      "mean_token_accuracy": 0.991577273607254,
+      "num_tokens": 60060486.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.7069303750991821,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.0339760780334473,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.021571077406406403,
+      "mean_token_accuracy": 0.9922139644622803,
+      "num_tokens": 60249686.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.7146372079849244,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.9761297702789307,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.022300006449222566,
+      "mean_token_accuracy": 0.9918746411800384,
+      "num_tokens": 60438481.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.7134014248847962,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 0.9729707837104797,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.02121303677558899,
+      "mean_token_accuracy": 0.9925692796707153,
+      "num_tokens": 60627501.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.7110478222370148,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.9456778168678284,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.020917078852653502,
+      "mean_token_accuracy": 0.9922434687614441,
+      "num_tokens": 60816135.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.7111951589584351,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8617656826972961,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.020851245522499083,
+      "mean_token_accuracy": 0.9924458563327789,
+      "num_tokens": 61005658.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.7193978548049926,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.8495420813560486,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.02084285914897919,
+      "mean_token_accuracy": 0.9927356660366058,
+      "num_tokens": 61194170.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.7188209414482116,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.728841245174408,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.02111642211675644,
+      "mean_token_accuracy": 0.9926697969436645,
+      "num_tokens": 61382852.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.7130864262580872,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.1274386644363403,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.021722708642482758,
+      "mean_token_accuracy": 0.9922453343868256,
+      "num_tokens": 61572195.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.7132770180702209,
+      "epoch": 4.401076716016151,
+      "grad_norm": 0.9251006245613098,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.021602503955364227,
+      "mean_token_accuracy": 0.9919042110443115,
+      "num_tokens": 61761119.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.7063322901725769,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.0182228088378906,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.020274630188941954,
+      "mean_token_accuracy": 0.9928637027740479,
+      "num_tokens": 61949931.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.7106044590473175,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.795170783996582,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.020594277977943422,
+      "mean_token_accuracy": 0.9925063610076904,
+      "num_tokens": 62138377.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7151111841201783,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1479791402816772,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.02117723524570465,
+      "mean_token_accuracy": 0.9921884059906005,
+      "num_tokens": 62327009.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.7172957301139832,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9795703291893005,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021917465329170226,
+      "mean_token_accuracy": 0.9915876984596252,
+      "num_tokens": 62515859.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.7175660550594329,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.9326112866401672,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.021142563223838805,
+      "mean_token_accuracy": 0.9922006845474243,
+      "num_tokens": 62704678.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.7085749089717865,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.8895787596702576,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.0222358375787735,
+      "mean_token_accuracy": 0.9918791234493256,
+      "num_tokens": 62893576.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.7059603810310364,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2298023700714111,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021999095380306245,
+      "mean_token_accuracy": 0.9920522570610046,
+      "num_tokens": 63082258.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.707107275724411,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.8295655250549316,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021844917535781862,
+      "mean_token_accuracy": 0.9922264873981476,
+      "num_tokens": 63271202.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.7119229018688202,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.7440339922904968,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.022164498269557954,
+      "mean_token_accuracy": 0.9918813884258271,
+      "num_tokens": 63460614.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.7065344154834747,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7275091409683228,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.019415955245494842,
+      "mean_token_accuracy": 0.9928511261940003,
+      "num_tokens": 63649327.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.6984272241592407,
+      "epoch": 4.549125168236878,
+      "grad_norm": 0.9989121556282043,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02176133990287781,
+      "mean_token_accuracy": 0.9923081934452057,
+      "num_tokens": 63837983.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.7084203243255616,
+      "epoch": 4.562584118438762,
+      "grad_norm": 0.9526837468147278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.02182276248931885,
+      "mean_token_accuracy": 0.9921061515808105,
+      "num_tokens": 64026855.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.7097050070762634,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.0156456232070923,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.020074066519737244,
+      "mean_token_accuracy": 0.9926879525184631,
+      "num_tokens": 64215889.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.7087334096431732,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.780275821685791,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.021703639626502992,
+      "mean_token_accuracy": 0.9917874097824096,
+      "num_tokens": 64404515.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.7142964243888855,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7382510304450989,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.019867205619812013,
+      "mean_token_accuracy": 0.9927354514598846,
+      "num_tokens": 64593147.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.7165492177009583,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.8259146213531494,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.021262936294078827,
+      "mean_token_accuracy": 0.9914182722568512,
+      "num_tokens": 64782041.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.7098793745040893,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.9540995359420776,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.0195529043674469,
+      "mean_token_accuracy": 0.9931159794330597,
+      "num_tokens": 64971390.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.7104229986667633,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.8698863387107849,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.019119253754615782,
+      "mean_token_accuracy": 0.9934523403644562,
+      "num_tokens": 65160253.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.7067269921302796,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.9622324705123901,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.02044745087623596,
+      "mean_token_accuracy": 0.992815512418747,
+      "num_tokens": 65349403.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.7016099870204926,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.999839723110199,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.019250848889350893,
+      "mean_token_accuracy": 0.9929339647293091,
+      "num_tokens": 65538164.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.7059290170669555,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.7539619207382202,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.02152409851551056,
+      "mean_token_accuracy": 0.99238161444664,
+      "num_tokens": 65727493.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.7103133797645569,
+      "epoch": 4.697173620457605,
+      "grad_norm": 0.9892463088035583,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020616935193538667,
+      "mean_token_accuracy": 0.9927784681320191,
+      "num_tokens": 65916233.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.703304260969162,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.1750195026397705,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021048617362976075,
+      "mean_token_accuracy": 0.9921808183193207,
+      "num_tokens": 66105229.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.7017257273197174,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.887427031993866,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02077500820159912,
+      "mean_token_accuracy": 0.992085075378418,
+      "num_tokens": 66294280.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.7027543246746063,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.0987571477890015,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.02167164385318756,
+      "mean_token_accuracy": 0.992034786939621,
+      "num_tokens": 66483182.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.7053992748260498,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.0075602531433105,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.022686071693897247,
+      "mean_token_accuracy": 0.9918968200683593,
+      "num_tokens": 66671795.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.7109473288059235,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.1655596494674683,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.022698774933815002,
+      "mean_token_accuracy": 0.9913642406463623,
+      "num_tokens": 66861166.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.7130867660045623,
+      "epoch": 4.77792732166891,
+      "grad_norm": 0.9386347532272339,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0205889493227005,
+      "mean_token_accuracy": 0.9924897074699401,
+      "num_tokens": 67049967.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.7099342107772827,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.8524489998817444,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.01985667645931244,
+      "mean_token_accuracy": 0.9928426086902619,
+      "num_tokens": 67238781.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.7100094079971313,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7522528171539307,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.02066437602043152,
+      "mean_token_accuracy": 0.9924224317073822,
+      "num_tokens": 67427802.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7172351717948914,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9865215420722961,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.021969664096832275,
+      "mean_token_accuracy": 0.9919721603393554,
+      "num_tokens": 67616675.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.7113273084163666,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9339645504951477,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.019794631004333495,
+      "mean_token_accuracy": 0.9925971984863281,
+      "num_tokens": 67805877.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.7041554272174835,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.337364912033081,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020412631332874298,
+      "mean_token_accuracy": 0.9923949122428894,
+      "num_tokens": 67994858.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.7090700149536133,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.7725968956947327,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.02121713161468506,
+      "mean_token_accuracy": 0.9920605599880219,
+      "num_tokens": 68183865.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.7141157209873199,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.8638854026794434,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.020726922154426574,
+      "mean_token_accuracy": 0.992779266834259,
+      "num_tokens": 68372954.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.7078199684619904,
+      "epoch": 4.885598923283984,
+      "grad_norm": 0.8576115965843201,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.019535741209983824,
+      "mean_token_accuracy": 0.9927346885204316,
+      "num_tokens": 68562535.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.7034667313098908,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8339338898658752,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.021528851985931397,
+      "mean_token_accuracy": 0.991979056596756,
+      "num_tokens": 68751258.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.6983348250389099,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8739761710166931,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020945993065834046,
+      "mean_token_accuracy": 0.9923328399658203,
+      "num_tokens": 68939586.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.6922877132892609,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.886886715888977,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.020974791049957274,
+      "mean_token_accuracy": 0.992312878370285,
+      "num_tokens": 69128270.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.6857815742492676,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.9415780901908875,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.020201480388641356,
+      "mean_token_accuracy": 0.9925193309783935,
+      "num_tokens": 69317066.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.6832823157310486,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.8311293721199036,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.02008202373981476,
+      "mean_token_accuracy": 0.9926081895828247,
+      "num_tokens": 69505698.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.6914379715919494,
+      "epoch": 4.96635262449529,
+      "grad_norm": 0.8799999356269836,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.0189486563205719,
+      "mean_token_accuracy": 0.9932616055011749,
+      "num_tokens": 69694164.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.692799162864685,
+      "epoch": 4.979811574697173,
+      "grad_norm": 0.9579370021820068,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.019743245840072633,
+      "mean_token_accuracy": 0.9932292222976684,
+      "num_tokens": 69882717.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.6969726324081421,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.0632585287094116,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.019562503695487975,
+      "mean_token_accuracy": 0.9929000854492187,
+      "num_tokens": 70071225.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.7056300662885047,
+      "eval_loss": 0.08702203631401062,
+      "eval_mean_token_accuracy": 0.9712980936287315,
+      "eval_num_tokens": 70165509.0,
+      "eval_runtime": 12.7642,
+      "eval_samples_per_second": 391.719,
+      "eval_steps_per_second": 12.3,
+      "step": 3715
+    },
+    {
+      "entropy": 0.6998905837535858,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8414336442947388,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01888625919818878,
+      "mean_token_accuracy": 0.9934305250644684,
+      "num_tokens": 70259811.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.6933371186256408,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.7618409395217896,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01257207989692688,
+      "mean_token_accuracy": 0.9958574831485748,
+      "num_tokens": 70448683.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.6861799776554107,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.935829758644104,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.0118876114487648,
+      "mean_token_accuracy": 0.9963350534439087,
+      "num_tokens": 70637078.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.6778797626495361,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.9417919516563416,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.012690700590610504,
+      "mean_token_accuracy": 0.9954731047153473,
+      "num_tokens": 70826145.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.6735071957111358,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.7570633292198181,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01321282535791397,
+      "mean_token_accuracy": 0.9955435931682587,
+      "num_tokens": 71015458.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.677568006515503,
+      "epoch": 5.074024226110363,
+      "grad_norm": 0.8157011866569519,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.012434301525354385,
+      "mean_token_accuracy": 0.9957576751708984,
+      "num_tokens": 71204332.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.6771604120731354,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.780483603477478,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.013237486779689788,
+      "mean_token_accuracy": 0.9957249999046326,
+      "num_tokens": 71393307.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.6761959791183472,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8249189257621765,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.011912986636161804,
+      "mean_token_accuracy": 0.9961272835731506,
+      "num_tokens": 71582233.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.6717506468296051,
+      "epoch": 5.114401076716016,
+      "grad_norm": 0.9667718410491943,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.012740559875965118,
+      "mean_token_accuracy": 0.9953394651412963,
+      "num_tokens": 71771906.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.6755871415138245,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.9068875908851624,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.012314559519290924,
+      "mean_token_accuracy": 0.9955949664115906,
+      "num_tokens": 71960698.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.6791335880756378,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.7650554776191711,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.011927495896816253,
+      "mean_token_accuracy": 0.9957858562469483,
+      "num_tokens": 72149627.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.6738516509532928,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 0.7267928123474121,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.011817440390586853,
+      "mean_token_accuracy": 0.996101975440979,
+      "num_tokens": 72338469.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.6701940059661865,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.9211037755012512,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012324941903352737,
+      "mean_token_accuracy": 0.9956477820873261,
+      "num_tokens": 72527499.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.6763210237026215,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.7772125601768494,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.013027346134185791,
+      "mean_token_accuracy": 0.9959511935710907,
+      "num_tokens": 72715639.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.6793705642223358,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.582334578037262,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.011401662230491638,
+      "mean_token_accuracy": 0.9961598336696624,
+      "num_tokens": 72904615.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.6798616409301758,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.8502725958824158,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011671387404203416,
+      "mean_token_accuracy": 0.9961067497730255,
+      "num_tokens": 73093522.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.682939213514328,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7421601414680481,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.011898189783096313,
+      "mean_token_accuracy": 0.9962145388126373,
+      "num_tokens": 73282415.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.6821306705474853,
+      "epoch": 5.235531628532974,
+      "grad_norm": 0.7552776336669922,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.01232050210237503,
+      "mean_token_accuracy": 0.9957845032215118,
+      "num_tokens": 73471211.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.6746087670326233,
+      "epoch": 5.248990578734858,
+      "grad_norm": 0.7144336700439453,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01240069791674614,
+      "mean_token_accuracy": 0.9960526466369629,
+      "num_tokens": 73660289.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.6723428070545197,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.8043408393859863,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.01204245388507843,
+      "mean_token_accuracy": 0.9960211217403412,
+      "num_tokens": 73849312.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.6766897737979889,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.8608251810073853,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012303075194358826,
+      "mean_token_accuracy": 0.9955043613910675,
+      "num_tokens": 74038344.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.6809355199337006,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.4872569739818573,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.010690021514892577,
+      "mean_token_accuracy": 0.9965942919254303,
+      "num_tokens": 74226981.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.6819032669067383,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.0376152992248535,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012148186564445496,
+      "mean_token_accuracy": 0.9958918631076813,
+      "num_tokens": 74415130.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.6856858193874359,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.7241474986076355,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012284471094608307,
+      "mean_token_accuracy": 0.9959208726882934,
+      "num_tokens": 74603788.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.6874188125133515,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.6505919098854065,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.012959125638008117,
+      "mean_token_accuracy": 0.9956629753112793,
+      "num_tokens": 74792306.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.6912538707256317,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7696103453636169,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012945668399333954,
+      "mean_token_accuracy": 0.9956756651401519,
+      "num_tokens": 74981265.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.6915560126304626,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.8245812058448792,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.01408768892288208,
+      "mean_token_accuracy": 0.9950980842113495,
+      "num_tokens": 75170337.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.6872740864753724,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8330141305923462,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.011667902022600174,
+      "mean_token_accuracy": 0.9960226893424988,
+      "num_tokens": 75359113.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.6825072944164277,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.9389140009880066,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012351009249687194,
+      "mean_token_accuracy": 0.9954748272895813,
+      "num_tokens": 75548102.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.6821270763874054,
+      "epoch": 5.397039030955585,
+      "grad_norm": 1.0638995170593262,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.0128641277551651,
+      "mean_token_accuracy": 0.995329612493515,
+      "num_tokens": 75736452.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.6857812821865081,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7838811278343201,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.011757558584213257,
+      "mean_token_accuracy": 0.9959722757339478,
+      "num_tokens": 75925013.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.6869213104248046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 0.5632259845733643,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.011284272372722625,
+      "mean_token_accuracy": 0.9962978959083557,
+      "num_tokens": 76113398.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.6807511866092681,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.7132044434547424,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012044224143028259,
+      "mean_token_accuracy": 0.9959294259548187,
+      "num_tokens": 76302629.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.6815520942211151,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.7640891075134277,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.012763911485671997,
+      "mean_token_accuracy": 0.9953242301940918,
+      "num_tokens": 76491675.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.6802713990211486,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.5691967010498047,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.0116192027926445,
+      "mean_token_accuracy": 0.9958603858947754,
+      "num_tokens": 76680426.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.676917290687561,
+      "epoch": 5.477792732166891,
+      "grad_norm": 0.8160638213157654,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.01229381114244461,
+      "mean_token_accuracy": 0.9956679701805115,
+      "num_tokens": 76869321.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.6745438575744629,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.8013969659805298,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.012171164155006409,
+      "mean_token_accuracy": 0.9958188533782959,
+      "num_tokens": 77058299.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.6774423599243165,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.77237868309021,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.013017389178276061,
+      "mean_token_accuracy": 0.9956206619739533,
+      "num_tokens": 77247567.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.6773760914802551,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.6185512542724609,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.011151286959648132,
+      "mean_token_accuracy": 0.9961191415786743,
+      "num_tokens": 77436417.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.6750624418258667,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.7832470536231995,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.011553966999053955,
+      "mean_token_accuracy": 0.9958207130432128,
+      "num_tokens": 77625255.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.6814670443534852,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.7947144508361816,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.011924812197685241,
+      "mean_token_accuracy": 0.9959113657474518,
+      "num_tokens": 77813632.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.68285973072052,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.6403945684432983,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.01066286414861679,
+      "mean_token_accuracy": 0.9966644048690796,
+      "num_tokens": 78002213.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.6764974176883698,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.8078592419624329,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.010837660729885101,
+      "mean_token_accuracy": 0.9961750328540802,
+      "num_tokens": 78191471.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.6726884067058563,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.7293245792388916,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.01087343692779541,
+      "mean_token_accuracy": 0.996295166015625,
+      "num_tokens": 78380424.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.6714434802532196,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0266199111938477,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.011094169318675995,
+      "mean_token_accuracy": 0.9960172355175019,
+      "num_tokens": 78569653.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.6689312815666199,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.8064524531364441,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.01098542958498001,
+      "mean_token_accuracy": 0.9961700737476349,
+      "num_tokens": 78759287.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.6728951573371887,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7976726293563843,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.012345843017101288,
+      "mean_token_accuracy": 0.9957349181175232,
+      "num_tokens": 78948032.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.6702533006668091,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.7385782599449158,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.011018334329128266,
+      "mean_token_accuracy": 0.9964040815830231,
+      "num_tokens": 79136972.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.67191202044487,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.6246873736381531,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.011075176298618317,
+      "mean_token_accuracy": 0.9962317526340485,
+      "num_tokens": 79325966.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.6736057937145233,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9479376077651978,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012818500399589539,
+      "mean_token_accuracy": 0.9951797604560852,
+      "num_tokens": 79514853.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.6723325371742248,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.6468302607536316,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.010920494794845581,
+      "mean_token_accuracy": 0.9961069703102112,
+      "num_tokens": 79703739.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.6762719631195069,
+      "epoch": 5.693135935397039,
+      "grad_norm": 0.6862959265708923,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.011675214767456055,
+      "mean_token_accuracy": 0.9958149492740631,
+      "num_tokens": 79892582.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.6739612460136414,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.8402584195137024,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.010931231081485748,
+      "mean_token_accuracy": 0.9960719287395478,
+      "num_tokens": 80081362.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.6767356872558594,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.6180281043052673,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.012499480694532394,
+      "mean_token_accuracy": 0.9957239329814911,
+      "num_tokens": 80269388.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.6773952782154083,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.8782367706298828,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012943127751350402,
+      "mean_token_accuracy": 0.9958240866661072,
+      "num_tokens": 80458191.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.6774130165576935,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.7854585647583008,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.012309956550598144,
+      "mean_token_accuracy": 0.99593386054039,
+      "num_tokens": 80647306.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.6813723623752594,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.6340908408164978,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.01146906390786171,
+      "mean_token_accuracy": 0.9963057219982148,
+      "num_tokens": 80836147.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.6804767549037933,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.5984897613525391,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.010843665152788163,
+      "mean_token_accuracy": 0.996556407213211,
+      "num_tokens": 81025302.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.6759022295475006,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.6014022827148438,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.010135546326637268,
+      "mean_token_accuracy": 0.9966514885425568,
+      "num_tokens": 81214251.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.675349909067154,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.6571088433265686,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01084519699215889,
+      "mean_token_accuracy": 0.9962432861328125,
+      "num_tokens": 81402608.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.6718680560588837,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.9250450730323792,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.01139993518590927,
+      "mean_token_accuracy": 0.9961999177932739,
+      "num_tokens": 81591841.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.674647331237793,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.8669414520263672,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.010886493325233459,
+      "mean_token_accuracy": 0.9962068378925324,
+      "num_tokens": 81780873.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.6801326870918274,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7395965456962585,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.012087566405534744,
+      "mean_token_accuracy": 0.9957775354385376,
+      "num_tokens": 81969835.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.676666659116745,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.7648837566375732,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.011214715242385865,
+      "mean_token_accuracy": 0.9963518619537354,
+      "num_tokens": 82159146.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.676488333940506,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.6776206493377686,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.011839983612298965,
+      "mean_token_accuracy": 0.9959003031253815,
+      "num_tokens": 82347867.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.6743977665901184,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.7158207893371582,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011066053062677383,
+      "mean_token_accuracy": 0.9966431736946106,
+      "num_tokens": 82536890.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.6764561176300049,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.7066213488578796,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.012056462466716766,
+      "mean_token_accuracy": 0.9958225846290588,
+      "num_tokens": 82725921.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.6813391983509064,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.763606607913971,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.011903972923755645,
+      "mean_token_accuracy": 0.9958972215652466,
+      "num_tokens": 82914838.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.6814220428466797,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.8407371044158936,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.011677531152963638,
+      "mean_token_accuracy": 0.9959126651287079,
+      "num_tokens": 83103937.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.6833201944828033,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8418315052986145,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012164801359176636,
+      "mean_token_accuracy": 0.9956748068332673,
+      "num_tokens": 83293106.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.6802135646343231,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.8431894183158875,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012858036160469054,
+      "mean_token_accuracy": 0.995720499753952,
+      "num_tokens": 83481559.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.6758989214897155,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7379446029663086,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.011657674610614777,
+      "mean_token_accuracy": 0.9959479212760926,
+      "num_tokens": 83670620.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.6832396805286407,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9370406866073608,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012775366008281708,
+      "mean_token_accuracy": 0.9958724677562714,
+      "num_tokens": 83858979.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.6826908767223359,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.5771452188491821,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011871224641799927,
+      "mean_token_accuracy": 0.9962578654289246,
+      "num_tokens": 84047664.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.6935603318700365,
+      "eval_loss": 0.08950015902519226,
+      "eval_mean_token_accuracy": 0.9720694904874085,
+      "eval_num_tokens": 84198646.0,
+      "eval_runtime": 12.7741,
+      "eval_samples_per_second": 391.416,
+      "eval_steps_per_second": 12.29,
+      "step": 4458
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.8917923305160704e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-4458/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-4458/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5201/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5201/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5201/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..34f5f643472ecdcbad0911c9bb971f5fb8604ec1
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7801c3eda22ba493e7ea5d14519804540590f6ad1a5cf9278105b41c0f21cc18
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5201/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5201/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5201/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5201/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..1bd82758cfef073a6e8f1b9bd9148c14cca143ab
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/trainer_state.json
@@ -0,0 +1,5311 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 5201,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    },
+    {
+      "entropy": 0.8332668900489807,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9914855360984802,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.06607928276062011,
+      "mean_token_accuracy": 0.974932211637497,
+      "num_tokens": 14165567.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8149146616458893,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.3080861568450928,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.057932257652282715,
+      "mean_token_accuracy": 0.9770712852478027,
+      "num_tokens": 14353912.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8197620987892151,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.0898637771606445,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.06010057330131531,
+      "mean_token_accuracy": 0.9758293390274048,
+      "num_tokens": 14542288.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8183245182037353,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.3120660781860352,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.0610861599445343,
+      "mean_token_accuracy": 0.9764986515045166,
+      "num_tokens": 14731028.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8135185301303863,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.138473391532898,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.06268800497055053,
+      "mean_token_accuracy": 0.9752912580966949,
+      "num_tokens": 14919682.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8142645537853241,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.1258258819580078,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.06271481513977051,
+      "mean_token_accuracy": 0.9757774949073792,
+      "num_tokens": 15107904.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8170483231544494,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.007297396659851,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.06338523030281067,
+      "mean_token_accuracy": 0.9754154622554779,
+      "num_tokens": 15297115.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8201254069805145,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 0.8240306973457336,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.06355815529823303,
+      "mean_token_accuracy": 0.9752808332443237,
+      "num_tokens": 15485921.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8140238881111145,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.0350368022918701,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.05969444513320923,
+      "mean_token_accuracy": 0.977089112997055,
+      "num_tokens": 15674824.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8133325934410095,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.0544878244400024,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.06062774658203125,
+      "mean_token_accuracy": 0.976290100812912,
+      "num_tokens": 15863526.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8185161769390106,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.1076316833496094,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.061567097902297974,
+      "mean_token_accuracy": 0.9755526125431061,
+      "num_tokens": 16051685.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8176139056682586,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.0057616233825684,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.06227490901947021,
+      "mean_token_accuracy": 0.9758114695549012,
+      "num_tokens": 16240451.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8217637002468109,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.97276771068573,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.06410331726074218,
+      "mean_token_accuracy": 0.9754030883312226,
+      "num_tokens": 16428961.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8141795754432678,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.0135555267333984,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.06253207921981811,
+      "mean_token_accuracy": 0.9760490536689759,
+      "num_tokens": 16617327.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8167653739452362,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2272467613220215,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.060455435514450075,
+      "mean_token_accuracy": 0.9758628964424133,
+      "num_tokens": 16806414.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8155323445796967,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 0.9328635334968567,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.06302794814109802,
+      "mean_token_accuracy": 0.9745101630687714,
+      "num_tokens": 16995016.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.818474805355072,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.1651655435562134,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.06043210029602051,
+      "mean_token_accuracy": 0.9766487181186676,
+      "num_tokens": 17184250.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8140560865402222,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.1005650758743286,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.06131819486618042,
+      "mean_token_accuracy": 0.9757712185382843,
+      "num_tokens": 17373075.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8204843640327454,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.0825532674789429,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.06296311616897583,
+      "mean_token_accuracy": 0.9745951414108276,
+      "num_tokens": 17561454.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8237834334373474,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.9842743873596191,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.061483919620513916,
+      "mean_token_accuracy": 0.9754831731319428,
+      "num_tokens": 17750758.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8165915131568908,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.0179396867752075,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.06237881183624268,
+      "mean_token_accuracy": 0.9750245690345765,
+      "num_tokens": 17940126.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8329708755016327,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0732359886169434,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.0644665539264679,
+      "mean_token_accuracy": 0.9743888020515442,
+      "num_tokens": 18129000.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8210074126720428,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.3696036338806152,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.06320792436599731,
+      "mean_token_accuracy": 0.9751390159130097,
+      "num_tokens": 18318232.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8248398661613464,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.327525019645691,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.06277769804000854,
+      "mean_token_accuracy": 0.9752413153648376,
+      "num_tokens": 18507403.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8284746050834656,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.044022560119629,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.06162315607070923,
+      "mean_token_accuracy": 0.9760516047477722,
+      "num_tokens": 18696242.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8233858644962311,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.062048077583313,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.059620130062103274,
+      "mean_token_accuracy": 0.9755307257175445,
+      "num_tokens": 18885199.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8236705601215363,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.2024412155151367,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.06304935216903687,
+      "mean_token_accuracy": 0.975091153383255,
+      "num_tokens": 19074028.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.828409218788147,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.1684212684631348,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.06451238393783569,
+      "mean_token_accuracy": 0.9743404746055603,
+      "num_tokens": 19262295.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8214354634284973,
+      "epoch": 1.386271870794078,
+      "grad_norm": 0.8512521386146545,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.06042641997337341,
+      "mean_token_accuracy": 0.9755558848381043,
+      "num_tokens": 19451154.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8223241448402405,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 0.8546568155288696,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.062134528160095216,
+      "mean_token_accuracy": 0.9750089883804322,
+      "num_tokens": 19639827.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.8217969954013824,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 0.972831666469574,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.060803139209747316,
+      "mean_token_accuracy": 0.9760692238807678,
+      "num_tokens": 19828174.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.8144341170787811,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1216644048690796,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.06458876132965088,
+      "mean_token_accuracy": 0.9738282442092896,
+      "num_tokens": 20016700.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.8168790876865387,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.0821806192398071,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.06258658170700074,
+      "mean_token_accuracy": 0.9749014675617218,
+      "num_tokens": 20205823.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8146390020847321,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.0067790746688843,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.06509006023406982,
+      "mean_token_accuracy": 0.9737810134887696,
+      "num_tokens": 20394611.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8224801063537598,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.0583593845367432,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.06343590021133423,
+      "mean_token_accuracy": 0.9750119090080261,
+      "num_tokens": 20583252.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.8083258986473083,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.8305535316467285,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.06573111414909363,
+      "mean_token_accuracy": 0.9736428141593934,
+      "num_tokens": 20772107.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8081986546516419,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5895942449569702,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.06376024484634399,
+      "mean_token_accuracy": 0.9751200795173645,
+      "num_tokens": 20960494.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.8026683986186981,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 0.9962654113769531,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.06282572746276856,
+      "mean_token_accuracy": 0.9753849327564239,
+      "num_tokens": 21149763.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8030014991760254,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9083836078643799,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.06473478078842163,
+      "mean_token_accuracy": 0.9747069597244262,
+      "num_tokens": 21338690.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.8008662164211273,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.097593069076538,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.05911685228347778,
+      "mean_token_accuracy": 0.9767139732837677,
+      "num_tokens": 21527420.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7981986284255982,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.1673516035079956,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.05957244038581848,
+      "mean_token_accuracy": 0.9761911332607269,
+      "num_tokens": 21716791.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.8004718601703644,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 0.7955246567726135,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.06378913521766663,
+      "mean_token_accuracy": 0.9747395157814026,
+      "num_tokens": 21905487.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.8039821267127991,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.3141396045684814,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.06209801435470581,
+      "mean_token_accuracy": 0.9754719436168671,
+      "num_tokens": 22094436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.8051789462566376,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.2227814197540283,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.06261310577392579,
+      "mean_token_accuracy": 0.9753049433231353,
+      "num_tokens": 22283243.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7941994190216064,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.071070671081543,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.05962263345718384,
+      "mean_token_accuracy": 0.9763956189155578,
+      "num_tokens": 22472084.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7937953472137451,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.0329782962799072,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.060611891746521,
+      "mean_token_accuracy": 0.9758659780025483,
+      "num_tokens": 22661614.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7913113713264466,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.032772183418274,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.06002308130264282,
+      "mean_token_accuracy": 0.9758916199207306,
+      "num_tokens": 22850634.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.7909658491611481,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3234598636627197,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.06296972036361695,
+      "mean_token_accuracy": 0.9741117715835571,
+      "num_tokens": 23039402.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7880761086940765,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.9563435316085815,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.06209538578987121,
+      "mean_token_accuracy": 0.9750530481338501,
+      "num_tokens": 23228363.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7845663070678711,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 0.8875390291213989,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.06272958517074585,
+      "mean_token_accuracy": 0.9751476228237153,
+      "num_tokens": 23417469.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7928634822368622,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.0629489421844482,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.060286092758178714,
+      "mean_token_accuracy": 0.976036274433136,
+      "num_tokens": 23606274.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7870154738426208,
+      "epoch": 1.695827725437416,
+      "grad_norm": 0.9164250493049622,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.06263296008110046,
+      "mean_token_accuracy": 0.9752088189125061,
+      "num_tokens": 23795255.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.7948406875133515,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.0909371376037598,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.06165047883987427,
+      "mean_token_accuracy": 0.9764420628547669,
+      "num_tokens": 23984534.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7862435758113862,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.1839547157287598,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.0640158474445343,
+      "mean_token_accuracy": 0.9741048216819763,
+      "num_tokens": 24173632.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.7887867331504822,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.0730741024017334,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.06260218620300292,
+      "mean_token_accuracy": 0.974741768836975,
+      "num_tokens": 24362671.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7896574556827545,
+      "epoch": 1.749663526244953,
+      "grad_norm": 0.9559605121612549,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.061297619342803956,
+      "mean_token_accuracy": 0.9752223670482636,
+      "num_tokens": 24551825.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.7902800619602204,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.914057195186615,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.06320110559463502,
+      "mean_token_accuracy": 0.9749482691287994,
+      "num_tokens": 24740867.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.799742591381073,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.06534289717674255,
+      "mean_token_accuracy": 0.9741291284561158,
+      "num_tokens": 24930132.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7974063038825989,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1188615560531616,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.060908573865890506,
+      "mean_token_accuracy": 0.9765798091888428,
+      "num_tokens": 25119678.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.786439061164856,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9185847640037537,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.06158917546272278,
+      "mean_token_accuracy": 0.9760335922241211,
+      "num_tokens": 25308876.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7888528347015381,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.1948838233947754,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.060635948181152345,
+      "mean_token_accuracy": 0.976032954454422,
+      "num_tokens": 25498102.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7968044698238372,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.4128663539886475,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.06448977589607238,
+      "mean_token_accuracy": 0.9741183638572692,
+      "num_tokens": 25686747.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7967679858207702,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.9797796010971069,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.06342286467552186,
+      "mean_token_accuracy": 0.9752787709236145,
+      "num_tokens": 25875368.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7932188332080841,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.0549253225326538,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.060981249809265135,
+      "mean_token_accuracy": 0.976200807094574,
+      "num_tokens": 26064480.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7899552762508393,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.1172103881835938,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.06555815935134887,
+      "mean_token_accuracy": 0.9736676931381225,
+      "num_tokens": 26253619.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.8019726693630218,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.94905686378479,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.0663321852684021,
+      "mean_token_accuracy": 0.9739456713199616,
+      "num_tokens": 26442736.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7939825117588043,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 0.9922711253166199,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05981417298316956,
+      "mean_token_accuracy": 0.9758030951023102,
+      "num_tokens": 26631525.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8085412085056305,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4970781803131104,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.06708773374557495,
+      "mean_token_accuracy": 0.9737519145011901,
+      "num_tokens": 26820483.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7950678527355194,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.7868127226829529,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.061734098196029666,
+      "mean_token_accuracy": 0.9747878730297088,
+      "num_tokens": 27009393.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7850766360759736,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.1386486291885376,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.05714603662490845,
+      "mean_token_accuracy": 0.9769004642963409,
+      "num_tokens": 27198251.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7933461844921113,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.0092092752456665,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.06240262985229492,
+      "mean_token_accuracy": 0.9742367327213287,
+      "num_tokens": 27387231.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8044621467590332,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.2477015256881714,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0642085313796997,
+      "mean_token_accuracy": 0.9748865723609924,
+      "num_tokens": 27576251.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7975574374198914,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.1167609691619873,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.06382874846458435,
+      "mean_token_accuracy": 0.9748350441455841,
+      "num_tokens": 27764560.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.8090806484222413,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 0.9714348912239075,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.06401748657226562,
+      "mean_token_accuracy": 0.9741845309734345,
+      "num_tokens": 27953532.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.8104567945383157,
+      "eval_loss": 0.07282773405313492,
+      "eval_mean_token_accuracy": 0.9701232534305305,
+      "eval_num_tokens": 28066151.0,
+      "eval_runtime": 12.7852,
+      "eval_samples_per_second": 391.078,
+      "eval_steps_per_second": 12.28,
+      "step": 1486
+    },
+    {
+      "entropy": 0.8090688765048981,
+      "epoch": 2.005383580080754,
+      "grad_norm": 0.9886816143989563,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.06012396812438965,
+      "mean_token_accuracy": 0.9764885842800141,
+      "num_tokens": 28141848.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.7821709334850311,
+      "epoch": 2.018842530282638,
+      "grad_norm": 0.9049969911575317,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.04819132387638092,
+      "mean_token_accuracy": 0.9816783249378205,
+      "num_tokens": 28330784.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.7845205307006836,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0231750011444092,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.04539612233638764,
+      "mean_token_accuracy": 0.9818619966506958,
+      "num_tokens": 28519580.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.7817754447460175,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.5244711637496948,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0429688960313797,
+      "mean_token_accuracy": 0.9834935545921326,
+      "num_tokens": 28707486.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.7733526527881622,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.0685060024261475,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.045852530002594,
+      "mean_token_accuracy": 0.9817728459835052,
+      "num_tokens": 28896759.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.7815343141555786,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.075853705406189,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.04565088748931885,
+      "mean_token_accuracy": 0.9823914349079133,
+      "num_tokens": 29085097.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.7751398742198944,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2573742866516113,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.04465666115283966,
+      "mean_token_accuracy": 0.9829874813556672,
+      "num_tokens": 29273868.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.7737197399139404,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 0.8681219220161438,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.042188242077827454,
+      "mean_token_accuracy": 0.9837059378623962,
+      "num_tokens": 29462985.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.7823441684246063,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.3503504991531372,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.04393318593502045,
+      "mean_token_accuracy": 0.9827627122402192,
+      "num_tokens": 29651156.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.7796198308467865,
+      "epoch": 2.126514131897712,
+      "grad_norm": 0.7799349427223206,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.04367418885231018,
+      "mean_token_accuracy": 0.9837401032447814,
+      "num_tokens": 29839789.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.7767675697803498,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.375465750694275,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.04427264630794525,
+      "mean_token_accuracy": 0.9832019031047821,
+      "num_tokens": 30028957.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.7912136912345886,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.9683088660240173,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.04428936243057251,
+      "mean_token_accuracy": 0.9832306385040284,
+      "num_tokens": 30217841.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.790750366449356,
+      "epoch": 2.166890982503365,
+      "grad_norm": 0.9393274188041687,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.04529332816600799,
+      "mean_token_accuracy": 0.9832263946533203,
+      "num_tokens": 30406540.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.7816023886203766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.1822642087936401,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.04192846417427063,
+      "mean_token_accuracy": 0.9846727311611175,
+      "num_tokens": 30595568.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7754982054233551,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.1961082220077515,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.043791168928146364,
+      "mean_token_accuracy": 0.9830836653709412,
+      "num_tokens": 30784300.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.7756961703300476,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 0.9719674587249756,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.04410521686077118,
+      "mean_token_accuracy": 0.9837198793888092,
+      "num_tokens": 30972950.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.7833336114883422,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.0175098180770874,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.044813218712806704,
+      "mean_token_accuracy": 0.9829723179340363,
+      "num_tokens": 31161783.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.7893047273159027,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.2077007293701172,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.04417986273765564,
+      "mean_token_accuracy": 0.9825236797332764,
+      "num_tokens": 31350487.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.794023334980011,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8530077934265137,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.043008115887641904,
+      "mean_token_accuracy": 0.9832422196865082,
+      "num_tokens": 31539217.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.785652619600296,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.4543156623840332,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0445009708404541,
+      "mean_token_accuracy": 0.9828064203262329,
+      "num_tokens": 31728308.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7841614365577698,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.297763466835022,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.045649731159210206,
+      "mean_token_accuracy": 0.9823782920837403,
+      "num_tokens": 31917293.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.7825099349021911,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.2237403392791748,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.047218257188797,
+      "mean_token_accuracy": 0.98156818151474,
+      "num_tokens": 32106628.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.7791289031505585,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8094402551651001,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.04178315699100495,
+      "mean_token_accuracy": 0.9846469461917877,
+      "num_tokens": 32295183.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.7743105947971344,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.3677793741226196,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.04540807008743286,
+      "mean_token_accuracy": 0.9827918827533721,
+      "num_tokens": 32484143.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.7804521083831787,
+      "epoch": 2.328398384925976,
+      "grad_norm": 0.8883333802223206,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.04598270654678345,
+      "mean_token_accuracy": 0.9823103487491608,
+      "num_tokens": 32673144.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.7760683119297027,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.989972710609436,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.043702703714370725,
+      "mean_token_accuracy": 0.9831363618373871,
+      "num_tokens": 32862142.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7726980447769165,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.0461052656173706,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.044418123364448545,
+      "mean_token_accuracy": 0.9821757376194,
+      "num_tokens": 33051382.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.7822885036468505,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.9027822017669678,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.04624638557434082,
+      "mean_token_accuracy": 0.9815774321556091,
+      "num_tokens": 33239729.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.7830030500888825,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.1743420362472534,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.04747769832611084,
+      "mean_token_accuracy": 0.9812638640403748,
+      "num_tokens": 33428615.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.7933187901973724,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 0.9334444403648376,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.047731178998947146,
+      "mean_token_accuracy": 0.9820823550224305,
+      "num_tokens": 33617331.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.7904782056808471,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.9396253824234009,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.04432434141635895,
+      "mean_token_accuracy": 0.9830163478851318,
+      "num_tokens": 33805978.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.7811533212661743,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 0.9742764234542847,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.046493279933929446,
+      "mean_token_accuracy": 0.9818211495876312,
+      "num_tokens": 33994957.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7848010778427124,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0378986597061157,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0454249233007431,
+      "mean_token_accuracy": 0.9817994356155395,
+      "num_tokens": 34184297.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.7754390180110932,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.1023566722869873,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.04380808770656586,
+      "mean_token_accuracy": 0.9829573810100556,
+      "num_tokens": 34373551.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.7767188012599945,
+      "epoch": 2.462987886944818,
+      "grad_norm": 0.8460658192634583,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04502716064453125,
+      "mean_token_accuracy": 0.9825617015361786,
+      "num_tokens": 34562169.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.7757977426052094,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.1013047695159912,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.045161890983581546,
+      "mean_token_accuracy": 0.9821914732456207,
+      "num_tokens": 34750863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.785107308626175,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.9510554671287537,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.045189899206161496,
+      "mean_token_accuracy": 0.9822288691997528,
+      "num_tokens": 34939528.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.7849742233753204,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0994503498077393,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.04333080947399139,
+      "mean_token_accuracy": 0.9830258965492249,
+      "num_tokens": 35128334.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.7875016331672668,
+      "epoch": 2.516823687752355,
+      "grad_norm": 0.9202979803085327,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.04838588833808899,
+      "mean_token_accuracy": 0.9815591931343078,
+      "num_tokens": 35317067.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.7813792943954467,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.9036716222763062,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.045662492513656616,
+      "mean_token_accuracy": 0.9817738354206085,
+      "num_tokens": 35506291.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.777835363149643,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.853438675403595,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.044686585664749146,
+      "mean_token_accuracy": 0.98329097032547,
+      "num_tokens": 35695141.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.7677761793136597,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.0699759721755981,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.045490306615829465,
+      "mean_token_accuracy": 0.9826831936836242,
+      "num_tokens": 35884173.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.7756118893623352,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.0356824398040771,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.0467751681804657,
+      "mean_token_accuracy": 0.9823338150978088,
+      "num_tokens": 36073241.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.7775443732738495,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.1619561910629272,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.04741084575653076,
+      "mean_token_accuracy": 0.9815192580223083,
+      "num_tokens": 36262133.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7899596333503723,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.3479284048080444,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.04588942229747772,
+      "mean_token_accuracy": 0.9821865320205688,
+      "num_tokens": 36451460.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.7832387804985046,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.6297204494476318,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.04711659252643585,
+      "mean_token_accuracy": 0.9815292179584503,
+      "num_tokens": 36640366.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.7871865928173065,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.3100138902664185,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.04919236898422241,
+      "mean_token_accuracy": 0.9809704780578613,
+      "num_tokens": 36829017.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.7844007611274719,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.2907201051712036,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.047159326076507566,
+      "mean_token_accuracy": 0.9816433548927307,
+      "num_tokens": 37018034.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.7885129511356354,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.1324766874313354,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.04420983195304871,
+      "mean_token_accuracy": 0.9828807294368744,
+      "num_tokens": 37206768.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.7761563122272491,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.9269904494285583,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.04585547745227814,
+      "mean_token_accuracy": 0.9823735177516937,
+      "num_tokens": 37396010.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7831155300140381,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5890840291976929,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.046450641751289365,
+      "mean_token_accuracy": 0.9817734777927398,
+      "num_tokens": 37584809.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.7729385912418365,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.4823863506317139,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.04653529226779938,
+      "mean_token_accuracy": 0.9815726220607758,
+      "num_tokens": 37773869.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.7752901673316955,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.9423630237579346,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.04599723219871521,
+      "mean_token_accuracy": 0.9816120624542236,
+      "num_tokens": 37962642.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.7750158667564392,
+      "epoch": 2.718707940780619,
+      "grad_norm": 0.9398621320724487,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.04729728400707245,
+      "mean_token_accuracy": 0.9816868543624878,
+      "num_tokens": 38151325.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.7693424999713898,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3207950592041016,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.04474019706249237,
+      "mean_token_accuracy": 0.9830808460712432,
+      "num_tokens": 38340719.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.7750859558582306,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.070200800895691,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.045914703607559205,
+      "mean_token_accuracy": 0.9820488631725312,
+      "num_tokens": 38529293.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7697387218475342,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.0545568466186523,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.04357335567474365,
+      "mean_token_accuracy": 0.9827888309955597,
+      "num_tokens": 38718464.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.7682756781578064,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.0945985317230225,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.0429091215133667,
+      "mean_token_accuracy": 0.9832668721675872,
+      "num_tokens": 38907573.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.7804523587226868,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8521431684494019,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.04662351012229919,
+      "mean_token_accuracy": 0.981964933872223,
+      "num_tokens": 39096598.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.7786886096000671,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.09945809841156,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.050166720151901247,
+      "mean_token_accuracy": 0.9816301107406616,
+      "num_tokens": 39285069.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.7790449261665344,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.1673868894577026,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.048508059978485105,
+      "mean_token_accuracy": 0.9809533059597015,
+      "num_tokens": 39474252.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.7774537563323974,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.0648490190505981,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.04540320634841919,
+      "mean_token_accuracy": 0.9822351098060608,
+      "num_tokens": 39663262.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7683943033218383,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.0106912851333618,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.04466369152069092,
+      "mean_token_accuracy": 0.9818881273269653,
+      "num_tokens": 39852133.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.7642774760723114,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0109530687332153,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.045978450775146486,
+      "mean_token_accuracy": 0.9819821894168854,
+      "num_tokens": 40041679.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.7647961139678955,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.9050237536430359,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.045694130659103396,
+      "mean_token_accuracy": 0.982342666387558,
+      "num_tokens": 40230019.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.7589850842952728,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.8879176378250122,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.046186423301696776,
+      "mean_token_accuracy": 0.9818771004676818,
+      "num_tokens": 40418932.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.765201997756958,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 0.9103795886039734,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.04475873112678528,
+      "mean_token_accuracy": 0.9821203172206878,
+      "num_tokens": 40607792.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.7653619945049286,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.0566699504852295,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.04391545057296753,
+      "mean_token_accuracy": 0.9828326046466828,
+      "num_tokens": 40797122.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.763550215959549,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.1280677318572998,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.045582184195518495,
+      "mean_token_accuracy": 0.9819066524505615,
+      "num_tokens": 40985689.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.7749062657356263,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.110323429107666,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.04598402082920074,
+      "mean_token_accuracy": 0.9815545618534088,
+      "num_tokens": 41174743.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.7805372893810272,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.0649195909500122,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.04705569744110107,
+      "mean_token_accuracy": 0.9815925776958465,
+      "num_tokens": 41363160.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.7711063683032989,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.1459592580795288,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.045355689525604245,
+      "mean_token_accuracy": 0.9832463562488556,
+      "num_tokens": 41551932.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.7747956931591033,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.1602799892425537,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.04494523704051971,
+      "mean_token_accuracy": 0.9819566667079925,
+      "num_tokens": 41740544.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.7584574401378632,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.054417371749878,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.04810510575771332,
+      "mean_token_accuracy": 0.9811540365219116,
+      "num_tokens": 41929344.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.7753648416251894,
+      "eval_loss": 0.07181113958358765,
+      "eval_mean_token_accuracy": 0.9716544128527307,
+      "eval_num_tokens": 42099366.0,
+      "eval_runtime": 12.748,
+      "eval_samples_per_second": 392.22,
+      "eval_steps_per_second": 12.316,
+      "step": 2229
+    },
+    {
+      "entropy": 0.7669954419136047,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.8211690783500671,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.04577431380748749,
+      "mean_token_accuracy": 0.9823363184928894,
+      "num_tokens": 42118188.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.7600219964981079,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.0873607397079468,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.032501548528671265,
+      "mean_token_accuracy": 0.9881355822086334,
+      "num_tokens": 42306598.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.7441959500312805,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.1175898313522339,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.031047609448432923,
+      "mean_token_accuracy": 0.9891582727432251,
+      "num_tokens": 42495250.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.7364335536956788,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.987751841545105,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.028629058599472047,
+      "mean_token_accuracy": 0.9894859492778778,
+      "num_tokens": 42684326.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.735680228471756,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.194305658340454,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.029283303022384643,
+      "mean_token_accuracy": 0.9887769162654877,
+      "num_tokens": 42873221.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.7342639684677124,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.254442811012268,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.0302200973033905,
+      "mean_token_accuracy": 0.988795381784439,
+      "num_tokens": 43062486.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.7391974747180938,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.2648969888687134,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.0316451758146286,
+      "mean_token_accuracy": 0.9879588901996612,
+      "num_tokens": 43251113.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.7401826322078705,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.0483295917510986,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.030878132581710814,
+      "mean_token_accuracy": 0.9886977672576904,
+      "num_tokens": 43440152.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.732647967338562,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.1125447750091553,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.02838427424430847,
+      "mean_token_accuracy": 0.9890213429927825,
+      "num_tokens": 43629353.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.7314932703971863,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.0979042053222656,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.0315592348575592,
+      "mean_token_accuracy": 0.9888832688331604,
+      "num_tokens": 43818981.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.7305827140808105,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9610060453414917,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.03112468421459198,
+      "mean_token_accuracy": 0.9890477001667023,
+      "num_tokens": 44008430.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.7333625435829163,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.0625078678131104,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.03105958104133606,
+      "mean_token_accuracy": 0.9884848177433014,
+      "num_tokens": 44197264.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.7373903751373291,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.1683465242385864,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.031627464294433597,
+      "mean_token_accuracy": 0.9879396498203278,
+      "num_tokens": 44386205.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.7391358911991119,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.05154287815094,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.03118889331817627,
+      "mean_token_accuracy": 0.9880312740802765,
+      "num_tokens": 44575099.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.7351223647594451,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.9531214237213135,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.030980610847473146,
+      "mean_token_accuracy": 0.9882127285003662,
+      "num_tokens": 44763965.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.7361315310001373,
+      "epoch": 3.203230148048452,
+      "grad_norm": 0.8818249702453613,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.029705989360809325,
+      "mean_token_accuracy": 0.9894622683525085,
+      "num_tokens": 44952983.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.7356473922729492,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 0.9263529777526855,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.03202669024467468,
+      "mean_token_accuracy": 0.9877129793167114,
+      "num_tokens": 45142080.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.7326373934745789,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.0517516136169434,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.031153690814971925,
+      "mean_token_accuracy": 0.9891136825084687,
+      "num_tokens": 45330769.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.7348711133003235,
+      "epoch": 3.243606998654105,
+      "grad_norm": 0.8774152994155884,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030574262142181396,
+      "mean_token_accuracy": 0.9888435959815979,
+      "num_tokens": 45519217.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.7379967510700226,
+      "epoch": 3.257065948855989,
+      "grad_norm": 0.9164374470710754,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.03096820116043091,
+      "mean_token_accuracy": 0.9884447932243348,
+      "num_tokens": 45708467.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.7369260907173156,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.3184924125671387,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03082413077354431,
+      "mean_token_accuracy": 0.9888757109642029,
+      "num_tokens": 45897424.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.7366503119468689,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.9073860049247742,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.03220943808555603,
+      "mean_token_accuracy": 0.9877902865409851,
+      "num_tokens": 46086351.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.7341453731060028,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.0940542221069336,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.03329495787620544,
+      "mean_token_accuracy": 0.9877542018890381,
+      "num_tokens": 46275394.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.7312689840793609,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 0.9651952385902405,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.031667912006378175,
+      "mean_token_accuracy": 0.9881450653076171,
+      "num_tokens": 46463821.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.7228557825088501,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 0.971441924571991,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.030608028173446655,
+      "mean_token_accuracy": 0.9882231116294861,
+      "num_tokens": 46653025.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.723542982339859,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.0060454607009888,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.030545371770858764,
+      "mean_token_accuracy": 0.9879125893115998,
+      "num_tokens": 46842068.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.7321009397506714,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.0572997331619263,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.032070663571357724,
+      "mean_token_accuracy": 0.9878279268741608,
+      "num_tokens": 47031045.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.735515958070755,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.9835748672485352,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.03338083028793335,
+      "mean_token_accuracy": 0.9877255499362946,
+      "num_tokens": 47219625.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.7396618247032165,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 0.9655317068099976,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.03589712679386139,
+      "mean_token_accuracy": 0.9869556546211242,
+      "num_tokens": 47408311.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.7358698546886444,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9225422143936157,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.02914820909500122,
+      "mean_token_accuracy": 0.9886718571186066,
+      "num_tokens": 47597297.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.7338236391544342,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.8253662586212158,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.02996352016925812,
+      "mean_token_accuracy": 0.9891344487667084,
+      "num_tokens": 47785737.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.7320856511592865,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.0716700553894043,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.032322123646736145,
+      "mean_token_accuracy": 0.9877110660076142,
+      "num_tokens": 47974241.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.7293457925319672,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.0915459394454956,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.031131434440612792,
+      "mean_token_accuracy": 0.9885926306247711,
+      "num_tokens": 48162582.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.7207926452159882,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0700562000274658,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03113105595111847,
+      "mean_token_accuracy": 0.9880475521087646,
+      "num_tokens": 48351627.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.7178894281387329,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.1952542066574097,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03204495906829834,
+      "mean_token_accuracy": 0.9880682706832886,
+      "num_tokens": 48540171.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.7166945099830627,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0920100212097168,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.03096899390220642,
+      "mean_token_accuracy": 0.9884315967559815,
+      "num_tokens": 48729737.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.7187432169914245,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.9596593976020813,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.03240742087364197,
+      "mean_token_accuracy": 0.9878904581069946,
+      "num_tokens": 48918599.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.7224443972110748,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9406276345252991,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.03253386616706848,
+      "mean_token_accuracy": 0.9873985171318054,
+      "num_tokens": 49107385.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.7226529896259308,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.2169438600540161,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.03282734453678131,
+      "mean_token_accuracy": 0.9876641929149628,
+      "num_tokens": 49295969.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.7263983964920044,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8808715343475342,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.030802983045578002,
+      "mean_token_accuracy": 0.9885481476783753,
+      "num_tokens": 49484706.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.7291677474975586,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2453594207763672,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.03403027057647705,
+      "mean_token_accuracy": 0.987172520160675,
+      "num_tokens": 49673484.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.7396404981613159,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.9215838313102722,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.03405522108078003,
+      "mean_token_accuracy": 0.9871597826480866,
+      "num_tokens": 49862087.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.7382308602333069,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0659712553024292,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.032089871168136594,
+      "mean_token_accuracy": 0.9874629497528076,
+      "num_tokens": 50051379.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.7337264895439148,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 0.9307538270950317,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.032426226139068606,
+      "mean_token_accuracy": 0.9879062473773956,
+      "num_tokens": 50239986.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.7381527841091156,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.1867440938949585,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.03276415467262268,
+      "mean_token_accuracy": 0.9878055810928345,
+      "num_tokens": 50428872.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.7337838113307953,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.9655186533927917,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.03141146302223206,
+      "mean_token_accuracy": 0.9882215678691864,
+      "num_tokens": 50618182.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.7354952573776246,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8848851919174194,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.03387039303779602,
+      "mean_token_accuracy": 0.9871303081512451,
+      "num_tokens": 50806430.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.7329667210578918,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 0.9976673722267151,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.03129216730594635,
+      "mean_token_accuracy": 0.9882283449172974,
+      "num_tokens": 50995500.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.7277809977531433,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.2560697793960571,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.03357822299003601,
+      "mean_token_accuracy": 0.9876434028148651,
+      "num_tokens": 51184436.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.727894538640976,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.1099125146865845,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.032482534646987915,
+      "mean_token_accuracy": 0.9878577649593353,
+      "num_tokens": 51373238.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.7318481564521789,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1980950832366943,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.03284388482570648,
+      "mean_token_accuracy": 0.9874962747097016,
+      "num_tokens": 51561631.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.7219811856746674,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 0.8730921745300293,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.030655479431152342,
+      "mean_token_accuracy": 0.9877778172492981,
+      "num_tokens": 51750845.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.7289035737514495,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.1195666790008545,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.03155539035797119,
+      "mean_token_accuracy": 0.9877057671546936,
+      "num_tokens": 51940009.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.7332375526428223,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9390997886657715,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.031307452917099,
+      "mean_token_accuracy": 0.987467783689499,
+      "num_tokens": 52128442.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.7293307244777679,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.0272324085235596,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.03318062424659729,
+      "mean_token_accuracy": 0.9873259484767913,
+      "num_tokens": 52317152.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.7320701718330384,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9884606003761292,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03461401164531708,
+      "mean_token_accuracy": 0.9867411673069,
+      "num_tokens": 52506106.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.7248260796070098,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.879682719707489,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.03406862616539001,
+      "mean_token_accuracy": 0.9872734069824218,
+      "num_tokens": 52695500.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.7237857282161713,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9409974217414856,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.029435336589813232,
+      "mean_token_accuracy": 0.9891225278377533,
+      "num_tokens": 52884940.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.7237291693687439,
+      "epoch": 3.781965006729475,
+      "grad_norm": 0.9532694220542908,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.030907022953033447,
+      "mean_token_accuracy": 0.9881488382816315,
+      "num_tokens": 53073909.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.7252696454524994,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 0.9157902002334595,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.032040882110595706,
+      "mean_token_accuracy": 0.9880847573280335,
+      "num_tokens": 53262962.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.7237608432769775,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.1450515985488892,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.03199286162853241,
+      "mean_token_accuracy": 0.987687611579895,
+      "num_tokens": 53451982.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.7237064421176911,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.9945085644721985,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.03347462117671966,
+      "mean_token_accuracy": 0.9875530540943146,
+      "num_tokens": 53640737.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.7271939218044281,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.9351736903190613,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.03140422701835632,
+      "mean_token_accuracy": 0.9877301037311554,
+      "num_tokens": 53829142.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.7265262603759766,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.1019784212112427,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.030748724937438965,
+      "mean_token_accuracy": 0.9881740033626556,
+      "num_tokens": 54018331.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.7270502150058746,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 0.9153032898902893,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.033472740650177,
+      "mean_token_accuracy": 0.9876999020576477,
+      "num_tokens": 54206935.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.7345028996467591,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.0144257545471191,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.033550959825515744,
+      "mean_token_accuracy": 0.9874906241893768,
+      "num_tokens": 54395443.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.7274066030979156,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0221407413482666,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.033388751745224,
+      "mean_token_accuracy": 0.9870682001113892,
+      "num_tokens": 54584552.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.7293415367603302,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.019165277481079,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.03568689227104187,
+      "mean_token_accuracy": 0.9861597299575806,
+      "num_tokens": 54773705.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.7281844198703766,
+      "epoch": 3.916554508748318,
+      "grad_norm": 0.933020830154419,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.030142486095428467,
+      "mean_token_accuracy": 0.988915103673935,
+      "num_tokens": 54962152.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.7242733418941498,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.9429674744606018,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.0337733268737793,
+      "mean_token_accuracy": 0.987263411283493,
+      "num_tokens": 55151179.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.7306612014770508,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.9022014737129211,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.03387616574764252,
+      "mean_token_accuracy": 0.9871586680412292,
+      "num_tokens": 55339658.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.7246532738208771,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2193421125411987,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.03193953037261963,
+      "mean_token_accuracy": 0.9879468739032745,
+      "num_tokens": 55528467.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.7268587052822113,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.2917358875274658,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.03584722280502319,
+      "mean_token_accuracy": 0.9867104470729828,
+      "num_tokens": 55717308.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.735340416431427,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.2229385375976562,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.03506708741188049,
+      "mean_token_accuracy": 0.9864643573760986,
+      "num_tokens": 55906059.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.7337038695812226,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.0747122764587402,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.03241217732429504,
+      "mean_token_accuracy": 0.9876143455505371,
+      "num_tokens": 56094606.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.7333687073106219,
+      "eval_loss": 0.07489899545907974,
+      "eval_mean_token_accuracy": 0.9719084019114257,
+      "eval_num_tokens": 56132350.0,
+      "eval_runtime": 12.7823,
+      "eval_samples_per_second": 391.166,
+      "eval_steps_per_second": 12.283,
+      "step": 2972
+    },
+    {
+      "entropy": 0.72461416721344,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7916642427444458,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.02286165803670883,
+      "mean_token_accuracy": 0.9920066714286804,
+      "num_tokens": 56283397.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.7150031447410583,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.9078621864318848,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.020906335115432738,
+      "mean_token_accuracy": 0.9925728499889374,
+      "num_tokens": 56472408.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.7091723501682281,
+      "epoch": 4.037685060565276,
+      "grad_norm": 0.8117756843566895,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.02036277800798416,
+      "mean_token_accuracy": 0.9928471207618713,
+      "num_tokens": 56660906.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.695214182138443,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7815372943878174,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019653698801994322,
+      "mean_token_accuracy": 0.992583978176117,
+      "num_tokens": 56850067.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.6937313377857208,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4769781827926636,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.02311961054801941,
+      "mean_token_accuracy": 0.9921472012996674,
+      "num_tokens": 57039005.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.6977694034576416,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.0028746128082275,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.021832706034183504,
+      "mean_token_accuracy": 0.9917877376079559,
+      "num_tokens": 57227782.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.698432594537735,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.0107522010803223,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.020331941545009613,
+      "mean_token_accuracy": 0.9932828366756439,
+      "num_tokens": 57416541.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.6941913783550262,
+      "epoch": 4.104979811574697,
+      "grad_norm": 0.8923842310905457,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019534927606582642,
+      "mean_token_accuracy": 0.9930490672588348,
+      "num_tokens": 57605162.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.6947623729705811,
+      "epoch": 4.118438761776582,
+      "grad_norm": 0.9093812108039856,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.02151128649711609,
+      "mean_token_accuracy": 0.9920440256595612,
+      "num_tokens": 57793756.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.6972509443759918,
+      "epoch": 4.131897711978466,
+      "grad_norm": 0.7549760341644287,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.020574498176574706,
+      "mean_token_accuracy": 0.9928867638111114,
+      "num_tokens": 57982678.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.6952514111995697,
+      "epoch": 4.14535666218035,
+      "grad_norm": 0.9476824402809143,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.02087453305721283,
+      "mean_token_accuracy": 0.9924874365329742,
+      "num_tokens": 58171508.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.6910251140594482,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.2548149824142456,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.020806866884231567,
+      "mean_token_accuracy": 0.9921584248542785,
+      "num_tokens": 58360594.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.6958921313285827,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.0649152994155884,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.020952591300010683,
+      "mean_token_accuracy": 0.9924699544906617,
+      "num_tokens": 58549637.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.6932485699653625,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.8891078233718872,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.020813333988189697,
+      "mean_token_accuracy": 0.9927419483661651,
+      "num_tokens": 58738595.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.6984799921512603,
+      "epoch": 4.199192462987887,
+      "grad_norm": 0.9459101557731628,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.01971060484647751,
+      "mean_token_accuracy": 0.9926240146160126,
+      "num_tokens": 58927507.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.7021916508674622,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.8318898677825928,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020860180258750916,
+      "mean_token_accuracy": 0.9927471876144409,
+      "num_tokens": 59116537.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.6972847402095794,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.9066984057426453,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.023059430718421935,
+      "mean_token_accuracy": 0.9919216752052307,
+      "num_tokens": 59305838.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.7049594700336457,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1277318000793457,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.022385355830192567,
+      "mean_token_accuracy": 0.99165118932724,
+      "num_tokens": 59494587.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.7045334160327912,
+      "epoch": 4.253028263795424,
+      "grad_norm": 0.8147489428520203,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.022358573973178864,
+      "mean_token_accuracy": 0.9918249011039734,
+      "num_tokens": 59683021.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.7007148265838623,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.9161480069160461,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.02088596671819687,
+      "mean_token_accuracy": 0.9925470650196075,
+      "num_tokens": 59871999.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.7006107687950134,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.9611289501190186,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.022619232535362244,
+      "mean_token_accuracy": 0.991577273607254,
+      "num_tokens": 60060486.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.7069303750991821,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.0339760780334473,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.021571077406406403,
+      "mean_token_accuracy": 0.9922139644622803,
+      "num_tokens": 60249686.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.7146372079849244,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.9761297702789307,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.022300006449222566,
+      "mean_token_accuracy": 0.9918746411800384,
+      "num_tokens": 60438481.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.7134014248847962,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 0.9729707837104797,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.02121303677558899,
+      "mean_token_accuracy": 0.9925692796707153,
+      "num_tokens": 60627501.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.7110478222370148,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.9456778168678284,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.020917078852653502,
+      "mean_token_accuracy": 0.9922434687614441,
+      "num_tokens": 60816135.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.7111951589584351,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8617656826972961,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.020851245522499083,
+      "mean_token_accuracy": 0.9924458563327789,
+      "num_tokens": 61005658.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.7193978548049926,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.8495420813560486,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.02084285914897919,
+      "mean_token_accuracy": 0.9927356660366058,
+      "num_tokens": 61194170.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.7188209414482116,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.728841245174408,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.02111642211675644,
+      "mean_token_accuracy": 0.9926697969436645,
+      "num_tokens": 61382852.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.7130864262580872,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.1274386644363403,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.021722708642482758,
+      "mean_token_accuracy": 0.9922453343868256,
+      "num_tokens": 61572195.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.7132770180702209,
+      "epoch": 4.401076716016151,
+      "grad_norm": 0.9251006245613098,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.021602503955364227,
+      "mean_token_accuracy": 0.9919042110443115,
+      "num_tokens": 61761119.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.7063322901725769,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.0182228088378906,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.020274630188941954,
+      "mean_token_accuracy": 0.9928637027740479,
+      "num_tokens": 61949931.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.7106044590473175,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.795170783996582,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.020594277977943422,
+      "mean_token_accuracy": 0.9925063610076904,
+      "num_tokens": 62138377.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7151111841201783,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1479791402816772,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.02117723524570465,
+      "mean_token_accuracy": 0.9921884059906005,
+      "num_tokens": 62327009.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.7172957301139832,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9795703291893005,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021917465329170226,
+      "mean_token_accuracy": 0.9915876984596252,
+      "num_tokens": 62515859.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.7175660550594329,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.9326112866401672,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.021142563223838805,
+      "mean_token_accuracy": 0.9922006845474243,
+      "num_tokens": 62704678.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.7085749089717865,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.8895787596702576,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.0222358375787735,
+      "mean_token_accuracy": 0.9918791234493256,
+      "num_tokens": 62893576.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.7059603810310364,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2298023700714111,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021999095380306245,
+      "mean_token_accuracy": 0.9920522570610046,
+      "num_tokens": 63082258.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.707107275724411,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.8295655250549316,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021844917535781862,
+      "mean_token_accuracy": 0.9922264873981476,
+      "num_tokens": 63271202.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.7119229018688202,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.7440339922904968,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.022164498269557954,
+      "mean_token_accuracy": 0.9918813884258271,
+      "num_tokens": 63460614.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.7065344154834747,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7275091409683228,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.019415955245494842,
+      "mean_token_accuracy": 0.9928511261940003,
+      "num_tokens": 63649327.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.6984272241592407,
+      "epoch": 4.549125168236878,
+      "grad_norm": 0.9989121556282043,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02176133990287781,
+      "mean_token_accuracy": 0.9923081934452057,
+      "num_tokens": 63837983.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.7084203243255616,
+      "epoch": 4.562584118438762,
+      "grad_norm": 0.9526837468147278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.02182276248931885,
+      "mean_token_accuracy": 0.9921061515808105,
+      "num_tokens": 64026855.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.7097050070762634,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.0156456232070923,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.020074066519737244,
+      "mean_token_accuracy": 0.9926879525184631,
+      "num_tokens": 64215889.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.7087334096431732,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.780275821685791,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.021703639626502992,
+      "mean_token_accuracy": 0.9917874097824096,
+      "num_tokens": 64404515.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.7142964243888855,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7382510304450989,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.019867205619812013,
+      "mean_token_accuracy": 0.9927354514598846,
+      "num_tokens": 64593147.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.7165492177009583,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.8259146213531494,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.021262936294078827,
+      "mean_token_accuracy": 0.9914182722568512,
+      "num_tokens": 64782041.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.7098793745040893,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.9540995359420776,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.0195529043674469,
+      "mean_token_accuracy": 0.9931159794330597,
+      "num_tokens": 64971390.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.7104229986667633,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.8698863387107849,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.019119253754615782,
+      "mean_token_accuracy": 0.9934523403644562,
+      "num_tokens": 65160253.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.7067269921302796,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.9622324705123901,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.02044745087623596,
+      "mean_token_accuracy": 0.992815512418747,
+      "num_tokens": 65349403.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.7016099870204926,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.999839723110199,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.019250848889350893,
+      "mean_token_accuracy": 0.9929339647293091,
+      "num_tokens": 65538164.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.7059290170669555,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.7539619207382202,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.02152409851551056,
+      "mean_token_accuracy": 0.99238161444664,
+      "num_tokens": 65727493.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.7103133797645569,
+      "epoch": 4.697173620457605,
+      "grad_norm": 0.9892463088035583,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020616935193538667,
+      "mean_token_accuracy": 0.9927784681320191,
+      "num_tokens": 65916233.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.703304260969162,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.1750195026397705,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021048617362976075,
+      "mean_token_accuracy": 0.9921808183193207,
+      "num_tokens": 66105229.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.7017257273197174,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.887427031993866,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02077500820159912,
+      "mean_token_accuracy": 0.992085075378418,
+      "num_tokens": 66294280.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.7027543246746063,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.0987571477890015,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.02167164385318756,
+      "mean_token_accuracy": 0.992034786939621,
+      "num_tokens": 66483182.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.7053992748260498,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.0075602531433105,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.022686071693897247,
+      "mean_token_accuracy": 0.9918968200683593,
+      "num_tokens": 66671795.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.7109473288059235,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.1655596494674683,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.022698774933815002,
+      "mean_token_accuracy": 0.9913642406463623,
+      "num_tokens": 66861166.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.7130867660045623,
+      "epoch": 4.77792732166891,
+      "grad_norm": 0.9386347532272339,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0205889493227005,
+      "mean_token_accuracy": 0.9924897074699401,
+      "num_tokens": 67049967.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.7099342107772827,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.8524489998817444,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.01985667645931244,
+      "mean_token_accuracy": 0.9928426086902619,
+      "num_tokens": 67238781.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.7100094079971313,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7522528171539307,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.02066437602043152,
+      "mean_token_accuracy": 0.9924224317073822,
+      "num_tokens": 67427802.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7172351717948914,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9865215420722961,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.021969664096832275,
+      "mean_token_accuracy": 0.9919721603393554,
+      "num_tokens": 67616675.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.7113273084163666,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9339645504951477,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.019794631004333495,
+      "mean_token_accuracy": 0.9925971984863281,
+      "num_tokens": 67805877.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.7041554272174835,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.337364912033081,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020412631332874298,
+      "mean_token_accuracy": 0.9923949122428894,
+      "num_tokens": 67994858.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.7090700149536133,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.7725968956947327,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.02121713161468506,
+      "mean_token_accuracy": 0.9920605599880219,
+      "num_tokens": 68183865.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.7141157209873199,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.8638854026794434,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.020726922154426574,
+      "mean_token_accuracy": 0.992779266834259,
+      "num_tokens": 68372954.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.7078199684619904,
+      "epoch": 4.885598923283984,
+      "grad_norm": 0.8576115965843201,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.019535741209983824,
+      "mean_token_accuracy": 0.9927346885204316,
+      "num_tokens": 68562535.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.7034667313098908,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8339338898658752,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.021528851985931397,
+      "mean_token_accuracy": 0.991979056596756,
+      "num_tokens": 68751258.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.6983348250389099,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8739761710166931,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020945993065834046,
+      "mean_token_accuracy": 0.9923328399658203,
+      "num_tokens": 68939586.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.6922877132892609,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.886886715888977,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.020974791049957274,
+      "mean_token_accuracy": 0.992312878370285,
+      "num_tokens": 69128270.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.6857815742492676,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.9415780901908875,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.020201480388641356,
+      "mean_token_accuracy": 0.9925193309783935,
+      "num_tokens": 69317066.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.6832823157310486,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.8311293721199036,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.02008202373981476,
+      "mean_token_accuracy": 0.9926081895828247,
+      "num_tokens": 69505698.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.6914379715919494,
+      "epoch": 4.96635262449529,
+      "grad_norm": 0.8799999356269836,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.0189486563205719,
+      "mean_token_accuracy": 0.9932616055011749,
+      "num_tokens": 69694164.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.692799162864685,
+      "epoch": 4.979811574697173,
+      "grad_norm": 0.9579370021820068,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.019743245840072633,
+      "mean_token_accuracy": 0.9932292222976684,
+      "num_tokens": 69882717.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.6969726324081421,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.0632585287094116,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.019562503695487975,
+      "mean_token_accuracy": 0.9929000854492187,
+      "num_tokens": 70071225.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.7056300662885047,
+      "eval_loss": 0.08702203631401062,
+      "eval_mean_token_accuracy": 0.9712980936287315,
+      "eval_num_tokens": 70165509.0,
+      "eval_runtime": 12.7642,
+      "eval_samples_per_second": 391.719,
+      "eval_steps_per_second": 12.3,
+      "step": 3715
+    },
+    {
+      "entropy": 0.6998905837535858,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8414336442947388,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01888625919818878,
+      "mean_token_accuracy": 0.9934305250644684,
+      "num_tokens": 70259811.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.6933371186256408,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.7618409395217896,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01257207989692688,
+      "mean_token_accuracy": 0.9958574831485748,
+      "num_tokens": 70448683.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.6861799776554107,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.935829758644104,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.0118876114487648,
+      "mean_token_accuracy": 0.9963350534439087,
+      "num_tokens": 70637078.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.6778797626495361,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.9417919516563416,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.012690700590610504,
+      "mean_token_accuracy": 0.9954731047153473,
+      "num_tokens": 70826145.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.6735071957111358,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.7570633292198181,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01321282535791397,
+      "mean_token_accuracy": 0.9955435931682587,
+      "num_tokens": 71015458.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.677568006515503,
+      "epoch": 5.074024226110363,
+      "grad_norm": 0.8157011866569519,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.012434301525354385,
+      "mean_token_accuracy": 0.9957576751708984,
+      "num_tokens": 71204332.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.6771604120731354,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.780483603477478,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.013237486779689788,
+      "mean_token_accuracy": 0.9957249999046326,
+      "num_tokens": 71393307.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.6761959791183472,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8249189257621765,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.011912986636161804,
+      "mean_token_accuracy": 0.9961272835731506,
+      "num_tokens": 71582233.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.6717506468296051,
+      "epoch": 5.114401076716016,
+      "grad_norm": 0.9667718410491943,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.012740559875965118,
+      "mean_token_accuracy": 0.9953394651412963,
+      "num_tokens": 71771906.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.6755871415138245,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.9068875908851624,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.012314559519290924,
+      "mean_token_accuracy": 0.9955949664115906,
+      "num_tokens": 71960698.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.6791335880756378,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.7650554776191711,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.011927495896816253,
+      "mean_token_accuracy": 0.9957858562469483,
+      "num_tokens": 72149627.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.6738516509532928,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 0.7267928123474121,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.011817440390586853,
+      "mean_token_accuracy": 0.996101975440979,
+      "num_tokens": 72338469.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.6701940059661865,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.9211037755012512,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012324941903352737,
+      "mean_token_accuracy": 0.9956477820873261,
+      "num_tokens": 72527499.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.6763210237026215,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.7772125601768494,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.013027346134185791,
+      "mean_token_accuracy": 0.9959511935710907,
+      "num_tokens": 72715639.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.6793705642223358,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.582334578037262,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.011401662230491638,
+      "mean_token_accuracy": 0.9961598336696624,
+      "num_tokens": 72904615.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.6798616409301758,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.8502725958824158,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011671387404203416,
+      "mean_token_accuracy": 0.9961067497730255,
+      "num_tokens": 73093522.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.682939213514328,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7421601414680481,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.011898189783096313,
+      "mean_token_accuracy": 0.9962145388126373,
+      "num_tokens": 73282415.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.6821306705474853,
+      "epoch": 5.235531628532974,
+      "grad_norm": 0.7552776336669922,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.01232050210237503,
+      "mean_token_accuracy": 0.9957845032215118,
+      "num_tokens": 73471211.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.6746087670326233,
+      "epoch": 5.248990578734858,
+      "grad_norm": 0.7144336700439453,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01240069791674614,
+      "mean_token_accuracy": 0.9960526466369629,
+      "num_tokens": 73660289.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.6723428070545197,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.8043408393859863,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.01204245388507843,
+      "mean_token_accuracy": 0.9960211217403412,
+      "num_tokens": 73849312.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.6766897737979889,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.8608251810073853,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012303075194358826,
+      "mean_token_accuracy": 0.9955043613910675,
+      "num_tokens": 74038344.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.6809355199337006,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.4872569739818573,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.010690021514892577,
+      "mean_token_accuracy": 0.9965942919254303,
+      "num_tokens": 74226981.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.6819032669067383,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.0376152992248535,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012148186564445496,
+      "mean_token_accuracy": 0.9958918631076813,
+      "num_tokens": 74415130.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.6856858193874359,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.7241474986076355,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012284471094608307,
+      "mean_token_accuracy": 0.9959208726882934,
+      "num_tokens": 74603788.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.6874188125133515,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.6505919098854065,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.012959125638008117,
+      "mean_token_accuracy": 0.9956629753112793,
+      "num_tokens": 74792306.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.6912538707256317,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7696103453636169,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012945668399333954,
+      "mean_token_accuracy": 0.9956756651401519,
+      "num_tokens": 74981265.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.6915560126304626,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.8245812058448792,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.01408768892288208,
+      "mean_token_accuracy": 0.9950980842113495,
+      "num_tokens": 75170337.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.6872740864753724,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8330141305923462,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.011667902022600174,
+      "mean_token_accuracy": 0.9960226893424988,
+      "num_tokens": 75359113.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.6825072944164277,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.9389140009880066,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012351009249687194,
+      "mean_token_accuracy": 0.9954748272895813,
+      "num_tokens": 75548102.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.6821270763874054,
+      "epoch": 5.397039030955585,
+      "grad_norm": 1.0638995170593262,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.0128641277551651,
+      "mean_token_accuracy": 0.995329612493515,
+      "num_tokens": 75736452.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.6857812821865081,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7838811278343201,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.011757558584213257,
+      "mean_token_accuracy": 0.9959722757339478,
+      "num_tokens": 75925013.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.6869213104248046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 0.5632259845733643,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.011284272372722625,
+      "mean_token_accuracy": 0.9962978959083557,
+      "num_tokens": 76113398.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.6807511866092681,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.7132044434547424,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012044224143028259,
+      "mean_token_accuracy": 0.9959294259548187,
+      "num_tokens": 76302629.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.6815520942211151,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.7640891075134277,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.012763911485671997,
+      "mean_token_accuracy": 0.9953242301940918,
+      "num_tokens": 76491675.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.6802713990211486,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.5691967010498047,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.0116192027926445,
+      "mean_token_accuracy": 0.9958603858947754,
+      "num_tokens": 76680426.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.676917290687561,
+      "epoch": 5.477792732166891,
+      "grad_norm": 0.8160638213157654,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.01229381114244461,
+      "mean_token_accuracy": 0.9956679701805115,
+      "num_tokens": 76869321.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.6745438575744629,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.8013969659805298,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.012171164155006409,
+      "mean_token_accuracy": 0.9958188533782959,
+      "num_tokens": 77058299.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.6774423599243165,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.77237868309021,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.013017389178276061,
+      "mean_token_accuracy": 0.9956206619739533,
+      "num_tokens": 77247567.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.6773760914802551,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.6185512542724609,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.011151286959648132,
+      "mean_token_accuracy": 0.9961191415786743,
+      "num_tokens": 77436417.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.6750624418258667,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.7832470536231995,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.011553966999053955,
+      "mean_token_accuracy": 0.9958207130432128,
+      "num_tokens": 77625255.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.6814670443534852,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.7947144508361816,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.011924812197685241,
+      "mean_token_accuracy": 0.9959113657474518,
+      "num_tokens": 77813632.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.68285973072052,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.6403945684432983,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.01066286414861679,
+      "mean_token_accuracy": 0.9966644048690796,
+      "num_tokens": 78002213.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.6764974176883698,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.8078592419624329,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.010837660729885101,
+      "mean_token_accuracy": 0.9961750328540802,
+      "num_tokens": 78191471.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.6726884067058563,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.7293245792388916,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.01087343692779541,
+      "mean_token_accuracy": 0.996295166015625,
+      "num_tokens": 78380424.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.6714434802532196,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0266199111938477,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.011094169318675995,
+      "mean_token_accuracy": 0.9960172355175019,
+      "num_tokens": 78569653.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.6689312815666199,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.8064524531364441,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.01098542958498001,
+      "mean_token_accuracy": 0.9961700737476349,
+      "num_tokens": 78759287.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.6728951573371887,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7976726293563843,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.012345843017101288,
+      "mean_token_accuracy": 0.9957349181175232,
+      "num_tokens": 78948032.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.6702533006668091,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.7385782599449158,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.011018334329128266,
+      "mean_token_accuracy": 0.9964040815830231,
+      "num_tokens": 79136972.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.67191202044487,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.6246873736381531,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.011075176298618317,
+      "mean_token_accuracy": 0.9962317526340485,
+      "num_tokens": 79325966.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.6736057937145233,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9479376077651978,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012818500399589539,
+      "mean_token_accuracy": 0.9951797604560852,
+      "num_tokens": 79514853.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.6723325371742248,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.6468302607536316,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.010920494794845581,
+      "mean_token_accuracy": 0.9961069703102112,
+      "num_tokens": 79703739.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.6762719631195069,
+      "epoch": 5.693135935397039,
+      "grad_norm": 0.6862959265708923,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.011675214767456055,
+      "mean_token_accuracy": 0.9958149492740631,
+      "num_tokens": 79892582.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.6739612460136414,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.8402584195137024,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.010931231081485748,
+      "mean_token_accuracy": 0.9960719287395478,
+      "num_tokens": 80081362.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.6767356872558594,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.6180281043052673,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.012499480694532394,
+      "mean_token_accuracy": 0.9957239329814911,
+      "num_tokens": 80269388.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.6773952782154083,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.8782367706298828,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012943127751350402,
+      "mean_token_accuracy": 0.9958240866661072,
+      "num_tokens": 80458191.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.6774130165576935,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.7854585647583008,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.012309956550598144,
+      "mean_token_accuracy": 0.99593386054039,
+      "num_tokens": 80647306.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.6813723623752594,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.6340908408164978,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.01146906390786171,
+      "mean_token_accuracy": 0.9963057219982148,
+      "num_tokens": 80836147.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.6804767549037933,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.5984897613525391,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.010843665152788163,
+      "mean_token_accuracy": 0.996556407213211,
+      "num_tokens": 81025302.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.6759022295475006,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.6014022827148438,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.010135546326637268,
+      "mean_token_accuracy": 0.9966514885425568,
+      "num_tokens": 81214251.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.675349909067154,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.6571088433265686,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01084519699215889,
+      "mean_token_accuracy": 0.9962432861328125,
+      "num_tokens": 81402608.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.6718680560588837,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.9250450730323792,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.01139993518590927,
+      "mean_token_accuracy": 0.9961999177932739,
+      "num_tokens": 81591841.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.674647331237793,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.8669414520263672,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.010886493325233459,
+      "mean_token_accuracy": 0.9962068378925324,
+      "num_tokens": 81780873.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.6801326870918274,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7395965456962585,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.012087566405534744,
+      "mean_token_accuracy": 0.9957775354385376,
+      "num_tokens": 81969835.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.676666659116745,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.7648837566375732,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.011214715242385865,
+      "mean_token_accuracy": 0.9963518619537354,
+      "num_tokens": 82159146.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.676488333940506,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.6776206493377686,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.011839983612298965,
+      "mean_token_accuracy": 0.9959003031253815,
+      "num_tokens": 82347867.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.6743977665901184,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.7158207893371582,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011066053062677383,
+      "mean_token_accuracy": 0.9966431736946106,
+      "num_tokens": 82536890.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.6764561176300049,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.7066213488578796,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.012056462466716766,
+      "mean_token_accuracy": 0.9958225846290588,
+      "num_tokens": 82725921.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.6813391983509064,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.763606607913971,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.011903972923755645,
+      "mean_token_accuracy": 0.9958972215652466,
+      "num_tokens": 82914838.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.6814220428466797,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.8407371044158936,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.011677531152963638,
+      "mean_token_accuracy": 0.9959126651287079,
+      "num_tokens": 83103937.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.6833201944828033,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8418315052986145,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012164801359176636,
+      "mean_token_accuracy": 0.9956748068332673,
+      "num_tokens": 83293106.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.6802135646343231,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.8431894183158875,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012858036160469054,
+      "mean_token_accuracy": 0.995720499753952,
+      "num_tokens": 83481559.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.6758989214897155,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7379446029663086,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.011657674610614777,
+      "mean_token_accuracy": 0.9959479212760926,
+      "num_tokens": 83670620.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.6832396805286407,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9370406866073608,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012775366008281708,
+      "mean_token_accuracy": 0.9958724677562714,
+      "num_tokens": 83858979.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.6826908767223359,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.5771452188491821,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011871224641799927,
+      "mean_token_accuracy": 0.9962578654289246,
+      "num_tokens": 84047664.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.6935603318700365,
+      "eval_loss": 0.08950015902519226,
+      "eval_mean_token_accuracy": 0.9720694904874085,
+      "eval_num_tokens": 84198646.0,
+      "eval_runtime": 12.7741,
+      "eval_samples_per_second": 391.416,
+      "eval_steps_per_second": 12.29,
+      "step": 4458
+    },
+    {
+      "entropy": 0.6853953242301941,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.4715442359447479,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.011727289110422135,
+      "mean_token_accuracy": 0.9963658571243286,
+      "num_tokens": 84236476.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.6872529745101928,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.6521468162536621,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.007476924359798432,
+      "mean_token_accuracy": 0.9976866364479064,
+      "num_tokens": 84425424.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.6764522135257721,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.5134896039962769,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.006954360008239746,
+      "mean_token_accuracy": 0.9981172740459442,
+      "num_tokens": 84614973.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.674551123380661,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.5777756571769714,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.0064676709473133085,
+      "mean_token_accuracy": 0.9981890261173249,
+      "num_tokens": 84803584.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.6743139624595642,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.5243175625801086,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.005892159789800644,
+      "mean_token_accuracy": 0.99809131026268,
+      "num_tokens": 84992194.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.6691218018531799,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.6378570795059204,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.006101621687412262,
+      "mean_token_accuracy": 0.9981180131435394,
+      "num_tokens": 85181570.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.6653562963008881,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.7931498885154724,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.0059667728841304776,
+      "mean_token_accuracy": 0.9982839524745941,
+      "num_tokens": 85370354.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.6660257399082183,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.6156598329544067,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.006237205862998962,
+      "mean_token_accuracy": 0.9980066597461701,
+      "num_tokens": 85559287.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.6661248922348022,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.39466625452041626,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.006016004830598831,
+      "mean_token_accuracy": 0.9984841406345367,
+      "num_tokens": 85748489.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.6644636154174804,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.556770920753479,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.0060367342084646225,
+      "mean_token_accuracy": 0.9981182813644409,
+      "num_tokens": 85937336.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.6630209028720856,
+      "epoch": 6.137281292059219,
+      "grad_norm": 0.5270616412162781,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.005779195204377174,
+      "mean_token_accuracy": 0.9981250524520874,
+      "num_tokens": 86126286.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.6656446874141693,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 0.44525614380836487,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.005375263467431069,
+      "mean_token_accuracy": 0.9983476758003235,
+      "num_tokens": 86314904.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.664457255601883,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.6622279286384583,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.006194194033741951,
+      "mean_token_accuracy": 0.9981902241706848,
+      "num_tokens": 86503447.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.6608214318752289,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.6172921657562256,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.0061217501759529116,
+      "mean_token_accuracy": 0.9983710527420044,
+      "num_tokens": 86693005.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.6578842639923096,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.6241856217384338,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.006712782382965088,
+      "mean_token_accuracy": 0.9979998171329498,
+      "num_tokens": 86882287.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.6561006605625153,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 0.6784538626670837,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.006533285975456238,
+      "mean_token_accuracy": 0.9977931976318359,
+      "num_tokens": 87071532.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.6582947254180909,
+      "epoch": 6.218034993270525,
+      "grad_norm": 0.7572171688079834,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.006016571819782257,
+      "mean_token_accuracy": 0.9979285359382629,
+      "num_tokens": 87260158.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.6552021980285645,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.7922662496566772,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.006150579452514649,
+      "mean_token_accuracy": 0.9982126474380493,
+      "num_tokens": 87448718.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.6578958451747894,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.5214764475822449,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.005079228430986404,
+      "mean_token_accuracy": 0.9984815359115601,
+      "num_tokens": 87637461.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.6603941440582275,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.5404514670372009,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.005841487646102905,
+      "mean_token_accuracy": 0.9983745515346527,
+      "num_tokens": 87826292.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.6628166377544403,
+      "epoch": 6.271870794078062,
+      "grad_norm": 0.47433987259864807,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.006364698708057404,
+      "mean_token_accuracy": 0.9982446670532227,
+      "num_tokens": 88015224.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.6635581731796265,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.8039814233779907,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.006831821799278259,
+      "mean_token_accuracy": 0.9982130885124206,
+      "num_tokens": 88203965.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.6615313947200775,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.5904889106750488,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.006949013471603394,
+      "mean_token_accuracy": 0.9976236402988434,
+      "num_tokens": 88392910.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.6613587439060211,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.6321392059326172,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.006564225256443024,
+      "mean_token_accuracy": 0.9980846107006073,
+      "num_tokens": 88581883.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.6642421364784241,
+      "epoch": 6.325706594885599,
+      "grad_norm": 0.6209113597869873,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.007112482190132141,
+      "mean_token_accuracy": 0.9977501094341278,
+      "num_tokens": 88770574.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.6629104316234589,
+      "epoch": 6.339165545087483,
+      "grad_norm": 0.8903343081474304,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.00698748379945755,
+      "mean_token_accuracy": 0.9977229118347168,
+      "num_tokens": 88959487.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.6647967278957367,
+      "epoch": 6.352624495289367,
+      "grad_norm": 0.842950165271759,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.008400703966617584,
+      "mean_token_accuracy": 0.9975436449050903,
+      "num_tokens": 89148302.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.6641393005847931,
+      "epoch": 6.366083445491252,
+      "grad_norm": 0.6241896152496338,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.006545399129390716,
+      "mean_token_accuracy": 0.9982439935207367,
+      "num_tokens": 89337168.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.6646088480949401,
+      "epoch": 6.379542395693136,
+      "grad_norm": 0.6284488439559937,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.005783513188362122,
+      "mean_token_accuracy": 0.9982234418392182,
+      "num_tokens": 89525632.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.6648603081703186,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.7705378532409668,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.006973850727081299,
+      "mean_token_accuracy": 0.9977744221687317,
+      "num_tokens": 89714436.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.6667861938476562,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.7033233642578125,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.0062720529735088345,
+      "mean_token_accuracy": 0.9978694558143616,
+      "num_tokens": 89903591.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.66892409324646,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.6684027910232544,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.006154683232307434,
+      "mean_token_accuracy": 0.998307728767395,
+      "num_tokens": 90092231.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.6692153334617614,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.7860879302024841,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.006273554265499115,
+      "mean_token_accuracy": 0.9981006205081939,
+      "num_tokens": 90280721.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.6636229634284974,
+      "epoch": 6.446837146702557,
+      "grad_norm": 0.5911756753921509,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.0067244857549667355,
+      "mean_token_accuracy": 0.9979658007621766,
+      "num_tokens": 90469835.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.6663640201091766,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.5844401717185974,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.005917467921972275,
+      "mean_token_accuracy": 0.9981400012969971,
+      "num_tokens": 90658439.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.6635376453399658,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.554587721824646,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.005905192345380783,
+      "mean_token_accuracy": 0.9983574509620666,
+      "num_tokens": 90847689.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.660474544763565,
+      "epoch": 6.48721399730821,
+      "grad_norm": 0.6828640103340149,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.006179597228765488,
+      "mean_token_accuracy": 0.9978089511394501,
+      "num_tokens": 91036945.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.6635875403881073,
+      "epoch": 6.500672947510094,
+      "grad_norm": 0.621636688709259,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.006481277197599411,
+      "mean_token_accuracy": 0.9976639211177826,
+      "num_tokens": 91225796.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.6625118017196655,
+      "epoch": 6.514131897711978,
+      "grad_norm": 0.6648721098899841,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.006836448609828949,
+      "mean_token_accuracy": 0.9981350302696228,
+      "num_tokens": 91414454.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.6632138609886169,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.6122362017631531,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.0057271294295787815,
+      "mean_token_accuracy": 0.9983117520809174,
+      "num_tokens": 91603019.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.6621270835399627,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.707145094871521,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.005987466499209404,
+      "mean_token_accuracy": 0.9983782589435577,
+      "num_tokens": 91791729.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.6593974709510804,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.7802532315254211,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.00661279633641243,
+      "mean_token_accuracy": 0.9978727638721466,
+      "num_tokens": 91980977.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.658482962846756,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 0.8637571334838867,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.005754219368100166,
+      "mean_token_accuracy": 0.9982581317424775,
+      "num_tokens": 92169829.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.6561014533042908,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.7007631659507751,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.006207703426480293,
+      "mean_token_accuracy": 0.9980942070484161,
+      "num_tokens": 92358635.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.6558376729488373,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.6180236339569092,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.005951264500617981,
+      "mean_token_accuracy": 0.9980935871601104,
+      "num_tokens": 92546955.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.6548421680927277,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.5694975852966309,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.0058226577937603,
+      "mean_token_accuracy": 0.9980980336666108,
+      "num_tokens": 92735651.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.6563037514686585,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 0.7417933344841003,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.006121259927749634,
+      "mean_token_accuracy": 0.9980963408946991,
+      "num_tokens": 92925010.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.654166966676712,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.705633282661438,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.005783010274171829,
+      "mean_token_accuracy": 0.9983141779899597,
+      "num_tokens": 93113992.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.6567890524864197,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.5444097518920898,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.005712105333805085,
+      "mean_token_accuracy": 0.9982166588306427,
+      "num_tokens": 93302433.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.6536349058151245,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.4291290044784546,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.0060494735836982725,
+      "mean_token_accuracy": 0.9979224085807801,
+      "num_tokens": 93491429.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.6516136825084686,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.5134868621826172,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.005212320387363434,
+      "mean_token_accuracy": 0.9984170615673065,
+      "num_tokens": 93679973.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.6528811275959014,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.8154370188713074,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.00622846707701683,
+      "mean_token_accuracy": 0.998017430305481,
+      "num_tokens": 93868735.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.6532892882823944,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.5574272871017456,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.005193561688065529,
+      "mean_token_accuracy": 0.9985900521278381,
+      "num_tokens": 94057428.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.6552170813083649,
+      "epoch": 6.716016150740242,
+      "grad_norm": 0.6016380190849304,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.005771375447511673,
+      "mean_token_accuracy": 0.9981073677539826,
+      "num_tokens": 94246297.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.6572140097618103,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.7832751870155334,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.005496300011873245,
+      "mean_token_accuracy": 0.9983509957790375,
+      "num_tokens": 94434253.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.6541008234024048,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.4329380989074707,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.006188442185521126,
+      "mean_token_accuracy": 0.9981851994991302,
+      "num_tokens": 94623327.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.6556220591068268,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.5326682329177856,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.0052257638424634935,
+      "mean_token_accuracy": 0.9983932614326477,
+      "num_tokens": 94811837.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.6557375073432923,
+      "epoch": 6.769851951547779,
+      "grad_norm": 0.6864677667617798,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.005970630794763565,
+      "mean_token_accuracy": 0.9983245611190796,
+      "num_tokens": 95000726.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.6545388638973236,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.6318016648292542,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.006516661494970322,
+      "mean_token_accuracy": 0.9979697644710541,
+      "num_tokens": 95189395.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.6542861998081207,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.3979317247867584,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.005074064806103706,
+      "mean_token_accuracy": 0.998512452840805,
+      "num_tokens": 95378855.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.6542766332626343,
+      "epoch": 6.810228802153432,
+      "grad_norm": 0.4778045117855072,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.005774036794900894,
+      "mean_token_accuracy": 0.9982702076435089,
+      "num_tokens": 95568111.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.6550499975681305,
+      "epoch": 6.823687752355316,
+      "grad_norm": 0.4805110991001129,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.005620773509144783,
+      "mean_token_accuracy": 0.9983035147190094,
+      "num_tokens": 95756885.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.6543555080890655,
+      "epoch": 6.837146702557201,
+      "grad_norm": 0.6604891419410706,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.005247855931520462,
+      "mean_token_accuracy": 0.998341029882431,
+      "num_tokens": 95946078.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.6581745266914367,
+      "epoch": 6.850605652759085,
+      "grad_norm": 0.6589891910552979,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.005882954970002174,
+      "mean_token_accuracy": 0.9982152104377746,
+      "num_tokens": 96134597.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.6597592771053314,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.4152396023273468,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.0054433666169643406,
+      "mean_token_accuracy": 0.9983908593654632,
+      "num_tokens": 96323327.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.658534574508667,
+      "epoch": 6.877523553162853,
+      "grad_norm": 0.6025403141975403,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.005522595718502998,
+      "mean_token_accuracy": 0.9983524858951569,
+      "num_tokens": 96512170.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.6568575739860535,
+      "epoch": 6.890982503364738,
+      "grad_norm": 0.5730848908424377,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.004784010350704193,
+      "mean_token_accuracy": 0.9984236299991608,
+      "num_tokens": 96701363.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.6535935163497925,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.5301522612571716,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.004592524468898773,
+      "mean_token_accuracy": 0.9987652719020843,
+      "num_tokens": 96890642.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.6544224679470062,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.639918863773346,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.005577490478754043,
+      "mean_token_accuracy": 0.9984355092048645,
+      "num_tokens": 97079552.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.6542503774166107,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.5202826857566833,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.005623985454440117,
+      "mean_token_accuracy": 0.9982783615589141,
+      "num_tokens": 97268346.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.6499157190322876,
+      "epoch": 6.944818304172275,
+      "grad_norm": 0.5668892860412598,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.005104385688900948,
+      "mean_token_accuracy": 0.9985612750053405,
+      "num_tokens": 97457514.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.6520473599433899,
+      "epoch": 6.958277254374159,
+      "grad_norm": 0.5878952145576477,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.005280185490846634,
+      "mean_token_accuracy": 0.9985166370868683,
+      "num_tokens": 97646014.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.6488265216350555,
+      "epoch": 6.971736204576043,
+      "grad_norm": 0.4630887806415558,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.005359554663300514,
+      "mean_token_accuracy": 0.998493617773056,
+      "num_tokens": 97835401.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.6526197671890259,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.38062795996665955,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.0054448477923870085,
+      "mean_token_accuracy": 0.9982813715934753,
+      "num_tokens": 98024194.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.654500275850296,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.5739313364028931,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.005691033974289894,
+      "mean_token_accuracy": 0.9982043981552124,
+      "num_tokens": 98212962.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.6537400142402406,
+      "eval_loss": 0.09857651591300964,
+      "eval_mean_token_accuracy": 0.9730726879113799,
+      "eval_num_tokens": 98231758.0,
+      "eval_runtime": 12.7391,
+      "eval_samples_per_second": 392.491,
+      "eval_steps_per_second": 12.324,
+      "step": 5201
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.5405593933132595e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5201/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5201/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5944/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5944/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5944/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..66942bac6e546df88aa32e7b1211ae694f7a3fc6
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3f80cc1881c3f89e5b12d4df243868fbe480d417ebb11c2fd8e672b19d0a15b
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5944/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5944/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5944/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5944/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..05b24acd20bbe3a20742b1deb68b14b133e31f9b
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/trainer_state.json
@@ -0,0 +1,6062 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.0,
+  "eval_steps": 500,
+  "global_step": 5944,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    },
+    {
+      "entropy": 0.8332668900489807,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9914855360984802,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.06607928276062011,
+      "mean_token_accuracy": 0.974932211637497,
+      "num_tokens": 14165567.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8149146616458893,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.3080861568450928,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.057932257652282715,
+      "mean_token_accuracy": 0.9770712852478027,
+      "num_tokens": 14353912.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8197620987892151,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.0898637771606445,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.06010057330131531,
+      "mean_token_accuracy": 0.9758293390274048,
+      "num_tokens": 14542288.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8183245182037353,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.3120660781860352,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.0610861599445343,
+      "mean_token_accuracy": 0.9764986515045166,
+      "num_tokens": 14731028.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8135185301303863,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.138473391532898,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.06268800497055053,
+      "mean_token_accuracy": 0.9752912580966949,
+      "num_tokens": 14919682.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8142645537853241,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.1258258819580078,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.06271481513977051,
+      "mean_token_accuracy": 0.9757774949073792,
+      "num_tokens": 15107904.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8170483231544494,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.007297396659851,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.06338523030281067,
+      "mean_token_accuracy": 0.9754154622554779,
+      "num_tokens": 15297115.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8201254069805145,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 0.8240306973457336,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.06355815529823303,
+      "mean_token_accuracy": 0.9752808332443237,
+      "num_tokens": 15485921.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8140238881111145,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.0350368022918701,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.05969444513320923,
+      "mean_token_accuracy": 0.977089112997055,
+      "num_tokens": 15674824.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8133325934410095,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.0544878244400024,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.06062774658203125,
+      "mean_token_accuracy": 0.976290100812912,
+      "num_tokens": 15863526.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8185161769390106,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.1076316833496094,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.061567097902297974,
+      "mean_token_accuracy": 0.9755526125431061,
+      "num_tokens": 16051685.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8176139056682586,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.0057616233825684,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.06227490901947021,
+      "mean_token_accuracy": 0.9758114695549012,
+      "num_tokens": 16240451.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8217637002468109,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.97276771068573,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.06410331726074218,
+      "mean_token_accuracy": 0.9754030883312226,
+      "num_tokens": 16428961.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8141795754432678,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.0135555267333984,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.06253207921981811,
+      "mean_token_accuracy": 0.9760490536689759,
+      "num_tokens": 16617327.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8167653739452362,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2272467613220215,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.060455435514450075,
+      "mean_token_accuracy": 0.9758628964424133,
+      "num_tokens": 16806414.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8155323445796967,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 0.9328635334968567,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.06302794814109802,
+      "mean_token_accuracy": 0.9745101630687714,
+      "num_tokens": 16995016.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.818474805355072,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.1651655435562134,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.06043210029602051,
+      "mean_token_accuracy": 0.9766487181186676,
+      "num_tokens": 17184250.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8140560865402222,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.1005650758743286,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.06131819486618042,
+      "mean_token_accuracy": 0.9757712185382843,
+      "num_tokens": 17373075.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8204843640327454,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.0825532674789429,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.06296311616897583,
+      "mean_token_accuracy": 0.9745951414108276,
+      "num_tokens": 17561454.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8237834334373474,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.9842743873596191,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.061483919620513916,
+      "mean_token_accuracy": 0.9754831731319428,
+      "num_tokens": 17750758.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8165915131568908,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.0179396867752075,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.06237881183624268,
+      "mean_token_accuracy": 0.9750245690345765,
+      "num_tokens": 17940126.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8329708755016327,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0732359886169434,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.0644665539264679,
+      "mean_token_accuracy": 0.9743888020515442,
+      "num_tokens": 18129000.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8210074126720428,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.3696036338806152,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.06320792436599731,
+      "mean_token_accuracy": 0.9751390159130097,
+      "num_tokens": 18318232.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8248398661613464,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.327525019645691,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.06277769804000854,
+      "mean_token_accuracy": 0.9752413153648376,
+      "num_tokens": 18507403.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8284746050834656,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.044022560119629,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.06162315607070923,
+      "mean_token_accuracy": 0.9760516047477722,
+      "num_tokens": 18696242.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8233858644962311,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.062048077583313,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.059620130062103274,
+      "mean_token_accuracy": 0.9755307257175445,
+      "num_tokens": 18885199.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8236705601215363,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.2024412155151367,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.06304935216903687,
+      "mean_token_accuracy": 0.975091153383255,
+      "num_tokens": 19074028.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.828409218788147,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.1684212684631348,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.06451238393783569,
+      "mean_token_accuracy": 0.9743404746055603,
+      "num_tokens": 19262295.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8214354634284973,
+      "epoch": 1.386271870794078,
+      "grad_norm": 0.8512521386146545,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.06042641997337341,
+      "mean_token_accuracy": 0.9755558848381043,
+      "num_tokens": 19451154.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8223241448402405,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 0.8546568155288696,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.062134528160095216,
+      "mean_token_accuracy": 0.9750089883804322,
+      "num_tokens": 19639827.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.8217969954013824,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 0.972831666469574,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.060803139209747316,
+      "mean_token_accuracy": 0.9760692238807678,
+      "num_tokens": 19828174.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.8144341170787811,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1216644048690796,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.06458876132965088,
+      "mean_token_accuracy": 0.9738282442092896,
+      "num_tokens": 20016700.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.8168790876865387,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.0821806192398071,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.06258658170700074,
+      "mean_token_accuracy": 0.9749014675617218,
+      "num_tokens": 20205823.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8146390020847321,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.0067790746688843,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.06509006023406982,
+      "mean_token_accuracy": 0.9737810134887696,
+      "num_tokens": 20394611.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8224801063537598,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.0583593845367432,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.06343590021133423,
+      "mean_token_accuracy": 0.9750119090080261,
+      "num_tokens": 20583252.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.8083258986473083,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.8305535316467285,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.06573111414909363,
+      "mean_token_accuracy": 0.9736428141593934,
+      "num_tokens": 20772107.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8081986546516419,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5895942449569702,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.06376024484634399,
+      "mean_token_accuracy": 0.9751200795173645,
+      "num_tokens": 20960494.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.8026683986186981,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 0.9962654113769531,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.06282572746276856,
+      "mean_token_accuracy": 0.9753849327564239,
+      "num_tokens": 21149763.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8030014991760254,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9083836078643799,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.06473478078842163,
+      "mean_token_accuracy": 0.9747069597244262,
+      "num_tokens": 21338690.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.8008662164211273,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.097593069076538,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.05911685228347778,
+      "mean_token_accuracy": 0.9767139732837677,
+      "num_tokens": 21527420.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7981986284255982,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.1673516035079956,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.05957244038581848,
+      "mean_token_accuracy": 0.9761911332607269,
+      "num_tokens": 21716791.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.8004718601703644,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 0.7955246567726135,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.06378913521766663,
+      "mean_token_accuracy": 0.9747395157814026,
+      "num_tokens": 21905487.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.8039821267127991,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.3141396045684814,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.06209801435470581,
+      "mean_token_accuracy": 0.9754719436168671,
+      "num_tokens": 22094436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.8051789462566376,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.2227814197540283,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.06261310577392579,
+      "mean_token_accuracy": 0.9753049433231353,
+      "num_tokens": 22283243.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7941994190216064,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.071070671081543,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.05962263345718384,
+      "mean_token_accuracy": 0.9763956189155578,
+      "num_tokens": 22472084.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7937953472137451,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.0329782962799072,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.060611891746521,
+      "mean_token_accuracy": 0.9758659780025483,
+      "num_tokens": 22661614.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7913113713264466,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.032772183418274,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.06002308130264282,
+      "mean_token_accuracy": 0.9758916199207306,
+      "num_tokens": 22850634.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.7909658491611481,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3234598636627197,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.06296972036361695,
+      "mean_token_accuracy": 0.9741117715835571,
+      "num_tokens": 23039402.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7880761086940765,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.9563435316085815,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.06209538578987121,
+      "mean_token_accuracy": 0.9750530481338501,
+      "num_tokens": 23228363.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7845663070678711,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 0.8875390291213989,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.06272958517074585,
+      "mean_token_accuracy": 0.9751476228237153,
+      "num_tokens": 23417469.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7928634822368622,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.0629489421844482,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.060286092758178714,
+      "mean_token_accuracy": 0.976036274433136,
+      "num_tokens": 23606274.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7870154738426208,
+      "epoch": 1.695827725437416,
+      "grad_norm": 0.9164250493049622,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.06263296008110046,
+      "mean_token_accuracy": 0.9752088189125061,
+      "num_tokens": 23795255.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.7948406875133515,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.0909371376037598,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.06165047883987427,
+      "mean_token_accuracy": 0.9764420628547669,
+      "num_tokens": 23984534.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7862435758113862,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.1839547157287598,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.0640158474445343,
+      "mean_token_accuracy": 0.9741048216819763,
+      "num_tokens": 24173632.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.7887867331504822,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.0730741024017334,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.06260218620300292,
+      "mean_token_accuracy": 0.974741768836975,
+      "num_tokens": 24362671.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7896574556827545,
+      "epoch": 1.749663526244953,
+      "grad_norm": 0.9559605121612549,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.061297619342803956,
+      "mean_token_accuracy": 0.9752223670482636,
+      "num_tokens": 24551825.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.7902800619602204,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.914057195186615,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.06320110559463502,
+      "mean_token_accuracy": 0.9749482691287994,
+      "num_tokens": 24740867.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.799742591381073,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.06534289717674255,
+      "mean_token_accuracy": 0.9741291284561158,
+      "num_tokens": 24930132.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7974063038825989,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1188615560531616,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.060908573865890506,
+      "mean_token_accuracy": 0.9765798091888428,
+      "num_tokens": 25119678.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.786439061164856,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9185847640037537,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.06158917546272278,
+      "mean_token_accuracy": 0.9760335922241211,
+      "num_tokens": 25308876.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7888528347015381,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.1948838233947754,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.060635948181152345,
+      "mean_token_accuracy": 0.976032954454422,
+      "num_tokens": 25498102.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7968044698238372,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.4128663539886475,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.06448977589607238,
+      "mean_token_accuracy": 0.9741183638572692,
+      "num_tokens": 25686747.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7967679858207702,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.9797796010971069,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.06342286467552186,
+      "mean_token_accuracy": 0.9752787709236145,
+      "num_tokens": 25875368.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7932188332080841,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.0549253225326538,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.060981249809265135,
+      "mean_token_accuracy": 0.976200807094574,
+      "num_tokens": 26064480.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7899552762508393,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.1172103881835938,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.06555815935134887,
+      "mean_token_accuracy": 0.9736676931381225,
+      "num_tokens": 26253619.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.8019726693630218,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.94905686378479,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.0663321852684021,
+      "mean_token_accuracy": 0.9739456713199616,
+      "num_tokens": 26442736.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7939825117588043,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 0.9922711253166199,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05981417298316956,
+      "mean_token_accuracy": 0.9758030951023102,
+      "num_tokens": 26631525.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8085412085056305,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4970781803131104,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.06708773374557495,
+      "mean_token_accuracy": 0.9737519145011901,
+      "num_tokens": 26820483.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7950678527355194,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.7868127226829529,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.061734098196029666,
+      "mean_token_accuracy": 0.9747878730297088,
+      "num_tokens": 27009393.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7850766360759736,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.1386486291885376,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.05714603662490845,
+      "mean_token_accuracy": 0.9769004642963409,
+      "num_tokens": 27198251.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7933461844921113,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.0092092752456665,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.06240262985229492,
+      "mean_token_accuracy": 0.9742367327213287,
+      "num_tokens": 27387231.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8044621467590332,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.2477015256881714,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0642085313796997,
+      "mean_token_accuracy": 0.9748865723609924,
+      "num_tokens": 27576251.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7975574374198914,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.1167609691619873,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.06382874846458435,
+      "mean_token_accuracy": 0.9748350441455841,
+      "num_tokens": 27764560.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.8090806484222413,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 0.9714348912239075,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.06401748657226562,
+      "mean_token_accuracy": 0.9741845309734345,
+      "num_tokens": 27953532.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.8104567945383157,
+      "eval_loss": 0.07282773405313492,
+      "eval_mean_token_accuracy": 0.9701232534305305,
+      "eval_num_tokens": 28066151.0,
+      "eval_runtime": 12.7852,
+      "eval_samples_per_second": 391.078,
+      "eval_steps_per_second": 12.28,
+      "step": 1486
+    },
+    {
+      "entropy": 0.8090688765048981,
+      "epoch": 2.005383580080754,
+      "grad_norm": 0.9886816143989563,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.06012396812438965,
+      "mean_token_accuracy": 0.9764885842800141,
+      "num_tokens": 28141848.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.7821709334850311,
+      "epoch": 2.018842530282638,
+      "grad_norm": 0.9049969911575317,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.04819132387638092,
+      "mean_token_accuracy": 0.9816783249378205,
+      "num_tokens": 28330784.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.7845205307006836,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0231750011444092,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.04539612233638764,
+      "mean_token_accuracy": 0.9818619966506958,
+      "num_tokens": 28519580.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.7817754447460175,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.5244711637496948,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0429688960313797,
+      "mean_token_accuracy": 0.9834935545921326,
+      "num_tokens": 28707486.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.7733526527881622,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.0685060024261475,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.045852530002594,
+      "mean_token_accuracy": 0.9817728459835052,
+      "num_tokens": 28896759.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.7815343141555786,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.075853705406189,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.04565088748931885,
+      "mean_token_accuracy": 0.9823914349079133,
+      "num_tokens": 29085097.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.7751398742198944,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2573742866516113,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.04465666115283966,
+      "mean_token_accuracy": 0.9829874813556672,
+      "num_tokens": 29273868.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.7737197399139404,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 0.8681219220161438,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.042188242077827454,
+      "mean_token_accuracy": 0.9837059378623962,
+      "num_tokens": 29462985.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.7823441684246063,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.3503504991531372,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.04393318593502045,
+      "mean_token_accuracy": 0.9827627122402192,
+      "num_tokens": 29651156.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.7796198308467865,
+      "epoch": 2.126514131897712,
+      "grad_norm": 0.7799349427223206,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.04367418885231018,
+      "mean_token_accuracy": 0.9837401032447814,
+      "num_tokens": 29839789.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.7767675697803498,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.375465750694275,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.04427264630794525,
+      "mean_token_accuracy": 0.9832019031047821,
+      "num_tokens": 30028957.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.7912136912345886,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.9683088660240173,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.04428936243057251,
+      "mean_token_accuracy": 0.9832306385040284,
+      "num_tokens": 30217841.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.790750366449356,
+      "epoch": 2.166890982503365,
+      "grad_norm": 0.9393274188041687,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.04529332816600799,
+      "mean_token_accuracy": 0.9832263946533203,
+      "num_tokens": 30406540.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.7816023886203766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.1822642087936401,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.04192846417427063,
+      "mean_token_accuracy": 0.9846727311611175,
+      "num_tokens": 30595568.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7754982054233551,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.1961082220077515,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.043791168928146364,
+      "mean_token_accuracy": 0.9830836653709412,
+      "num_tokens": 30784300.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.7756961703300476,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 0.9719674587249756,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.04410521686077118,
+      "mean_token_accuracy": 0.9837198793888092,
+      "num_tokens": 30972950.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.7833336114883422,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.0175098180770874,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.044813218712806704,
+      "mean_token_accuracy": 0.9829723179340363,
+      "num_tokens": 31161783.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.7893047273159027,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.2077007293701172,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.04417986273765564,
+      "mean_token_accuracy": 0.9825236797332764,
+      "num_tokens": 31350487.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.794023334980011,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8530077934265137,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.043008115887641904,
+      "mean_token_accuracy": 0.9832422196865082,
+      "num_tokens": 31539217.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.785652619600296,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.4543156623840332,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0445009708404541,
+      "mean_token_accuracy": 0.9828064203262329,
+      "num_tokens": 31728308.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7841614365577698,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.297763466835022,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.045649731159210206,
+      "mean_token_accuracy": 0.9823782920837403,
+      "num_tokens": 31917293.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.7825099349021911,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.2237403392791748,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.047218257188797,
+      "mean_token_accuracy": 0.98156818151474,
+      "num_tokens": 32106628.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.7791289031505585,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8094402551651001,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.04178315699100495,
+      "mean_token_accuracy": 0.9846469461917877,
+      "num_tokens": 32295183.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.7743105947971344,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.3677793741226196,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.04540807008743286,
+      "mean_token_accuracy": 0.9827918827533721,
+      "num_tokens": 32484143.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.7804521083831787,
+      "epoch": 2.328398384925976,
+      "grad_norm": 0.8883333802223206,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.04598270654678345,
+      "mean_token_accuracy": 0.9823103487491608,
+      "num_tokens": 32673144.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.7760683119297027,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.989972710609436,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.043702703714370725,
+      "mean_token_accuracy": 0.9831363618373871,
+      "num_tokens": 32862142.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7726980447769165,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.0461052656173706,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.044418123364448545,
+      "mean_token_accuracy": 0.9821757376194,
+      "num_tokens": 33051382.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.7822885036468505,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.9027822017669678,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.04624638557434082,
+      "mean_token_accuracy": 0.9815774321556091,
+      "num_tokens": 33239729.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.7830030500888825,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.1743420362472534,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.04747769832611084,
+      "mean_token_accuracy": 0.9812638640403748,
+      "num_tokens": 33428615.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.7933187901973724,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 0.9334444403648376,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.047731178998947146,
+      "mean_token_accuracy": 0.9820823550224305,
+      "num_tokens": 33617331.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.7904782056808471,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.9396253824234009,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.04432434141635895,
+      "mean_token_accuracy": 0.9830163478851318,
+      "num_tokens": 33805978.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.7811533212661743,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 0.9742764234542847,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.046493279933929446,
+      "mean_token_accuracy": 0.9818211495876312,
+      "num_tokens": 33994957.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7848010778427124,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0378986597061157,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0454249233007431,
+      "mean_token_accuracy": 0.9817994356155395,
+      "num_tokens": 34184297.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.7754390180110932,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.1023566722869873,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.04380808770656586,
+      "mean_token_accuracy": 0.9829573810100556,
+      "num_tokens": 34373551.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.7767188012599945,
+      "epoch": 2.462987886944818,
+      "grad_norm": 0.8460658192634583,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04502716064453125,
+      "mean_token_accuracy": 0.9825617015361786,
+      "num_tokens": 34562169.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.7757977426052094,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.1013047695159912,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.045161890983581546,
+      "mean_token_accuracy": 0.9821914732456207,
+      "num_tokens": 34750863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.785107308626175,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.9510554671287537,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.045189899206161496,
+      "mean_token_accuracy": 0.9822288691997528,
+      "num_tokens": 34939528.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.7849742233753204,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0994503498077393,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.04333080947399139,
+      "mean_token_accuracy": 0.9830258965492249,
+      "num_tokens": 35128334.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.7875016331672668,
+      "epoch": 2.516823687752355,
+      "grad_norm": 0.9202979803085327,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.04838588833808899,
+      "mean_token_accuracy": 0.9815591931343078,
+      "num_tokens": 35317067.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.7813792943954467,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.9036716222763062,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.045662492513656616,
+      "mean_token_accuracy": 0.9817738354206085,
+      "num_tokens": 35506291.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.777835363149643,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.853438675403595,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.044686585664749146,
+      "mean_token_accuracy": 0.98329097032547,
+      "num_tokens": 35695141.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.7677761793136597,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.0699759721755981,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.045490306615829465,
+      "mean_token_accuracy": 0.9826831936836242,
+      "num_tokens": 35884173.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.7756118893623352,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.0356824398040771,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.0467751681804657,
+      "mean_token_accuracy": 0.9823338150978088,
+      "num_tokens": 36073241.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.7775443732738495,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.1619561910629272,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.04741084575653076,
+      "mean_token_accuracy": 0.9815192580223083,
+      "num_tokens": 36262133.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7899596333503723,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.3479284048080444,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.04588942229747772,
+      "mean_token_accuracy": 0.9821865320205688,
+      "num_tokens": 36451460.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.7832387804985046,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.6297204494476318,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.04711659252643585,
+      "mean_token_accuracy": 0.9815292179584503,
+      "num_tokens": 36640366.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.7871865928173065,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.3100138902664185,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.04919236898422241,
+      "mean_token_accuracy": 0.9809704780578613,
+      "num_tokens": 36829017.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.7844007611274719,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.2907201051712036,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.047159326076507566,
+      "mean_token_accuracy": 0.9816433548927307,
+      "num_tokens": 37018034.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.7885129511356354,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.1324766874313354,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.04420983195304871,
+      "mean_token_accuracy": 0.9828807294368744,
+      "num_tokens": 37206768.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.7761563122272491,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.9269904494285583,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.04585547745227814,
+      "mean_token_accuracy": 0.9823735177516937,
+      "num_tokens": 37396010.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7831155300140381,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5890840291976929,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.046450641751289365,
+      "mean_token_accuracy": 0.9817734777927398,
+      "num_tokens": 37584809.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.7729385912418365,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.4823863506317139,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.04653529226779938,
+      "mean_token_accuracy": 0.9815726220607758,
+      "num_tokens": 37773869.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.7752901673316955,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.9423630237579346,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.04599723219871521,
+      "mean_token_accuracy": 0.9816120624542236,
+      "num_tokens": 37962642.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.7750158667564392,
+      "epoch": 2.718707940780619,
+      "grad_norm": 0.9398621320724487,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.04729728400707245,
+      "mean_token_accuracy": 0.9816868543624878,
+      "num_tokens": 38151325.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.7693424999713898,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3207950592041016,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.04474019706249237,
+      "mean_token_accuracy": 0.9830808460712432,
+      "num_tokens": 38340719.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.7750859558582306,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.070200800895691,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.045914703607559205,
+      "mean_token_accuracy": 0.9820488631725312,
+      "num_tokens": 38529293.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7697387218475342,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.0545568466186523,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.04357335567474365,
+      "mean_token_accuracy": 0.9827888309955597,
+      "num_tokens": 38718464.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.7682756781578064,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.0945985317230225,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.0429091215133667,
+      "mean_token_accuracy": 0.9832668721675872,
+      "num_tokens": 38907573.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.7804523587226868,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8521431684494019,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.04662351012229919,
+      "mean_token_accuracy": 0.981964933872223,
+      "num_tokens": 39096598.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.7786886096000671,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.09945809841156,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.050166720151901247,
+      "mean_token_accuracy": 0.9816301107406616,
+      "num_tokens": 39285069.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.7790449261665344,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.1673868894577026,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.048508059978485105,
+      "mean_token_accuracy": 0.9809533059597015,
+      "num_tokens": 39474252.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.7774537563323974,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.0648490190505981,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.04540320634841919,
+      "mean_token_accuracy": 0.9822351098060608,
+      "num_tokens": 39663262.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7683943033218383,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.0106912851333618,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.04466369152069092,
+      "mean_token_accuracy": 0.9818881273269653,
+      "num_tokens": 39852133.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.7642774760723114,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0109530687332153,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.045978450775146486,
+      "mean_token_accuracy": 0.9819821894168854,
+      "num_tokens": 40041679.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.7647961139678955,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.9050237536430359,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.045694130659103396,
+      "mean_token_accuracy": 0.982342666387558,
+      "num_tokens": 40230019.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.7589850842952728,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.8879176378250122,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.046186423301696776,
+      "mean_token_accuracy": 0.9818771004676818,
+      "num_tokens": 40418932.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.765201997756958,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 0.9103795886039734,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.04475873112678528,
+      "mean_token_accuracy": 0.9821203172206878,
+      "num_tokens": 40607792.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.7653619945049286,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.0566699504852295,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.04391545057296753,
+      "mean_token_accuracy": 0.9828326046466828,
+      "num_tokens": 40797122.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.763550215959549,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.1280677318572998,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.045582184195518495,
+      "mean_token_accuracy": 0.9819066524505615,
+      "num_tokens": 40985689.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.7749062657356263,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.110323429107666,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.04598402082920074,
+      "mean_token_accuracy": 0.9815545618534088,
+      "num_tokens": 41174743.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.7805372893810272,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.0649195909500122,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.04705569744110107,
+      "mean_token_accuracy": 0.9815925776958465,
+      "num_tokens": 41363160.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.7711063683032989,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.1459592580795288,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.045355689525604245,
+      "mean_token_accuracy": 0.9832463562488556,
+      "num_tokens": 41551932.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.7747956931591033,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.1602799892425537,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.04494523704051971,
+      "mean_token_accuracy": 0.9819566667079925,
+      "num_tokens": 41740544.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.7584574401378632,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.054417371749878,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.04810510575771332,
+      "mean_token_accuracy": 0.9811540365219116,
+      "num_tokens": 41929344.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.7753648416251894,
+      "eval_loss": 0.07181113958358765,
+      "eval_mean_token_accuracy": 0.9716544128527307,
+      "eval_num_tokens": 42099366.0,
+      "eval_runtime": 12.748,
+      "eval_samples_per_second": 392.22,
+      "eval_steps_per_second": 12.316,
+      "step": 2229
+    },
+    {
+      "entropy": 0.7669954419136047,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.8211690783500671,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.04577431380748749,
+      "mean_token_accuracy": 0.9823363184928894,
+      "num_tokens": 42118188.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.7600219964981079,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.0873607397079468,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.032501548528671265,
+      "mean_token_accuracy": 0.9881355822086334,
+      "num_tokens": 42306598.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.7441959500312805,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.1175898313522339,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.031047609448432923,
+      "mean_token_accuracy": 0.9891582727432251,
+      "num_tokens": 42495250.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.7364335536956788,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.987751841545105,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.028629058599472047,
+      "mean_token_accuracy": 0.9894859492778778,
+      "num_tokens": 42684326.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.735680228471756,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.194305658340454,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.029283303022384643,
+      "mean_token_accuracy": 0.9887769162654877,
+      "num_tokens": 42873221.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.7342639684677124,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.254442811012268,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.0302200973033905,
+      "mean_token_accuracy": 0.988795381784439,
+      "num_tokens": 43062486.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.7391974747180938,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.2648969888687134,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.0316451758146286,
+      "mean_token_accuracy": 0.9879588901996612,
+      "num_tokens": 43251113.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.7401826322078705,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.0483295917510986,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.030878132581710814,
+      "mean_token_accuracy": 0.9886977672576904,
+      "num_tokens": 43440152.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.732647967338562,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.1125447750091553,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.02838427424430847,
+      "mean_token_accuracy": 0.9890213429927825,
+      "num_tokens": 43629353.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.7314932703971863,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.0979042053222656,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.0315592348575592,
+      "mean_token_accuracy": 0.9888832688331604,
+      "num_tokens": 43818981.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.7305827140808105,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9610060453414917,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.03112468421459198,
+      "mean_token_accuracy": 0.9890477001667023,
+      "num_tokens": 44008430.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.7333625435829163,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.0625078678131104,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.03105958104133606,
+      "mean_token_accuracy": 0.9884848177433014,
+      "num_tokens": 44197264.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.7373903751373291,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.1683465242385864,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.031627464294433597,
+      "mean_token_accuracy": 0.9879396498203278,
+      "num_tokens": 44386205.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.7391358911991119,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.05154287815094,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.03118889331817627,
+      "mean_token_accuracy": 0.9880312740802765,
+      "num_tokens": 44575099.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.7351223647594451,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.9531214237213135,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.030980610847473146,
+      "mean_token_accuracy": 0.9882127285003662,
+      "num_tokens": 44763965.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.7361315310001373,
+      "epoch": 3.203230148048452,
+      "grad_norm": 0.8818249702453613,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.029705989360809325,
+      "mean_token_accuracy": 0.9894622683525085,
+      "num_tokens": 44952983.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.7356473922729492,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 0.9263529777526855,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.03202669024467468,
+      "mean_token_accuracy": 0.9877129793167114,
+      "num_tokens": 45142080.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.7326373934745789,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.0517516136169434,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.031153690814971925,
+      "mean_token_accuracy": 0.9891136825084687,
+      "num_tokens": 45330769.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.7348711133003235,
+      "epoch": 3.243606998654105,
+      "grad_norm": 0.8774152994155884,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030574262142181396,
+      "mean_token_accuracy": 0.9888435959815979,
+      "num_tokens": 45519217.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.7379967510700226,
+      "epoch": 3.257065948855989,
+      "grad_norm": 0.9164374470710754,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.03096820116043091,
+      "mean_token_accuracy": 0.9884447932243348,
+      "num_tokens": 45708467.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.7369260907173156,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.3184924125671387,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03082413077354431,
+      "mean_token_accuracy": 0.9888757109642029,
+      "num_tokens": 45897424.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.7366503119468689,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.9073860049247742,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.03220943808555603,
+      "mean_token_accuracy": 0.9877902865409851,
+      "num_tokens": 46086351.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.7341453731060028,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.0940542221069336,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.03329495787620544,
+      "mean_token_accuracy": 0.9877542018890381,
+      "num_tokens": 46275394.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.7312689840793609,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 0.9651952385902405,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.031667912006378175,
+      "mean_token_accuracy": 0.9881450653076171,
+      "num_tokens": 46463821.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.7228557825088501,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 0.971441924571991,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.030608028173446655,
+      "mean_token_accuracy": 0.9882231116294861,
+      "num_tokens": 46653025.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.723542982339859,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.0060454607009888,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.030545371770858764,
+      "mean_token_accuracy": 0.9879125893115998,
+      "num_tokens": 46842068.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.7321009397506714,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.0572997331619263,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.032070663571357724,
+      "mean_token_accuracy": 0.9878279268741608,
+      "num_tokens": 47031045.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.735515958070755,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.9835748672485352,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.03338083028793335,
+      "mean_token_accuracy": 0.9877255499362946,
+      "num_tokens": 47219625.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.7396618247032165,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 0.9655317068099976,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.03589712679386139,
+      "mean_token_accuracy": 0.9869556546211242,
+      "num_tokens": 47408311.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.7358698546886444,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9225422143936157,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.02914820909500122,
+      "mean_token_accuracy": 0.9886718571186066,
+      "num_tokens": 47597297.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.7338236391544342,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.8253662586212158,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.02996352016925812,
+      "mean_token_accuracy": 0.9891344487667084,
+      "num_tokens": 47785737.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.7320856511592865,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.0716700553894043,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.032322123646736145,
+      "mean_token_accuracy": 0.9877110660076142,
+      "num_tokens": 47974241.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.7293457925319672,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.0915459394454956,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.031131434440612792,
+      "mean_token_accuracy": 0.9885926306247711,
+      "num_tokens": 48162582.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.7207926452159882,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0700562000274658,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03113105595111847,
+      "mean_token_accuracy": 0.9880475521087646,
+      "num_tokens": 48351627.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.7178894281387329,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.1952542066574097,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03204495906829834,
+      "mean_token_accuracy": 0.9880682706832886,
+      "num_tokens": 48540171.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.7166945099830627,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0920100212097168,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.03096899390220642,
+      "mean_token_accuracy": 0.9884315967559815,
+      "num_tokens": 48729737.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.7187432169914245,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.9596593976020813,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.03240742087364197,
+      "mean_token_accuracy": 0.9878904581069946,
+      "num_tokens": 48918599.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.7224443972110748,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9406276345252991,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.03253386616706848,
+      "mean_token_accuracy": 0.9873985171318054,
+      "num_tokens": 49107385.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.7226529896259308,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.2169438600540161,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.03282734453678131,
+      "mean_token_accuracy": 0.9876641929149628,
+      "num_tokens": 49295969.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.7263983964920044,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8808715343475342,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.030802983045578002,
+      "mean_token_accuracy": 0.9885481476783753,
+      "num_tokens": 49484706.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.7291677474975586,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2453594207763672,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.03403027057647705,
+      "mean_token_accuracy": 0.987172520160675,
+      "num_tokens": 49673484.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.7396404981613159,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.9215838313102722,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.03405522108078003,
+      "mean_token_accuracy": 0.9871597826480866,
+      "num_tokens": 49862087.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.7382308602333069,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0659712553024292,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.032089871168136594,
+      "mean_token_accuracy": 0.9874629497528076,
+      "num_tokens": 50051379.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.7337264895439148,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 0.9307538270950317,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.032426226139068606,
+      "mean_token_accuracy": 0.9879062473773956,
+      "num_tokens": 50239986.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.7381527841091156,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.1867440938949585,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.03276415467262268,
+      "mean_token_accuracy": 0.9878055810928345,
+      "num_tokens": 50428872.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.7337838113307953,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.9655186533927917,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.03141146302223206,
+      "mean_token_accuracy": 0.9882215678691864,
+      "num_tokens": 50618182.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.7354952573776246,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8848851919174194,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.03387039303779602,
+      "mean_token_accuracy": 0.9871303081512451,
+      "num_tokens": 50806430.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.7329667210578918,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 0.9976673722267151,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.03129216730594635,
+      "mean_token_accuracy": 0.9882283449172974,
+      "num_tokens": 50995500.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.7277809977531433,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.2560697793960571,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.03357822299003601,
+      "mean_token_accuracy": 0.9876434028148651,
+      "num_tokens": 51184436.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.727894538640976,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.1099125146865845,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.032482534646987915,
+      "mean_token_accuracy": 0.9878577649593353,
+      "num_tokens": 51373238.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.7318481564521789,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1980950832366943,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.03284388482570648,
+      "mean_token_accuracy": 0.9874962747097016,
+      "num_tokens": 51561631.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.7219811856746674,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 0.8730921745300293,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.030655479431152342,
+      "mean_token_accuracy": 0.9877778172492981,
+      "num_tokens": 51750845.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.7289035737514495,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.1195666790008545,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.03155539035797119,
+      "mean_token_accuracy": 0.9877057671546936,
+      "num_tokens": 51940009.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.7332375526428223,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9390997886657715,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.031307452917099,
+      "mean_token_accuracy": 0.987467783689499,
+      "num_tokens": 52128442.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.7293307244777679,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.0272324085235596,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.03318062424659729,
+      "mean_token_accuracy": 0.9873259484767913,
+      "num_tokens": 52317152.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.7320701718330384,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9884606003761292,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03461401164531708,
+      "mean_token_accuracy": 0.9867411673069,
+      "num_tokens": 52506106.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.7248260796070098,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.879682719707489,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.03406862616539001,
+      "mean_token_accuracy": 0.9872734069824218,
+      "num_tokens": 52695500.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.7237857282161713,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9409974217414856,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.029435336589813232,
+      "mean_token_accuracy": 0.9891225278377533,
+      "num_tokens": 52884940.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.7237291693687439,
+      "epoch": 3.781965006729475,
+      "grad_norm": 0.9532694220542908,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.030907022953033447,
+      "mean_token_accuracy": 0.9881488382816315,
+      "num_tokens": 53073909.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.7252696454524994,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 0.9157902002334595,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.032040882110595706,
+      "mean_token_accuracy": 0.9880847573280335,
+      "num_tokens": 53262962.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.7237608432769775,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.1450515985488892,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.03199286162853241,
+      "mean_token_accuracy": 0.987687611579895,
+      "num_tokens": 53451982.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.7237064421176911,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.9945085644721985,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.03347462117671966,
+      "mean_token_accuracy": 0.9875530540943146,
+      "num_tokens": 53640737.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.7271939218044281,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.9351736903190613,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.03140422701835632,
+      "mean_token_accuracy": 0.9877301037311554,
+      "num_tokens": 53829142.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.7265262603759766,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.1019784212112427,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.030748724937438965,
+      "mean_token_accuracy": 0.9881740033626556,
+      "num_tokens": 54018331.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.7270502150058746,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 0.9153032898902893,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.033472740650177,
+      "mean_token_accuracy": 0.9876999020576477,
+      "num_tokens": 54206935.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.7345028996467591,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.0144257545471191,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.033550959825515744,
+      "mean_token_accuracy": 0.9874906241893768,
+      "num_tokens": 54395443.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.7274066030979156,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0221407413482666,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.033388751745224,
+      "mean_token_accuracy": 0.9870682001113892,
+      "num_tokens": 54584552.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.7293415367603302,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.019165277481079,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.03568689227104187,
+      "mean_token_accuracy": 0.9861597299575806,
+      "num_tokens": 54773705.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.7281844198703766,
+      "epoch": 3.916554508748318,
+      "grad_norm": 0.933020830154419,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.030142486095428467,
+      "mean_token_accuracy": 0.988915103673935,
+      "num_tokens": 54962152.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.7242733418941498,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.9429674744606018,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.0337733268737793,
+      "mean_token_accuracy": 0.987263411283493,
+      "num_tokens": 55151179.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.7306612014770508,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.9022014737129211,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.03387616574764252,
+      "mean_token_accuracy": 0.9871586680412292,
+      "num_tokens": 55339658.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.7246532738208771,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2193421125411987,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.03193953037261963,
+      "mean_token_accuracy": 0.9879468739032745,
+      "num_tokens": 55528467.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.7268587052822113,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.2917358875274658,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.03584722280502319,
+      "mean_token_accuracy": 0.9867104470729828,
+      "num_tokens": 55717308.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.735340416431427,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.2229385375976562,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.03506708741188049,
+      "mean_token_accuracy": 0.9864643573760986,
+      "num_tokens": 55906059.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.7337038695812226,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.0747122764587402,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.03241217732429504,
+      "mean_token_accuracy": 0.9876143455505371,
+      "num_tokens": 56094606.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.7333687073106219,
+      "eval_loss": 0.07489899545907974,
+      "eval_mean_token_accuracy": 0.9719084019114257,
+      "eval_num_tokens": 56132350.0,
+      "eval_runtime": 12.7823,
+      "eval_samples_per_second": 391.166,
+      "eval_steps_per_second": 12.283,
+      "step": 2972
+    },
+    {
+      "entropy": 0.72461416721344,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7916642427444458,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.02286165803670883,
+      "mean_token_accuracy": 0.9920066714286804,
+      "num_tokens": 56283397.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.7150031447410583,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.9078621864318848,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.020906335115432738,
+      "mean_token_accuracy": 0.9925728499889374,
+      "num_tokens": 56472408.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.7091723501682281,
+      "epoch": 4.037685060565276,
+      "grad_norm": 0.8117756843566895,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.02036277800798416,
+      "mean_token_accuracy": 0.9928471207618713,
+      "num_tokens": 56660906.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.695214182138443,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7815372943878174,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019653698801994322,
+      "mean_token_accuracy": 0.992583978176117,
+      "num_tokens": 56850067.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.6937313377857208,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4769781827926636,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.02311961054801941,
+      "mean_token_accuracy": 0.9921472012996674,
+      "num_tokens": 57039005.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.6977694034576416,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.0028746128082275,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.021832706034183504,
+      "mean_token_accuracy": 0.9917877376079559,
+      "num_tokens": 57227782.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.698432594537735,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.0107522010803223,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.020331941545009613,
+      "mean_token_accuracy": 0.9932828366756439,
+      "num_tokens": 57416541.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.6941913783550262,
+      "epoch": 4.104979811574697,
+      "grad_norm": 0.8923842310905457,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019534927606582642,
+      "mean_token_accuracy": 0.9930490672588348,
+      "num_tokens": 57605162.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.6947623729705811,
+      "epoch": 4.118438761776582,
+      "grad_norm": 0.9093812108039856,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.02151128649711609,
+      "mean_token_accuracy": 0.9920440256595612,
+      "num_tokens": 57793756.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.6972509443759918,
+      "epoch": 4.131897711978466,
+      "grad_norm": 0.7549760341644287,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.020574498176574706,
+      "mean_token_accuracy": 0.9928867638111114,
+      "num_tokens": 57982678.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.6952514111995697,
+      "epoch": 4.14535666218035,
+      "grad_norm": 0.9476824402809143,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.02087453305721283,
+      "mean_token_accuracy": 0.9924874365329742,
+      "num_tokens": 58171508.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.6910251140594482,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.2548149824142456,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.020806866884231567,
+      "mean_token_accuracy": 0.9921584248542785,
+      "num_tokens": 58360594.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.6958921313285827,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.0649152994155884,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.020952591300010683,
+      "mean_token_accuracy": 0.9924699544906617,
+      "num_tokens": 58549637.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.6932485699653625,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.8891078233718872,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.020813333988189697,
+      "mean_token_accuracy": 0.9927419483661651,
+      "num_tokens": 58738595.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.6984799921512603,
+      "epoch": 4.199192462987887,
+      "grad_norm": 0.9459101557731628,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.01971060484647751,
+      "mean_token_accuracy": 0.9926240146160126,
+      "num_tokens": 58927507.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.7021916508674622,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.8318898677825928,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020860180258750916,
+      "mean_token_accuracy": 0.9927471876144409,
+      "num_tokens": 59116537.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.6972847402095794,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.9066984057426453,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.023059430718421935,
+      "mean_token_accuracy": 0.9919216752052307,
+      "num_tokens": 59305838.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.7049594700336457,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1277318000793457,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.022385355830192567,
+      "mean_token_accuracy": 0.99165118932724,
+      "num_tokens": 59494587.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.7045334160327912,
+      "epoch": 4.253028263795424,
+      "grad_norm": 0.8147489428520203,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.022358573973178864,
+      "mean_token_accuracy": 0.9918249011039734,
+      "num_tokens": 59683021.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.7007148265838623,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.9161480069160461,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.02088596671819687,
+      "mean_token_accuracy": 0.9925470650196075,
+      "num_tokens": 59871999.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.7006107687950134,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.9611289501190186,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.022619232535362244,
+      "mean_token_accuracy": 0.991577273607254,
+      "num_tokens": 60060486.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.7069303750991821,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.0339760780334473,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.021571077406406403,
+      "mean_token_accuracy": 0.9922139644622803,
+      "num_tokens": 60249686.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.7146372079849244,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.9761297702789307,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.022300006449222566,
+      "mean_token_accuracy": 0.9918746411800384,
+      "num_tokens": 60438481.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.7134014248847962,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 0.9729707837104797,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.02121303677558899,
+      "mean_token_accuracy": 0.9925692796707153,
+      "num_tokens": 60627501.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.7110478222370148,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.9456778168678284,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.020917078852653502,
+      "mean_token_accuracy": 0.9922434687614441,
+      "num_tokens": 60816135.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.7111951589584351,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8617656826972961,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.020851245522499083,
+      "mean_token_accuracy": 0.9924458563327789,
+      "num_tokens": 61005658.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.7193978548049926,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.8495420813560486,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.02084285914897919,
+      "mean_token_accuracy": 0.9927356660366058,
+      "num_tokens": 61194170.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.7188209414482116,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.728841245174408,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.02111642211675644,
+      "mean_token_accuracy": 0.9926697969436645,
+      "num_tokens": 61382852.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.7130864262580872,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.1274386644363403,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.021722708642482758,
+      "mean_token_accuracy": 0.9922453343868256,
+      "num_tokens": 61572195.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.7132770180702209,
+      "epoch": 4.401076716016151,
+      "grad_norm": 0.9251006245613098,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.021602503955364227,
+      "mean_token_accuracy": 0.9919042110443115,
+      "num_tokens": 61761119.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.7063322901725769,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.0182228088378906,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.020274630188941954,
+      "mean_token_accuracy": 0.9928637027740479,
+      "num_tokens": 61949931.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.7106044590473175,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.795170783996582,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.020594277977943422,
+      "mean_token_accuracy": 0.9925063610076904,
+      "num_tokens": 62138377.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7151111841201783,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1479791402816772,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.02117723524570465,
+      "mean_token_accuracy": 0.9921884059906005,
+      "num_tokens": 62327009.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.7172957301139832,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9795703291893005,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021917465329170226,
+      "mean_token_accuracy": 0.9915876984596252,
+      "num_tokens": 62515859.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.7175660550594329,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.9326112866401672,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.021142563223838805,
+      "mean_token_accuracy": 0.9922006845474243,
+      "num_tokens": 62704678.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.7085749089717865,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.8895787596702576,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.0222358375787735,
+      "mean_token_accuracy": 0.9918791234493256,
+      "num_tokens": 62893576.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.7059603810310364,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2298023700714111,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021999095380306245,
+      "mean_token_accuracy": 0.9920522570610046,
+      "num_tokens": 63082258.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.707107275724411,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.8295655250549316,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021844917535781862,
+      "mean_token_accuracy": 0.9922264873981476,
+      "num_tokens": 63271202.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.7119229018688202,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.7440339922904968,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.022164498269557954,
+      "mean_token_accuracy": 0.9918813884258271,
+      "num_tokens": 63460614.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.7065344154834747,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7275091409683228,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.019415955245494842,
+      "mean_token_accuracy": 0.9928511261940003,
+      "num_tokens": 63649327.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.6984272241592407,
+      "epoch": 4.549125168236878,
+      "grad_norm": 0.9989121556282043,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02176133990287781,
+      "mean_token_accuracy": 0.9923081934452057,
+      "num_tokens": 63837983.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.7084203243255616,
+      "epoch": 4.562584118438762,
+      "grad_norm": 0.9526837468147278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.02182276248931885,
+      "mean_token_accuracy": 0.9921061515808105,
+      "num_tokens": 64026855.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.7097050070762634,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.0156456232070923,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.020074066519737244,
+      "mean_token_accuracy": 0.9926879525184631,
+      "num_tokens": 64215889.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.7087334096431732,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.780275821685791,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.021703639626502992,
+      "mean_token_accuracy": 0.9917874097824096,
+      "num_tokens": 64404515.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.7142964243888855,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7382510304450989,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.019867205619812013,
+      "mean_token_accuracy": 0.9927354514598846,
+      "num_tokens": 64593147.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.7165492177009583,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.8259146213531494,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.021262936294078827,
+      "mean_token_accuracy": 0.9914182722568512,
+      "num_tokens": 64782041.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.7098793745040893,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.9540995359420776,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.0195529043674469,
+      "mean_token_accuracy": 0.9931159794330597,
+      "num_tokens": 64971390.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.7104229986667633,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.8698863387107849,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.019119253754615782,
+      "mean_token_accuracy": 0.9934523403644562,
+      "num_tokens": 65160253.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.7067269921302796,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.9622324705123901,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.02044745087623596,
+      "mean_token_accuracy": 0.992815512418747,
+      "num_tokens": 65349403.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.7016099870204926,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.999839723110199,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.019250848889350893,
+      "mean_token_accuracy": 0.9929339647293091,
+      "num_tokens": 65538164.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.7059290170669555,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.7539619207382202,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.02152409851551056,
+      "mean_token_accuracy": 0.99238161444664,
+      "num_tokens": 65727493.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.7103133797645569,
+      "epoch": 4.697173620457605,
+      "grad_norm": 0.9892463088035583,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020616935193538667,
+      "mean_token_accuracy": 0.9927784681320191,
+      "num_tokens": 65916233.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.703304260969162,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.1750195026397705,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021048617362976075,
+      "mean_token_accuracy": 0.9921808183193207,
+      "num_tokens": 66105229.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.7017257273197174,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.887427031993866,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02077500820159912,
+      "mean_token_accuracy": 0.992085075378418,
+      "num_tokens": 66294280.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.7027543246746063,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.0987571477890015,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.02167164385318756,
+      "mean_token_accuracy": 0.992034786939621,
+      "num_tokens": 66483182.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.7053992748260498,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.0075602531433105,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.022686071693897247,
+      "mean_token_accuracy": 0.9918968200683593,
+      "num_tokens": 66671795.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.7109473288059235,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.1655596494674683,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.022698774933815002,
+      "mean_token_accuracy": 0.9913642406463623,
+      "num_tokens": 66861166.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.7130867660045623,
+      "epoch": 4.77792732166891,
+      "grad_norm": 0.9386347532272339,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0205889493227005,
+      "mean_token_accuracy": 0.9924897074699401,
+      "num_tokens": 67049967.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.7099342107772827,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.8524489998817444,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.01985667645931244,
+      "mean_token_accuracy": 0.9928426086902619,
+      "num_tokens": 67238781.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.7100094079971313,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7522528171539307,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.02066437602043152,
+      "mean_token_accuracy": 0.9924224317073822,
+      "num_tokens": 67427802.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7172351717948914,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9865215420722961,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.021969664096832275,
+      "mean_token_accuracy": 0.9919721603393554,
+      "num_tokens": 67616675.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.7113273084163666,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9339645504951477,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.019794631004333495,
+      "mean_token_accuracy": 0.9925971984863281,
+      "num_tokens": 67805877.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.7041554272174835,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.337364912033081,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020412631332874298,
+      "mean_token_accuracy": 0.9923949122428894,
+      "num_tokens": 67994858.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.7090700149536133,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.7725968956947327,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.02121713161468506,
+      "mean_token_accuracy": 0.9920605599880219,
+      "num_tokens": 68183865.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.7141157209873199,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.8638854026794434,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.020726922154426574,
+      "mean_token_accuracy": 0.992779266834259,
+      "num_tokens": 68372954.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.7078199684619904,
+      "epoch": 4.885598923283984,
+      "grad_norm": 0.8576115965843201,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.019535741209983824,
+      "mean_token_accuracy": 0.9927346885204316,
+      "num_tokens": 68562535.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.7034667313098908,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8339338898658752,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.021528851985931397,
+      "mean_token_accuracy": 0.991979056596756,
+      "num_tokens": 68751258.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.6983348250389099,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8739761710166931,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020945993065834046,
+      "mean_token_accuracy": 0.9923328399658203,
+      "num_tokens": 68939586.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.6922877132892609,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.886886715888977,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.020974791049957274,
+      "mean_token_accuracy": 0.992312878370285,
+      "num_tokens": 69128270.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.6857815742492676,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.9415780901908875,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.020201480388641356,
+      "mean_token_accuracy": 0.9925193309783935,
+      "num_tokens": 69317066.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.6832823157310486,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.8311293721199036,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.02008202373981476,
+      "mean_token_accuracy": 0.9926081895828247,
+      "num_tokens": 69505698.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.6914379715919494,
+      "epoch": 4.96635262449529,
+      "grad_norm": 0.8799999356269836,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.0189486563205719,
+      "mean_token_accuracy": 0.9932616055011749,
+      "num_tokens": 69694164.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.692799162864685,
+      "epoch": 4.979811574697173,
+      "grad_norm": 0.9579370021820068,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.019743245840072633,
+      "mean_token_accuracy": 0.9932292222976684,
+      "num_tokens": 69882717.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.6969726324081421,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.0632585287094116,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.019562503695487975,
+      "mean_token_accuracy": 0.9929000854492187,
+      "num_tokens": 70071225.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.7056300662885047,
+      "eval_loss": 0.08702203631401062,
+      "eval_mean_token_accuracy": 0.9712980936287315,
+      "eval_num_tokens": 70165509.0,
+      "eval_runtime": 12.7642,
+      "eval_samples_per_second": 391.719,
+      "eval_steps_per_second": 12.3,
+      "step": 3715
+    },
+    {
+      "entropy": 0.6998905837535858,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8414336442947388,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01888625919818878,
+      "mean_token_accuracy": 0.9934305250644684,
+      "num_tokens": 70259811.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.6933371186256408,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.7618409395217896,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01257207989692688,
+      "mean_token_accuracy": 0.9958574831485748,
+      "num_tokens": 70448683.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.6861799776554107,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.935829758644104,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.0118876114487648,
+      "mean_token_accuracy": 0.9963350534439087,
+      "num_tokens": 70637078.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.6778797626495361,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.9417919516563416,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.012690700590610504,
+      "mean_token_accuracy": 0.9954731047153473,
+      "num_tokens": 70826145.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.6735071957111358,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.7570633292198181,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01321282535791397,
+      "mean_token_accuracy": 0.9955435931682587,
+      "num_tokens": 71015458.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.677568006515503,
+      "epoch": 5.074024226110363,
+      "grad_norm": 0.8157011866569519,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.012434301525354385,
+      "mean_token_accuracy": 0.9957576751708984,
+      "num_tokens": 71204332.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.6771604120731354,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.780483603477478,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.013237486779689788,
+      "mean_token_accuracy": 0.9957249999046326,
+      "num_tokens": 71393307.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.6761959791183472,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8249189257621765,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.011912986636161804,
+      "mean_token_accuracy": 0.9961272835731506,
+      "num_tokens": 71582233.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.6717506468296051,
+      "epoch": 5.114401076716016,
+      "grad_norm": 0.9667718410491943,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.012740559875965118,
+      "mean_token_accuracy": 0.9953394651412963,
+      "num_tokens": 71771906.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.6755871415138245,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.9068875908851624,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.012314559519290924,
+      "mean_token_accuracy": 0.9955949664115906,
+      "num_tokens": 71960698.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.6791335880756378,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.7650554776191711,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.011927495896816253,
+      "mean_token_accuracy": 0.9957858562469483,
+      "num_tokens": 72149627.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.6738516509532928,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 0.7267928123474121,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.011817440390586853,
+      "mean_token_accuracy": 0.996101975440979,
+      "num_tokens": 72338469.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.6701940059661865,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.9211037755012512,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012324941903352737,
+      "mean_token_accuracy": 0.9956477820873261,
+      "num_tokens": 72527499.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.6763210237026215,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.7772125601768494,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.013027346134185791,
+      "mean_token_accuracy": 0.9959511935710907,
+      "num_tokens": 72715639.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.6793705642223358,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.582334578037262,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.011401662230491638,
+      "mean_token_accuracy": 0.9961598336696624,
+      "num_tokens": 72904615.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.6798616409301758,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.8502725958824158,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011671387404203416,
+      "mean_token_accuracy": 0.9961067497730255,
+      "num_tokens": 73093522.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.682939213514328,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7421601414680481,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.011898189783096313,
+      "mean_token_accuracy": 0.9962145388126373,
+      "num_tokens": 73282415.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.6821306705474853,
+      "epoch": 5.235531628532974,
+      "grad_norm": 0.7552776336669922,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.01232050210237503,
+      "mean_token_accuracy": 0.9957845032215118,
+      "num_tokens": 73471211.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.6746087670326233,
+      "epoch": 5.248990578734858,
+      "grad_norm": 0.7144336700439453,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01240069791674614,
+      "mean_token_accuracy": 0.9960526466369629,
+      "num_tokens": 73660289.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.6723428070545197,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.8043408393859863,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.01204245388507843,
+      "mean_token_accuracy": 0.9960211217403412,
+      "num_tokens": 73849312.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.6766897737979889,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.8608251810073853,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012303075194358826,
+      "mean_token_accuracy": 0.9955043613910675,
+      "num_tokens": 74038344.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.6809355199337006,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.4872569739818573,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.010690021514892577,
+      "mean_token_accuracy": 0.9965942919254303,
+      "num_tokens": 74226981.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.6819032669067383,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.0376152992248535,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012148186564445496,
+      "mean_token_accuracy": 0.9958918631076813,
+      "num_tokens": 74415130.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.6856858193874359,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.7241474986076355,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012284471094608307,
+      "mean_token_accuracy": 0.9959208726882934,
+      "num_tokens": 74603788.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.6874188125133515,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.6505919098854065,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.012959125638008117,
+      "mean_token_accuracy": 0.9956629753112793,
+      "num_tokens": 74792306.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.6912538707256317,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7696103453636169,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012945668399333954,
+      "mean_token_accuracy": 0.9956756651401519,
+      "num_tokens": 74981265.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.6915560126304626,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.8245812058448792,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.01408768892288208,
+      "mean_token_accuracy": 0.9950980842113495,
+      "num_tokens": 75170337.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.6872740864753724,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8330141305923462,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.011667902022600174,
+      "mean_token_accuracy": 0.9960226893424988,
+      "num_tokens": 75359113.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.6825072944164277,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.9389140009880066,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012351009249687194,
+      "mean_token_accuracy": 0.9954748272895813,
+      "num_tokens": 75548102.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.6821270763874054,
+      "epoch": 5.397039030955585,
+      "grad_norm": 1.0638995170593262,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.0128641277551651,
+      "mean_token_accuracy": 0.995329612493515,
+      "num_tokens": 75736452.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.6857812821865081,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7838811278343201,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.011757558584213257,
+      "mean_token_accuracy": 0.9959722757339478,
+      "num_tokens": 75925013.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.6869213104248046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 0.5632259845733643,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.011284272372722625,
+      "mean_token_accuracy": 0.9962978959083557,
+      "num_tokens": 76113398.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.6807511866092681,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.7132044434547424,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012044224143028259,
+      "mean_token_accuracy": 0.9959294259548187,
+      "num_tokens": 76302629.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.6815520942211151,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.7640891075134277,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.012763911485671997,
+      "mean_token_accuracy": 0.9953242301940918,
+      "num_tokens": 76491675.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.6802713990211486,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.5691967010498047,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.0116192027926445,
+      "mean_token_accuracy": 0.9958603858947754,
+      "num_tokens": 76680426.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.676917290687561,
+      "epoch": 5.477792732166891,
+      "grad_norm": 0.8160638213157654,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.01229381114244461,
+      "mean_token_accuracy": 0.9956679701805115,
+      "num_tokens": 76869321.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.6745438575744629,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.8013969659805298,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.012171164155006409,
+      "mean_token_accuracy": 0.9958188533782959,
+      "num_tokens": 77058299.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.6774423599243165,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.77237868309021,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.013017389178276061,
+      "mean_token_accuracy": 0.9956206619739533,
+      "num_tokens": 77247567.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.6773760914802551,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.6185512542724609,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.011151286959648132,
+      "mean_token_accuracy": 0.9961191415786743,
+      "num_tokens": 77436417.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.6750624418258667,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.7832470536231995,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.011553966999053955,
+      "mean_token_accuracy": 0.9958207130432128,
+      "num_tokens": 77625255.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.6814670443534852,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.7947144508361816,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.011924812197685241,
+      "mean_token_accuracy": 0.9959113657474518,
+      "num_tokens": 77813632.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.68285973072052,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.6403945684432983,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.01066286414861679,
+      "mean_token_accuracy": 0.9966644048690796,
+      "num_tokens": 78002213.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.6764974176883698,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.8078592419624329,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.010837660729885101,
+      "mean_token_accuracy": 0.9961750328540802,
+      "num_tokens": 78191471.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.6726884067058563,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.7293245792388916,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.01087343692779541,
+      "mean_token_accuracy": 0.996295166015625,
+      "num_tokens": 78380424.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.6714434802532196,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0266199111938477,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.011094169318675995,
+      "mean_token_accuracy": 0.9960172355175019,
+      "num_tokens": 78569653.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.6689312815666199,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.8064524531364441,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.01098542958498001,
+      "mean_token_accuracy": 0.9961700737476349,
+      "num_tokens": 78759287.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.6728951573371887,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7976726293563843,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.012345843017101288,
+      "mean_token_accuracy": 0.9957349181175232,
+      "num_tokens": 78948032.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.6702533006668091,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.7385782599449158,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.011018334329128266,
+      "mean_token_accuracy": 0.9964040815830231,
+      "num_tokens": 79136972.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.67191202044487,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.6246873736381531,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.011075176298618317,
+      "mean_token_accuracy": 0.9962317526340485,
+      "num_tokens": 79325966.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.6736057937145233,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9479376077651978,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012818500399589539,
+      "mean_token_accuracy": 0.9951797604560852,
+      "num_tokens": 79514853.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.6723325371742248,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.6468302607536316,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.010920494794845581,
+      "mean_token_accuracy": 0.9961069703102112,
+      "num_tokens": 79703739.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.6762719631195069,
+      "epoch": 5.693135935397039,
+      "grad_norm": 0.6862959265708923,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.011675214767456055,
+      "mean_token_accuracy": 0.9958149492740631,
+      "num_tokens": 79892582.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.6739612460136414,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.8402584195137024,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.010931231081485748,
+      "mean_token_accuracy": 0.9960719287395478,
+      "num_tokens": 80081362.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.6767356872558594,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.6180281043052673,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.012499480694532394,
+      "mean_token_accuracy": 0.9957239329814911,
+      "num_tokens": 80269388.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.6773952782154083,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.8782367706298828,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012943127751350402,
+      "mean_token_accuracy": 0.9958240866661072,
+      "num_tokens": 80458191.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.6774130165576935,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.7854585647583008,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.012309956550598144,
+      "mean_token_accuracy": 0.99593386054039,
+      "num_tokens": 80647306.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.6813723623752594,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.6340908408164978,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.01146906390786171,
+      "mean_token_accuracy": 0.9963057219982148,
+      "num_tokens": 80836147.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.6804767549037933,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.5984897613525391,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.010843665152788163,
+      "mean_token_accuracy": 0.996556407213211,
+      "num_tokens": 81025302.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.6759022295475006,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.6014022827148438,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.010135546326637268,
+      "mean_token_accuracy": 0.9966514885425568,
+      "num_tokens": 81214251.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.675349909067154,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.6571088433265686,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01084519699215889,
+      "mean_token_accuracy": 0.9962432861328125,
+      "num_tokens": 81402608.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.6718680560588837,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.9250450730323792,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.01139993518590927,
+      "mean_token_accuracy": 0.9961999177932739,
+      "num_tokens": 81591841.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.674647331237793,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.8669414520263672,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.010886493325233459,
+      "mean_token_accuracy": 0.9962068378925324,
+      "num_tokens": 81780873.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.6801326870918274,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7395965456962585,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.012087566405534744,
+      "mean_token_accuracy": 0.9957775354385376,
+      "num_tokens": 81969835.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.676666659116745,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.7648837566375732,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.011214715242385865,
+      "mean_token_accuracy": 0.9963518619537354,
+      "num_tokens": 82159146.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.676488333940506,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.6776206493377686,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.011839983612298965,
+      "mean_token_accuracy": 0.9959003031253815,
+      "num_tokens": 82347867.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.6743977665901184,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.7158207893371582,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011066053062677383,
+      "mean_token_accuracy": 0.9966431736946106,
+      "num_tokens": 82536890.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.6764561176300049,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.7066213488578796,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.012056462466716766,
+      "mean_token_accuracy": 0.9958225846290588,
+      "num_tokens": 82725921.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.6813391983509064,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.763606607913971,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.011903972923755645,
+      "mean_token_accuracy": 0.9958972215652466,
+      "num_tokens": 82914838.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.6814220428466797,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.8407371044158936,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.011677531152963638,
+      "mean_token_accuracy": 0.9959126651287079,
+      "num_tokens": 83103937.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.6833201944828033,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8418315052986145,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012164801359176636,
+      "mean_token_accuracy": 0.9956748068332673,
+      "num_tokens": 83293106.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.6802135646343231,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.8431894183158875,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012858036160469054,
+      "mean_token_accuracy": 0.995720499753952,
+      "num_tokens": 83481559.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.6758989214897155,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7379446029663086,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.011657674610614777,
+      "mean_token_accuracy": 0.9959479212760926,
+      "num_tokens": 83670620.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.6832396805286407,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9370406866073608,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012775366008281708,
+      "mean_token_accuracy": 0.9958724677562714,
+      "num_tokens": 83858979.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.6826908767223359,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.5771452188491821,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011871224641799927,
+      "mean_token_accuracy": 0.9962578654289246,
+      "num_tokens": 84047664.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.6935603318700365,
+      "eval_loss": 0.08950015902519226,
+      "eval_mean_token_accuracy": 0.9720694904874085,
+      "eval_num_tokens": 84198646.0,
+      "eval_runtime": 12.7741,
+      "eval_samples_per_second": 391.416,
+      "eval_steps_per_second": 12.29,
+      "step": 4458
+    },
+    {
+      "entropy": 0.6853953242301941,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.4715442359447479,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.011727289110422135,
+      "mean_token_accuracy": 0.9963658571243286,
+      "num_tokens": 84236476.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.6872529745101928,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.6521468162536621,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.007476924359798432,
+      "mean_token_accuracy": 0.9976866364479064,
+      "num_tokens": 84425424.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.6764522135257721,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.5134896039962769,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.006954360008239746,
+      "mean_token_accuracy": 0.9981172740459442,
+      "num_tokens": 84614973.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.674551123380661,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.5777756571769714,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.0064676709473133085,
+      "mean_token_accuracy": 0.9981890261173249,
+      "num_tokens": 84803584.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.6743139624595642,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.5243175625801086,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.005892159789800644,
+      "mean_token_accuracy": 0.99809131026268,
+      "num_tokens": 84992194.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.6691218018531799,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.6378570795059204,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.006101621687412262,
+      "mean_token_accuracy": 0.9981180131435394,
+      "num_tokens": 85181570.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.6653562963008881,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.7931498885154724,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.0059667728841304776,
+      "mean_token_accuracy": 0.9982839524745941,
+      "num_tokens": 85370354.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.6660257399082183,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.6156598329544067,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.006237205862998962,
+      "mean_token_accuracy": 0.9980066597461701,
+      "num_tokens": 85559287.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.6661248922348022,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.39466625452041626,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.006016004830598831,
+      "mean_token_accuracy": 0.9984841406345367,
+      "num_tokens": 85748489.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.6644636154174804,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.556770920753479,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.0060367342084646225,
+      "mean_token_accuracy": 0.9981182813644409,
+      "num_tokens": 85937336.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.6630209028720856,
+      "epoch": 6.137281292059219,
+      "grad_norm": 0.5270616412162781,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.005779195204377174,
+      "mean_token_accuracy": 0.9981250524520874,
+      "num_tokens": 86126286.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.6656446874141693,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 0.44525614380836487,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.005375263467431069,
+      "mean_token_accuracy": 0.9983476758003235,
+      "num_tokens": 86314904.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.664457255601883,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.6622279286384583,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.006194194033741951,
+      "mean_token_accuracy": 0.9981902241706848,
+      "num_tokens": 86503447.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.6608214318752289,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.6172921657562256,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.0061217501759529116,
+      "mean_token_accuracy": 0.9983710527420044,
+      "num_tokens": 86693005.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.6578842639923096,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.6241856217384338,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.006712782382965088,
+      "mean_token_accuracy": 0.9979998171329498,
+      "num_tokens": 86882287.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.6561006605625153,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 0.6784538626670837,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.006533285975456238,
+      "mean_token_accuracy": 0.9977931976318359,
+      "num_tokens": 87071532.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.6582947254180909,
+      "epoch": 6.218034993270525,
+      "grad_norm": 0.7572171688079834,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.006016571819782257,
+      "mean_token_accuracy": 0.9979285359382629,
+      "num_tokens": 87260158.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.6552021980285645,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.7922662496566772,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.006150579452514649,
+      "mean_token_accuracy": 0.9982126474380493,
+      "num_tokens": 87448718.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.6578958451747894,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.5214764475822449,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.005079228430986404,
+      "mean_token_accuracy": 0.9984815359115601,
+      "num_tokens": 87637461.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.6603941440582275,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.5404514670372009,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.005841487646102905,
+      "mean_token_accuracy": 0.9983745515346527,
+      "num_tokens": 87826292.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.6628166377544403,
+      "epoch": 6.271870794078062,
+      "grad_norm": 0.47433987259864807,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.006364698708057404,
+      "mean_token_accuracy": 0.9982446670532227,
+      "num_tokens": 88015224.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.6635581731796265,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.8039814233779907,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.006831821799278259,
+      "mean_token_accuracy": 0.9982130885124206,
+      "num_tokens": 88203965.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.6615313947200775,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.5904889106750488,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.006949013471603394,
+      "mean_token_accuracy": 0.9976236402988434,
+      "num_tokens": 88392910.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.6613587439060211,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.6321392059326172,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.006564225256443024,
+      "mean_token_accuracy": 0.9980846107006073,
+      "num_tokens": 88581883.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.6642421364784241,
+      "epoch": 6.325706594885599,
+      "grad_norm": 0.6209113597869873,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.007112482190132141,
+      "mean_token_accuracy": 0.9977501094341278,
+      "num_tokens": 88770574.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.6629104316234589,
+      "epoch": 6.339165545087483,
+      "grad_norm": 0.8903343081474304,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.00698748379945755,
+      "mean_token_accuracy": 0.9977229118347168,
+      "num_tokens": 88959487.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.6647967278957367,
+      "epoch": 6.352624495289367,
+      "grad_norm": 0.842950165271759,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.008400703966617584,
+      "mean_token_accuracy": 0.9975436449050903,
+      "num_tokens": 89148302.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.6641393005847931,
+      "epoch": 6.366083445491252,
+      "grad_norm": 0.6241896152496338,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.006545399129390716,
+      "mean_token_accuracy": 0.9982439935207367,
+      "num_tokens": 89337168.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.6646088480949401,
+      "epoch": 6.379542395693136,
+      "grad_norm": 0.6284488439559937,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.005783513188362122,
+      "mean_token_accuracy": 0.9982234418392182,
+      "num_tokens": 89525632.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.6648603081703186,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.7705378532409668,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.006973850727081299,
+      "mean_token_accuracy": 0.9977744221687317,
+      "num_tokens": 89714436.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.6667861938476562,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.7033233642578125,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.0062720529735088345,
+      "mean_token_accuracy": 0.9978694558143616,
+      "num_tokens": 89903591.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.66892409324646,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.6684027910232544,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.006154683232307434,
+      "mean_token_accuracy": 0.998307728767395,
+      "num_tokens": 90092231.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.6692153334617614,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.7860879302024841,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.006273554265499115,
+      "mean_token_accuracy": 0.9981006205081939,
+      "num_tokens": 90280721.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.6636229634284974,
+      "epoch": 6.446837146702557,
+      "grad_norm": 0.5911756753921509,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.0067244857549667355,
+      "mean_token_accuracy": 0.9979658007621766,
+      "num_tokens": 90469835.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.6663640201091766,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.5844401717185974,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.005917467921972275,
+      "mean_token_accuracy": 0.9981400012969971,
+      "num_tokens": 90658439.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.6635376453399658,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.554587721824646,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.005905192345380783,
+      "mean_token_accuracy": 0.9983574509620666,
+      "num_tokens": 90847689.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.660474544763565,
+      "epoch": 6.48721399730821,
+      "grad_norm": 0.6828640103340149,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.006179597228765488,
+      "mean_token_accuracy": 0.9978089511394501,
+      "num_tokens": 91036945.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.6635875403881073,
+      "epoch": 6.500672947510094,
+      "grad_norm": 0.621636688709259,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.006481277197599411,
+      "mean_token_accuracy": 0.9976639211177826,
+      "num_tokens": 91225796.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.6625118017196655,
+      "epoch": 6.514131897711978,
+      "grad_norm": 0.6648721098899841,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.006836448609828949,
+      "mean_token_accuracy": 0.9981350302696228,
+      "num_tokens": 91414454.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.6632138609886169,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.6122362017631531,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.0057271294295787815,
+      "mean_token_accuracy": 0.9983117520809174,
+      "num_tokens": 91603019.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.6621270835399627,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.707145094871521,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.005987466499209404,
+      "mean_token_accuracy": 0.9983782589435577,
+      "num_tokens": 91791729.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.6593974709510804,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.7802532315254211,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.00661279633641243,
+      "mean_token_accuracy": 0.9978727638721466,
+      "num_tokens": 91980977.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.658482962846756,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 0.8637571334838867,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.005754219368100166,
+      "mean_token_accuracy": 0.9982581317424775,
+      "num_tokens": 92169829.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.6561014533042908,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.7007631659507751,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.006207703426480293,
+      "mean_token_accuracy": 0.9980942070484161,
+      "num_tokens": 92358635.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.6558376729488373,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.6180236339569092,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.005951264500617981,
+      "mean_token_accuracy": 0.9980935871601104,
+      "num_tokens": 92546955.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.6548421680927277,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.5694975852966309,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.0058226577937603,
+      "mean_token_accuracy": 0.9980980336666108,
+      "num_tokens": 92735651.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.6563037514686585,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 0.7417933344841003,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.006121259927749634,
+      "mean_token_accuracy": 0.9980963408946991,
+      "num_tokens": 92925010.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.654166966676712,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.705633282661438,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.005783010274171829,
+      "mean_token_accuracy": 0.9983141779899597,
+      "num_tokens": 93113992.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.6567890524864197,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.5444097518920898,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.005712105333805085,
+      "mean_token_accuracy": 0.9982166588306427,
+      "num_tokens": 93302433.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.6536349058151245,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.4291290044784546,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.0060494735836982725,
+      "mean_token_accuracy": 0.9979224085807801,
+      "num_tokens": 93491429.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.6516136825084686,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.5134868621826172,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.005212320387363434,
+      "mean_token_accuracy": 0.9984170615673065,
+      "num_tokens": 93679973.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.6528811275959014,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.8154370188713074,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.00622846707701683,
+      "mean_token_accuracy": 0.998017430305481,
+      "num_tokens": 93868735.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.6532892882823944,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.5574272871017456,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.005193561688065529,
+      "mean_token_accuracy": 0.9985900521278381,
+      "num_tokens": 94057428.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.6552170813083649,
+      "epoch": 6.716016150740242,
+      "grad_norm": 0.6016380190849304,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.005771375447511673,
+      "mean_token_accuracy": 0.9981073677539826,
+      "num_tokens": 94246297.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.6572140097618103,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.7832751870155334,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.005496300011873245,
+      "mean_token_accuracy": 0.9983509957790375,
+      "num_tokens": 94434253.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.6541008234024048,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.4329380989074707,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.006188442185521126,
+      "mean_token_accuracy": 0.9981851994991302,
+      "num_tokens": 94623327.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.6556220591068268,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.5326682329177856,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.0052257638424634935,
+      "mean_token_accuracy": 0.9983932614326477,
+      "num_tokens": 94811837.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.6557375073432923,
+      "epoch": 6.769851951547779,
+      "grad_norm": 0.6864677667617798,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.005970630794763565,
+      "mean_token_accuracy": 0.9983245611190796,
+      "num_tokens": 95000726.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.6545388638973236,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.6318016648292542,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.006516661494970322,
+      "mean_token_accuracy": 0.9979697644710541,
+      "num_tokens": 95189395.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.6542861998081207,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.3979317247867584,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.005074064806103706,
+      "mean_token_accuracy": 0.998512452840805,
+      "num_tokens": 95378855.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.6542766332626343,
+      "epoch": 6.810228802153432,
+      "grad_norm": 0.4778045117855072,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.005774036794900894,
+      "mean_token_accuracy": 0.9982702076435089,
+      "num_tokens": 95568111.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.6550499975681305,
+      "epoch": 6.823687752355316,
+      "grad_norm": 0.4805110991001129,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.005620773509144783,
+      "mean_token_accuracy": 0.9983035147190094,
+      "num_tokens": 95756885.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.6543555080890655,
+      "epoch": 6.837146702557201,
+      "grad_norm": 0.6604891419410706,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.005247855931520462,
+      "mean_token_accuracy": 0.998341029882431,
+      "num_tokens": 95946078.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.6581745266914367,
+      "epoch": 6.850605652759085,
+      "grad_norm": 0.6589891910552979,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.005882954970002174,
+      "mean_token_accuracy": 0.9982152104377746,
+      "num_tokens": 96134597.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.6597592771053314,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.4152396023273468,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.0054433666169643406,
+      "mean_token_accuracy": 0.9983908593654632,
+      "num_tokens": 96323327.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.658534574508667,
+      "epoch": 6.877523553162853,
+      "grad_norm": 0.6025403141975403,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.005522595718502998,
+      "mean_token_accuracy": 0.9983524858951569,
+      "num_tokens": 96512170.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.6568575739860535,
+      "epoch": 6.890982503364738,
+      "grad_norm": 0.5730848908424377,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.004784010350704193,
+      "mean_token_accuracy": 0.9984236299991608,
+      "num_tokens": 96701363.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.6535935163497925,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.5301522612571716,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.004592524468898773,
+      "mean_token_accuracy": 0.9987652719020843,
+      "num_tokens": 96890642.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.6544224679470062,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.639918863773346,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.005577490478754043,
+      "mean_token_accuracy": 0.9984355092048645,
+      "num_tokens": 97079552.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.6542503774166107,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.5202826857566833,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.005623985454440117,
+      "mean_token_accuracy": 0.9982783615589141,
+      "num_tokens": 97268346.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.6499157190322876,
+      "epoch": 6.944818304172275,
+      "grad_norm": 0.5668892860412598,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.005104385688900948,
+      "mean_token_accuracy": 0.9985612750053405,
+      "num_tokens": 97457514.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.6520473599433899,
+      "epoch": 6.958277254374159,
+      "grad_norm": 0.5878952145576477,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.005280185490846634,
+      "mean_token_accuracy": 0.9985166370868683,
+      "num_tokens": 97646014.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.6488265216350555,
+      "epoch": 6.971736204576043,
+      "grad_norm": 0.4630887806415558,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.005359554663300514,
+      "mean_token_accuracy": 0.998493617773056,
+      "num_tokens": 97835401.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.6526197671890259,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.38062795996665955,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.0054448477923870085,
+      "mean_token_accuracy": 0.9982813715934753,
+      "num_tokens": 98024194.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.654500275850296,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.5739313364028931,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.005691033974289894,
+      "mean_token_accuracy": 0.9982043981552124,
+      "num_tokens": 98212962.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.6537400142402406,
+      "eval_loss": 0.09857651591300964,
+      "eval_mean_token_accuracy": 0.9730726879113799,
+      "eval_num_tokens": 98231758.0,
+      "eval_runtime": 12.7391,
+      "eval_samples_per_second": 392.491,
+      "eval_steps_per_second": 12.324,
+      "step": 5201
+    },
+    {
+      "entropy": 0.6512321710586548,
+      "epoch": 7.012113055181696,
+      "grad_norm": 0.2961742877960205,
+      "learning_rate": 2.2506648060191042e-05,
+      "loss": 0.0032418236136436463,
+      "mean_token_accuracy": 0.999194347858429,
+      "num_tokens": 98401402.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.6491944074630738,
+      "epoch": 7.02557200538358,
+      "grad_norm": 0.35240232944488525,
+      "learning_rate": 2.2321031102941066e-05,
+      "loss": 0.0028009919449687006,
+      "mean_token_accuracy": 0.9995193898677825,
+      "num_tokens": 98590150.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.6500660896301269,
+      "epoch": 7.039030955585464,
+      "grad_norm": 0.3501753509044647,
+      "learning_rate": 2.2135962530768055e-05,
+      "loss": 0.0029570702463388443,
+      "mean_token_accuracy": 0.9993301749229431,
+      "num_tokens": 98778612.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.6460368275642395,
+      "epoch": 7.052489905787349,
+      "grad_norm": 0.41639405488967896,
+      "learning_rate": 2.1951446010313053e-05,
+      "loss": 0.003057733364403248,
+      "mean_token_accuracy": 0.9994841694831849,
+      "num_tokens": 98967879.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.6441825687885284,
+      "epoch": 7.065948855989233,
+      "grad_norm": 0.3744174540042877,
+      "learning_rate": 2.1767485197279662e-05,
+      "loss": 0.0028053496032953264,
+      "mean_token_accuracy": 0.9994258165359498,
+      "num_tokens": 99157371.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.6460713386535645,
+      "epoch": 7.079407806191117,
+      "grad_norm": 0.22786536812782288,
+      "learning_rate": 2.1584083736361617e-05,
+      "loss": 0.004050987213850022,
+      "mean_token_accuracy": 0.9992221534252167,
+      "num_tokens": 99346074.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.6439846336841584,
+      "epoch": 7.0928667563930015,
+      "grad_norm": 0.33427926898002625,
+      "learning_rate": 2.1401245261170582e-05,
+      "loss": 0.002635418064892292,
+      "mean_token_accuracy": 0.9993825376033783,
+      "num_tokens": 99535282.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.6429140031337738,
+      "epoch": 7.106325706594886,
+      "grad_norm": 0.35418426990509033,
+      "learning_rate": 2.121897339416416e-05,
+      "loss": 0.002664094232022762,
+      "mean_token_accuracy": 0.999459445476532,
+      "num_tokens": 99724199.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.6431400418281555,
+      "epoch": 7.11978465679677,
+      "grad_norm": 0.2773686945438385,
+      "learning_rate": 2.103727174657412e-05,
+      "loss": 0.0029107578098773955,
+      "mean_token_accuracy": 0.999220222234726,
+      "num_tokens": 99913190.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.644054365158081,
+      "epoch": 7.133243606998654,
+      "grad_norm": 0.466249942779541,
+      "learning_rate": 2.085614391833486e-05,
+      "loss": 0.0026346595957875254,
+      "mean_token_accuracy": 0.9993867635726928,
+      "num_tokens": 100102083.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.6469399213790894,
+      "epoch": 7.1467025572005385,
+      "grad_norm": 0.37575817108154297,
+      "learning_rate": 2.0675593498012052e-05,
+      "loss": 0.0028098784387111664,
+      "mean_token_accuracy": 0.9992999136447906,
+      "num_tokens": 100290764.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.6454071402549744,
+      "epoch": 7.160161507402423,
+      "grad_norm": 0.5268272161483765,
+      "learning_rate": 2.049562406273159e-05,
+      "loss": 0.0025207167491316795,
+      "mean_token_accuracy": 0.9995582520961761,
+      "num_tokens": 100479571.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.6419129431247711,
+      "epoch": 7.173620457604307,
+      "grad_norm": 0.28378117084503174,
+      "learning_rate": 2.0316239178108688e-05,
+      "loss": 0.00287601538002491,
+      "mean_token_accuracy": 0.9993176221847534,
+      "num_tokens": 100668516.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.6420379161834717,
+      "epoch": 7.187079407806191,
+      "grad_norm": 0.24231773614883423,
+      "learning_rate": 2.013744239817725e-05,
+      "loss": 0.002684992924332619,
+      "mean_token_accuracy": 0.9993604898452759,
+      "num_tokens": 100857247.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.6403457880020141,
+      "epoch": 7.2005383580080755,
+      "grad_norm": 0.350322961807251,
+      "learning_rate": 1.9959237265319452e-05,
+      "loss": 0.002843806892633438,
+      "mean_token_accuracy": 0.9992822408676147,
+      "num_tokens": 101046251.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.6437962651252747,
+      "epoch": 7.21399730820996,
+      "grad_norm": 0.504984974861145,
+      "learning_rate": 1.9781627310195528e-05,
+      "loss": 0.002904067933559418,
+      "mean_token_accuracy": 0.9991836249828339,
+      "num_tokens": 101235009.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.6444802343845367,
+      "epoch": 7.227456258411844,
+      "grad_norm": 0.5015905499458313,
+      "learning_rate": 1.960461605167389e-05,
+      "loss": 0.0031951837241649627,
+      "mean_token_accuracy": 0.9991663873195649,
+      "num_tokens": 101423795.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.6445916533470154,
+      "epoch": 7.240915208613728,
+      "grad_norm": 0.8163104057312012,
+      "learning_rate": 1.9428206996761357e-05,
+      "loss": 0.0031633909791707994,
+      "mean_token_accuracy": 0.9990993022918702,
+      "num_tokens": 101612550.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.6424523174762726,
+      "epoch": 7.2543741588156125,
+      "grad_norm": 0.7167068123817444,
+      "learning_rate": 1.9252403640533678e-05,
+      "loss": 0.0027146343141794204,
+      "mean_token_accuracy": 0.9993792653083802,
+      "num_tokens": 101801308.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.6425661742687225,
+      "epoch": 7.267833109017497,
+      "grad_norm": 0.3235819935798645,
+      "learning_rate": 1.9077209466066315e-05,
+      "loss": 0.002999866008758545,
+      "mean_token_accuracy": 0.9993207931518555,
+      "num_tokens": 101990167.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.6397831857204437,
+      "epoch": 7.281292059219381,
+      "grad_norm": 0.2559046149253845,
+      "learning_rate": 1.8902627944365396e-05,
+      "loss": 0.0023009315133094787,
+      "mean_token_accuracy": 0.9995379090309143,
+      "num_tokens": 102179185.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.638386732339859,
+      "epoch": 7.294751009421265,
+      "grad_norm": 0.4241917133331299,
+      "learning_rate": 1.872866253429897e-05,
+      "loss": 0.003075192496180534,
+      "mean_token_accuracy": 0.9992788910865784,
+      "num_tokens": 102368161.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.6398141801357269,
+      "epoch": 7.30820995962315,
+      "grad_norm": 0.4451645612716675,
+      "learning_rate": 1.8555316682528485e-05,
+      "loss": 0.0027588184922933577,
+      "mean_token_accuracy": 0.9993425786495209,
+      "num_tokens": 102557229.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.6400523662567139,
+      "epoch": 7.321668909825034,
+      "grad_norm": 0.2873964309692383,
+      "learning_rate": 1.8382593823440474e-05,
+      "loss": 0.0026803961023688315,
+      "mean_token_accuracy": 0.9994625568389892,
+      "num_tokens": 102745928.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.6401661276817322,
+      "epoch": 7.335127860026918,
+      "grad_norm": 0.5918443202972412,
+      "learning_rate": 1.821049737907854e-05,
+      "loss": 0.0029163889586925505,
+      "mean_token_accuracy": 0.9992423593997956,
+      "num_tokens": 102934875.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.6410463333129883,
+      "epoch": 7.348586810228802,
+      "grad_norm": 0.39631369709968567,
+      "learning_rate": 1.803903075907554e-05,
+      "loss": 0.002800934761762619,
+      "mean_token_accuracy": 0.9993205904960633,
+      "num_tokens": 103123665.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.6384492158889771,
+      "epoch": 7.362045760430687,
+      "grad_norm": 0.44129809737205505,
+      "learning_rate": 1.7868197360586054e-05,
+      "loss": 0.0027701158076524736,
+      "mean_token_accuracy": 0.9992826044559479,
+      "num_tokens": 103312769.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.6399481356143951,
+      "epoch": 7.375504710632571,
+      "grad_norm": 0.3396782875061035,
+      "learning_rate": 1.769800056821904e-05,
+      "loss": 0.002843455225229263,
+      "mean_token_accuracy": 0.9993624866008759,
+      "num_tokens": 103501428.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.6391602873802185,
+      "epoch": 7.388963660834455,
+      "grad_norm": 0.508573055267334,
+      "learning_rate": 1.7528443753970814e-05,
+      "loss": 0.0030046245083212852,
+      "mean_token_accuracy": 0.9992032408714294,
+      "num_tokens": 103690316.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.641320925951004,
+      "epoch": 7.402422611036339,
+      "grad_norm": 0.372712641954422,
+      "learning_rate": 1.7359530277158227e-05,
+      "loss": 0.0029981501400470735,
+      "mean_token_accuracy": 0.9992169618606568,
+      "num_tokens": 103879262.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.6416053056716919,
+      "epoch": 7.415881561238224,
+      "grad_norm": 0.47443583607673645,
+      "learning_rate": 1.7191263484352106e-05,
+      "loss": 0.00280308797955513,
+      "mean_token_accuracy": 0.9993564009666442,
+      "num_tokens": 104067769.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.6400010466575623,
+      "epoch": 7.429340511440108,
+      "grad_norm": 0.26680561900138855,
+      "learning_rate": 1.702364670931097e-05,
+      "loss": 0.0029377201572060585,
+      "mean_token_accuracy": 0.9992177426815033,
+      "num_tokens": 104256396.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.6378615140914917,
+      "epoch": 7.442799461641992,
+      "grad_norm": 0.3017747700214386,
+      "learning_rate": 1.685668327291495e-05,
+      "loss": 0.0025883980095386507,
+      "mean_token_accuracy": 0.9993183672428131,
+      "num_tokens": 104445275.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.6393269956111908,
+      "epoch": 7.456258411843876,
+      "grad_norm": 0.33738431334495544,
+      "learning_rate": 1.6690376483100034e-05,
+      "loss": 0.0023764621466398237,
+      "mean_token_accuracy": 0.999459034204483,
+      "num_tokens": 104633705.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.6352120459079742,
+      "epoch": 7.469717362045761,
+      "grad_norm": 0.30927929282188416,
+      "learning_rate": 1.6524729634792475e-05,
+      "loss": 0.002412043884396553,
+      "mean_token_accuracy": 0.9993467569351197,
+      "num_tokens": 104822986.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.6355067729949951,
+      "epoch": 7.483176312247645,
+      "grad_norm": 0.3294868469238281,
+      "learning_rate": 1.6359746009843567e-05,
+      "loss": 0.002515685372054577,
+      "mean_token_accuracy": 0.999462366104126,
+      "num_tokens": 105012029.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.6385121285915375,
+      "epoch": 7.496635262449529,
+      "grad_norm": 0.4267696738243103,
+      "learning_rate": 1.6195428876964592e-05,
+      "loss": 0.0027210000902414323,
+      "mean_token_accuracy": 0.9993004381656647,
+      "num_tokens": 105200942.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.641570907831192,
+      "epoch": 7.510094212651413,
+      "grad_norm": 0.28043922781944275,
+      "learning_rate": 1.6031781491662056e-05,
+      "loss": 0.0025119861587882043,
+      "mean_token_accuracy": 0.9993769824504852,
+      "num_tokens": 105389627.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.6401131987571717,
+      "epoch": 7.523553162853298,
+      "grad_norm": 0.5159914493560791,
+      "learning_rate": 1.58688070961732e-05,
+      "loss": 0.0026847902685403825,
+      "mean_token_accuracy": 0.9993621706962585,
+      "num_tokens": 105578832.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.6401657581329345,
+      "epoch": 7.537012113055182,
+      "grad_norm": 0.47359421849250793,
+      "learning_rate": 1.5706508919401804e-05,
+      "loss": 0.0025615351274609565,
+      "mean_token_accuracy": 0.999418193101883,
+      "num_tokens": 105767407.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.640064799785614,
+      "epoch": 7.550471063257066,
+      "grad_norm": 0.2965790927410126,
+      "learning_rate": 1.554489017685412e-05,
+      "loss": 0.002729836292564869,
+      "mean_token_accuracy": 0.999356073141098,
+      "num_tokens": 105955938.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.6364653050899506,
+      "epoch": 7.56393001345895,
+      "grad_norm": 0.3772422969341278,
+      "learning_rate": 1.538395407057525e-05,
+      "loss": 0.002501054294407368,
+      "mean_token_accuracy": 0.99944589138031,
+      "num_tokens": 106145287.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.6365915596485138,
+      "epoch": 7.577388963660835,
+      "grad_norm": 0.34281232953071594,
+      "learning_rate": 1.522370378908567e-05,
+      "loss": 0.0024556241929531097,
+      "mean_token_accuracy": 0.9995415151119232,
+      "num_tokens": 106334420.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.6392146408557892,
+      "epoch": 7.590847913862719,
+      "grad_norm": 0.289888858795166,
+      "learning_rate": 1.5064142507318051e-05,
+      "loss": 0.0024054452776908876,
+      "mean_token_accuracy": 0.999481338262558,
+      "num_tokens": 106523346.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.63958261013031,
+      "epoch": 7.604306864064603,
+      "grad_norm": 0.4470290541648865,
+      "learning_rate": 1.4905273386554386e-05,
+      "loss": 0.002950721047818661,
+      "mean_token_accuracy": 0.9994794666767121,
+      "num_tokens": 106711623.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.6364846646785736,
+      "epoch": 7.6177658142664875,
+      "grad_norm": 0.3294987678527832,
+      "learning_rate": 1.4747099574363332e-05,
+      "loss": 0.0029568497091531755,
+      "mean_token_accuracy": 0.99925497174263,
+      "num_tokens": 106900044.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.6371748626232148,
+      "epoch": 7.631224764468372,
+      "grad_norm": 0.3742607533931732,
+      "learning_rate": 1.4589624204537866e-05,
+      "loss": 0.002776237577199936,
+      "mean_token_accuracy": 0.9994770407676696,
+      "num_tokens": 107088499.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.6374151885509491,
+      "epoch": 7.644683714670256,
+      "grad_norm": 0.34489092230796814,
+      "learning_rate": 1.4432850397033176e-05,
+      "loss": 0.0025846634060144426,
+      "mean_token_accuracy": 0.9992987811565399,
+      "num_tokens": 107277120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.6350409626960755,
+      "epoch": 7.65814266487214,
+      "grad_norm": 0.4673531651496887,
+      "learning_rate": 1.427678125790487e-05,
+      "loss": 0.002661401592195034,
+      "mean_token_accuracy": 0.9993412852287292,
+      "num_tokens": 107466227.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.6376935839653015,
+      "epoch": 7.6716016150740245,
+      "grad_norm": 0.2515313923358917,
+      "learning_rate": 1.4121419879247422e-05,
+      "loss": 0.0023796845227479933,
+      "mean_token_accuracy": 0.999422425031662,
+      "num_tokens": 107654984.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.6349337220191955,
+      "epoch": 7.685060565275909,
+      "grad_norm": 0.38676151633262634,
+      "learning_rate": 1.3966769339132929e-05,
+      "loss": 0.002421387284994125,
+      "mean_token_accuracy": 0.999384069442749,
+      "num_tokens": 107844126.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.6366126656532287,
+      "epoch": 7.698519515477793,
+      "grad_norm": 0.39246416091918945,
+      "learning_rate": 1.3812832701550099e-05,
+      "loss": 0.0024261830374598505,
+      "mean_token_accuracy": 0.9993186116218566,
+      "num_tokens": 108032874.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.6357373297214508,
+      "epoch": 7.711978465679677,
+      "grad_norm": 0.319026917219162,
+      "learning_rate": 1.365961301634357e-05,
+      "loss": 0.0023328609764575957,
+      "mean_token_accuracy": 0.9995430946350098,
+      "num_tokens": 108222192.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.6360769748687745,
+      "epoch": 7.7254374158815615,
+      "grad_norm": 0.38521459698677063,
+      "learning_rate": 1.350711331915348e-05,
+      "loss": 0.0026227174326777457,
+      "mean_token_accuracy": 0.9992436349391938,
+      "num_tokens": 108410750.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.6360529124736786,
+      "epoch": 7.738896366083446,
+      "grad_norm": 0.2718389928340912,
+      "learning_rate": 1.3355336631355314e-05,
+      "loss": 0.0024306148290634154,
+      "mean_token_accuracy": 0.9995024859905243,
+      "num_tokens": 108599812.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.6351405322551728,
+      "epoch": 7.75235531628533,
+      "grad_norm": 0.38458114862442017,
+      "learning_rate": 1.3204285960000051e-05,
+      "loss": 0.002783242240548134,
+      "mean_token_accuracy": 0.9993249535560608,
+      "num_tokens": 108788623.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.6342482805252075,
+      "epoch": 7.765814266487214,
+      "grad_norm": 0.37706634402275085,
+      "learning_rate": 1.30539642977546e-05,
+      "loss": 0.002872874028980732,
+      "mean_token_accuracy": 0.9993605136871337,
+      "num_tokens": 108977505.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.6347688436508179,
+      "epoch": 7.7792732166890985,
+      "grad_norm": 0.44668373465538025,
+      "learning_rate": 1.2904374622842486e-05,
+      "loss": 0.002442394569516182,
+      "mean_token_accuracy": 0.999343854188919,
+      "num_tokens": 109166589.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.6364605367183686,
+      "epoch": 7.792732166890983,
+      "grad_norm": 0.2349705696105957,
+      "learning_rate": 1.2755519898984846e-05,
+      "loss": 0.00250922292470932,
+      "mean_token_accuracy": 0.9995181024074554,
+      "num_tokens": 109355292.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.6371697664260865,
+      "epoch": 7.806191117092867,
+      "grad_norm": 0.3123619556427002,
+      "learning_rate": 1.2607403075341728e-05,
+      "loss": 0.002688922733068466,
+      "mean_token_accuracy": 0.9993325173854828,
+      "num_tokens": 109543865.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.6380978763103485,
+      "epoch": 7.819650067294751,
+      "grad_norm": 0.2686324417591095,
+      "learning_rate": 1.2460027086453658e-05,
+      "loss": 0.0023211821913719176,
+      "mean_token_accuracy": 0.9995401680469513,
+      "num_tokens": 109732486.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.6382215797901154,
+      "epoch": 7.833109017496636,
+      "grad_norm": 0.4553178548812866,
+      "learning_rate": 1.2313394852183473e-05,
+      "loss": 0.0024582967162132263,
+      "mean_token_accuracy": 0.9994239926338195,
+      "num_tokens": 109921491.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.6385389983654022,
+      "epoch": 7.84656796769852,
+      "grad_norm": 0.2880251109600067,
+      "learning_rate": 1.2167509277658518e-05,
+      "loss": 0.0024548277258872988,
+      "mean_token_accuracy": 0.9994565665721893,
+      "num_tokens": 110110256.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.633701753616333,
+      "epoch": 7.860026917900404,
+      "grad_norm": 0.4071319103240967,
+      "learning_rate": 1.2022373253213049e-05,
+      "loss": 0.0025199253112077715,
+      "mean_token_accuracy": 0.9993486106395721,
+      "num_tokens": 110299624.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.6337139904499054,
+      "epoch": 7.873485868102288,
+      "grad_norm": 0.4106242060661316,
+      "learning_rate": 1.1877989654330973e-05,
+      "loss": 0.0022239934653043745,
+      "mean_token_accuracy": 0.999601423740387,
+      "num_tokens": 110488951.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.6346798777580261,
+      "epoch": 7.886944818304173,
+      "grad_norm": 0.5445306301116943,
+      "learning_rate": 1.1734361341588912e-05,
+      "loss": 0.0027007974684238436,
+      "mean_token_accuracy": 0.9993659555912018,
+      "num_tokens": 110678171.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.6366956651210784,
+      "epoch": 7.900403768506057,
+      "grad_norm": 0.2735433280467987,
+      "learning_rate": 1.1591491160599488e-05,
+      "loss": 0.0026022626087069513,
+      "mean_token_accuracy": 0.9993394792079926,
+      "num_tokens": 110867003.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.6380104124546051,
+      "epoch": 7.913862718707941,
+      "grad_norm": 0.5340652465820312,
+      "learning_rate": 1.1449381941954956e-05,
+      "loss": 0.002648356184363365,
+      "mean_token_accuracy": 0.9994826078414917,
+      "num_tokens": 111055985.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.6362045764923095,
+      "epoch": 7.927321668909825,
+      "grad_norm": 0.3767072856426239,
+      "learning_rate": 1.130803650117116e-05,
+      "loss": 0.0024559229612350466,
+      "mean_token_accuracy": 0.9995233714580536,
+      "num_tokens": 111244949.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.6355053067207337,
+      "epoch": 7.94078061911171,
+      "grad_norm": 0.3835017681121826,
+      "learning_rate": 1.1167457638631656e-05,
+      "loss": 0.002397604286670685,
+      "mean_token_accuracy": 0.9995415270328522,
+      "num_tokens": 111433946.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.6365890502929688,
+      "epoch": 7.954239569313594,
+      "grad_norm": 0.3102132976055145,
+      "learning_rate": 1.1027648139532355e-05,
+      "loss": 0.002390575036406517,
+      "mean_token_accuracy": 0.9994417130947113,
+      "num_tokens": 111622676.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.6346444308757782,
+      "epoch": 7.967698519515478,
+      "grad_norm": 0.49554574489593506,
+      "learning_rate": 1.0888610773826253e-05,
+      "loss": 0.0021300587803125383,
+      "mean_token_accuracy": 0.9996205389499664,
+      "num_tokens": 111811705.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.6345613896846771,
+      "epoch": 7.981157469717362,
+      "grad_norm": 0.2676335871219635,
+      "learning_rate": 1.0750348296168583e-05,
+      "loss": 0.0024733129888772965,
+      "mean_token_accuracy": 0.9994426965713501,
+      "num_tokens": 112000972.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.6350580513477325,
+      "epoch": 7.994616419919247,
+      "grad_norm": 0.20178340375423431,
+      "learning_rate": 1.0612863445862236e-05,
+      "loss": 0.002096991054713726,
+      "mean_token_accuracy": 0.9996170699596405,
+      "num_tokens": 112189178.0,
+      "step": 5940
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.6389722296386767,
+      "eval_loss": 0.10558179020881653,
+      "eval_mean_token_accuracy": 0.9726916976795075,
+      "eval_num_tokens": 112264896.0,
+      "eval_runtime": 12.7264,
+      "eval_samples_per_second": 392.885,
+      "eval_steps_per_second": 12.337,
+      "step": 5944
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.189139942055346e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-5944/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-5944/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-6687/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-6687/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-6687/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2bfa96535ff2e770c3c2c5bf0e562b4911528e37
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8bad00e5d77e18aacde8baf843c2e1272d8dda9c458ab51684dd76e6e6433a9
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-6687/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-6687/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-6687/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-6687/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bddd60c1220d890190dd136bfa8fde48caa3350
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/trainer_state.json
@@ -0,0 +1,6813 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.0,
+  "eval_steps": 500,
+  "global_step": 6687,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    },
+    {
+      "entropy": 0.8332668900489807,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9914855360984802,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.06607928276062011,
+      "mean_token_accuracy": 0.974932211637497,
+      "num_tokens": 14165567.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8149146616458893,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.3080861568450928,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.057932257652282715,
+      "mean_token_accuracy": 0.9770712852478027,
+      "num_tokens": 14353912.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8197620987892151,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.0898637771606445,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.06010057330131531,
+      "mean_token_accuracy": 0.9758293390274048,
+      "num_tokens": 14542288.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8183245182037353,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.3120660781860352,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.0610861599445343,
+      "mean_token_accuracy": 0.9764986515045166,
+      "num_tokens": 14731028.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8135185301303863,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.138473391532898,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.06268800497055053,
+      "mean_token_accuracy": 0.9752912580966949,
+      "num_tokens": 14919682.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8142645537853241,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.1258258819580078,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.06271481513977051,
+      "mean_token_accuracy": 0.9757774949073792,
+      "num_tokens": 15107904.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8170483231544494,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.007297396659851,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.06338523030281067,
+      "mean_token_accuracy": 0.9754154622554779,
+      "num_tokens": 15297115.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8201254069805145,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 0.8240306973457336,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.06355815529823303,
+      "mean_token_accuracy": 0.9752808332443237,
+      "num_tokens": 15485921.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8140238881111145,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.0350368022918701,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.05969444513320923,
+      "mean_token_accuracy": 0.977089112997055,
+      "num_tokens": 15674824.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8133325934410095,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.0544878244400024,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.06062774658203125,
+      "mean_token_accuracy": 0.976290100812912,
+      "num_tokens": 15863526.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8185161769390106,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.1076316833496094,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.061567097902297974,
+      "mean_token_accuracy": 0.9755526125431061,
+      "num_tokens": 16051685.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8176139056682586,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.0057616233825684,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.06227490901947021,
+      "mean_token_accuracy": 0.9758114695549012,
+      "num_tokens": 16240451.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8217637002468109,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.97276771068573,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.06410331726074218,
+      "mean_token_accuracy": 0.9754030883312226,
+      "num_tokens": 16428961.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8141795754432678,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.0135555267333984,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.06253207921981811,
+      "mean_token_accuracy": 0.9760490536689759,
+      "num_tokens": 16617327.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8167653739452362,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2272467613220215,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.060455435514450075,
+      "mean_token_accuracy": 0.9758628964424133,
+      "num_tokens": 16806414.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8155323445796967,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 0.9328635334968567,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.06302794814109802,
+      "mean_token_accuracy": 0.9745101630687714,
+      "num_tokens": 16995016.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.818474805355072,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.1651655435562134,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.06043210029602051,
+      "mean_token_accuracy": 0.9766487181186676,
+      "num_tokens": 17184250.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8140560865402222,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.1005650758743286,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.06131819486618042,
+      "mean_token_accuracy": 0.9757712185382843,
+      "num_tokens": 17373075.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8204843640327454,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.0825532674789429,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.06296311616897583,
+      "mean_token_accuracy": 0.9745951414108276,
+      "num_tokens": 17561454.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8237834334373474,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.9842743873596191,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.061483919620513916,
+      "mean_token_accuracy": 0.9754831731319428,
+      "num_tokens": 17750758.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8165915131568908,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.0179396867752075,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.06237881183624268,
+      "mean_token_accuracy": 0.9750245690345765,
+      "num_tokens": 17940126.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8329708755016327,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0732359886169434,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.0644665539264679,
+      "mean_token_accuracy": 0.9743888020515442,
+      "num_tokens": 18129000.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8210074126720428,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.3696036338806152,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.06320792436599731,
+      "mean_token_accuracy": 0.9751390159130097,
+      "num_tokens": 18318232.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8248398661613464,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.327525019645691,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.06277769804000854,
+      "mean_token_accuracy": 0.9752413153648376,
+      "num_tokens": 18507403.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8284746050834656,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.044022560119629,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.06162315607070923,
+      "mean_token_accuracy": 0.9760516047477722,
+      "num_tokens": 18696242.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8233858644962311,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.062048077583313,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.059620130062103274,
+      "mean_token_accuracy": 0.9755307257175445,
+      "num_tokens": 18885199.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8236705601215363,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.2024412155151367,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.06304935216903687,
+      "mean_token_accuracy": 0.975091153383255,
+      "num_tokens": 19074028.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.828409218788147,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.1684212684631348,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.06451238393783569,
+      "mean_token_accuracy": 0.9743404746055603,
+      "num_tokens": 19262295.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8214354634284973,
+      "epoch": 1.386271870794078,
+      "grad_norm": 0.8512521386146545,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.06042641997337341,
+      "mean_token_accuracy": 0.9755558848381043,
+      "num_tokens": 19451154.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8223241448402405,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 0.8546568155288696,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.062134528160095216,
+      "mean_token_accuracy": 0.9750089883804322,
+      "num_tokens": 19639827.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.8217969954013824,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 0.972831666469574,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.060803139209747316,
+      "mean_token_accuracy": 0.9760692238807678,
+      "num_tokens": 19828174.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.8144341170787811,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1216644048690796,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.06458876132965088,
+      "mean_token_accuracy": 0.9738282442092896,
+      "num_tokens": 20016700.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.8168790876865387,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.0821806192398071,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.06258658170700074,
+      "mean_token_accuracy": 0.9749014675617218,
+      "num_tokens": 20205823.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8146390020847321,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.0067790746688843,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.06509006023406982,
+      "mean_token_accuracy": 0.9737810134887696,
+      "num_tokens": 20394611.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8224801063537598,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.0583593845367432,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.06343590021133423,
+      "mean_token_accuracy": 0.9750119090080261,
+      "num_tokens": 20583252.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.8083258986473083,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.8305535316467285,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.06573111414909363,
+      "mean_token_accuracy": 0.9736428141593934,
+      "num_tokens": 20772107.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8081986546516419,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5895942449569702,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.06376024484634399,
+      "mean_token_accuracy": 0.9751200795173645,
+      "num_tokens": 20960494.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.8026683986186981,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 0.9962654113769531,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.06282572746276856,
+      "mean_token_accuracy": 0.9753849327564239,
+      "num_tokens": 21149763.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8030014991760254,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9083836078643799,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.06473478078842163,
+      "mean_token_accuracy": 0.9747069597244262,
+      "num_tokens": 21338690.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.8008662164211273,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.097593069076538,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.05911685228347778,
+      "mean_token_accuracy": 0.9767139732837677,
+      "num_tokens": 21527420.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7981986284255982,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.1673516035079956,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.05957244038581848,
+      "mean_token_accuracy": 0.9761911332607269,
+      "num_tokens": 21716791.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.8004718601703644,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 0.7955246567726135,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.06378913521766663,
+      "mean_token_accuracy": 0.9747395157814026,
+      "num_tokens": 21905487.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.8039821267127991,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.3141396045684814,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.06209801435470581,
+      "mean_token_accuracy": 0.9754719436168671,
+      "num_tokens": 22094436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.8051789462566376,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.2227814197540283,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.06261310577392579,
+      "mean_token_accuracy": 0.9753049433231353,
+      "num_tokens": 22283243.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7941994190216064,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.071070671081543,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.05962263345718384,
+      "mean_token_accuracy": 0.9763956189155578,
+      "num_tokens": 22472084.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7937953472137451,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.0329782962799072,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.060611891746521,
+      "mean_token_accuracy": 0.9758659780025483,
+      "num_tokens": 22661614.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7913113713264466,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.032772183418274,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.06002308130264282,
+      "mean_token_accuracy": 0.9758916199207306,
+      "num_tokens": 22850634.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.7909658491611481,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3234598636627197,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.06296972036361695,
+      "mean_token_accuracy": 0.9741117715835571,
+      "num_tokens": 23039402.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7880761086940765,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.9563435316085815,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.06209538578987121,
+      "mean_token_accuracy": 0.9750530481338501,
+      "num_tokens": 23228363.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7845663070678711,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 0.8875390291213989,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.06272958517074585,
+      "mean_token_accuracy": 0.9751476228237153,
+      "num_tokens": 23417469.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7928634822368622,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.0629489421844482,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.060286092758178714,
+      "mean_token_accuracy": 0.976036274433136,
+      "num_tokens": 23606274.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7870154738426208,
+      "epoch": 1.695827725437416,
+      "grad_norm": 0.9164250493049622,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.06263296008110046,
+      "mean_token_accuracy": 0.9752088189125061,
+      "num_tokens": 23795255.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.7948406875133515,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.0909371376037598,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.06165047883987427,
+      "mean_token_accuracy": 0.9764420628547669,
+      "num_tokens": 23984534.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7862435758113862,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.1839547157287598,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.0640158474445343,
+      "mean_token_accuracy": 0.9741048216819763,
+      "num_tokens": 24173632.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.7887867331504822,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.0730741024017334,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.06260218620300292,
+      "mean_token_accuracy": 0.974741768836975,
+      "num_tokens": 24362671.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7896574556827545,
+      "epoch": 1.749663526244953,
+      "grad_norm": 0.9559605121612549,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.061297619342803956,
+      "mean_token_accuracy": 0.9752223670482636,
+      "num_tokens": 24551825.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.7902800619602204,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.914057195186615,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.06320110559463502,
+      "mean_token_accuracy": 0.9749482691287994,
+      "num_tokens": 24740867.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.799742591381073,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.06534289717674255,
+      "mean_token_accuracy": 0.9741291284561158,
+      "num_tokens": 24930132.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7974063038825989,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1188615560531616,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.060908573865890506,
+      "mean_token_accuracy": 0.9765798091888428,
+      "num_tokens": 25119678.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.786439061164856,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9185847640037537,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.06158917546272278,
+      "mean_token_accuracy": 0.9760335922241211,
+      "num_tokens": 25308876.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7888528347015381,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.1948838233947754,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.060635948181152345,
+      "mean_token_accuracy": 0.976032954454422,
+      "num_tokens": 25498102.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7968044698238372,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.4128663539886475,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.06448977589607238,
+      "mean_token_accuracy": 0.9741183638572692,
+      "num_tokens": 25686747.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7967679858207702,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.9797796010971069,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.06342286467552186,
+      "mean_token_accuracy": 0.9752787709236145,
+      "num_tokens": 25875368.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7932188332080841,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.0549253225326538,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.060981249809265135,
+      "mean_token_accuracy": 0.976200807094574,
+      "num_tokens": 26064480.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7899552762508393,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.1172103881835938,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.06555815935134887,
+      "mean_token_accuracy": 0.9736676931381225,
+      "num_tokens": 26253619.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.8019726693630218,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.94905686378479,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.0663321852684021,
+      "mean_token_accuracy": 0.9739456713199616,
+      "num_tokens": 26442736.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7939825117588043,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 0.9922711253166199,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05981417298316956,
+      "mean_token_accuracy": 0.9758030951023102,
+      "num_tokens": 26631525.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8085412085056305,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4970781803131104,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.06708773374557495,
+      "mean_token_accuracy": 0.9737519145011901,
+      "num_tokens": 26820483.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7950678527355194,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.7868127226829529,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.061734098196029666,
+      "mean_token_accuracy": 0.9747878730297088,
+      "num_tokens": 27009393.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7850766360759736,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.1386486291885376,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.05714603662490845,
+      "mean_token_accuracy": 0.9769004642963409,
+      "num_tokens": 27198251.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7933461844921113,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.0092092752456665,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.06240262985229492,
+      "mean_token_accuracy": 0.9742367327213287,
+      "num_tokens": 27387231.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8044621467590332,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.2477015256881714,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0642085313796997,
+      "mean_token_accuracy": 0.9748865723609924,
+      "num_tokens": 27576251.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7975574374198914,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.1167609691619873,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.06382874846458435,
+      "mean_token_accuracy": 0.9748350441455841,
+      "num_tokens": 27764560.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.8090806484222413,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 0.9714348912239075,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.06401748657226562,
+      "mean_token_accuracy": 0.9741845309734345,
+      "num_tokens": 27953532.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.8104567945383157,
+      "eval_loss": 0.07282773405313492,
+      "eval_mean_token_accuracy": 0.9701232534305305,
+      "eval_num_tokens": 28066151.0,
+      "eval_runtime": 12.7852,
+      "eval_samples_per_second": 391.078,
+      "eval_steps_per_second": 12.28,
+      "step": 1486
+    },
+    {
+      "entropy": 0.8090688765048981,
+      "epoch": 2.005383580080754,
+      "grad_norm": 0.9886816143989563,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.06012396812438965,
+      "mean_token_accuracy": 0.9764885842800141,
+      "num_tokens": 28141848.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.7821709334850311,
+      "epoch": 2.018842530282638,
+      "grad_norm": 0.9049969911575317,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.04819132387638092,
+      "mean_token_accuracy": 0.9816783249378205,
+      "num_tokens": 28330784.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.7845205307006836,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0231750011444092,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.04539612233638764,
+      "mean_token_accuracy": 0.9818619966506958,
+      "num_tokens": 28519580.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.7817754447460175,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.5244711637496948,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0429688960313797,
+      "mean_token_accuracy": 0.9834935545921326,
+      "num_tokens": 28707486.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.7733526527881622,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.0685060024261475,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.045852530002594,
+      "mean_token_accuracy": 0.9817728459835052,
+      "num_tokens": 28896759.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.7815343141555786,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.075853705406189,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.04565088748931885,
+      "mean_token_accuracy": 0.9823914349079133,
+      "num_tokens": 29085097.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.7751398742198944,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2573742866516113,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.04465666115283966,
+      "mean_token_accuracy": 0.9829874813556672,
+      "num_tokens": 29273868.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.7737197399139404,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 0.8681219220161438,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.042188242077827454,
+      "mean_token_accuracy": 0.9837059378623962,
+      "num_tokens": 29462985.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.7823441684246063,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.3503504991531372,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.04393318593502045,
+      "mean_token_accuracy": 0.9827627122402192,
+      "num_tokens": 29651156.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.7796198308467865,
+      "epoch": 2.126514131897712,
+      "grad_norm": 0.7799349427223206,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.04367418885231018,
+      "mean_token_accuracy": 0.9837401032447814,
+      "num_tokens": 29839789.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.7767675697803498,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.375465750694275,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.04427264630794525,
+      "mean_token_accuracy": 0.9832019031047821,
+      "num_tokens": 30028957.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.7912136912345886,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.9683088660240173,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.04428936243057251,
+      "mean_token_accuracy": 0.9832306385040284,
+      "num_tokens": 30217841.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.790750366449356,
+      "epoch": 2.166890982503365,
+      "grad_norm": 0.9393274188041687,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.04529332816600799,
+      "mean_token_accuracy": 0.9832263946533203,
+      "num_tokens": 30406540.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.7816023886203766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.1822642087936401,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.04192846417427063,
+      "mean_token_accuracy": 0.9846727311611175,
+      "num_tokens": 30595568.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7754982054233551,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.1961082220077515,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.043791168928146364,
+      "mean_token_accuracy": 0.9830836653709412,
+      "num_tokens": 30784300.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.7756961703300476,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 0.9719674587249756,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.04410521686077118,
+      "mean_token_accuracy": 0.9837198793888092,
+      "num_tokens": 30972950.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.7833336114883422,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.0175098180770874,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.044813218712806704,
+      "mean_token_accuracy": 0.9829723179340363,
+      "num_tokens": 31161783.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.7893047273159027,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.2077007293701172,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.04417986273765564,
+      "mean_token_accuracy": 0.9825236797332764,
+      "num_tokens": 31350487.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.794023334980011,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8530077934265137,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.043008115887641904,
+      "mean_token_accuracy": 0.9832422196865082,
+      "num_tokens": 31539217.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.785652619600296,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.4543156623840332,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0445009708404541,
+      "mean_token_accuracy": 0.9828064203262329,
+      "num_tokens": 31728308.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7841614365577698,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.297763466835022,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.045649731159210206,
+      "mean_token_accuracy": 0.9823782920837403,
+      "num_tokens": 31917293.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.7825099349021911,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.2237403392791748,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.047218257188797,
+      "mean_token_accuracy": 0.98156818151474,
+      "num_tokens": 32106628.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.7791289031505585,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8094402551651001,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.04178315699100495,
+      "mean_token_accuracy": 0.9846469461917877,
+      "num_tokens": 32295183.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.7743105947971344,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.3677793741226196,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.04540807008743286,
+      "mean_token_accuracy": 0.9827918827533721,
+      "num_tokens": 32484143.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.7804521083831787,
+      "epoch": 2.328398384925976,
+      "grad_norm": 0.8883333802223206,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.04598270654678345,
+      "mean_token_accuracy": 0.9823103487491608,
+      "num_tokens": 32673144.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.7760683119297027,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.989972710609436,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.043702703714370725,
+      "mean_token_accuracy": 0.9831363618373871,
+      "num_tokens": 32862142.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7726980447769165,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.0461052656173706,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.044418123364448545,
+      "mean_token_accuracy": 0.9821757376194,
+      "num_tokens": 33051382.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.7822885036468505,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.9027822017669678,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.04624638557434082,
+      "mean_token_accuracy": 0.9815774321556091,
+      "num_tokens": 33239729.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.7830030500888825,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.1743420362472534,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.04747769832611084,
+      "mean_token_accuracy": 0.9812638640403748,
+      "num_tokens": 33428615.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.7933187901973724,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 0.9334444403648376,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.047731178998947146,
+      "mean_token_accuracy": 0.9820823550224305,
+      "num_tokens": 33617331.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.7904782056808471,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.9396253824234009,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.04432434141635895,
+      "mean_token_accuracy": 0.9830163478851318,
+      "num_tokens": 33805978.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.7811533212661743,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 0.9742764234542847,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.046493279933929446,
+      "mean_token_accuracy": 0.9818211495876312,
+      "num_tokens": 33994957.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7848010778427124,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0378986597061157,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0454249233007431,
+      "mean_token_accuracy": 0.9817994356155395,
+      "num_tokens": 34184297.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.7754390180110932,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.1023566722869873,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.04380808770656586,
+      "mean_token_accuracy": 0.9829573810100556,
+      "num_tokens": 34373551.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.7767188012599945,
+      "epoch": 2.462987886944818,
+      "grad_norm": 0.8460658192634583,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04502716064453125,
+      "mean_token_accuracy": 0.9825617015361786,
+      "num_tokens": 34562169.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.7757977426052094,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.1013047695159912,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.045161890983581546,
+      "mean_token_accuracy": 0.9821914732456207,
+      "num_tokens": 34750863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.785107308626175,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.9510554671287537,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.045189899206161496,
+      "mean_token_accuracy": 0.9822288691997528,
+      "num_tokens": 34939528.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.7849742233753204,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0994503498077393,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.04333080947399139,
+      "mean_token_accuracy": 0.9830258965492249,
+      "num_tokens": 35128334.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.7875016331672668,
+      "epoch": 2.516823687752355,
+      "grad_norm": 0.9202979803085327,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.04838588833808899,
+      "mean_token_accuracy": 0.9815591931343078,
+      "num_tokens": 35317067.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.7813792943954467,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.9036716222763062,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.045662492513656616,
+      "mean_token_accuracy": 0.9817738354206085,
+      "num_tokens": 35506291.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.777835363149643,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.853438675403595,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.044686585664749146,
+      "mean_token_accuracy": 0.98329097032547,
+      "num_tokens": 35695141.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.7677761793136597,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.0699759721755981,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.045490306615829465,
+      "mean_token_accuracy": 0.9826831936836242,
+      "num_tokens": 35884173.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.7756118893623352,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.0356824398040771,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.0467751681804657,
+      "mean_token_accuracy": 0.9823338150978088,
+      "num_tokens": 36073241.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.7775443732738495,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.1619561910629272,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.04741084575653076,
+      "mean_token_accuracy": 0.9815192580223083,
+      "num_tokens": 36262133.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7899596333503723,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.3479284048080444,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.04588942229747772,
+      "mean_token_accuracy": 0.9821865320205688,
+      "num_tokens": 36451460.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.7832387804985046,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.6297204494476318,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.04711659252643585,
+      "mean_token_accuracy": 0.9815292179584503,
+      "num_tokens": 36640366.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.7871865928173065,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.3100138902664185,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.04919236898422241,
+      "mean_token_accuracy": 0.9809704780578613,
+      "num_tokens": 36829017.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.7844007611274719,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.2907201051712036,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.047159326076507566,
+      "mean_token_accuracy": 0.9816433548927307,
+      "num_tokens": 37018034.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.7885129511356354,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.1324766874313354,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.04420983195304871,
+      "mean_token_accuracy": 0.9828807294368744,
+      "num_tokens": 37206768.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.7761563122272491,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.9269904494285583,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.04585547745227814,
+      "mean_token_accuracy": 0.9823735177516937,
+      "num_tokens": 37396010.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7831155300140381,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5890840291976929,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.046450641751289365,
+      "mean_token_accuracy": 0.9817734777927398,
+      "num_tokens": 37584809.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.7729385912418365,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.4823863506317139,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.04653529226779938,
+      "mean_token_accuracy": 0.9815726220607758,
+      "num_tokens": 37773869.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.7752901673316955,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.9423630237579346,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.04599723219871521,
+      "mean_token_accuracy": 0.9816120624542236,
+      "num_tokens": 37962642.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.7750158667564392,
+      "epoch": 2.718707940780619,
+      "grad_norm": 0.9398621320724487,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.04729728400707245,
+      "mean_token_accuracy": 0.9816868543624878,
+      "num_tokens": 38151325.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.7693424999713898,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3207950592041016,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.04474019706249237,
+      "mean_token_accuracy": 0.9830808460712432,
+      "num_tokens": 38340719.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.7750859558582306,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.070200800895691,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.045914703607559205,
+      "mean_token_accuracy": 0.9820488631725312,
+      "num_tokens": 38529293.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7697387218475342,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.0545568466186523,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.04357335567474365,
+      "mean_token_accuracy": 0.9827888309955597,
+      "num_tokens": 38718464.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.7682756781578064,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.0945985317230225,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.0429091215133667,
+      "mean_token_accuracy": 0.9832668721675872,
+      "num_tokens": 38907573.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.7804523587226868,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8521431684494019,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.04662351012229919,
+      "mean_token_accuracy": 0.981964933872223,
+      "num_tokens": 39096598.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.7786886096000671,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.09945809841156,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.050166720151901247,
+      "mean_token_accuracy": 0.9816301107406616,
+      "num_tokens": 39285069.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.7790449261665344,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.1673868894577026,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.048508059978485105,
+      "mean_token_accuracy": 0.9809533059597015,
+      "num_tokens": 39474252.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.7774537563323974,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.0648490190505981,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.04540320634841919,
+      "mean_token_accuracy": 0.9822351098060608,
+      "num_tokens": 39663262.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7683943033218383,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.0106912851333618,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.04466369152069092,
+      "mean_token_accuracy": 0.9818881273269653,
+      "num_tokens": 39852133.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.7642774760723114,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0109530687332153,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.045978450775146486,
+      "mean_token_accuracy": 0.9819821894168854,
+      "num_tokens": 40041679.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.7647961139678955,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.9050237536430359,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.045694130659103396,
+      "mean_token_accuracy": 0.982342666387558,
+      "num_tokens": 40230019.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.7589850842952728,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.8879176378250122,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.046186423301696776,
+      "mean_token_accuracy": 0.9818771004676818,
+      "num_tokens": 40418932.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.765201997756958,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 0.9103795886039734,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.04475873112678528,
+      "mean_token_accuracy": 0.9821203172206878,
+      "num_tokens": 40607792.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.7653619945049286,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.0566699504852295,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.04391545057296753,
+      "mean_token_accuracy": 0.9828326046466828,
+      "num_tokens": 40797122.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.763550215959549,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.1280677318572998,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.045582184195518495,
+      "mean_token_accuracy": 0.9819066524505615,
+      "num_tokens": 40985689.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.7749062657356263,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.110323429107666,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.04598402082920074,
+      "mean_token_accuracy": 0.9815545618534088,
+      "num_tokens": 41174743.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.7805372893810272,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.0649195909500122,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.04705569744110107,
+      "mean_token_accuracy": 0.9815925776958465,
+      "num_tokens": 41363160.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.7711063683032989,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.1459592580795288,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.045355689525604245,
+      "mean_token_accuracy": 0.9832463562488556,
+      "num_tokens": 41551932.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.7747956931591033,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.1602799892425537,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.04494523704051971,
+      "mean_token_accuracy": 0.9819566667079925,
+      "num_tokens": 41740544.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.7584574401378632,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.054417371749878,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.04810510575771332,
+      "mean_token_accuracy": 0.9811540365219116,
+      "num_tokens": 41929344.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.7753648416251894,
+      "eval_loss": 0.07181113958358765,
+      "eval_mean_token_accuracy": 0.9716544128527307,
+      "eval_num_tokens": 42099366.0,
+      "eval_runtime": 12.748,
+      "eval_samples_per_second": 392.22,
+      "eval_steps_per_second": 12.316,
+      "step": 2229
+    },
+    {
+      "entropy": 0.7669954419136047,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.8211690783500671,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.04577431380748749,
+      "mean_token_accuracy": 0.9823363184928894,
+      "num_tokens": 42118188.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.7600219964981079,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.0873607397079468,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.032501548528671265,
+      "mean_token_accuracy": 0.9881355822086334,
+      "num_tokens": 42306598.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.7441959500312805,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.1175898313522339,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.031047609448432923,
+      "mean_token_accuracy": 0.9891582727432251,
+      "num_tokens": 42495250.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.7364335536956788,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.987751841545105,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.028629058599472047,
+      "mean_token_accuracy": 0.9894859492778778,
+      "num_tokens": 42684326.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.735680228471756,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.194305658340454,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.029283303022384643,
+      "mean_token_accuracy": 0.9887769162654877,
+      "num_tokens": 42873221.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.7342639684677124,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.254442811012268,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.0302200973033905,
+      "mean_token_accuracy": 0.988795381784439,
+      "num_tokens": 43062486.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.7391974747180938,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.2648969888687134,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.0316451758146286,
+      "mean_token_accuracy": 0.9879588901996612,
+      "num_tokens": 43251113.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.7401826322078705,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.0483295917510986,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.030878132581710814,
+      "mean_token_accuracy": 0.9886977672576904,
+      "num_tokens": 43440152.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.732647967338562,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.1125447750091553,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.02838427424430847,
+      "mean_token_accuracy": 0.9890213429927825,
+      "num_tokens": 43629353.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.7314932703971863,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.0979042053222656,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.0315592348575592,
+      "mean_token_accuracy": 0.9888832688331604,
+      "num_tokens": 43818981.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.7305827140808105,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9610060453414917,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.03112468421459198,
+      "mean_token_accuracy": 0.9890477001667023,
+      "num_tokens": 44008430.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.7333625435829163,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.0625078678131104,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.03105958104133606,
+      "mean_token_accuracy": 0.9884848177433014,
+      "num_tokens": 44197264.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.7373903751373291,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.1683465242385864,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.031627464294433597,
+      "mean_token_accuracy": 0.9879396498203278,
+      "num_tokens": 44386205.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.7391358911991119,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.05154287815094,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.03118889331817627,
+      "mean_token_accuracy": 0.9880312740802765,
+      "num_tokens": 44575099.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.7351223647594451,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.9531214237213135,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.030980610847473146,
+      "mean_token_accuracy": 0.9882127285003662,
+      "num_tokens": 44763965.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.7361315310001373,
+      "epoch": 3.203230148048452,
+      "grad_norm": 0.8818249702453613,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.029705989360809325,
+      "mean_token_accuracy": 0.9894622683525085,
+      "num_tokens": 44952983.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.7356473922729492,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 0.9263529777526855,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.03202669024467468,
+      "mean_token_accuracy": 0.9877129793167114,
+      "num_tokens": 45142080.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.7326373934745789,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.0517516136169434,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.031153690814971925,
+      "mean_token_accuracy": 0.9891136825084687,
+      "num_tokens": 45330769.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.7348711133003235,
+      "epoch": 3.243606998654105,
+      "grad_norm": 0.8774152994155884,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030574262142181396,
+      "mean_token_accuracy": 0.9888435959815979,
+      "num_tokens": 45519217.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.7379967510700226,
+      "epoch": 3.257065948855989,
+      "grad_norm": 0.9164374470710754,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.03096820116043091,
+      "mean_token_accuracy": 0.9884447932243348,
+      "num_tokens": 45708467.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.7369260907173156,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.3184924125671387,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03082413077354431,
+      "mean_token_accuracy": 0.9888757109642029,
+      "num_tokens": 45897424.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.7366503119468689,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.9073860049247742,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.03220943808555603,
+      "mean_token_accuracy": 0.9877902865409851,
+      "num_tokens": 46086351.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.7341453731060028,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.0940542221069336,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.03329495787620544,
+      "mean_token_accuracy": 0.9877542018890381,
+      "num_tokens": 46275394.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.7312689840793609,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 0.9651952385902405,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.031667912006378175,
+      "mean_token_accuracy": 0.9881450653076171,
+      "num_tokens": 46463821.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.7228557825088501,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 0.971441924571991,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.030608028173446655,
+      "mean_token_accuracy": 0.9882231116294861,
+      "num_tokens": 46653025.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.723542982339859,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.0060454607009888,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.030545371770858764,
+      "mean_token_accuracy": 0.9879125893115998,
+      "num_tokens": 46842068.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.7321009397506714,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.0572997331619263,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.032070663571357724,
+      "mean_token_accuracy": 0.9878279268741608,
+      "num_tokens": 47031045.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.735515958070755,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.9835748672485352,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.03338083028793335,
+      "mean_token_accuracy": 0.9877255499362946,
+      "num_tokens": 47219625.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.7396618247032165,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 0.9655317068099976,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.03589712679386139,
+      "mean_token_accuracy": 0.9869556546211242,
+      "num_tokens": 47408311.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.7358698546886444,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9225422143936157,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.02914820909500122,
+      "mean_token_accuracy": 0.9886718571186066,
+      "num_tokens": 47597297.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.7338236391544342,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.8253662586212158,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.02996352016925812,
+      "mean_token_accuracy": 0.9891344487667084,
+      "num_tokens": 47785737.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.7320856511592865,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.0716700553894043,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.032322123646736145,
+      "mean_token_accuracy": 0.9877110660076142,
+      "num_tokens": 47974241.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.7293457925319672,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.0915459394454956,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.031131434440612792,
+      "mean_token_accuracy": 0.9885926306247711,
+      "num_tokens": 48162582.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.7207926452159882,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0700562000274658,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03113105595111847,
+      "mean_token_accuracy": 0.9880475521087646,
+      "num_tokens": 48351627.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.7178894281387329,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.1952542066574097,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03204495906829834,
+      "mean_token_accuracy": 0.9880682706832886,
+      "num_tokens": 48540171.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.7166945099830627,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0920100212097168,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.03096899390220642,
+      "mean_token_accuracy": 0.9884315967559815,
+      "num_tokens": 48729737.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.7187432169914245,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.9596593976020813,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.03240742087364197,
+      "mean_token_accuracy": 0.9878904581069946,
+      "num_tokens": 48918599.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.7224443972110748,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9406276345252991,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.03253386616706848,
+      "mean_token_accuracy": 0.9873985171318054,
+      "num_tokens": 49107385.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.7226529896259308,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.2169438600540161,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.03282734453678131,
+      "mean_token_accuracy": 0.9876641929149628,
+      "num_tokens": 49295969.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.7263983964920044,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8808715343475342,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.030802983045578002,
+      "mean_token_accuracy": 0.9885481476783753,
+      "num_tokens": 49484706.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.7291677474975586,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2453594207763672,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.03403027057647705,
+      "mean_token_accuracy": 0.987172520160675,
+      "num_tokens": 49673484.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.7396404981613159,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.9215838313102722,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.03405522108078003,
+      "mean_token_accuracy": 0.9871597826480866,
+      "num_tokens": 49862087.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.7382308602333069,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0659712553024292,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.032089871168136594,
+      "mean_token_accuracy": 0.9874629497528076,
+      "num_tokens": 50051379.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.7337264895439148,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 0.9307538270950317,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.032426226139068606,
+      "mean_token_accuracy": 0.9879062473773956,
+      "num_tokens": 50239986.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.7381527841091156,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.1867440938949585,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.03276415467262268,
+      "mean_token_accuracy": 0.9878055810928345,
+      "num_tokens": 50428872.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.7337838113307953,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.9655186533927917,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.03141146302223206,
+      "mean_token_accuracy": 0.9882215678691864,
+      "num_tokens": 50618182.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.7354952573776246,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8848851919174194,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.03387039303779602,
+      "mean_token_accuracy": 0.9871303081512451,
+      "num_tokens": 50806430.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.7329667210578918,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 0.9976673722267151,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.03129216730594635,
+      "mean_token_accuracy": 0.9882283449172974,
+      "num_tokens": 50995500.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.7277809977531433,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.2560697793960571,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.03357822299003601,
+      "mean_token_accuracy": 0.9876434028148651,
+      "num_tokens": 51184436.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.727894538640976,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.1099125146865845,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.032482534646987915,
+      "mean_token_accuracy": 0.9878577649593353,
+      "num_tokens": 51373238.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.7318481564521789,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1980950832366943,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.03284388482570648,
+      "mean_token_accuracy": 0.9874962747097016,
+      "num_tokens": 51561631.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.7219811856746674,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 0.8730921745300293,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.030655479431152342,
+      "mean_token_accuracy": 0.9877778172492981,
+      "num_tokens": 51750845.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.7289035737514495,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.1195666790008545,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.03155539035797119,
+      "mean_token_accuracy": 0.9877057671546936,
+      "num_tokens": 51940009.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.7332375526428223,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9390997886657715,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.031307452917099,
+      "mean_token_accuracy": 0.987467783689499,
+      "num_tokens": 52128442.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.7293307244777679,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.0272324085235596,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.03318062424659729,
+      "mean_token_accuracy": 0.9873259484767913,
+      "num_tokens": 52317152.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.7320701718330384,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9884606003761292,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03461401164531708,
+      "mean_token_accuracy": 0.9867411673069,
+      "num_tokens": 52506106.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.7248260796070098,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.879682719707489,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.03406862616539001,
+      "mean_token_accuracy": 0.9872734069824218,
+      "num_tokens": 52695500.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.7237857282161713,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9409974217414856,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.029435336589813232,
+      "mean_token_accuracy": 0.9891225278377533,
+      "num_tokens": 52884940.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.7237291693687439,
+      "epoch": 3.781965006729475,
+      "grad_norm": 0.9532694220542908,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.030907022953033447,
+      "mean_token_accuracy": 0.9881488382816315,
+      "num_tokens": 53073909.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.7252696454524994,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 0.9157902002334595,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.032040882110595706,
+      "mean_token_accuracy": 0.9880847573280335,
+      "num_tokens": 53262962.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.7237608432769775,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.1450515985488892,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.03199286162853241,
+      "mean_token_accuracy": 0.987687611579895,
+      "num_tokens": 53451982.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.7237064421176911,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.9945085644721985,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.03347462117671966,
+      "mean_token_accuracy": 0.9875530540943146,
+      "num_tokens": 53640737.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.7271939218044281,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.9351736903190613,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.03140422701835632,
+      "mean_token_accuracy": 0.9877301037311554,
+      "num_tokens": 53829142.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.7265262603759766,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.1019784212112427,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.030748724937438965,
+      "mean_token_accuracy": 0.9881740033626556,
+      "num_tokens": 54018331.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.7270502150058746,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 0.9153032898902893,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.033472740650177,
+      "mean_token_accuracy": 0.9876999020576477,
+      "num_tokens": 54206935.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.7345028996467591,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.0144257545471191,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.033550959825515744,
+      "mean_token_accuracy": 0.9874906241893768,
+      "num_tokens": 54395443.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.7274066030979156,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0221407413482666,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.033388751745224,
+      "mean_token_accuracy": 0.9870682001113892,
+      "num_tokens": 54584552.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.7293415367603302,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.019165277481079,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.03568689227104187,
+      "mean_token_accuracy": 0.9861597299575806,
+      "num_tokens": 54773705.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.7281844198703766,
+      "epoch": 3.916554508748318,
+      "grad_norm": 0.933020830154419,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.030142486095428467,
+      "mean_token_accuracy": 0.988915103673935,
+      "num_tokens": 54962152.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.7242733418941498,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.9429674744606018,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.0337733268737793,
+      "mean_token_accuracy": 0.987263411283493,
+      "num_tokens": 55151179.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.7306612014770508,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.9022014737129211,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.03387616574764252,
+      "mean_token_accuracy": 0.9871586680412292,
+      "num_tokens": 55339658.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.7246532738208771,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2193421125411987,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.03193953037261963,
+      "mean_token_accuracy": 0.9879468739032745,
+      "num_tokens": 55528467.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.7268587052822113,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.2917358875274658,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.03584722280502319,
+      "mean_token_accuracy": 0.9867104470729828,
+      "num_tokens": 55717308.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.735340416431427,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.2229385375976562,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.03506708741188049,
+      "mean_token_accuracy": 0.9864643573760986,
+      "num_tokens": 55906059.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.7337038695812226,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.0747122764587402,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.03241217732429504,
+      "mean_token_accuracy": 0.9876143455505371,
+      "num_tokens": 56094606.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.7333687073106219,
+      "eval_loss": 0.07489899545907974,
+      "eval_mean_token_accuracy": 0.9719084019114257,
+      "eval_num_tokens": 56132350.0,
+      "eval_runtime": 12.7823,
+      "eval_samples_per_second": 391.166,
+      "eval_steps_per_second": 12.283,
+      "step": 2972
+    },
+    {
+      "entropy": 0.72461416721344,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7916642427444458,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.02286165803670883,
+      "mean_token_accuracy": 0.9920066714286804,
+      "num_tokens": 56283397.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.7150031447410583,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.9078621864318848,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.020906335115432738,
+      "mean_token_accuracy": 0.9925728499889374,
+      "num_tokens": 56472408.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.7091723501682281,
+      "epoch": 4.037685060565276,
+      "grad_norm": 0.8117756843566895,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.02036277800798416,
+      "mean_token_accuracy": 0.9928471207618713,
+      "num_tokens": 56660906.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.695214182138443,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7815372943878174,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019653698801994322,
+      "mean_token_accuracy": 0.992583978176117,
+      "num_tokens": 56850067.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.6937313377857208,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4769781827926636,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.02311961054801941,
+      "mean_token_accuracy": 0.9921472012996674,
+      "num_tokens": 57039005.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.6977694034576416,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.0028746128082275,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.021832706034183504,
+      "mean_token_accuracy": 0.9917877376079559,
+      "num_tokens": 57227782.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.698432594537735,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.0107522010803223,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.020331941545009613,
+      "mean_token_accuracy": 0.9932828366756439,
+      "num_tokens": 57416541.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.6941913783550262,
+      "epoch": 4.104979811574697,
+      "grad_norm": 0.8923842310905457,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019534927606582642,
+      "mean_token_accuracy": 0.9930490672588348,
+      "num_tokens": 57605162.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.6947623729705811,
+      "epoch": 4.118438761776582,
+      "grad_norm": 0.9093812108039856,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.02151128649711609,
+      "mean_token_accuracy": 0.9920440256595612,
+      "num_tokens": 57793756.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.6972509443759918,
+      "epoch": 4.131897711978466,
+      "grad_norm": 0.7549760341644287,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.020574498176574706,
+      "mean_token_accuracy": 0.9928867638111114,
+      "num_tokens": 57982678.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.6952514111995697,
+      "epoch": 4.14535666218035,
+      "grad_norm": 0.9476824402809143,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.02087453305721283,
+      "mean_token_accuracy": 0.9924874365329742,
+      "num_tokens": 58171508.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.6910251140594482,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.2548149824142456,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.020806866884231567,
+      "mean_token_accuracy": 0.9921584248542785,
+      "num_tokens": 58360594.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.6958921313285827,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.0649152994155884,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.020952591300010683,
+      "mean_token_accuracy": 0.9924699544906617,
+      "num_tokens": 58549637.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.6932485699653625,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.8891078233718872,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.020813333988189697,
+      "mean_token_accuracy": 0.9927419483661651,
+      "num_tokens": 58738595.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.6984799921512603,
+      "epoch": 4.199192462987887,
+      "grad_norm": 0.9459101557731628,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.01971060484647751,
+      "mean_token_accuracy": 0.9926240146160126,
+      "num_tokens": 58927507.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.7021916508674622,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.8318898677825928,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020860180258750916,
+      "mean_token_accuracy": 0.9927471876144409,
+      "num_tokens": 59116537.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.6972847402095794,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.9066984057426453,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.023059430718421935,
+      "mean_token_accuracy": 0.9919216752052307,
+      "num_tokens": 59305838.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.7049594700336457,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1277318000793457,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.022385355830192567,
+      "mean_token_accuracy": 0.99165118932724,
+      "num_tokens": 59494587.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.7045334160327912,
+      "epoch": 4.253028263795424,
+      "grad_norm": 0.8147489428520203,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.022358573973178864,
+      "mean_token_accuracy": 0.9918249011039734,
+      "num_tokens": 59683021.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.7007148265838623,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.9161480069160461,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.02088596671819687,
+      "mean_token_accuracy": 0.9925470650196075,
+      "num_tokens": 59871999.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.7006107687950134,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.9611289501190186,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.022619232535362244,
+      "mean_token_accuracy": 0.991577273607254,
+      "num_tokens": 60060486.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.7069303750991821,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.0339760780334473,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.021571077406406403,
+      "mean_token_accuracy": 0.9922139644622803,
+      "num_tokens": 60249686.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.7146372079849244,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.9761297702789307,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.022300006449222566,
+      "mean_token_accuracy": 0.9918746411800384,
+      "num_tokens": 60438481.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.7134014248847962,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 0.9729707837104797,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.02121303677558899,
+      "mean_token_accuracy": 0.9925692796707153,
+      "num_tokens": 60627501.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.7110478222370148,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.9456778168678284,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.020917078852653502,
+      "mean_token_accuracy": 0.9922434687614441,
+      "num_tokens": 60816135.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.7111951589584351,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8617656826972961,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.020851245522499083,
+      "mean_token_accuracy": 0.9924458563327789,
+      "num_tokens": 61005658.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.7193978548049926,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.8495420813560486,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.02084285914897919,
+      "mean_token_accuracy": 0.9927356660366058,
+      "num_tokens": 61194170.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.7188209414482116,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.728841245174408,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.02111642211675644,
+      "mean_token_accuracy": 0.9926697969436645,
+      "num_tokens": 61382852.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.7130864262580872,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.1274386644363403,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.021722708642482758,
+      "mean_token_accuracy": 0.9922453343868256,
+      "num_tokens": 61572195.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.7132770180702209,
+      "epoch": 4.401076716016151,
+      "grad_norm": 0.9251006245613098,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.021602503955364227,
+      "mean_token_accuracy": 0.9919042110443115,
+      "num_tokens": 61761119.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.7063322901725769,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.0182228088378906,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.020274630188941954,
+      "mean_token_accuracy": 0.9928637027740479,
+      "num_tokens": 61949931.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.7106044590473175,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.795170783996582,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.020594277977943422,
+      "mean_token_accuracy": 0.9925063610076904,
+      "num_tokens": 62138377.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7151111841201783,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1479791402816772,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.02117723524570465,
+      "mean_token_accuracy": 0.9921884059906005,
+      "num_tokens": 62327009.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.7172957301139832,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9795703291893005,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021917465329170226,
+      "mean_token_accuracy": 0.9915876984596252,
+      "num_tokens": 62515859.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.7175660550594329,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.9326112866401672,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.021142563223838805,
+      "mean_token_accuracy": 0.9922006845474243,
+      "num_tokens": 62704678.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.7085749089717865,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.8895787596702576,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.0222358375787735,
+      "mean_token_accuracy": 0.9918791234493256,
+      "num_tokens": 62893576.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.7059603810310364,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2298023700714111,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021999095380306245,
+      "mean_token_accuracy": 0.9920522570610046,
+      "num_tokens": 63082258.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.707107275724411,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.8295655250549316,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021844917535781862,
+      "mean_token_accuracy": 0.9922264873981476,
+      "num_tokens": 63271202.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.7119229018688202,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.7440339922904968,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.022164498269557954,
+      "mean_token_accuracy": 0.9918813884258271,
+      "num_tokens": 63460614.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.7065344154834747,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7275091409683228,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.019415955245494842,
+      "mean_token_accuracy": 0.9928511261940003,
+      "num_tokens": 63649327.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.6984272241592407,
+      "epoch": 4.549125168236878,
+      "grad_norm": 0.9989121556282043,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02176133990287781,
+      "mean_token_accuracy": 0.9923081934452057,
+      "num_tokens": 63837983.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.7084203243255616,
+      "epoch": 4.562584118438762,
+      "grad_norm": 0.9526837468147278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.02182276248931885,
+      "mean_token_accuracy": 0.9921061515808105,
+      "num_tokens": 64026855.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.7097050070762634,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.0156456232070923,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.020074066519737244,
+      "mean_token_accuracy": 0.9926879525184631,
+      "num_tokens": 64215889.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.7087334096431732,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.780275821685791,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.021703639626502992,
+      "mean_token_accuracy": 0.9917874097824096,
+      "num_tokens": 64404515.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.7142964243888855,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7382510304450989,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.019867205619812013,
+      "mean_token_accuracy": 0.9927354514598846,
+      "num_tokens": 64593147.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.7165492177009583,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.8259146213531494,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.021262936294078827,
+      "mean_token_accuracy": 0.9914182722568512,
+      "num_tokens": 64782041.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.7098793745040893,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.9540995359420776,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.0195529043674469,
+      "mean_token_accuracy": 0.9931159794330597,
+      "num_tokens": 64971390.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.7104229986667633,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.8698863387107849,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.019119253754615782,
+      "mean_token_accuracy": 0.9934523403644562,
+      "num_tokens": 65160253.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.7067269921302796,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.9622324705123901,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.02044745087623596,
+      "mean_token_accuracy": 0.992815512418747,
+      "num_tokens": 65349403.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.7016099870204926,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.999839723110199,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.019250848889350893,
+      "mean_token_accuracy": 0.9929339647293091,
+      "num_tokens": 65538164.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.7059290170669555,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.7539619207382202,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.02152409851551056,
+      "mean_token_accuracy": 0.99238161444664,
+      "num_tokens": 65727493.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.7103133797645569,
+      "epoch": 4.697173620457605,
+      "grad_norm": 0.9892463088035583,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020616935193538667,
+      "mean_token_accuracy": 0.9927784681320191,
+      "num_tokens": 65916233.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.703304260969162,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.1750195026397705,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021048617362976075,
+      "mean_token_accuracy": 0.9921808183193207,
+      "num_tokens": 66105229.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.7017257273197174,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.887427031993866,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02077500820159912,
+      "mean_token_accuracy": 0.992085075378418,
+      "num_tokens": 66294280.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.7027543246746063,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.0987571477890015,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.02167164385318756,
+      "mean_token_accuracy": 0.992034786939621,
+      "num_tokens": 66483182.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.7053992748260498,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.0075602531433105,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.022686071693897247,
+      "mean_token_accuracy": 0.9918968200683593,
+      "num_tokens": 66671795.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.7109473288059235,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.1655596494674683,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.022698774933815002,
+      "mean_token_accuracy": 0.9913642406463623,
+      "num_tokens": 66861166.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.7130867660045623,
+      "epoch": 4.77792732166891,
+      "grad_norm": 0.9386347532272339,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0205889493227005,
+      "mean_token_accuracy": 0.9924897074699401,
+      "num_tokens": 67049967.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.7099342107772827,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.8524489998817444,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.01985667645931244,
+      "mean_token_accuracy": 0.9928426086902619,
+      "num_tokens": 67238781.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.7100094079971313,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7522528171539307,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.02066437602043152,
+      "mean_token_accuracy": 0.9924224317073822,
+      "num_tokens": 67427802.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7172351717948914,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9865215420722961,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.021969664096832275,
+      "mean_token_accuracy": 0.9919721603393554,
+      "num_tokens": 67616675.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.7113273084163666,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9339645504951477,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.019794631004333495,
+      "mean_token_accuracy": 0.9925971984863281,
+      "num_tokens": 67805877.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.7041554272174835,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.337364912033081,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020412631332874298,
+      "mean_token_accuracy": 0.9923949122428894,
+      "num_tokens": 67994858.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.7090700149536133,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.7725968956947327,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.02121713161468506,
+      "mean_token_accuracy": 0.9920605599880219,
+      "num_tokens": 68183865.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.7141157209873199,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.8638854026794434,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.020726922154426574,
+      "mean_token_accuracy": 0.992779266834259,
+      "num_tokens": 68372954.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.7078199684619904,
+      "epoch": 4.885598923283984,
+      "grad_norm": 0.8576115965843201,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.019535741209983824,
+      "mean_token_accuracy": 0.9927346885204316,
+      "num_tokens": 68562535.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.7034667313098908,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8339338898658752,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.021528851985931397,
+      "mean_token_accuracy": 0.991979056596756,
+      "num_tokens": 68751258.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.6983348250389099,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8739761710166931,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020945993065834046,
+      "mean_token_accuracy": 0.9923328399658203,
+      "num_tokens": 68939586.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.6922877132892609,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.886886715888977,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.020974791049957274,
+      "mean_token_accuracy": 0.992312878370285,
+      "num_tokens": 69128270.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.6857815742492676,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.9415780901908875,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.020201480388641356,
+      "mean_token_accuracy": 0.9925193309783935,
+      "num_tokens": 69317066.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.6832823157310486,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.8311293721199036,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.02008202373981476,
+      "mean_token_accuracy": 0.9926081895828247,
+      "num_tokens": 69505698.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.6914379715919494,
+      "epoch": 4.96635262449529,
+      "grad_norm": 0.8799999356269836,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.0189486563205719,
+      "mean_token_accuracy": 0.9932616055011749,
+      "num_tokens": 69694164.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.692799162864685,
+      "epoch": 4.979811574697173,
+      "grad_norm": 0.9579370021820068,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.019743245840072633,
+      "mean_token_accuracy": 0.9932292222976684,
+      "num_tokens": 69882717.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.6969726324081421,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.0632585287094116,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.019562503695487975,
+      "mean_token_accuracy": 0.9929000854492187,
+      "num_tokens": 70071225.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.7056300662885047,
+      "eval_loss": 0.08702203631401062,
+      "eval_mean_token_accuracy": 0.9712980936287315,
+      "eval_num_tokens": 70165509.0,
+      "eval_runtime": 12.7642,
+      "eval_samples_per_second": 391.719,
+      "eval_steps_per_second": 12.3,
+      "step": 3715
+    },
+    {
+      "entropy": 0.6998905837535858,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8414336442947388,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01888625919818878,
+      "mean_token_accuracy": 0.9934305250644684,
+      "num_tokens": 70259811.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.6933371186256408,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.7618409395217896,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01257207989692688,
+      "mean_token_accuracy": 0.9958574831485748,
+      "num_tokens": 70448683.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.6861799776554107,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.935829758644104,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.0118876114487648,
+      "mean_token_accuracy": 0.9963350534439087,
+      "num_tokens": 70637078.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.6778797626495361,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.9417919516563416,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.012690700590610504,
+      "mean_token_accuracy": 0.9954731047153473,
+      "num_tokens": 70826145.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.6735071957111358,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.7570633292198181,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01321282535791397,
+      "mean_token_accuracy": 0.9955435931682587,
+      "num_tokens": 71015458.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.677568006515503,
+      "epoch": 5.074024226110363,
+      "grad_norm": 0.8157011866569519,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.012434301525354385,
+      "mean_token_accuracy": 0.9957576751708984,
+      "num_tokens": 71204332.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.6771604120731354,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.780483603477478,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.013237486779689788,
+      "mean_token_accuracy": 0.9957249999046326,
+      "num_tokens": 71393307.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.6761959791183472,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8249189257621765,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.011912986636161804,
+      "mean_token_accuracy": 0.9961272835731506,
+      "num_tokens": 71582233.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.6717506468296051,
+      "epoch": 5.114401076716016,
+      "grad_norm": 0.9667718410491943,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.012740559875965118,
+      "mean_token_accuracy": 0.9953394651412963,
+      "num_tokens": 71771906.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.6755871415138245,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.9068875908851624,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.012314559519290924,
+      "mean_token_accuracy": 0.9955949664115906,
+      "num_tokens": 71960698.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.6791335880756378,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.7650554776191711,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.011927495896816253,
+      "mean_token_accuracy": 0.9957858562469483,
+      "num_tokens": 72149627.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.6738516509532928,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 0.7267928123474121,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.011817440390586853,
+      "mean_token_accuracy": 0.996101975440979,
+      "num_tokens": 72338469.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.6701940059661865,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.9211037755012512,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012324941903352737,
+      "mean_token_accuracy": 0.9956477820873261,
+      "num_tokens": 72527499.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.6763210237026215,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.7772125601768494,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.013027346134185791,
+      "mean_token_accuracy": 0.9959511935710907,
+      "num_tokens": 72715639.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.6793705642223358,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.582334578037262,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.011401662230491638,
+      "mean_token_accuracy": 0.9961598336696624,
+      "num_tokens": 72904615.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.6798616409301758,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.8502725958824158,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011671387404203416,
+      "mean_token_accuracy": 0.9961067497730255,
+      "num_tokens": 73093522.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.682939213514328,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7421601414680481,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.011898189783096313,
+      "mean_token_accuracy": 0.9962145388126373,
+      "num_tokens": 73282415.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.6821306705474853,
+      "epoch": 5.235531628532974,
+      "grad_norm": 0.7552776336669922,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.01232050210237503,
+      "mean_token_accuracy": 0.9957845032215118,
+      "num_tokens": 73471211.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.6746087670326233,
+      "epoch": 5.248990578734858,
+      "grad_norm": 0.7144336700439453,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01240069791674614,
+      "mean_token_accuracy": 0.9960526466369629,
+      "num_tokens": 73660289.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.6723428070545197,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.8043408393859863,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.01204245388507843,
+      "mean_token_accuracy": 0.9960211217403412,
+      "num_tokens": 73849312.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.6766897737979889,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.8608251810073853,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012303075194358826,
+      "mean_token_accuracy": 0.9955043613910675,
+      "num_tokens": 74038344.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.6809355199337006,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.4872569739818573,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.010690021514892577,
+      "mean_token_accuracy": 0.9965942919254303,
+      "num_tokens": 74226981.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.6819032669067383,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.0376152992248535,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012148186564445496,
+      "mean_token_accuracy": 0.9958918631076813,
+      "num_tokens": 74415130.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.6856858193874359,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.7241474986076355,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012284471094608307,
+      "mean_token_accuracy": 0.9959208726882934,
+      "num_tokens": 74603788.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.6874188125133515,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.6505919098854065,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.012959125638008117,
+      "mean_token_accuracy": 0.9956629753112793,
+      "num_tokens": 74792306.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.6912538707256317,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7696103453636169,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012945668399333954,
+      "mean_token_accuracy": 0.9956756651401519,
+      "num_tokens": 74981265.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.6915560126304626,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.8245812058448792,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.01408768892288208,
+      "mean_token_accuracy": 0.9950980842113495,
+      "num_tokens": 75170337.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.6872740864753724,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8330141305923462,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.011667902022600174,
+      "mean_token_accuracy": 0.9960226893424988,
+      "num_tokens": 75359113.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.6825072944164277,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.9389140009880066,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012351009249687194,
+      "mean_token_accuracy": 0.9954748272895813,
+      "num_tokens": 75548102.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.6821270763874054,
+      "epoch": 5.397039030955585,
+      "grad_norm": 1.0638995170593262,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.0128641277551651,
+      "mean_token_accuracy": 0.995329612493515,
+      "num_tokens": 75736452.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.6857812821865081,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7838811278343201,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.011757558584213257,
+      "mean_token_accuracy": 0.9959722757339478,
+      "num_tokens": 75925013.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.6869213104248046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 0.5632259845733643,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.011284272372722625,
+      "mean_token_accuracy": 0.9962978959083557,
+      "num_tokens": 76113398.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.6807511866092681,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.7132044434547424,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012044224143028259,
+      "mean_token_accuracy": 0.9959294259548187,
+      "num_tokens": 76302629.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.6815520942211151,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.7640891075134277,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.012763911485671997,
+      "mean_token_accuracy": 0.9953242301940918,
+      "num_tokens": 76491675.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.6802713990211486,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.5691967010498047,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.0116192027926445,
+      "mean_token_accuracy": 0.9958603858947754,
+      "num_tokens": 76680426.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.676917290687561,
+      "epoch": 5.477792732166891,
+      "grad_norm": 0.8160638213157654,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.01229381114244461,
+      "mean_token_accuracy": 0.9956679701805115,
+      "num_tokens": 76869321.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.6745438575744629,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.8013969659805298,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.012171164155006409,
+      "mean_token_accuracy": 0.9958188533782959,
+      "num_tokens": 77058299.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.6774423599243165,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.77237868309021,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.013017389178276061,
+      "mean_token_accuracy": 0.9956206619739533,
+      "num_tokens": 77247567.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.6773760914802551,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.6185512542724609,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.011151286959648132,
+      "mean_token_accuracy": 0.9961191415786743,
+      "num_tokens": 77436417.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.6750624418258667,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.7832470536231995,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.011553966999053955,
+      "mean_token_accuracy": 0.9958207130432128,
+      "num_tokens": 77625255.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.6814670443534852,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.7947144508361816,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.011924812197685241,
+      "mean_token_accuracy": 0.9959113657474518,
+      "num_tokens": 77813632.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.68285973072052,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.6403945684432983,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.01066286414861679,
+      "mean_token_accuracy": 0.9966644048690796,
+      "num_tokens": 78002213.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.6764974176883698,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.8078592419624329,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.010837660729885101,
+      "mean_token_accuracy": 0.9961750328540802,
+      "num_tokens": 78191471.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.6726884067058563,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.7293245792388916,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.01087343692779541,
+      "mean_token_accuracy": 0.996295166015625,
+      "num_tokens": 78380424.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.6714434802532196,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0266199111938477,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.011094169318675995,
+      "mean_token_accuracy": 0.9960172355175019,
+      "num_tokens": 78569653.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.6689312815666199,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.8064524531364441,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.01098542958498001,
+      "mean_token_accuracy": 0.9961700737476349,
+      "num_tokens": 78759287.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.6728951573371887,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7976726293563843,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.012345843017101288,
+      "mean_token_accuracy": 0.9957349181175232,
+      "num_tokens": 78948032.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.6702533006668091,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.7385782599449158,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.011018334329128266,
+      "mean_token_accuracy": 0.9964040815830231,
+      "num_tokens": 79136972.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.67191202044487,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.6246873736381531,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.011075176298618317,
+      "mean_token_accuracy": 0.9962317526340485,
+      "num_tokens": 79325966.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.6736057937145233,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9479376077651978,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012818500399589539,
+      "mean_token_accuracy": 0.9951797604560852,
+      "num_tokens": 79514853.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.6723325371742248,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.6468302607536316,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.010920494794845581,
+      "mean_token_accuracy": 0.9961069703102112,
+      "num_tokens": 79703739.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.6762719631195069,
+      "epoch": 5.693135935397039,
+      "grad_norm": 0.6862959265708923,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.011675214767456055,
+      "mean_token_accuracy": 0.9958149492740631,
+      "num_tokens": 79892582.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.6739612460136414,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.8402584195137024,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.010931231081485748,
+      "mean_token_accuracy": 0.9960719287395478,
+      "num_tokens": 80081362.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.6767356872558594,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.6180281043052673,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.012499480694532394,
+      "mean_token_accuracy": 0.9957239329814911,
+      "num_tokens": 80269388.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.6773952782154083,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.8782367706298828,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012943127751350402,
+      "mean_token_accuracy": 0.9958240866661072,
+      "num_tokens": 80458191.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.6774130165576935,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.7854585647583008,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.012309956550598144,
+      "mean_token_accuracy": 0.99593386054039,
+      "num_tokens": 80647306.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.6813723623752594,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.6340908408164978,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.01146906390786171,
+      "mean_token_accuracy": 0.9963057219982148,
+      "num_tokens": 80836147.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.6804767549037933,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.5984897613525391,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.010843665152788163,
+      "mean_token_accuracy": 0.996556407213211,
+      "num_tokens": 81025302.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.6759022295475006,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.6014022827148438,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.010135546326637268,
+      "mean_token_accuracy": 0.9966514885425568,
+      "num_tokens": 81214251.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.675349909067154,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.6571088433265686,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01084519699215889,
+      "mean_token_accuracy": 0.9962432861328125,
+      "num_tokens": 81402608.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.6718680560588837,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.9250450730323792,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.01139993518590927,
+      "mean_token_accuracy": 0.9961999177932739,
+      "num_tokens": 81591841.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.674647331237793,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.8669414520263672,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.010886493325233459,
+      "mean_token_accuracy": 0.9962068378925324,
+      "num_tokens": 81780873.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.6801326870918274,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7395965456962585,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.012087566405534744,
+      "mean_token_accuracy": 0.9957775354385376,
+      "num_tokens": 81969835.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.676666659116745,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.7648837566375732,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.011214715242385865,
+      "mean_token_accuracy": 0.9963518619537354,
+      "num_tokens": 82159146.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.676488333940506,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.6776206493377686,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.011839983612298965,
+      "mean_token_accuracy": 0.9959003031253815,
+      "num_tokens": 82347867.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.6743977665901184,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.7158207893371582,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011066053062677383,
+      "mean_token_accuracy": 0.9966431736946106,
+      "num_tokens": 82536890.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.6764561176300049,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.7066213488578796,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.012056462466716766,
+      "mean_token_accuracy": 0.9958225846290588,
+      "num_tokens": 82725921.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.6813391983509064,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.763606607913971,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.011903972923755645,
+      "mean_token_accuracy": 0.9958972215652466,
+      "num_tokens": 82914838.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.6814220428466797,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.8407371044158936,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.011677531152963638,
+      "mean_token_accuracy": 0.9959126651287079,
+      "num_tokens": 83103937.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.6833201944828033,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8418315052986145,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012164801359176636,
+      "mean_token_accuracy": 0.9956748068332673,
+      "num_tokens": 83293106.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.6802135646343231,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.8431894183158875,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012858036160469054,
+      "mean_token_accuracy": 0.995720499753952,
+      "num_tokens": 83481559.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.6758989214897155,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7379446029663086,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.011657674610614777,
+      "mean_token_accuracy": 0.9959479212760926,
+      "num_tokens": 83670620.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.6832396805286407,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9370406866073608,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012775366008281708,
+      "mean_token_accuracy": 0.9958724677562714,
+      "num_tokens": 83858979.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.6826908767223359,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.5771452188491821,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011871224641799927,
+      "mean_token_accuracy": 0.9962578654289246,
+      "num_tokens": 84047664.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.6935603318700365,
+      "eval_loss": 0.08950015902519226,
+      "eval_mean_token_accuracy": 0.9720694904874085,
+      "eval_num_tokens": 84198646.0,
+      "eval_runtime": 12.7741,
+      "eval_samples_per_second": 391.416,
+      "eval_steps_per_second": 12.29,
+      "step": 4458
+    },
+    {
+      "entropy": 0.6853953242301941,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.4715442359447479,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.011727289110422135,
+      "mean_token_accuracy": 0.9963658571243286,
+      "num_tokens": 84236476.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.6872529745101928,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.6521468162536621,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.007476924359798432,
+      "mean_token_accuracy": 0.9976866364479064,
+      "num_tokens": 84425424.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.6764522135257721,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.5134896039962769,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.006954360008239746,
+      "mean_token_accuracy": 0.9981172740459442,
+      "num_tokens": 84614973.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.674551123380661,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.5777756571769714,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.0064676709473133085,
+      "mean_token_accuracy": 0.9981890261173249,
+      "num_tokens": 84803584.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.6743139624595642,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.5243175625801086,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.005892159789800644,
+      "mean_token_accuracy": 0.99809131026268,
+      "num_tokens": 84992194.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.6691218018531799,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.6378570795059204,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.006101621687412262,
+      "mean_token_accuracy": 0.9981180131435394,
+      "num_tokens": 85181570.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.6653562963008881,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.7931498885154724,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.0059667728841304776,
+      "mean_token_accuracy": 0.9982839524745941,
+      "num_tokens": 85370354.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.6660257399082183,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.6156598329544067,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.006237205862998962,
+      "mean_token_accuracy": 0.9980066597461701,
+      "num_tokens": 85559287.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.6661248922348022,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.39466625452041626,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.006016004830598831,
+      "mean_token_accuracy": 0.9984841406345367,
+      "num_tokens": 85748489.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.6644636154174804,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.556770920753479,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.0060367342084646225,
+      "mean_token_accuracy": 0.9981182813644409,
+      "num_tokens": 85937336.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.6630209028720856,
+      "epoch": 6.137281292059219,
+      "grad_norm": 0.5270616412162781,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.005779195204377174,
+      "mean_token_accuracy": 0.9981250524520874,
+      "num_tokens": 86126286.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.6656446874141693,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 0.44525614380836487,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.005375263467431069,
+      "mean_token_accuracy": 0.9983476758003235,
+      "num_tokens": 86314904.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.664457255601883,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.6622279286384583,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.006194194033741951,
+      "mean_token_accuracy": 0.9981902241706848,
+      "num_tokens": 86503447.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.6608214318752289,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.6172921657562256,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.0061217501759529116,
+      "mean_token_accuracy": 0.9983710527420044,
+      "num_tokens": 86693005.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.6578842639923096,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.6241856217384338,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.006712782382965088,
+      "mean_token_accuracy": 0.9979998171329498,
+      "num_tokens": 86882287.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.6561006605625153,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 0.6784538626670837,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.006533285975456238,
+      "mean_token_accuracy": 0.9977931976318359,
+      "num_tokens": 87071532.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.6582947254180909,
+      "epoch": 6.218034993270525,
+      "grad_norm": 0.7572171688079834,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.006016571819782257,
+      "mean_token_accuracy": 0.9979285359382629,
+      "num_tokens": 87260158.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.6552021980285645,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.7922662496566772,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.006150579452514649,
+      "mean_token_accuracy": 0.9982126474380493,
+      "num_tokens": 87448718.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.6578958451747894,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.5214764475822449,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.005079228430986404,
+      "mean_token_accuracy": 0.9984815359115601,
+      "num_tokens": 87637461.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.6603941440582275,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.5404514670372009,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.005841487646102905,
+      "mean_token_accuracy": 0.9983745515346527,
+      "num_tokens": 87826292.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.6628166377544403,
+      "epoch": 6.271870794078062,
+      "grad_norm": 0.47433987259864807,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.006364698708057404,
+      "mean_token_accuracy": 0.9982446670532227,
+      "num_tokens": 88015224.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.6635581731796265,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.8039814233779907,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.006831821799278259,
+      "mean_token_accuracy": 0.9982130885124206,
+      "num_tokens": 88203965.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.6615313947200775,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.5904889106750488,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.006949013471603394,
+      "mean_token_accuracy": 0.9976236402988434,
+      "num_tokens": 88392910.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.6613587439060211,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.6321392059326172,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.006564225256443024,
+      "mean_token_accuracy": 0.9980846107006073,
+      "num_tokens": 88581883.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.6642421364784241,
+      "epoch": 6.325706594885599,
+      "grad_norm": 0.6209113597869873,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.007112482190132141,
+      "mean_token_accuracy": 0.9977501094341278,
+      "num_tokens": 88770574.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.6629104316234589,
+      "epoch": 6.339165545087483,
+      "grad_norm": 0.8903343081474304,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.00698748379945755,
+      "mean_token_accuracy": 0.9977229118347168,
+      "num_tokens": 88959487.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.6647967278957367,
+      "epoch": 6.352624495289367,
+      "grad_norm": 0.842950165271759,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.008400703966617584,
+      "mean_token_accuracy": 0.9975436449050903,
+      "num_tokens": 89148302.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.6641393005847931,
+      "epoch": 6.366083445491252,
+      "grad_norm": 0.6241896152496338,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.006545399129390716,
+      "mean_token_accuracy": 0.9982439935207367,
+      "num_tokens": 89337168.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.6646088480949401,
+      "epoch": 6.379542395693136,
+      "grad_norm": 0.6284488439559937,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.005783513188362122,
+      "mean_token_accuracy": 0.9982234418392182,
+      "num_tokens": 89525632.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.6648603081703186,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.7705378532409668,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.006973850727081299,
+      "mean_token_accuracy": 0.9977744221687317,
+      "num_tokens": 89714436.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.6667861938476562,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.7033233642578125,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.0062720529735088345,
+      "mean_token_accuracy": 0.9978694558143616,
+      "num_tokens": 89903591.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.66892409324646,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.6684027910232544,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.006154683232307434,
+      "mean_token_accuracy": 0.998307728767395,
+      "num_tokens": 90092231.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.6692153334617614,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.7860879302024841,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.006273554265499115,
+      "mean_token_accuracy": 0.9981006205081939,
+      "num_tokens": 90280721.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.6636229634284974,
+      "epoch": 6.446837146702557,
+      "grad_norm": 0.5911756753921509,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.0067244857549667355,
+      "mean_token_accuracy": 0.9979658007621766,
+      "num_tokens": 90469835.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.6663640201091766,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.5844401717185974,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.005917467921972275,
+      "mean_token_accuracy": 0.9981400012969971,
+      "num_tokens": 90658439.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.6635376453399658,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.554587721824646,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.005905192345380783,
+      "mean_token_accuracy": 0.9983574509620666,
+      "num_tokens": 90847689.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.660474544763565,
+      "epoch": 6.48721399730821,
+      "grad_norm": 0.6828640103340149,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.006179597228765488,
+      "mean_token_accuracy": 0.9978089511394501,
+      "num_tokens": 91036945.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.6635875403881073,
+      "epoch": 6.500672947510094,
+      "grad_norm": 0.621636688709259,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.006481277197599411,
+      "mean_token_accuracy": 0.9976639211177826,
+      "num_tokens": 91225796.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.6625118017196655,
+      "epoch": 6.514131897711978,
+      "grad_norm": 0.6648721098899841,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.006836448609828949,
+      "mean_token_accuracy": 0.9981350302696228,
+      "num_tokens": 91414454.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.6632138609886169,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.6122362017631531,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.0057271294295787815,
+      "mean_token_accuracy": 0.9983117520809174,
+      "num_tokens": 91603019.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.6621270835399627,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.707145094871521,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.005987466499209404,
+      "mean_token_accuracy": 0.9983782589435577,
+      "num_tokens": 91791729.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.6593974709510804,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.7802532315254211,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.00661279633641243,
+      "mean_token_accuracy": 0.9978727638721466,
+      "num_tokens": 91980977.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.658482962846756,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 0.8637571334838867,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.005754219368100166,
+      "mean_token_accuracy": 0.9982581317424775,
+      "num_tokens": 92169829.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.6561014533042908,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.7007631659507751,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.006207703426480293,
+      "mean_token_accuracy": 0.9980942070484161,
+      "num_tokens": 92358635.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.6558376729488373,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.6180236339569092,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.005951264500617981,
+      "mean_token_accuracy": 0.9980935871601104,
+      "num_tokens": 92546955.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.6548421680927277,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.5694975852966309,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.0058226577937603,
+      "mean_token_accuracy": 0.9980980336666108,
+      "num_tokens": 92735651.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.6563037514686585,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 0.7417933344841003,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.006121259927749634,
+      "mean_token_accuracy": 0.9980963408946991,
+      "num_tokens": 92925010.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.654166966676712,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.705633282661438,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.005783010274171829,
+      "mean_token_accuracy": 0.9983141779899597,
+      "num_tokens": 93113992.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.6567890524864197,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.5444097518920898,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.005712105333805085,
+      "mean_token_accuracy": 0.9982166588306427,
+      "num_tokens": 93302433.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.6536349058151245,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.4291290044784546,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.0060494735836982725,
+      "mean_token_accuracy": 0.9979224085807801,
+      "num_tokens": 93491429.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.6516136825084686,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.5134868621826172,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.005212320387363434,
+      "mean_token_accuracy": 0.9984170615673065,
+      "num_tokens": 93679973.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.6528811275959014,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.8154370188713074,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.00622846707701683,
+      "mean_token_accuracy": 0.998017430305481,
+      "num_tokens": 93868735.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.6532892882823944,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.5574272871017456,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.005193561688065529,
+      "mean_token_accuracy": 0.9985900521278381,
+      "num_tokens": 94057428.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.6552170813083649,
+      "epoch": 6.716016150740242,
+      "grad_norm": 0.6016380190849304,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.005771375447511673,
+      "mean_token_accuracy": 0.9981073677539826,
+      "num_tokens": 94246297.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.6572140097618103,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.7832751870155334,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.005496300011873245,
+      "mean_token_accuracy": 0.9983509957790375,
+      "num_tokens": 94434253.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.6541008234024048,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.4329380989074707,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.006188442185521126,
+      "mean_token_accuracy": 0.9981851994991302,
+      "num_tokens": 94623327.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.6556220591068268,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.5326682329177856,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.0052257638424634935,
+      "mean_token_accuracy": 0.9983932614326477,
+      "num_tokens": 94811837.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.6557375073432923,
+      "epoch": 6.769851951547779,
+      "grad_norm": 0.6864677667617798,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.005970630794763565,
+      "mean_token_accuracy": 0.9983245611190796,
+      "num_tokens": 95000726.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.6545388638973236,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.6318016648292542,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.006516661494970322,
+      "mean_token_accuracy": 0.9979697644710541,
+      "num_tokens": 95189395.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.6542861998081207,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.3979317247867584,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.005074064806103706,
+      "mean_token_accuracy": 0.998512452840805,
+      "num_tokens": 95378855.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.6542766332626343,
+      "epoch": 6.810228802153432,
+      "grad_norm": 0.4778045117855072,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.005774036794900894,
+      "mean_token_accuracy": 0.9982702076435089,
+      "num_tokens": 95568111.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.6550499975681305,
+      "epoch": 6.823687752355316,
+      "grad_norm": 0.4805110991001129,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.005620773509144783,
+      "mean_token_accuracy": 0.9983035147190094,
+      "num_tokens": 95756885.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.6543555080890655,
+      "epoch": 6.837146702557201,
+      "grad_norm": 0.6604891419410706,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.005247855931520462,
+      "mean_token_accuracy": 0.998341029882431,
+      "num_tokens": 95946078.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.6581745266914367,
+      "epoch": 6.850605652759085,
+      "grad_norm": 0.6589891910552979,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.005882954970002174,
+      "mean_token_accuracy": 0.9982152104377746,
+      "num_tokens": 96134597.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.6597592771053314,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.4152396023273468,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.0054433666169643406,
+      "mean_token_accuracy": 0.9983908593654632,
+      "num_tokens": 96323327.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.658534574508667,
+      "epoch": 6.877523553162853,
+      "grad_norm": 0.6025403141975403,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.005522595718502998,
+      "mean_token_accuracy": 0.9983524858951569,
+      "num_tokens": 96512170.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.6568575739860535,
+      "epoch": 6.890982503364738,
+      "grad_norm": 0.5730848908424377,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.004784010350704193,
+      "mean_token_accuracy": 0.9984236299991608,
+      "num_tokens": 96701363.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.6535935163497925,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.5301522612571716,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.004592524468898773,
+      "mean_token_accuracy": 0.9987652719020843,
+      "num_tokens": 96890642.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.6544224679470062,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.639918863773346,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.005577490478754043,
+      "mean_token_accuracy": 0.9984355092048645,
+      "num_tokens": 97079552.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.6542503774166107,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.5202826857566833,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.005623985454440117,
+      "mean_token_accuracy": 0.9982783615589141,
+      "num_tokens": 97268346.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.6499157190322876,
+      "epoch": 6.944818304172275,
+      "grad_norm": 0.5668892860412598,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.005104385688900948,
+      "mean_token_accuracy": 0.9985612750053405,
+      "num_tokens": 97457514.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.6520473599433899,
+      "epoch": 6.958277254374159,
+      "grad_norm": 0.5878952145576477,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.005280185490846634,
+      "mean_token_accuracy": 0.9985166370868683,
+      "num_tokens": 97646014.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.6488265216350555,
+      "epoch": 6.971736204576043,
+      "grad_norm": 0.4630887806415558,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.005359554663300514,
+      "mean_token_accuracy": 0.998493617773056,
+      "num_tokens": 97835401.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.6526197671890259,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.38062795996665955,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.0054448477923870085,
+      "mean_token_accuracy": 0.9982813715934753,
+      "num_tokens": 98024194.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.654500275850296,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.5739313364028931,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.005691033974289894,
+      "mean_token_accuracy": 0.9982043981552124,
+      "num_tokens": 98212962.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.6537400142402406,
+      "eval_loss": 0.09857651591300964,
+      "eval_mean_token_accuracy": 0.9730726879113799,
+      "eval_num_tokens": 98231758.0,
+      "eval_runtime": 12.7391,
+      "eval_samples_per_second": 392.491,
+      "eval_steps_per_second": 12.324,
+      "step": 5201
+    },
+    {
+      "entropy": 0.6512321710586548,
+      "epoch": 7.012113055181696,
+      "grad_norm": 0.2961742877960205,
+      "learning_rate": 2.2506648060191042e-05,
+      "loss": 0.0032418236136436463,
+      "mean_token_accuracy": 0.999194347858429,
+      "num_tokens": 98401402.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.6491944074630738,
+      "epoch": 7.02557200538358,
+      "grad_norm": 0.35240232944488525,
+      "learning_rate": 2.2321031102941066e-05,
+      "loss": 0.0028009919449687006,
+      "mean_token_accuracy": 0.9995193898677825,
+      "num_tokens": 98590150.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.6500660896301269,
+      "epoch": 7.039030955585464,
+      "grad_norm": 0.3501753509044647,
+      "learning_rate": 2.2135962530768055e-05,
+      "loss": 0.0029570702463388443,
+      "mean_token_accuracy": 0.9993301749229431,
+      "num_tokens": 98778612.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.6460368275642395,
+      "epoch": 7.052489905787349,
+      "grad_norm": 0.41639405488967896,
+      "learning_rate": 2.1951446010313053e-05,
+      "loss": 0.003057733364403248,
+      "mean_token_accuracy": 0.9994841694831849,
+      "num_tokens": 98967879.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.6441825687885284,
+      "epoch": 7.065948855989233,
+      "grad_norm": 0.3744174540042877,
+      "learning_rate": 2.1767485197279662e-05,
+      "loss": 0.0028053496032953264,
+      "mean_token_accuracy": 0.9994258165359498,
+      "num_tokens": 99157371.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.6460713386535645,
+      "epoch": 7.079407806191117,
+      "grad_norm": 0.22786536812782288,
+      "learning_rate": 2.1584083736361617e-05,
+      "loss": 0.004050987213850022,
+      "mean_token_accuracy": 0.9992221534252167,
+      "num_tokens": 99346074.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.6439846336841584,
+      "epoch": 7.0928667563930015,
+      "grad_norm": 0.33427926898002625,
+      "learning_rate": 2.1401245261170582e-05,
+      "loss": 0.002635418064892292,
+      "mean_token_accuracy": 0.9993825376033783,
+      "num_tokens": 99535282.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.6429140031337738,
+      "epoch": 7.106325706594886,
+      "grad_norm": 0.35418426990509033,
+      "learning_rate": 2.121897339416416e-05,
+      "loss": 0.002664094232022762,
+      "mean_token_accuracy": 0.999459445476532,
+      "num_tokens": 99724199.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.6431400418281555,
+      "epoch": 7.11978465679677,
+      "grad_norm": 0.2773686945438385,
+      "learning_rate": 2.103727174657412e-05,
+      "loss": 0.0029107578098773955,
+      "mean_token_accuracy": 0.999220222234726,
+      "num_tokens": 99913190.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.644054365158081,
+      "epoch": 7.133243606998654,
+      "grad_norm": 0.466249942779541,
+      "learning_rate": 2.085614391833486e-05,
+      "loss": 0.0026346595957875254,
+      "mean_token_accuracy": 0.9993867635726928,
+      "num_tokens": 100102083.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.6469399213790894,
+      "epoch": 7.1467025572005385,
+      "grad_norm": 0.37575817108154297,
+      "learning_rate": 2.0675593498012052e-05,
+      "loss": 0.0028098784387111664,
+      "mean_token_accuracy": 0.9992999136447906,
+      "num_tokens": 100290764.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.6454071402549744,
+      "epoch": 7.160161507402423,
+      "grad_norm": 0.5268272161483765,
+      "learning_rate": 2.049562406273159e-05,
+      "loss": 0.0025207167491316795,
+      "mean_token_accuracy": 0.9995582520961761,
+      "num_tokens": 100479571.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.6419129431247711,
+      "epoch": 7.173620457604307,
+      "grad_norm": 0.28378117084503174,
+      "learning_rate": 2.0316239178108688e-05,
+      "loss": 0.00287601538002491,
+      "mean_token_accuracy": 0.9993176221847534,
+      "num_tokens": 100668516.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.6420379161834717,
+      "epoch": 7.187079407806191,
+      "grad_norm": 0.24231773614883423,
+      "learning_rate": 2.013744239817725e-05,
+      "loss": 0.002684992924332619,
+      "mean_token_accuracy": 0.9993604898452759,
+      "num_tokens": 100857247.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.6403457880020141,
+      "epoch": 7.2005383580080755,
+      "grad_norm": 0.350322961807251,
+      "learning_rate": 1.9959237265319452e-05,
+      "loss": 0.002843806892633438,
+      "mean_token_accuracy": 0.9992822408676147,
+      "num_tokens": 101046251.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.6437962651252747,
+      "epoch": 7.21399730820996,
+      "grad_norm": 0.504984974861145,
+      "learning_rate": 1.9781627310195528e-05,
+      "loss": 0.002904067933559418,
+      "mean_token_accuracy": 0.9991836249828339,
+      "num_tokens": 101235009.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.6444802343845367,
+      "epoch": 7.227456258411844,
+      "grad_norm": 0.5015905499458313,
+      "learning_rate": 1.960461605167389e-05,
+      "loss": 0.0031951837241649627,
+      "mean_token_accuracy": 0.9991663873195649,
+      "num_tokens": 101423795.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.6445916533470154,
+      "epoch": 7.240915208613728,
+      "grad_norm": 0.8163104057312012,
+      "learning_rate": 1.9428206996761357e-05,
+      "loss": 0.0031633909791707994,
+      "mean_token_accuracy": 0.9990993022918702,
+      "num_tokens": 101612550.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.6424523174762726,
+      "epoch": 7.2543741588156125,
+      "grad_norm": 0.7167068123817444,
+      "learning_rate": 1.9252403640533678e-05,
+      "loss": 0.0027146343141794204,
+      "mean_token_accuracy": 0.9993792653083802,
+      "num_tokens": 101801308.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.6425661742687225,
+      "epoch": 7.267833109017497,
+      "grad_norm": 0.3235819935798645,
+      "learning_rate": 1.9077209466066315e-05,
+      "loss": 0.002999866008758545,
+      "mean_token_accuracy": 0.9993207931518555,
+      "num_tokens": 101990167.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.6397831857204437,
+      "epoch": 7.281292059219381,
+      "grad_norm": 0.2559046149253845,
+      "learning_rate": 1.8902627944365396e-05,
+      "loss": 0.0023009315133094787,
+      "mean_token_accuracy": 0.9995379090309143,
+      "num_tokens": 102179185.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.638386732339859,
+      "epoch": 7.294751009421265,
+      "grad_norm": 0.4241917133331299,
+      "learning_rate": 1.872866253429897e-05,
+      "loss": 0.003075192496180534,
+      "mean_token_accuracy": 0.9992788910865784,
+      "num_tokens": 102368161.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.6398141801357269,
+      "epoch": 7.30820995962315,
+      "grad_norm": 0.4451645612716675,
+      "learning_rate": 1.8555316682528485e-05,
+      "loss": 0.0027588184922933577,
+      "mean_token_accuracy": 0.9993425786495209,
+      "num_tokens": 102557229.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.6400523662567139,
+      "epoch": 7.321668909825034,
+      "grad_norm": 0.2873964309692383,
+      "learning_rate": 1.8382593823440474e-05,
+      "loss": 0.0026803961023688315,
+      "mean_token_accuracy": 0.9994625568389892,
+      "num_tokens": 102745928.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.6401661276817322,
+      "epoch": 7.335127860026918,
+      "grad_norm": 0.5918443202972412,
+      "learning_rate": 1.821049737907854e-05,
+      "loss": 0.0029163889586925505,
+      "mean_token_accuracy": 0.9992423593997956,
+      "num_tokens": 102934875.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.6410463333129883,
+      "epoch": 7.348586810228802,
+      "grad_norm": 0.39631369709968567,
+      "learning_rate": 1.803903075907554e-05,
+      "loss": 0.002800934761762619,
+      "mean_token_accuracy": 0.9993205904960633,
+      "num_tokens": 103123665.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.6384492158889771,
+      "epoch": 7.362045760430687,
+      "grad_norm": 0.44129809737205505,
+      "learning_rate": 1.7868197360586054e-05,
+      "loss": 0.0027701158076524736,
+      "mean_token_accuracy": 0.9992826044559479,
+      "num_tokens": 103312769.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.6399481356143951,
+      "epoch": 7.375504710632571,
+      "grad_norm": 0.3396782875061035,
+      "learning_rate": 1.769800056821904e-05,
+      "loss": 0.002843455225229263,
+      "mean_token_accuracy": 0.9993624866008759,
+      "num_tokens": 103501428.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.6391602873802185,
+      "epoch": 7.388963660834455,
+      "grad_norm": 0.508573055267334,
+      "learning_rate": 1.7528443753970814e-05,
+      "loss": 0.0030046245083212852,
+      "mean_token_accuracy": 0.9992032408714294,
+      "num_tokens": 103690316.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.641320925951004,
+      "epoch": 7.402422611036339,
+      "grad_norm": 0.372712641954422,
+      "learning_rate": 1.7359530277158227e-05,
+      "loss": 0.0029981501400470735,
+      "mean_token_accuracy": 0.9992169618606568,
+      "num_tokens": 103879262.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.6416053056716919,
+      "epoch": 7.415881561238224,
+      "grad_norm": 0.47443583607673645,
+      "learning_rate": 1.7191263484352106e-05,
+      "loss": 0.00280308797955513,
+      "mean_token_accuracy": 0.9993564009666442,
+      "num_tokens": 104067769.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.6400010466575623,
+      "epoch": 7.429340511440108,
+      "grad_norm": 0.26680561900138855,
+      "learning_rate": 1.702364670931097e-05,
+      "loss": 0.0029377201572060585,
+      "mean_token_accuracy": 0.9992177426815033,
+      "num_tokens": 104256396.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.6378615140914917,
+      "epoch": 7.442799461641992,
+      "grad_norm": 0.3017747700214386,
+      "learning_rate": 1.685668327291495e-05,
+      "loss": 0.0025883980095386507,
+      "mean_token_accuracy": 0.9993183672428131,
+      "num_tokens": 104445275.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.6393269956111908,
+      "epoch": 7.456258411843876,
+      "grad_norm": 0.33738431334495544,
+      "learning_rate": 1.6690376483100034e-05,
+      "loss": 0.0023764621466398237,
+      "mean_token_accuracy": 0.999459034204483,
+      "num_tokens": 104633705.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.6352120459079742,
+      "epoch": 7.469717362045761,
+      "grad_norm": 0.30927929282188416,
+      "learning_rate": 1.6524729634792475e-05,
+      "loss": 0.002412043884396553,
+      "mean_token_accuracy": 0.9993467569351197,
+      "num_tokens": 104822986.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.6355067729949951,
+      "epoch": 7.483176312247645,
+      "grad_norm": 0.3294868469238281,
+      "learning_rate": 1.6359746009843567e-05,
+      "loss": 0.002515685372054577,
+      "mean_token_accuracy": 0.999462366104126,
+      "num_tokens": 105012029.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.6385121285915375,
+      "epoch": 7.496635262449529,
+      "grad_norm": 0.4267696738243103,
+      "learning_rate": 1.6195428876964592e-05,
+      "loss": 0.0027210000902414323,
+      "mean_token_accuracy": 0.9993004381656647,
+      "num_tokens": 105200942.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.641570907831192,
+      "epoch": 7.510094212651413,
+      "grad_norm": 0.28043922781944275,
+      "learning_rate": 1.6031781491662056e-05,
+      "loss": 0.0025119861587882043,
+      "mean_token_accuracy": 0.9993769824504852,
+      "num_tokens": 105389627.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.6401131987571717,
+      "epoch": 7.523553162853298,
+      "grad_norm": 0.5159914493560791,
+      "learning_rate": 1.58688070961732e-05,
+      "loss": 0.0026847902685403825,
+      "mean_token_accuracy": 0.9993621706962585,
+      "num_tokens": 105578832.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.6401657581329345,
+      "epoch": 7.537012113055182,
+      "grad_norm": 0.47359421849250793,
+      "learning_rate": 1.5706508919401804e-05,
+      "loss": 0.0025615351274609565,
+      "mean_token_accuracy": 0.999418193101883,
+      "num_tokens": 105767407.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.640064799785614,
+      "epoch": 7.550471063257066,
+      "grad_norm": 0.2965790927410126,
+      "learning_rate": 1.554489017685412e-05,
+      "loss": 0.002729836292564869,
+      "mean_token_accuracy": 0.999356073141098,
+      "num_tokens": 105955938.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.6364653050899506,
+      "epoch": 7.56393001345895,
+      "grad_norm": 0.3772422969341278,
+      "learning_rate": 1.538395407057525e-05,
+      "loss": 0.002501054294407368,
+      "mean_token_accuracy": 0.99944589138031,
+      "num_tokens": 106145287.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.6365915596485138,
+      "epoch": 7.577388963660835,
+      "grad_norm": 0.34281232953071594,
+      "learning_rate": 1.522370378908567e-05,
+      "loss": 0.0024556241929531097,
+      "mean_token_accuracy": 0.9995415151119232,
+      "num_tokens": 106334420.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.6392146408557892,
+      "epoch": 7.590847913862719,
+      "grad_norm": 0.289888858795166,
+      "learning_rate": 1.5064142507318051e-05,
+      "loss": 0.0024054452776908876,
+      "mean_token_accuracy": 0.999481338262558,
+      "num_tokens": 106523346.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.63958261013031,
+      "epoch": 7.604306864064603,
+      "grad_norm": 0.4470290541648865,
+      "learning_rate": 1.4905273386554386e-05,
+      "loss": 0.002950721047818661,
+      "mean_token_accuracy": 0.9994794666767121,
+      "num_tokens": 106711623.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.6364846646785736,
+      "epoch": 7.6177658142664875,
+      "grad_norm": 0.3294987678527832,
+      "learning_rate": 1.4747099574363332e-05,
+      "loss": 0.0029568497091531755,
+      "mean_token_accuracy": 0.99925497174263,
+      "num_tokens": 106900044.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.6371748626232148,
+      "epoch": 7.631224764468372,
+      "grad_norm": 0.3742607533931732,
+      "learning_rate": 1.4589624204537866e-05,
+      "loss": 0.002776237577199936,
+      "mean_token_accuracy": 0.9994770407676696,
+      "num_tokens": 107088499.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.6374151885509491,
+      "epoch": 7.644683714670256,
+      "grad_norm": 0.34489092230796814,
+      "learning_rate": 1.4432850397033176e-05,
+      "loss": 0.0025846634060144426,
+      "mean_token_accuracy": 0.9992987811565399,
+      "num_tokens": 107277120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.6350409626960755,
+      "epoch": 7.65814266487214,
+      "grad_norm": 0.4673531651496887,
+      "learning_rate": 1.427678125790487e-05,
+      "loss": 0.002661401592195034,
+      "mean_token_accuracy": 0.9993412852287292,
+      "num_tokens": 107466227.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.6376935839653015,
+      "epoch": 7.6716016150740245,
+      "grad_norm": 0.2515313923358917,
+      "learning_rate": 1.4121419879247422e-05,
+      "loss": 0.0023796845227479933,
+      "mean_token_accuracy": 0.999422425031662,
+      "num_tokens": 107654984.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.6349337220191955,
+      "epoch": 7.685060565275909,
+      "grad_norm": 0.38676151633262634,
+      "learning_rate": 1.3966769339132929e-05,
+      "loss": 0.002421387284994125,
+      "mean_token_accuracy": 0.999384069442749,
+      "num_tokens": 107844126.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.6366126656532287,
+      "epoch": 7.698519515477793,
+      "grad_norm": 0.39246416091918945,
+      "learning_rate": 1.3812832701550099e-05,
+      "loss": 0.0024261830374598505,
+      "mean_token_accuracy": 0.9993186116218566,
+      "num_tokens": 108032874.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.6357373297214508,
+      "epoch": 7.711978465679677,
+      "grad_norm": 0.319026917219162,
+      "learning_rate": 1.365961301634357e-05,
+      "loss": 0.0023328609764575957,
+      "mean_token_accuracy": 0.9995430946350098,
+      "num_tokens": 108222192.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.6360769748687745,
+      "epoch": 7.7254374158815615,
+      "grad_norm": 0.38521459698677063,
+      "learning_rate": 1.350711331915348e-05,
+      "loss": 0.0026227174326777457,
+      "mean_token_accuracy": 0.9992436349391938,
+      "num_tokens": 108410750.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.6360529124736786,
+      "epoch": 7.738896366083446,
+      "grad_norm": 0.2718389928340912,
+      "learning_rate": 1.3355336631355314e-05,
+      "loss": 0.0024306148290634154,
+      "mean_token_accuracy": 0.9995024859905243,
+      "num_tokens": 108599812.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.6351405322551728,
+      "epoch": 7.75235531628533,
+      "grad_norm": 0.38458114862442017,
+      "learning_rate": 1.3204285960000051e-05,
+      "loss": 0.002783242240548134,
+      "mean_token_accuracy": 0.9993249535560608,
+      "num_tokens": 108788623.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.6342482805252075,
+      "epoch": 7.765814266487214,
+      "grad_norm": 0.37706634402275085,
+      "learning_rate": 1.30539642977546e-05,
+      "loss": 0.002872874028980732,
+      "mean_token_accuracy": 0.9993605136871337,
+      "num_tokens": 108977505.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.6347688436508179,
+      "epoch": 7.7792732166890985,
+      "grad_norm": 0.44668373465538025,
+      "learning_rate": 1.2904374622842486e-05,
+      "loss": 0.002442394569516182,
+      "mean_token_accuracy": 0.999343854188919,
+      "num_tokens": 109166589.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.6364605367183686,
+      "epoch": 7.792732166890983,
+      "grad_norm": 0.2349705696105957,
+      "learning_rate": 1.2755519898984846e-05,
+      "loss": 0.00250922292470932,
+      "mean_token_accuracy": 0.9995181024074554,
+      "num_tokens": 109355292.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.6371697664260865,
+      "epoch": 7.806191117092867,
+      "grad_norm": 0.3123619556427002,
+      "learning_rate": 1.2607403075341728e-05,
+      "loss": 0.002688922733068466,
+      "mean_token_accuracy": 0.9993325173854828,
+      "num_tokens": 109543865.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.6380978763103485,
+      "epoch": 7.819650067294751,
+      "grad_norm": 0.2686324417591095,
+      "learning_rate": 1.2460027086453658e-05,
+      "loss": 0.0023211821913719176,
+      "mean_token_accuracy": 0.9995401680469513,
+      "num_tokens": 109732486.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.6382215797901154,
+      "epoch": 7.833109017496636,
+      "grad_norm": 0.4553178548812866,
+      "learning_rate": 1.2313394852183473e-05,
+      "loss": 0.0024582967162132263,
+      "mean_token_accuracy": 0.9994239926338195,
+      "num_tokens": 109921491.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.6385389983654022,
+      "epoch": 7.84656796769852,
+      "grad_norm": 0.2880251109600067,
+      "learning_rate": 1.2167509277658518e-05,
+      "loss": 0.0024548277258872988,
+      "mean_token_accuracy": 0.9994565665721893,
+      "num_tokens": 110110256.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.633701753616333,
+      "epoch": 7.860026917900404,
+      "grad_norm": 0.4071319103240967,
+      "learning_rate": 1.2022373253213049e-05,
+      "loss": 0.0025199253112077715,
+      "mean_token_accuracy": 0.9993486106395721,
+      "num_tokens": 110299624.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.6337139904499054,
+      "epoch": 7.873485868102288,
+      "grad_norm": 0.4106242060661316,
+      "learning_rate": 1.1877989654330973e-05,
+      "loss": 0.0022239934653043745,
+      "mean_token_accuracy": 0.999601423740387,
+      "num_tokens": 110488951.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.6346798777580261,
+      "epoch": 7.886944818304173,
+      "grad_norm": 0.5445306301116943,
+      "learning_rate": 1.1734361341588912e-05,
+      "loss": 0.0027007974684238436,
+      "mean_token_accuracy": 0.9993659555912018,
+      "num_tokens": 110678171.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.6366956651210784,
+      "epoch": 7.900403768506057,
+      "grad_norm": 0.2735433280467987,
+      "learning_rate": 1.1591491160599488e-05,
+      "loss": 0.0026022626087069513,
+      "mean_token_accuracy": 0.9993394792079926,
+      "num_tokens": 110867003.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.6380104124546051,
+      "epoch": 7.913862718707941,
+      "grad_norm": 0.5340652465820312,
+      "learning_rate": 1.1449381941954956e-05,
+      "loss": 0.002648356184363365,
+      "mean_token_accuracy": 0.9994826078414917,
+      "num_tokens": 111055985.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.6362045764923095,
+      "epoch": 7.927321668909825,
+      "grad_norm": 0.3767072856426239,
+      "learning_rate": 1.130803650117116e-05,
+      "loss": 0.0024559229612350466,
+      "mean_token_accuracy": 0.9995233714580536,
+      "num_tokens": 111244949.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.6355053067207337,
+      "epoch": 7.94078061911171,
+      "grad_norm": 0.3835017681121826,
+      "learning_rate": 1.1167457638631656e-05,
+      "loss": 0.002397604286670685,
+      "mean_token_accuracy": 0.9995415270328522,
+      "num_tokens": 111433946.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.6365890502929688,
+      "epoch": 7.954239569313594,
+      "grad_norm": 0.3102132976055145,
+      "learning_rate": 1.1027648139532355e-05,
+      "loss": 0.002390575036406517,
+      "mean_token_accuracy": 0.9994417130947113,
+      "num_tokens": 111622676.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.6346444308757782,
+      "epoch": 7.967698519515478,
+      "grad_norm": 0.49554574489593506,
+      "learning_rate": 1.0888610773826253e-05,
+      "loss": 0.0021300587803125383,
+      "mean_token_accuracy": 0.9996205389499664,
+      "num_tokens": 111811705.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.6345613896846771,
+      "epoch": 7.981157469717362,
+      "grad_norm": 0.2676335871219635,
+      "learning_rate": 1.0750348296168583e-05,
+      "loss": 0.0024733129888772965,
+      "mean_token_accuracy": 0.9994426965713501,
+      "num_tokens": 112000972.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.6350580513477325,
+      "epoch": 7.994616419919247,
+      "grad_norm": 0.20178340375423431,
+      "learning_rate": 1.0612863445862236e-05,
+      "loss": 0.002096991054713726,
+      "mean_token_accuracy": 0.9996170699596405,
+      "num_tokens": 112189178.0,
+      "step": 5940
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.6389722296386767,
+      "eval_loss": 0.10558179020881653,
+      "eval_mean_token_accuracy": 0.9726916976795075,
+      "eval_num_tokens": 112264896.0,
+      "eval_runtime": 12.7264,
+      "eval_samples_per_second": 392.885,
+      "eval_steps_per_second": 12.337,
+      "step": 5944
+    },
+    {
+      "entropy": 0.6323651373386383,
+      "epoch": 8.00807537012113,
+      "grad_norm": 0.2156553864479065,
+      "learning_rate": 1.0476158946803482e-05,
+      "loss": 0.0017782203853130341,
+      "mean_token_accuracy": 0.9996409475803375,
+      "num_tokens": 112378248.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.6356893301010131,
+      "epoch": 8.021534320323015,
+      "grad_norm": 0.4005093574523926,
+      "learning_rate": 1.0340237507428009e-05,
+      "loss": 0.0015996076166629792,
+      "mean_token_accuracy": 0.9998390734195709,
+      "num_tokens": 112566565.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.6323986113071441,
+      "epoch": 8.034993270524899,
+      "grad_norm": 0.2522565722465515,
+      "learning_rate": 1.020510182065727e-05,
+      "loss": 0.0014581691473722457,
+      "mean_token_accuracy": 0.9998226821422577,
+      "num_tokens": 112755790.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.6342778027057647,
+      "epoch": 8.048452220726784,
+      "grad_norm": 0.14223679900169373,
+      "learning_rate": 1.0070754563845114e-05,
+      "loss": 0.0015433312393724919,
+      "mean_token_accuracy": 0.9998199999332428,
+      "num_tokens": 112944666.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.6335026860237122,
+      "epoch": 8.061911170928667,
+      "grad_norm": 0.45429059863090515,
+      "learning_rate": 9.937198398724751e-06,
+      "loss": 0.0015122929587960243,
+      "mean_token_accuracy": 0.9997402548789978,
+      "num_tokens": 113133242.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.6328038454055787,
+      "epoch": 8.075370121130552,
+      "grad_norm": 0.12108061462640762,
+      "learning_rate": 9.804435971356025e-06,
+      "loss": 0.0013736591674387455,
+      "mean_token_accuracy": 0.9998795092105865,
+      "num_tokens": 113321861.0,
+      "step": 6000
+    },
+    {
+      "entropy": 0.6300597488880157,
+      "epoch": 8.088829071332436,
+      "grad_norm": 0.2497449666261673,
+      "learning_rate": 9.672469912072956e-06,
+      "loss": 0.0014812872745096684,
+      "mean_token_accuracy": 0.9998218238353729,
+      "num_tokens": 113510984.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.6325613379478454,
+      "epoch": 8.10228802153432,
+      "grad_norm": 0.2513059973716736,
+      "learning_rate": 9.541302835431681e-06,
+      "loss": 0.0014441429637372493,
+      "mean_token_accuracy": 0.9998395144939423,
+      "num_tokens": 113699794.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.6325579524040222,
+      "epoch": 8.115746971736204,
+      "grad_norm": 0.15703566372394562,
+      "learning_rate": 9.410937340158599e-06,
+      "loss": 0.0015422176569700242,
+      "mean_token_accuracy": 0.9997781276702881,
+      "num_tokens": 113888200.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.6323439538478851,
+      "epoch": 8.12920592193809,
+      "grad_norm": 0.18420128524303436,
+      "learning_rate": 9.28137600909892e-06,
+      "loss": 0.0013185075484216213,
+      "mean_token_accuracy": 0.9998004496097564,
+      "num_tokens": 114077149.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.6302162289619446,
+      "epoch": 8.142664872139973,
+      "grad_norm": 0.18233658373355865,
+      "learning_rate": 9.15262140916548e-06,
+      "loss": 0.001340259239077568,
+      "mean_token_accuracy": 0.9998397171497345,
+      "num_tokens": 114266569.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.6314728736877442,
+      "epoch": 8.156123822341858,
+      "grad_norm": 0.34443631768226624,
+      "learning_rate": 9.024676091287881e-06,
+      "loss": 0.0015510787256062032,
+      "mean_token_accuracy": 0.9997199475765228,
+      "num_tokens": 114455310.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.6306299686431884,
+      "epoch": 8.169582772543741,
+      "grad_norm": 0.23195648193359375,
+      "learning_rate": 8.897542590361963e-06,
+      "loss": 0.001497127115726471,
+      "mean_token_accuracy": 0.999662047624588,
+      "num_tokens": 114644329.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.6310850799083709,
+      "epoch": 8.183041722745626,
+      "grad_norm": 0.32436603307724,
+      "learning_rate": 8.77122342519956e-06,
+      "loss": 0.0013079997152090073,
+      "mean_token_accuracy": 0.9997977912425995,
+      "num_tokens": 114833260.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.6300573408603668,
+      "epoch": 8.19650067294751,
+      "grad_norm": 0.21875229477882385,
+      "learning_rate": 8.64572109847862e-06,
+      "loss": 0.001645646058022976,
+      "mean_token_accuracy": 0.9996179223060608,
+      "num_tokens": 115022065.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.6323310673236847,
+      "epoch": 8.209959623149395,
+      "grad_norm": 0.27595558762550354,
+      "learning_rate": 8.521038096693635e-06,
+      "loss": 0.0014854520559310913,
+      "mean_token_accuracy": 0.9997394740581512,
+      "num_tokens": 115210794.0,
+      "step": 6100
+    },
+    {
+      "entropy": 0.631768387556076,
+      "epoch": 8.223418573351278,
+      "grad_norm": 0.2772620618343353,
+      "learning_rate": 8.397176890106317e-06,
+      "loss": 0.0014293460175395013,
+      "mean_token_accuracy": 0.9999000608921051,
+      "num_tokens": 115399823.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.6337137937545776,
+      "epoch": 8.236877523553163,
+      "grad_norm": 0.15117399394512177,
+      "learning_rate": 8.274139932696717e-06,
+      "loss": 0.001172219216823578,
+      "mean_token_accuracy": 0.9999594986438751,
+      "num_tokens": 115588301.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.6334636569023132,
+      "epoch": 8.250336473755047,
+      "grad_norm": 0.22133320569992065,
+      "learning_rate": 8.151929662114566e-06,
+      "loss": 0.0014448422007262707,
+      "mean_token_accuracy": 0.9997801542282104,
+      "num_tokens": 115776977.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.6315647959709167,
+      "epoch": 8.263795423956932,
+      "grad_norm": 0.23076151311397552,
+      "learning_rate": 8.030548499631014e-06,
+      "loss": 0.0013002446852624417,
+      "mean_token_accuracy": 0.9998793780803681,
+      "num_tokens": 115966052.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.6293221652507782,
+      "epoch": 8.277254374158815,
+      "grad_norm": 0.2104390561580658,
+      "learning_rate": 7.909998850090628e-06,
+      "loss": 0.0014004193246364594,
+      "mean_token_accuracy": 0.9997619271278382,
+      "num_tokens": 116155401.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.629232895374298,
+      "epoch": 8.2907133243607,
+      "grad_norm": 0.24602340161800385,
+      "learning_rate": 7.790283101863771e-06,
+      "loss": 0.0013899436220526696,
+      "mean_token_accuracy": 0.9998601615428925,
+      "num_tokens": 116344618.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.628896701335907,
+      "epoch": 8.304172274562584,
+      "grad_norm": 0.36142218112945557,
+      "learning_rate": 7.671403626799262e-06,
+      "loss": 0.001389096863567829,
+      "mean_token_accuracy": 0.9997016251087188,
+      "num_tokens": 116533953.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.6315865159034729,
+      "epoch": 8.317631224764469,
+      "grad_norm": 0.15314172208309174,
+      "learning_rate": 7.553362780177398e-06,
+      "loss": 0.0014957386068999768,
+      "mean_token_accuracy": 0.9997191250324249,
+      "num_tokens": 116722247.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.6303807675838471,
+      "epoch": 8.331090174966352,
+      "grad_norm": 0.16988365352153778,
+      "learning_rate": 7.43616290066329e-06,
+      "loss": 0.0012819443829357624,
+      "mean_token_accuracy": 0.9998224377632141,
+      "num_tokens": 116911239.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.6324020445346832,
+      "epoch": 8.344549125168237,
+      "grad_norm": 0.214530348777771,
+      "learning_rate": 7.3198063102605185e-06,
+      "loss": 0.0013960381969809533,
+      "mean_token_accuracy": 0.9998191654682159,
+      "num_tokens": 117099644.0,
+      "step": 6200
+    },
+    {
+      "entropy": 0.6305644690990448,
+      "epoch": 8.35800807537012,
+      "grad_norm": 0.6130462288856506,
+      "learning_rate": 7.204295314265136e-06,
+      "loss": 0.001624145545065403,
+      "mean_token_accuracy": 0.999723345041275,
+      "num_tokens": 117288440.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.6282017886638641,
+      "epoch": 8.371467025572006,
+      "grad_norm": 0.33955836296081543,
+      "learning_rate": 7.089632201219997e-06,
+      "loss": 0.0014544738456606864,
+      "mean_token_accuracy": 0.9996826589107514,
+      "num_tokens": 117478031.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.6303719699382782,
+      "epoch": 8.38492597577389,
+      "grad_norm": 0.21833766996860504,
+      "learning_rate": 6.975819242869414e-06,
+      "loss": 0.001375422067940235,
+      "mean_token_accuracy": 0.9996990442276001,
+      "num_tokens": 117666823.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.6290243804454804,
+      "epoch": 8.398384925975774,
+      "grad_norm": 0.18415814638137817,
+      "learning_rate": 6.862858694114138e-06,
+      "loss": 0.0013824976980686189,
+      "mean_token_accuracy": 0.9998021006584168,
+      "num_tokens": 117855959.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.6298940062522889,
+      "epoch": 8.411843876177658,
+      "grad_norm": 0.38978856801986694,
+      "learning_rate": 6.750752792966702e-06,
+      "loss": 0.001461424771696329,
+      "mean_token_accuracy": 0.9998207986354828,
+      "num_tokens": 118044550.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.6267169773578644,
+      "epoch": 8.425302826379543,
+      "grad_norm": 0.2565508186817169,
+      "learning_rate": 6.639503760507065e-06,
+      "loss": 0.0014107320457696915,
+      "mean_token_accuracy": 0.9998013794422149,
+      "num_tokens": 118233720.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.6296021044254303,
+      "epoch": 8.438761776581426,
+      "grad_norm": 0.14881721138954163,
+      "learning_rate": 6.529113800838632e-06,
+      "loss": 0.0013083923608064652,
+      "mean_token_accuracy": 0.9998387336730957,
+      "num_tokens": 118422118.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.6294852197170258,
+      "epoch": 8.452220726783311,
+      "grad_norm": 0.22628630697727203,
+      "learning_rate": 6.41958510104454e-06,
+      "loss": 0.0013258070684969425,
+      "mean_token_accuracy": 0.9998400807380676,
+      "num_tokens": 118610757.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.6305896222591401,
+      "epoch": 8.465679676985195,
+      "grad_norm": 0.3069230020046234,
+      "learning_rate": 6.31091983114438e-06,
+      "loss": 0.0014844583347439766,
+      "mean_token_accuracy": 0.999760502576828,
+      "num_tokens": 118799460.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.6299797236919403,
+      "epoch": 8.47913862718708,
+      "grad_norm": 0.2504591941833496,
+      "learning_rate": 6.203120144051161e-06,
+      "loss": 0.0013100499287247657,
+      "mean_token_accuracy": 0.9998400688171387,
+      "num_tokens": 118988395.0,
+      "step": 6300
+    },
+    {
+      "entropy": 0.6294752597808838,
+      "epoch": 8.492597577388963,
+      "grad_norm": 0.21706247329711914,
+      "learning_rate": 6.0961881755286775e-06,
+      "loss": 0.0014089202508330346,
+      "mean_token_accuracy": 0.9998202979564667,
+      "num_tokens": 119177293.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.630080658197403,
+      "epoch": 8.506056527590848,
+      "grad_norm": 0.24263928830623627,
+      "learning_rate": 5.990126044149203e-06,
+      "loss": 0.0013817671686410903,
+      "mean_token_accuracy": 0.9997431695461273,
+      "num_tokens": 119366224.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.6312020123004913,
+      "epoch": 8.519515477792732,
+      "grad_norm": 0.2104533314704895,
+      "learning_rate": 5.884935851251483e-06,
+      "loss": 0.0014581452123820782,
+      "mean_token_accuracy": 0.9996995806694031,
+      "num_tokens": 119554641.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.6307013928890228,
+      "epoch": 8.532974427994617,
+      "grad_norm": 0.29689937829971313,
+      "learning_rate": 5.780619680899146e-06,
+      "loss": 0.0015367996878921986,
+      "mean_token_accuracy": 0.9997993230819702,
+      "num_tokens": 119743217.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.6292266964912414,
+      "epoch": 8.5464333781965,
+      "grad_norm": 0.20508502423763275,
+      "learning_rate": 5.677179599839366e-06,
+      "loss": 0.0012161459773778916,
+      "mean_token_accuracy": 0.9998607873916626,
+      "num_tokens": 119932436.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.6305240333080292,
+      "epoch": 8.559892328398385,
+      "grad_norm": 0.22822189331054688,
+      "learning_rate": 5.574617657461967e-06,
+      "loss": 0.0013693704269826413,
+      "mean_token_accuracy": 0.9998015820980072,
+      "num_tokens": 120121506.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.6288493514060974,
+      "epoch": 8.573351278600269,
+      "grad_norm": 0.15999595820903778,
+      "learning_rate": 5.472935885758767e-06,
+      "loss": 0.0011829210445284844,
+      "mean_token_accuracy": 0.9998811721801758,
+      "num_tokens": 120310792.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.6302126765251159,
+      "epoch": 8.586810228802154,
+      "grad_norm": 0.48678338527679443,
+      "learning_rate": 5.372136299283365e-06,
+      "loss": 0.0014185780659317971,
+      "mean_token_accuracy": 0.999719786643982,
+      "num_tokens": 120499610.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.6311411023139953,
+      "epoch": 8.600269179004037,
+      "grad_norm": 0.10369251668453217,
+      "learning_rate": 5.272220895111196e-06,
+      "loss": 0.0013072849251329899,
+      "mean_token_accuracy": 0.9997986853122711,
+      "num_tokens": 120688546.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.6289890885353089,
+      "epoch": 8.613728129205922,
+      "grad_norm": 0.18738436698913574,
+      "learning_rate": 5.173191652799986e-06,
+      "loss": 0.0012681365013122558,
+      "mean_token_accuracy": 0.9997800529003144,
+      "num_tokens": 120877609.0,
+      "step": 6400
+    },
+    {
+      "entropy": 0.6279601216316223,
+      "epoch": 8.627187079407806,
+      "grad_norm": 0.19525793194770813,
+      "learning_rate": 5.075050534350512e-06,
+      "loss": 0.0012957751750946045,
+      "mean_token_accuracy": 0.9998006701469422,
+      "num_tokens": 121066294.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.6293328821659088,
+      "epoch": 8.640646029609691,
+      "grad_norm": 0.184369757771492,
+      "learning_rate": 4.977799484167767e-06,
+      "loss": 0.0012527795508503914,
+      "mean_token_accuracy": 0.999919193983078,
+      "num_tokens": 121255109.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.629350233078003,
+      "epoch": 8.654104979811574,
+      "grad_norm": 0.3674270510673523,
+      "learning_rate": 4.8814404290223655e-06,
+      "loss": 0.001326523907482624,
+      "mean_token_accuracy": 0.9998202204704285,
+      "num_tokens": 121443604.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.6280995666980743,
+      "epoch": 8.66756393001346,
+      "grad_norm": 0.19100305438041687,
+      "learning_rate": 4.7859752780124555e-06,
+      "loss": 0.0011802183464169503,
+      "mean_token_accuracy": 0.9998997509479522,
+      "num_tokens": 121632712.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.6302553176879883,
+      "epoch": 8.681022880215343,
+      "grad_norm": 0.21766118705272675,
+      "learning_rate": 4.691405922525838e-06,
+      "loss": 0.00140056349337101,
+      "mean_token_accuracy": 0.9998192071914673,
+      "num_tokens": 121821234.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.628807520866394,
+      "epoch": 8.694481830417228,
+      "grad_norm": 0.30580392479896545,
+      "learning_rate": 4.597734236202511e-06,
+      "loss": 0.00123771820217371,
+      "mean_token_accuracy": 0.9998995006084442,
+      "num_tokens": 122009998.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.6287366271018981,
+      "epoch": 8.707940780619111,
+      "grad_norm": 0.14816957712173462,
+      "learning_rate": 4.504962074897556e-06,
+      "loss": 0.0014824405312538146,
+      "mean_token_accuracy": 0.999697607755661,
+      "num_tokens": 122198378.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.6262040913105011,
+      "epoch": 8.721399730820997,
+      "grad_norm": 0.17951056361198425,
+      "learning_rate": 4.413091276644349e-06,
+      "loss": 0.00132729671895504,
+      "mean_token_accuracy": 0.9998216688632965,
+      "num_tokens": 122387614.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.6280441462993622,
+      "epoch": 8.73485868102288,
+      "grad_norm": 0.1389010101556778,
+      "learning_rate": 4.322123661618166e-06,
+      "loss": 0.001298490446060896,
+      "mean_token_accuracy": 0.999879801273346,
+      "num_tokens": 122576224.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.6271527647972107,
+      "epoch": 8.748317631224765,
+      "grad_norm": 0.24173741042613983,
+      "learning_rate": 4.23206103210011e-06,
+      "loss": 0.0011219406500458718,
+      "mean_token_accuracy": 0.999880450963974,
+      "num_tokens": 122765123.0,
+      "step": 6500
+    },
+    {
+      "entropy": 0.6269227385520935,
+      "epoch": 8.761776581426648,
+      "grad_norm": 0.22519297897815704,
+      "learning_rate": 4.142905172441408e-06,
+      "loss": 0.0012161782942712307,
+      "mean_token_accuracy": 0.9998603820800781,
+      "num_tokens": 122954344.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.6288948476314544,
+      "epoch": 8.775235531628534,
+      "grad_norm": 0.21760156750679016,
+      "learning_rate": 4.054657849028043e-06,
+      "loss": 0.001162637397646904,
+      "mean_token_accuracy": 0.9999193131923676,
+      "num_tokens": 123142872.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.6281030297279357,
+      "epoch": 8.788694481830417,
+      "grad_norm": 0.3592071831226349,
+      "learning_rate": 3.9673208102457895e-06,
+      "loss": 0.0013433885760605335,
+      "mean_token_accuracy": 0.9997989416122437,
+      "num_tokens": 123331858.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.6263146758079529,
+      "epoch": 8.802153432032302,
+      "grad_norm": 0.3514918386936188,
+      "learning_rate": 3.880895786445543e-06,
+      "loss": 0.0014502016827464103,
+      "mean_token_accuracy": 0.9997799456119537,
+      "num_tokens": 123521321.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.6269251763820648,
+      "epoch": 8.815612382234185,
+      "grad_norm": 0.22137221693992615,
+      "learning_rate": 3.795384489909054e-06,
+      "loss": 0.001280687004327774,
+      "mean_token_accuracy": 0.9998396694660187,
+      "num_tokens": 123710572.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.6270365953445435,
+      "epoch": 8.82907133243607,
+      "grad_norm": 0.36301663517951965,
+      "learning_rate": 3.7107886148150084e-06,
+      "loss": 0.0012877551838755608,
+      "mean_token_accuracy": 0.9998403787612915,
+      "num_tokens": 123899764.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.6268510282039642,
+      "epoch": 8.842530282637954,
+      "grad_norm": 0.19353987276554108,
+      "learning_rate": 3.6271098372054334e-06,
+      "loss": 0.00124650401994586,
+      "mean_token_accuracy": 0.9998400628566741,
+      "num_tokens": 124088703.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.6260306179523468,
+      "epoch": 8.855989232839839,
+      "grad_norm": 0.11702294647693634,
+      "learning_rate": 3.5443498149525357e-06,
+      "loss": 0.0011134844273328782,
+      "mean_token_accuracy": 0.9998809397220612,
+      "num_tokens": 124277949.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.6265114545822144,
+      "epoch": 8.869448183041722,
+      "grad_norm": 0.24068744480609894,
+      "learning_rate": 3.462510187725809e-06,
+      "loss": 0.0013966705650091172,
+      "mean_token_accuracy": 0.9997197926044464,
+      "num_tokens": 124466937.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.6271577537059784,
+      "epoch": 8.882907133243608,
+      "grad_norm": 0.1259060949087143,
+      "learning_rate": 3.3815925769595745e-06,
+      "loss": 0.0013276103883981705,
+      "mean_token_accuracy": 0.9997398614883423,
+      "num_tokens": 124655536.0,
+      "step": 6600
+    },
+    {
+      "entropy": 0.626012533903122,
+      "epoch": 8.896366083445491,
+      "grad_norm": 0.21367371082305908,
+      "learning_rate": 3.30159858582087e-06,
+      "loss": 0.0012806886807084084,
+      "mean_token_accuracy": 0.9998196005821228,
+      "num_tokens": 124844587.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.6261564731597901,
+      "epoch": 8.909825033647376,
+      "grad_norm": 0.2548461854457855,
+      "learning_rate": 3.2225297991776536e-06,
+      "loss": 0.0012241953983902932,
+      "mean_token_accuracy": 0.9998611867427826,
+      "num_tokens": 125033688.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.6268511712551117,
+      "epoch": 8.92328398384926,
+      "grad_norm": 0.23393066227436066,
+      "learning_rate": 3.1443877835674208e-06,
+      "loss": 0.0010460015386343003,
+      "mean_token_accuracy": 0.9998401999473572,
+      "num_tokens": 125222397.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.6260151982307434,
+      "epoch": 8.936742934051145,
+      "grad_norm": 0.09390900284051895,
+      "learning_rate": 3.0671740871661635e-06,
+      "loss": 0.0011951563879847526,
+      "mean_token_accuracy": 0.9998811960220337,
+      "num_tokens": 125411589.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.625042176246643,
+      "epoch": 8.950201884253028,
+      "grad_norm": 0.1279630959033966,
+      "learning_rate": 2.9908902397577043e-06,
+      "loss": 0.001325780339539051,
+      "mean_token_accuracy": 0.9998415470123291,
+      "num_tokens": 125600890.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.6269629180431366,
+      "epoch": 8.963660834454913,
+      "grad_norm": 0.16514816880226135,
+      "learning_rate": 2.91553775270339e-06,
+      "loss": 0.0012059432454407215,
+      "mean_token_accuracy": 0.9999194324016571,
+      "num_tokens": 125789165.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.6265843152999878,
+      "epoch": 8.977119784656796,
+      "grad_norm": 0.16388432681560516,
+      "learning_rate": 2.8411181189121284e-06,
+      "loss": 0.001122188288718462,
+      "mean_token_accuracy": 0.9998812198638916,
+      "num_tokens": 125977920.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.6274819135665893,
+      "epoch": 8.990578734858682,
+      "grad_norm": 0.11002290993928909,
+      "learning_rate": 2.767632812810839e-06,
+      "loss": 0.001199054718017578,
+      "mean_token_accuracy": 0.999840772151947,
+      "num_tokens": 126166297.0,
+      "step": 6680
+    },
+    {
+      "epoch": 9.0,
+      "eval_entropy": 0.63310007494726,
+      "eval_loss": 0.10951486974954605,
+      "eval_mean_token_accuracy": 0.9730027690055264,
+      "eval_num_tokens": 126297929.0,
+      "eval_runtime": 12.7449,
+      "eval_samples_per_second": 392.315,
+      "eval_steps_per_second": 12.319,
+      "step": 6687
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.837766778733724e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-6687/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-6687/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-743/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-743/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-743/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-743/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-743/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-743/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-743/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-743/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..6a8673e9f6877db9181be660c3dc29250891a441
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-743/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a96cc7375568827f54f0dc8479fa1bcac7ad371a86ec54625ce468aa29d4b96
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-743/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-743/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-743/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-743/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-743/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-743/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-743/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-743/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-743/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-743/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-743/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..15681d0cd09e531d5704e1b0528b9892eef2e0f9
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-743/trainer_state.json
@@ -0,0 +1,785 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 743,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.485751027235553e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-743/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-743/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-743/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-7430/README.md b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-7430/adapter_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..56c11616c4a6e695288dce8b4b53766b984780eb
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-7430/adapter_model.safetensors b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..6a9068a7e2617c3749cdb1285f06e7b462156e73
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f727d104436a5e9ef43e5ce5832d6e6e30e45297cb19bd4edb53754ec57373df
+size 80792096
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-7430/chat_template.jinja b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-7430/tokenizer.json b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-7430/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-7430/trainer_state.json b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f7ef2c58ac1466de5bde84265b016c1f813fb970
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/trainer_state.json
@@ -0,0 +1,7574 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 7430,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.262857687473297,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.8653019666671753,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.4183002471923828,
+      "mean_token_accuracy": 0.8609529256820678,
+      "num_tokens": 188766.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.2641610026359558,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.8475091457366943,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.4031085014343262,
+      "mean_token_accuracy": 0.8657449543476105,
+      "num_tokens": 377824.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2695851802825928,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 1.252985954284668,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.34129221439361573,
+      "mean_token_accuracy": 0.8839877903461456,
+      "num_tokens": 566046.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2462505459785462,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 0.9305229783058167,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.2515744209289551,
+      "mean_token_accuracy": 0.9147139430046082,
+      "num_tokens": 755000.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2354708194732666,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.5796370506286621,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2124643087387085,
+      "mean_token_accuracy": 0.9246431946754455,
+      "num_tokens": 944080.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2379411935806275,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.6359605193138123,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.19829050302505494,
+      "mean_token_accuracy": 0.9282503485679626,
+      "num_tokens": 1132745.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2273941040039062,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.6423936486244202,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1811063766479492,
+      "mean_token_accuracy": 0.9327361822128296,
+      "num_tokens": 1321873.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2149020314216614,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 0.6821616888046265,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.1640337347984314,
+      "mean_token_accuracy": 0.9373818874359131,
+      "num_tokens": 1511241.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2144461631774903,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 0.6947558522224426,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.15713095664978027,
+      "mean_token_accuracy": 0.9404333114624024,
+      "num_tokens": 1700078.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.210549235343933,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 0.7010672688484192,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.14957726001739502,
+      "mean_token_accuracy": 0.9445081353187561,
+      "num_tokens": 1888891.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.203629207611084,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 0.8623549938201904,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.14619266986846924,
+      "mean_token_accuracy": 0.9446018993854522,
+      "num_tokens": 2077874.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2045462131500244,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 0.7523015141487122,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.13490312099456786,
+      "mean_token_accuracy": 0.9499653518199921,
+      "num_tokens": 2267014.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1907317399978639,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 0.957962691783905,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.13256092071533204,
+      "mean_token_accuracy": 0.9484152019023895,
+      "num_tokens": 2455750.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.1792213916778564,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.7358795404434204,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.12550873756408693,
+      "mean_token_accuracy": 0.9506115138530731,
+      "num_tokens": 2644556.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.1749429821968078,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.07390558719635,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.12108533382415772,
+      "mean_token_accuracy": 0.9525261521339417,
+      "num_tokens": 2832789.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.154041826725006,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.256065845489502,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.11878299713134766,
+      "mean_token_accuracy": 0.9539411306381226,
+      "num_tokens": 3021802.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.1430203318595886,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 0.8330121636390686,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.11740827560424805,
+      "mean_token_accuracy": 0.9544454514980316,
+      "num_tokens": 3210479.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1300233721733093,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.1392722129821777,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.10987560749053955,
+      "mean_token_accuracy": 0.9574419140815735,
+      "num_tokens": 3399701.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.1173638939857482,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.0542868375778198,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.10590486526489258,
+      "mean_token_accuracy": 0.9591289818286896,
+      "num_tokens": 3588258.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1004456758499146,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.1485332250595093,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.10361509323120117,
+      "mean_token_accuracy": 0.9592108130455017,
+      "num_tokens": 3776978.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0869829535484314,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.0169382095336914,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.10033822059631348,
+      "mean_token_accuracy": 0.9602985560894013,
+      "num_tokens": 3965629.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0619446635246277,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.1220632791519165,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.09786752462387086,
+      "mean_token_accuracy": 0.9610431730747223,
+      "num_tokens": 4154775.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.0505661725997926,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 0.8749313354492188,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0965538501739502,
+      "mean_token_accuracy": 0.9614811658859252,
+      "num_tokens": 4343835.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0416118502616882,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.3980506658554077,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.10011651515960693,
+      "mean_token_accuracy": 0.9602194666862488,
+      "num_tokens": 4532640.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0293668389320374,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.285994529724121,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.09453256726264954,
+      "mean_token_accuracy": 0.9619364976882935,
+      "num_tokens": 4721431.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.0149331629276275,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 0.8904402256011963,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.09606351852416992,
+      "mean_token_accuracy": 0.9622808158397674,
+      "num_tokens": 4910540.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9930494785308838,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.159342646598816,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.09172534346580505,
+      "mean_token_accuracy": 0.9637623131275177,
+      "num_tokens": 5099310.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9804227173328399,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.2235705852508545,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.09083572626113892,
+      "mean_token_accuracy": 0.9640021800994873,
+      "num_tokens": 5288311.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9682882308959961,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.1984940767288208,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.0892288088798523,
+      "mean_token_accuracy": 0.9648444414138794,
+      "num_tokens": 5476810.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9664222657680511,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.029746413230896,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.0873763084411621,
+      "mean_token_accuracy": 0.9654866337776185,
+      "num_tokens": 5665659.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9546509027481079,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 0.8723888993263245,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.08508302569389344,
+      "mean_token_accuracy": 0.9655120432376861,
+      "num_tokens": 5854756.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9635786533355712,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.0173970460891724,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.08573030829429626,
+      "mean_token_accuracy": 0.9661867380142212,
+      "num_tokens": 6043569.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9461539149284363,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 0.9700696468353271,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.0906404435634613,
+      "mean_token_accuracy": 0.9645311295986175,
+      "num_tokens": 6232193.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9317194223403931,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.3392207622528076,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.08504368662834168,
+      "mean_token_accuracy": 0.9667904317378998,
+      "num_tokens": 6421576.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9278222620487213,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.1564866304397583,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.08676948547363281,
+      "mean_token_accuracy": 0.9657181918621063,
+      "num_tokens": 6610060.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.9280106008052826,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.133396863937378,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.08761489391326904,
+      "mean_token_accuracy": 0.9646157205104828,
+      "num_tokens": 6798161.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9222531199455262,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 0.8363345861434937,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.08614999055862427,
+      "mean_token_accuracy": 0.9664735674858094,
+      "num_tokens": 6987325.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9103430569171905,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2359459400177002,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.08534908294677734,
+      "mean_token_accuracy": 0.9659685432910919,
+      "num_tokens": 7176275.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9022238194942475,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 0.9123634696006775,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.08124258518218994,
+      "mean_token_accuracy": 0.9670641124248505,
+      "num_tokens": 7365931.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.9111613273620606,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.0706427097320557,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.08465015888214111,
+      "mean_token_accuracy": 0.9668884217739105,
+      "num_tokens": 7555216.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.8935062766075135,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.139824628829956,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.0857190489768982,
+      "mean_token_accuracy": 0.9656417548656464,
+      "num_tokens": 7744043.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8897038161754608,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 0.9223172068595886,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.0820292055606842,
+      "mean_token_accuracy": 0.9668228268623352,
+      "num_tokens": 7932739.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9031975388526916,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 0.9153602123260498,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.08561227321624756,
+      "mean_token_accuracy": 0.9665785551071167,
+      "num_tokens": 8121534.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8915951550006866,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.0522831678390503,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.08352552056312561,
+      "mean_token_accuracy": 0.9667543470859528,
+      "num_tokens": 8310333.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.88769411444664,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.2566636800765991,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.08170981407165527,
+      "mean_token_accuracy": 0.9679584920406341,
+      "num_tokens": 8499079.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.8826716780662537,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 0.7404268980026245,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.07975507378578187,
+      "mean_token_accuracy": 0.9676892876625061,
+      "num_tokens": 8687847.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8825231611728668,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.0333958864212036,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.08148443698883057,
+      "mean_token_accuracy": 0.967664510011673,
+      "num_tokens": 8877539.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8867287635803223,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 0.8842368125915527,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.08020110130310058,
+      "mean_token_accuracy": 0.9680711150169372,
+      "num_tokens": 9066253.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8916994333267212,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.1372274160385132,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.07954617738723754,
+      "mean_token_accuracy": 0.9682127356529235,
+      "num_tokens": 9255445.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8834196865558624,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 0.9504761099815369,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.08058012723922729,
+      "mean_token_accuracy": 0.9679273366928101,
+      "num_tokens": 9443806.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8739849209785462,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.2260017395019531,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.0798442006111145,
+      "mean_token_accuracy": 0.9688306748867035,
+      "num_tokens": 9632985.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8745266020298004,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.4060558080673218,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.07968794107437134,
+      "mean_token_accuracy": 0.9677379071712494,
+      "num_tokens": 9822040.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8789939701557159,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.0593465566635132,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.07835215330123901,
+      "mean_token_accuracy": 0.9681119978427887,
+      "num_tokens": 10010549.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8595301926136016,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 0.902250349521637,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.07510886788368225,
+      "mean_token_accuracy": 0.9700966715812683,
+      "num_tokens": 10199633.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8628671288490295,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.3397527933120728,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.07530273795127869,
+      "mean_token_accuracy": 0.969423794746399,
+      "num_tokens": 10388522.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.856841766834259,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.1785457134246826,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.07783904671669006,
+      "mean_token_accuracy": 0.9692493975162506,
+      "num_tokens": 10577336.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8589329421520233,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 0.8644716739654541,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.07657524943351746,
+      "mean_token_accuracy": 0.9696632564067841,
+      "num_tokens": 10766306.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8540739715099335,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.343432903289795,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.07348920702934265,
+      "mean_token_accuracy": 0.9698194086551666,
+      "num_tokens": 10954999.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8668401718139649,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 0.8398891091346741,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.07891384363174439,
+      "mean_token_accuracy": 0.9690254986286163,
+      "num_tokens": 11143585.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8587459862232208,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 0.894865870475769,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.07987629771232604,
+      "mean_token_accuracy": 0.9679262936115265,
+      "num_tokens": 11332680.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8630038559436798,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 0.9651695489883423,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.0761431872844696,
+      "mean_token_accuracy": 0.969614040851593,
+      "num_tokens": 11520976.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8410441219806671,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.0581592321395874,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.07258764505386353,
+      "mean_token_accuracy": 0.9708519279956818,
+      "num_tokens": 11709601.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8466249167919159,
+      "epoch": 0.847913862718708,
+      "grad_norm": 0.957252025604248,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.07713524103164673,
+      "mean_token_accuracy": 0.9694346964359284,
+      "num_tokens": 11898013.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8469240963459015,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.641280174255371,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.07863805890083313,
+      "mean_token_accuracy": 0.9682413160800933,
+      "num_tokens": 12086913.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8583328485488891,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.249804139137268,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.08108658194541932,
+      "mean_token_accuracy": 0.9681722402572632,
+      "num_tokens": 12275906.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.849593561887741,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.0288732051849365,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.0736888587474823,
+      "mean_token_accuracy": 0.9697738111019134,
+      "num_tokens": 12465250.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8390313148498535,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.4289873838424683,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.07528446912765503,
+      "mean_token_accuracy": 0.9698788702487946,
+      "num_tokens": 12654691.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8532404243946076,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.0830386877059937,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.07614066004753113,
+      "mean_token_accuracy": 0.9693071842193604,
+      "num_tokens": 12843517.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8432701885700226,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.2883894443511963,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.07694239020347596,
+      "mean_token_accuracy": 0.9688229441642762,
+      "num_tokens": 13032637.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8445841610431671,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.0825793743133545,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.07297058701515198,
+      "mean_token_accuracy": 0.9707355678081513,
+      "num_tokens": 13221747.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.844417005777359,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.0527976751327515,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.07573122978210449,
+      "mean_token_accuracy": 0.9695912897586823,
+      "num_tokens": 13410634.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8467911183834076,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 0.970084011554718,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.0760277509689331,
+      "mean_token_accuracy": 0.9699971795082092,
+      "num_tokens": 13599297.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8429471254348755,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 0.8835607171058655,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.07751307487487794,
+      "mean_token_accuracy": 0.9680874586105347,
+      "num_tokens": 13788007.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8457451581954956,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.0810054540634155,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.07772084474563598,
+      "mean_token_accuracy": 0.9694566845893859,
+      "num_tokens": 13976478.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 0.8389141521636089,
+      "eval_loss": 0.07403359562158585,
+      "eval_mean_token_accuracy": 0.9703246442375669,
+      "eval_num_tokens": 14033155.0,
+      "eval_runtime": 13.1584,
+      "eval_samples_per_second": 379.987,
+      "eval_steps_per_second": 11.932,
+      "step": 743
+    },
+    {
+      "entropy": 0.8332668900489807,
+      "epoch": 1.009421265141319,
+      "grad_norm": 0.9914855360984802,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.06607928276062011,
+      "mean_token_accuracy": 0.974932211637497,
+      "num_tokens": 14165567.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8149146616458893,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.3080861568450928,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.057932257652282715,
+      "mean_token_accuracy": 0.9770712852478027,
+      "num_tokens": 14353912.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.8197620987892151,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.0898637771606445,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.06010057330131531,
+      "mean_token_accuracy": 0.9758293390274048,
+      "num_tokens": 14542288.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8183245182037353,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.3120660781860352,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.0610861599445343,
+      "mean_token_accuracy": 0.9764986515045166,
+      "num_tokens": 14731028.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8135185301303863,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.138473391532898,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.06268800497055053,
+      "mean_token_accuracy": 0.9752912580966949,
+      "num_tokens": 14919682.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8142645537853241,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.1258258819580078,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.06271481513977051,
+      "mean_token_accuracy": 0.9757774949073792,
+      "num_tokens": 15107904.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8170483231544494,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.007297396659851,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.06338523030281067,
+      "mean_token_accuracy": 0.9754154622554779,
+      "num_tokens": 15297115.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8201254069805145,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 0.8240306973457336,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.06355815529823303,
+      "mean_token_accuracy": 0.9752808332443237,
+      "num_tokens": 15485921.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8140238881111145,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.0350368022918701,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.05969444513320923,
+      "mean_token_accuracy": 0.977089112997055,
+      "num_tokens": 15674824.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8133325934410095,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.0544878244400024,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.06062774658203125,
+      "mean_token_accuracy": 0.976290100812912,
+      "num_tokens": 15863526.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8185161769390106,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.1076316833496094,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.061567097902297974,
+      "mean_token_accuracy": 0.9755526125431061,
+      "num_tokens": 16051685.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8176139056682586,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.0057616233825684,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.06227490901947021,
+      "mean_token_accuracy": 0.9758114695549012,
+      "num_tokens": 16240451.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8217637002468109,
+      "epoch": 1.17092866756393,
+      "grad_norm": 0.97276771068573,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.06410331726074218,
+      "mean_token_accuracy": 0.9754030883312226,
+      "num_tokens": 16428961.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8141795754432678,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.0135555267333984,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.06253207921981811,
+      "mean_token_accuracy": 0.9760490536689759,
+      "num_tokens": 16617327.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8167653739452362,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2272467613220215,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.060455435514450075,
+      "mean_token_accuracy": 0.9758628964424133,
+      "num_tokens": 16806414.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8155323445796967,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 0.9328635334968567,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.06302794814109802,
+      "mean_token_accuracy": 0.9745101630687714,
+      "num_tokens": 16995016.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.818474805355072,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.1651655435562134,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.06043210029602051,
+      "mean_token_accuracy": 0.9766487181186676,
+      "num_tokens": 17184250.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8140560865402222,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.1005650758743286,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.06131819486618042,
+      "mean_token_accuracy": 0.9757712185382843,
+      "num_tokens": 17373075.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8204843640327454,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.0825532674789429,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.06296311616897583,
+      "mean_token_accuracy": 0.9745951414108276,
+      "num_tokens": 17561454.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8237834334373474,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 0.9842743873596191,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.061483919620513916,
+      "mean_token_accuracy": 0.9754831731319428,
+      "num_tokens": 17750758.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8165915131568908,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.0179396867752075,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.06237881183624268,
+      "mean_token_accuracy": 0.9750245690345765,
+      "num_tokens": 17940126.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8329708755016327,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.0732359886169434,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.0644665539264679,
+      "mean_token_accuracy": 0.9743888020515442,
+      "num_tokens": 18129000.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8210074126720428,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.3696036338806152,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.06320792436599731,
+      "mean_token_accuracy": 0.9751390159130097,
+      "num_tokens": 18318232.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8248398661613464,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.327525019645691,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.06277769804000854,
+      "mean_token_accuracy": 0.9752413153648376,
+      "num_tokens": 18507403.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8284746050834656,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.044022560119629,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.06162315607070923,
+      "mean_token_accuracy": 0.9760516047477722,
+      "num_tokens": 18696242.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8233858644962311,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.062048077583313,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.059620130062103274,
+      "mean_token_accuracy": 0.9755307257175445,
+      "num_tokens": 18885199.0,
+      "step": 1000
+    },
+    {
+      "entropy": 0.8236705601215363,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.2024412155151367,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.06304935216903687,
+      "mean_token_accuracy": 0.975091153383255,
+      "num_tokens": 19074028.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.828409218788147,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.1684212684631348,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.06451238393783569,
+      "mean_token_accuracy": 0.9743404746055603,
+      "num_tokens": 19262295.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.8214354634284973,
+      "epoch": 1.386271870794078,
+      "grad_norm": 0.8512521386146545,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.06042641997337341,
+      "mean_token_accuracy": 0.9755558848381043,
+      "num_tokens": 19451154.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.8223241448402405,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 0.8546568155288696,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.062134528160095216,
+      "mean_token_accuracy": 0.9750089883804322,
+      "num_tokens": 19639827.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.8217969954013824,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 0.972831666469574,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.060803139209747316,
+      "mean_token_accuracy": 0.9760692238807678,
+      "num_tokens": 19828174.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.8144341170787811,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.1216644048690796,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.06458876132965088,
+      "mean_token_accuracy": 0.9738282442092896,
+      "num_tokens": 20016700.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.8168790876865387,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.0821806192398071,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.06258658170700074,
+      "mean_token_accuracy": 0.9749014675617218,
+      "num_tokens": 20205823.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.8146390020847321,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.0067790746688843,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.06509006023406982,
+      "mean_token_accuracy": 0.9737810134887696,
+      "num_tokens": 20394611.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.8224801063537598,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.0583593845367432,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.06343590021133423,
+      "mean_token_accuracy": 0.9750119090080261,
+      "num_tokens": 20583252.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.8083258986473083,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 0.8305535316467285,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.06573111414909363,
+      "mean_token_accuracy": 0.9736428141593934,
+      "num_tokens": 20772107.0,
+      "step": 1100
+    },
+    {
+      "entropy": 0.8081986546516419,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5895942449569702,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.06376024484634399,
+      "mean_token_accuracy": 0.9751200795173645,
+      "num_tokens": 20960494.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.8026683986186981,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 0.9962654113769531,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.06282572746276856,
+      "mean_token_accuracy": 0.9753849327564239,
+      "num_tokens": 21149763.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.8030014991760254,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 0.9083836078643799,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.06473478078842163,
+      "mean_token_accuracy": 0.9747069597244262,
+      "num_tokens": 21338690.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.8008662164211273,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.097593069076538,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.05911685228347778,
+      "mean_token_accuracy": 0.9767139732837677,
+      "num_tokens": 21527420.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7981986284255982,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.1673516035079956,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.05957244038581848,
+      "mean_token_accuracy": 0.9761911332607269,
+      "num_tokens": 21716791.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.8004718601703644,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 0.7955246567726135,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.06378913521766663,
+      "mean_token_accuracy": 0.9747395157814026,
+      "num_tokens": 21905487.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.8039821267127991,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.3141396045684814,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.06209801435470581,
+      "mean_token_accuracy": 0.9754719436168671,
+      "num_tokens": 22094436.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.8051789462566376,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.2227814197540283,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.06261310577392579,
+      "mean_token_accuracy": 0.9753049433231353,
+      "num_tokens": 22283243.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7941994190216064,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.071070671081543,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.05962263345718384,
+      "mean_token_accuracy": 0.9763956189155578,
+      "num_tokens": 22472084.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7937953472137451,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.0329782962799072,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.060611891746521,
+      "mean_token_accuracy": 0.9758659780025483,
+      "num_tokens": 22661614.0,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7913113713264466,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.032772183418274,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.06002308130264282,
+      "mean_token_accuracy": 0.9758916199207306,
+      "num_tokens": 22850634.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.7909658491611481,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3234598636627197,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.06296972036361695,
+      "mean_token_accuracy": 0.9741117715835571,
+      "num_tokens": 23039402.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.7880761086940765,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 0.9563435316085815,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.06209538578987121,
+      "mean_token_accuracy": 0.9750530481338501,
+      "num_tokens": 23228363.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7845663070678711,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 0.8875390291213989,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.06272958517074585,
+      "mean_token_accuracy": 0.9751476228237153,
+      "num_tokens": 23417469.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7928634822368622,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.0629489421844482,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.060286092758178714,
+      "mean_token_accuracy": 0.976036274433136,
+      "num_tokens": 23606274.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.7870154738426208,
+      "epoch": 1.695827725437416,
+      "grad_norm": 0.9164250493049622,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.06263296008110046,
+      "mean_token_accuracy": 0.9752088189125061,
+      "num_tokens": 23795255.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.7948406875133515,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.0909371376037598,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.06165047883987427,
+      "mean_token_accuracy": 0.9764420628547669,
+      "num_tokens": 23984534.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.7862435758113862,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.1839547157287598,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.0640158474445343,
+      "mean_token_accuracy": 0.9741048216819763,
+      "num_tokens": 24173632.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.7887867331504822,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.0730741024017334,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.06260218620300292,
+      "mean_token_accuracy": 0.974741768836975,
+      "num_tokens": 24362671.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.7896574556827545,
+      "epoch": 1.749663526244953,
+      "grad_norm": 0.9559605121612549,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.061297619342803956,
+      "mean_token_accuracy": 0.9752223670482636,
+      "num_tokens": 24551825.0,
+      "step": 1300
+    },
+    {
+      "entropy": 0.7902800619602204,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 0.914057195186615,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.06320110559463502,
+      "mean_token_accuracy": 0.9749482691287994,
+      "num_tokens": 24740867.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.799742591381073,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.0199551582336426,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.06534289717674255,
+      "mean_token_accuracy": 0.9741291284561158,
+      "num_tokens": 24930132.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.7974063038825989,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.1188615560531616,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.060908573865890506,
+      "mean_token_accuracy": 0.9765798091888428,
+      "num_tokens": 25119678.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.786439061164856,
+      "epoch": 1.80349932705249,
+      "grad_norm": 0.9185847640037537,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.06158917546272278,
+      "mean_token_accuracy": 0.9760335922241211,
+      "num_tokens": 25308876.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.7888528347015381,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.1948838233947754,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.060635948181152345,
+      "mean_token_accuracy": 0.976032954454422,
+      "num_tokens": 25498102.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.7968044698238372,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.4128663539886475,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.06448977589607238,
+      "mean_token_accuracy": 0.9741183638572692,
+      "num_tokens": 25686747.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.7967679858207702,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 0.9797796010971069,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.06342286467552186,
+      "mean_token_accuracy": 0.9752787709236145,
+      "num_tokens": 25875368.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.7932188332080841,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.0549253225326538,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.060981249809265135,
+      "mean_token_accuracy": 0.976200807094574,
+      "num_tokens": 26064480.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.7899552762508393,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.1172103881835938,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.06555815935134887,
+      "mean_token_accuracy": 0.9736676931381225,
+      "num_tokens": 26253619.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.8019726693630218,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 0.94905686378479,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.0663321852684021,
+      "mean_token_accuracy": 0.9739456713199616,
+      "num_tokens": 26442736.0,
+      "step": 1400
+    },
+    {
+      "entropy": 0.7939825117588043,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 0.9922711253166199,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.05981417298316956,
+      "mean_token_accuracy": 0.9758030951023102,
+      "num_tokens": 26631525.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.8085412085056305,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4970781803131104,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.06708773374557495,
+      "mean_token_accuracy": 0.9737519145011901,
+      "num_tokens": 26820483.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.7950678527355194,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 0.7868127226829529,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.061734098196029666,
+      "mean_token_accuracy": 0.9747878730297088,
+      "num_tokens": 27009393.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.7850766360759736,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.1386486291885376,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.05714603662490845,
+      "mean_token_accuracy": 0.9769004642963409,
+      "num_tokens": 27198251.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.7933461844921113,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.0092092752456665,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.06240262985229492,
+      "mean_token_accuracy": 0.9742367327213287,
+      "num_tokens": 27387231.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.8044621467590332,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.2477015256881714,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.0642085313796997,
+      "mean_token_accuracy": 0.9748865723609924,
+      "num_tokens": 27576251.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.7975574374198914,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.1167609691619873,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.06382874846458435,
+      "mean_token_accuracy": 0.9748350441455841,
+      "num_tokens": 27764560.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.8090806484222413,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 0.9714348912239075,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.06401748657226562,
+      "mean_token_accuracy": 0.9741845309734345,
+      "num_tokens": 27953532.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 0.8104567945383157,
+      "eval_loss": 0.07282773405313492,
+      "eval_mean_token_accuracy": 0.9701232534305305,
+      "eval_num_tokens": 28066151.0,
+      "eval_runtime": 12.7852,
+      "eval_samples_per_second": 391.078,
+      "eval_steps_per_second": 12.28,
+      "step": 1486
+    },
+    {
+      "entropy": 0.8090688765048981,
+      "epoch": 2.005383580080754,
+      "grad_norm": 0.9886816143989563,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.06012396812438965,
+      "mean_token_accuracy": 0.9764885842800141,
+      "num_tokens": 28141848.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.7821709334850311,
+      "epoch": 2.018842530282638,
+      "grad_norm": 0.9049969911575317,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.04819132387638092,
+      "mean_token_accuracy": 0.9816783249378205,
+      "num_tokens": 28330784.0,
+      "step": 1500
+    },
+    {
+      "entropy": 0.7845205307006836,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0231750011444092,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.04539612233638764,
+      "mean_token_accuracy": 0.9818619966506958,
+      "num_tokens": 28519580.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.7817754447460175,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.5244711637496948,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0429688960313797,
+      "mean_token_accuracy": 0.9834935545921326,
+      "num_tokens": 28707486.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.7733526527881622,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.0685060024261475,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.045852530002594,
+      "mean_token_accuracy": 0.9817728459835052,
+      "num_tokens": 28896759.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.7815343141555786,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.075853705406189,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.04565088748931885,
+      "mean_token_accuracy": 0.9823914349079133,
+      "num_tokens": 29085097.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.7751398742198944,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2573742866516113,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.04465666115283966,
+      "mean_token_accuracy": 0.9829874813556672,
+      "num_tokens": 29273868.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.7737197399139404,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 0.8681219220161438,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.042188242077827454,
+      "mean_token_accuracy": 0.9837059378623962,
+      "num_tokens": 29462985.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.7823441684246063,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.3503504991531372,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.04393318593502045,
+      "mean_token_accuracy": 0.9827627122402192,
+      "num_tokens": 29651156.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.7796198308467865,
+      "epoch": 2.126514131897712,
+      "grad_norm": 0.7799349427223206,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.04367418885231018,
+      "mean_token_accuracy": 0.9837401032447814,
+      "num_tokens": 29839789.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.7767675697803498,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.375465750694275,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.04427264630794525,
+      "mean_token_accuracy": 0.9832019031047821,
+      "num_tokens": 30028957.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.7912136912345886,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 0.9683088660240173,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.04428936243057251,
+      "mean_token_accuracy": 0.9832306385040284,
+      "num_tokens": 30217841.0,
+      "step": 1600
+    },
+    {
+      "entropy": 0.790750366449356,
+      "epoch": 2.166890982503365,
+      "grad_norm": 0.9393274188041687,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.04529332816600799,
+      "mean_token_accuracy": 0.9832263946533203,
+      "num_tokens": 30406540.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.7816023886203766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.1822642087936401,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.04192846417427063,
+      "mean_token_accuracy": 0.9846727311611175,
+      "num_tokens": 30595568.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.7754982054233551,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.1961082220077515,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.043791168928146364,
+      "mean_token_accuracy": 0.9830836653709412,
+      "num_tokens": 30784300.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.7756961703300476,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 0.9719674587249756,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.04410521686077118,
+      "mean_token_accuracy": 0.9837198793888092,
+      "num_tokens": 30972950.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.7833336114883422,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.0175098180770874,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.044813218712806704,
+      "mean_token_accuracy": 0.9829723179340363,
+      "num_tokens": 31161783.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.7893047273159027,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.2077007293701172,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.04417986273765564,
+      "mean_token_accuracy": 0.9825236797332764,
+      "num_tokens": 31350487.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.794023334980011,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 0.8530077934265137,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.043008115887641904,
+      "mean_token_accuracy": 0.9832422196865082,
+      "num_tokens": 31539217.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.785652619600296,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.4543156623840332,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.0445009708404541,
+      "mean_token_accuracy": 0.9828064203262329,
+      "num_tokens": 31728308.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.7841614365577698,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.297763466835022,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.045649731159210206,
+      "mean_token_accuracy": 0.9823782920837403,
+      "num_tokens": 31917293.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.7825099349021911,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.2237403392791748,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.047218257188797,
+      "mean_token_accuracy": 0.98156818151474,
+      "num_tokens": 32106628.0,
+      "step": 1700
+    },
+    {
+      "entropy": 0.7791289031505585,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 0.8094402551651001,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.04178315699100495,
+      "mean_token_accuracy": 0.9846469461917877,
+      "num_tokens": 32295183.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.7743105947971344,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.3677793741226196,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.04540807008743286,
+      "mean_token_accuracy": 0.9827918827533721,
+      "num_tokens": 32484143.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.7804521083831787,
+      "epoch": 2.328398384925976,
+      "grad_norm": 0.8883333802223206,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.04598270654678345,
+      "mean_token_accuracy": 0.9823103487491608,
+      "num_tokens": 32673144.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.7760683119297027,
+      "epoch": 2.34185733512786,
+      "grad_norm": 0.989972710609436,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.043702703714370725,
+      "mean_token_accuracy": 0.9831363618373871,
+      "num_tokens": 32862142.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.7726980447769165,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.0461052656173706,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.044418123364448545,
+      "mean_token_accuracy": 0.9821757376194,
+      "num_tokens": 33051382.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.7822885036468505,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 0.9027822017669678,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.04624638557434082,
+      "mean_token_accuracy": 0.9815774321556091,
+      "num_tokens": 33239729.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.7830030500888825,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.1743420362472534,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.04747769832611084,
+      "mean_token_accuracy": 0.9812638640403748,
+      "num_tokens": 33428615.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.7933187901973724,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 0.9334444403648376,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.047731178998947146,
+      "mean_token_accuracy": 0.9820823550224305,
+      "num_tokens": 33617331.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.7904782056808471,
+      "epoch": 2.409152086137281,
+      "grad_norm": 0.9396253824234009,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.04432434141635895,
+      "mean_token_accuracy": 0.9830163478851318,
+      "num_tokens": 33805978.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.7811533212661743,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 0.9742764234542847,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.046493279933929446,
+      "mean_token_accuracy": 0.9818211495876312,
+      "num_tokens": 33994957.0,
+      "step": 1800
+    },
+    {
+      "entropy": 0.7848010778427124,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.0378986597061157,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0454249233007431,
+      "mean_token_accuracy": 0.9817994356155395,
+      "num_tokens": 34184297.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.7754390180110932,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.1023566722869873,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.04380808770656586,
+      "mean_token_accuracy": 0.9829573810100556,
+      "num_tokens": 34373551.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.7767188012599945,
+      "epoch": 2.462987886944818,
+      "grad_norm": 0.8460658192634583,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.04502716064453125,
+      "mean_token_accuracy": 0.9825617015361786,
+      "num_tokens": 34562169.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.7757977426052094,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.1013047695159912,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.045161890983581546,
+      "mean_token_accuracy": 0.9821914732456207,
+      "num_tokens": 34750863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.785107308626175,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 0.9510554671287537,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.045189899206161496,
+      "mean_token_accuracy": 0.9822288691997528,
+      "num_tokens": 34939528.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.7849742233753204,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.0994503498077393,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.04333080947399139,
+      "mean_token_accuracy": 0.9830258965492249,
+      "num_tokens": 35128334.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.7875016331672668,
+      "epoch": 2.516823687752355,
+      "grad_norm": 0.9202979803085327,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.04838588833808899,
+      "mean_token_accuracy": 0.9815591931343078,
+      "num_tokens": 35317067.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.7813792943954467,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 0.9036716222763062,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.045662492513656616,
+      "mean_token_accuracy": 0.9817738354206085,
+      "num_tokens": 35506291.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.777835363149643,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 0.853438675403595,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.044686585664749146,
+      "mean_token_accuracy": 0.98329097032547,
+      "num_tokens": 35695141.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.7677761793136597,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.0699759721755981,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.045490306615829465,
+      "mean_token_accuracy": 0.9826831936836242,
+      "num_tokens": 35884173.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.7756118893623352,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.0356824398040771,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.0467751681804657,
+      "mean_token_accuracy": 0.9823338150978088,
+      "num_tokens": 36073241.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.7775443732738495,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.1619561910629272,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.04741084575653076,
+      "mean_token_accuracy": 0.9815192580223083,
+      "num_tokens": 36262133.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.7899596333503723,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.3479284048080444,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.04588942229747772,
+      "mean_token_accuracy": 0.9821865320205688,
+      "num_tokens": 36451460.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.7832387804985046,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.6297204494476318,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.04711659252643585,
+      "mean_token_accuracy": 0.9815292179584503,
+      "num_tokens": 36640366.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.7871865928173065,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.3100138902664185,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.04919236898422241,
+      "mean_token_accuracy": 0.9809704780578613,
+      "num_tokens": 36829017.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.7844007611274719,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.2907201051712036,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.047159326076507566,
+      "mean_token_accuracy": 0.9816433548927307,
+      "num_tokens": 37018034.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.7885129511356354,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.1324766874313354,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.04420983195304871,
+      "mean_token_accuracy": 0.9828807294368744,
+      "num_tokens": 37206768.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.7761563122272491,
+      "epoch": 2.664872139973082,
+      "grad_norm": 0.9269904494285583,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.04585547745227814,
+      "mean_token_accuracy": 0.9823735177516937,
+      "num_tokens": 37396010.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.7831155300140381,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5890840291976929,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.046450641751289365,
+      "mean_token_accuracy": 0.9817734777927398,
+      "num_tokens": 37584809.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.7729385912418365,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.4823863506317139,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.04653529226779938,
+      "mean_token_accuracy": 0.9815726220607758,
+      "num_tokens": 37773869.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.7752901673316955,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 0.9423630237579346,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.04599723219871521,
+      "mean_token_accuracy": 0.9816120624542236,
+      "num_tokens": 37962642.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.7750158667564392,
+      "epoch": 2.718707940780619,
+      "grad_norm": 0.9398621320724487,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.04729728400707245,
+      "mean_token_accuracy": 0.9816868543624878,
+      "num_tokens": 38151325.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.7693424999713898,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3207950592041016,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.04474019706249237,
+      "mean_token_accuracy": 0.9830808460712432,
+      "num_tokens": 38340719.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.7750859558582306,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.070200800895691,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.045914703607559205,
+      "mean_token_accuracy": 0.9820488631725312,
+      "num_tokens": 38529293.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.7697387218475342,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.0545568466186523,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.04357335567474365,
+      "mean_token_accuracy": 0.9827888309955597,
+      "num_tokens": 38718464.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.7682756781578064,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.0945985317230225,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.0429091215133667,
+      "mean_token_accuracy": 0.9832668721675872,
+      "num_tokens": 38907573.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.7804523587226868,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 0.8521431684494019,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.04662351012229919,
+      "mean_token_accuracy": 0.981964933872223,
+      "num_tokens": 39096598.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.7786886096000671,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.09945809841156,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.050166720151901247,
+      "mean_token_accuracy": 0.9816301107406616,
+      "num_tokens": 39285069.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.7790449261665344,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.1673868894577026,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.048508059978485105,
+      "mean_token_accuracy": 0.9809533059597015,
+      "num_tokens": 39474252.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.7774537563323974,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.0648490190505981,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.04540320634841919,
+      "mean_token_accuracy": 0.9822351098060608,
+      "num_tokens": 39663262.0,
+      "step": 2100
+    },
+    {
+      "entropy": 0.7683943033218383,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.0106912851333618,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.04466369152069092,
+      "mean_token_accuracy": 0.9818881273269653,
+      "num_tokens": 39852133.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.7642774760723114,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.0109530687332153,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.045978450775146486,
+      "mean_token_accuracy": 0.9819821894168854,
+      "num_tokens": 40041679.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.7647961139678955,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 0.9050237536430359,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.045694130659103396,
+      "mean_token_accuracy": 0.982342666387558,
+      "num_tokens": 40230019.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.7589850842952728,
+      "epoch": 2.88021534320323,
+      "grad_norm": 0.8879176378250122,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.046186423301696776,
+      "mean_token_accuracy": 0.9818771004676818,
+      "num_tokens": 40418932.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.765201997756958,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 0.9103795886039734,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.04475873112678528,
+      "mean_token_accuracy": 0.9821203172206878,
+      "num_tokens": 40607792.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.7653619945049286,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.0566699504852295,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.04391545057296753,
+      "mean_token_accuracy": 0.9828326046466828,
+      "num_tokens": 40797122.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.763550215959549,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.1280677318572998,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.045582184195518495,
+      "mean_token_accuracy": 0.9819066524505615,
+      "num_tokens": 40985689.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.7749062657356263,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.110323429107666,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.04598402082920074,
+      "mean_token_accuracy": 0.9815545618534088,
+      "num_tokens": 41174743.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.7805372893810272,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.0649195909500122,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.04705569744110107,
+      "mean_token_accuracy": 0.9815925776958465,
+      "num_tokens": 41363160.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.7711063683032989,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.1459592580795288,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.045355689525604245,
+      "mean_token_accuracy": 0.9832463562488556,
+      "num_tokens": 41551932.0,
+      "step": 2200
+    },
+    {
+      "entropy": 0.7747956931591033,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.1602799892425537,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.04494523704051971,
+      "mean_token_accuracy": 0.9819566667079925,
+      "num_tokens": 41740544.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.7584574401378632,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.054417371749878,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.04810510575771332,
+      "mean_token_accuracy": 0.9811540365219116,
+      "num_tokens": 41929344.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 0.7753648416251894,
+      "eval_loss": 0.07181113958358765,
+      "eval_mean_token_accuracy": 0.9716544128527307,
+      "eval_num_tokens": 42099366.0,
+      "eval_runtime": 12.748,
+      "eval_samples_per_second": 392.22,
+      "eval_steps_per_second": 12.316,
+      "step": 2229
+    },
+    {
+      "entropy": 0.7669954419136047,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 0.8211690783500671,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.04577431380748749,
+      "mean_token_accuracy": 0.9823363184928894,
+      "num_tokens": 42118188.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.7600219964981079,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.0873607397079468,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.032501548528671265,
+      "mean_token_accuracy": 0.9881355822086334,
+      "num_tokens": 42306598.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.7441959500312805,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.1175898313522339,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.031047609448432923,
+      "mean_token_accuracy": 0.9891582727432251,
+      "num_tokens": 42495250.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.7364335536956788,
+      "epoch": 3.041722745625841,
+      "grad_norm": 0.987751841545105,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.028629058599472047,
+      "mean_token_accuracy": 0.9894859492778778,
+      "num_tokens": 42684326.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.735680228471756,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.194305658340454,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.029283303022384643,
+      "mean_token_accuracy": 0.9887769162654877,
+      "num_tokens": 42873221.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.7342639684677124,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.254442811012268,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.0302200973033905,
+      "mean_token_accuracy": 0.988795381784439,
+      "num_tokens": 43062486.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.7391974747180938,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.2648969888687134,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.0316451758146286,
+      "mean_token_accuracy": 0.9879588901996612,
+      "num_tokens": 43251113.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.7401826322078705,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.0483295917510986,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.030878132581710814,
+      "mean_token_accuracy": 0.9886977672576904,
+      "num_tokens": 43440152.0,
+      "step": 2300
+    },
+    {
+      "entropy": 0.732647967338562,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.1125447750091553,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.02838427424430847,
+      "mean_token_accuracy": 0.9890213429927825,
+      "num_tokens": 43629353.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.7314932703971863,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.0979042053222656,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.0315592348575592,
+      "mean_token_accuracy": 0.9888832688331604,
+      "num_tokens": 43818981.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.7305827140808105,
+      "epoch": 3.135935397039031,
+      "grad_norm": 0.9610060453414917,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.03112468421459198,
+      "mean_token_accuracy": 0.9890477001667023,
+      "num_tokens": 44008430.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.7333625435829163,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.0625078678131104,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.03105958104133606,
+      "mean_token_accuracy": 0.9884848177433014,
+      "num_tokens": 44197264.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.7373903751373291,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.1683465242385864,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.031627464294433597,
+      "mean_token_accuracy": 0.9879396498203278,
+      "num_tokens": 44386205.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.7391358911991119,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.05154287815094,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.03118889331817627,
+      "mean_token_accuracy": 0.9880312740802765,
+      "num_tokens": 44575099.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.7351223647594451,
+      "epoch": 3.189771197846568,
+      "grad_norm": 0.9531214237213135,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.030980610847473146,
+      "mean_token_accuracy": 0.9882127285003662,
+      "num_tokens": 44763965.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.7361315310001373,
+      "epoch": 3.203230148048452,
+      "grad_norm": 0.8818249702453613,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.029705989360809325,
+      "mean_token_accuracy": 0.9894622683525085,
+      "num_tokens": 44952983.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.7356473922729492,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 0.9263529777526855,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.03202669024467468,
+      "mean_token_accuracy": 0.9877129793167114,
+      "num_tokens": 45142080.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.7326373934745789,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.0517516136169434,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.031153690814971925,
+      "mean_token_accuracy": 0.9891136825084687,
+      "num_tokens": 45330769.0,
+      "step": 2400
+    },
+    {
+      "entropy": 0.7348711133003235,
+      "epoch": 3.243606998654105,
+      "grad_norm": 0.8774152994155884,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.030574262142181396,
+      "mean_token_accuracy": 0.9888435959815979,
+      "num_tokens": 45519217.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.7379967510700226,
+      "epoch": 3.257065948855989,
+      "grad_norm": 0.9164374470710754,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.03096820116043091,
+      "mean_token_accuracy": 0.9884447932243348,
+      "num_tokens": 45708467.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.7369260907173156,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.3184924125671387,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.03082413077354431,
+      "mean_token_accuracy": 0.9888757109642029,
+      "num_tokens": 45897424.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.7366503119468689,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 0.9073860049247742,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.03220943808555603,
+      "mean_token_accuracy": 0.9877902865409851,
+      "num_tokens": 46086351.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.7341453731060028,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.0940542221069336,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.03329495787620544,
+      "mean_token_accuracy": 0.9877542018890381,
+      "num_tokens": 46275394.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.7312689840793609,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 0.9651952385902405,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.031667912006378175,
+      "mean_token_accuracy": 0.9881450653076171,
+      "num_tokens": 46463821.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.7228557825088501,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 0.971441924571991,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.030608028173446655,
+      "mean_token_accuracy": 0.9882231116294861,
+      "num_tokens": 46653025.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.723542982339859,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.0060454607009888,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.030545371770858764,
+      "mean_token_accuracy": 0.9879125893115998,
+      "num_tokens": 46842068.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.7321009397506714,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.0572997331619263,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.032070663571357724,
+      "mean_token_accuracy": 0.9878279268741608,
+      "num_tokens": 47031045.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.735515958070755,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 0.9835748672485352,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.03338083028793335,
+      "mean_token_accuracy": 0.9877255499362946,
+      "num_tokens": 47219625.0,
+      "step": 2500
+    },
+    {
+      "entropy": 0.7396618247032165,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 0.9655317068099976,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.03589712679386139,
+      "mean_token_accuracy": 0.9869556546211242,
+      "num_tokens": 47408311.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.7358698546886444,
+      "epoch": 3.391655450874832,
+      "grad_norm": 0.9225422143936157,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.02914820909500122,
+      "mean_token_accuracy": 0.9886718571186066,
+      "num_tokens": 47597297.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.7338236391544342,
+      "epoch": 3.405114401076716,
+      "grad_norm": 0.8253662586212158,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.02996352016925812,
+      "mean_token_accuracy": 0.9891344487667084,
+      "num_tokens": 47785737.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.7320856511592865,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.0716700553894043,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.032322123646736145,
+      "mean_token_accuracy": 0.9877110660076142,
+      "num_tokens": 47974241.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.7293457925319672,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.0915459394454956,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.031131434440612792,
+      "mean_token_accuracy": 0.9885926306247711,
+      "num_tokens": 48162582.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.7207926452159882,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.0700562000274658,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.03113105595111847,
+      "mean_token_accuracy": 0.9880475521087646,
+      "num_tokens": 48351627.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.7178894281387329,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.1952542066574097,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.03204495906829834,
+      "mean_token_accuracy": 0.9880682706832886,
+      "num_tokens": 48540171.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.7166945099830627,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.0920100212097168,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.03096899390220642,
+      "mean_token_accuracy": 0.9884315967559815,
+      "num_tokens": 48729737.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.7187432169914245,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 0.9596593976020813,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.03240742087364197,
+      "mean_token_accuracy": 0.9878904581069946,
+      "num_tokens": 48918599.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.7224443972110748,
+      "epoch": 3.499327052489906,
+      "grad_norm": 0.9406276345252991,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.03253386616706848,
+      "mean_token_accuracy": 0.9873985171318054,
+      "num_tokens": 49107385.0,
+      "step": 2600
+    },
+    {
+      "entropy": 0.7226529896259308,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.2169438600540161,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.03282734453678131,
+      "mean_token_accuracy": 0.9876641929149628,
+      "num_tokens": 49295969.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.7263983964920044,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 0.8808715343475342,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.030802983045578002,
+      "mean_token_accuracy": 0.9885481476783753,
+      "num_tokens": 49484706.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.7291677474975586,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2453594207763672,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.03403027057647705,
+      "mean_token_accuracy": 0.987172520160675,
+      "num_tokens": 49673484.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.7396404981613159,
+      "epoch": 3.553162853297443,
+      "grad_norm": 0.9215838313102722,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.03405522108078003,
+      "mean_token_accuracy": 0.9871597826480866,
+      "num_tokens": 49862087.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.7382308602333069,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.0659712553024292,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.032089871168136594,
+      "mean_token_accuracy": 0.9874629497528076,
+      "num_tokens": 50051379.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.7337264895439148,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 0.9307538270950317,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.032426226139068606,
+      "mean_token_accuracy": 0.9879062473773956,
+      "num_tokens": 50239986.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.7381527841091156,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.1867440938949585,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.03276415467262268,
+      "mean_token_accuracy": 0.9878055810928345,
+      "num_tokens": 50428872.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.7337838113307953,
+      "epoch": 3.60699865410498,
+      "grad_norm": 0.9655186533927917,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.03141146302223206,
+      "mean_token_accuracy": 0.9882215678691864,
+      "num_tokens": 50618182.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.7354952573776246,
+      "epoch": 3.620457604306864,
+      "grad_norm": 0.8848851919174194,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.03387039303779602,
+      "mean_token_accuracy": 0.9871303081512451,
+      "num_tokens": 50806430.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.7329667210578918,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 0.9976673722267151,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.03129216730594635,
+      "mean_token_accuracy": 0.9882283449172974,
+      "num_tokens": 50995500.0,
+      "step": 2700
+    },
+    {
+      "entropy": 0.7277809977531433,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.2560697793960571,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.03357822299003601,
+      "mean_token_accuracy": 0.9876434028148651,
+      "num_tokens": 51184436.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.727894538640976,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.1099125146865845,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.032482534646987915,
+      "mean_token_accuracy": 0.9878577649593353,
+      "num_tokens": 51373238.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.7318481564521789,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.1980950832366943,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.03284388482570648,
+      "mean_token_accuracy": 0.9874962747097016,
+      "num_tokens": 51561631.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.7219811856746674,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 0.8730921745300293,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.030655479431152342,
+      "mean_token_accuracy": 0.9877778172492981,
+      "num_tokens": 51750845.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.7289035737514495,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.1195666790008545,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.03155539035797119,
+      "mean_token_accuracy": 0.9877057671546936,
+      "num_tokens": 51940009.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.7332375526428223,
+      "epoch": 3.714670255720054,
+      "grad_norm": 0.9390997886657715,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.031307452917099,
+      "mean_token_accuracy": 0.987467783689499,
+      "num_tokens": 52128442.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.7293307244777679,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.0272324085235596,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.03318062424659729,
+      "mean_token_accuracy": 0.9873259484767913,
+      "num_tokens": 52317152.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.7320701718330384,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 0.9884606003761292,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.03461401164531708,
+      "mean_token_accuracy": 0.9867411673069,
+      "num_tokens": 52506106.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.7248260796070098,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 0.879682719707489,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.03406862616539001,
+      "mean_token_accuracy": 0.9872734069824218,
+      "num_tokens": 52695500.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.7237857282161713,
+      "epoch": 3.768506056527591,
+      "grad_norm": 0.9409974217414856,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.029435336589813232,
+      "mean_token_accuracy": 0.9891225278377533,
+      "num_tokens": 52884940.0,
+      "step": 2800
+    },
+    {
+      "entropy": 0.7237291693687439,
+      "epoch": 3.781965006729475,
+      "grad_norm": 0.9532694220542908,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.030907022953033447,
+      "mean_token_accuracy": 0.9881488382816315,
+      "num_tokens": 53073909.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.7252696454524994,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 0.9157902002334595,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.032040882110595706,
+      "mean_token_accuracy": 0.9880847573280335,
+      "num_tokens": 53262962.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.7237608432769775,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.1450515985488892,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.03199286162853241,
+      "mean_token_accuracy": 0.987687611579895,
+      "num_tokens": 53451982.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.7237064421176911,
+      "epoch": 3.822341857335128,
+      "grad_norm": 0.9945085644721985,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.03347462117671966,
+      "mean_token_accuracy": 0.9875530540943146,
+      "num_tokens": 53640737.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.7271939218044281,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 0.9351736903190613,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.03140422701835632,
+      "mean_token_accuracy": 0.9877301037311554,
+      "num_tokens": 53829142.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.7265262603759766,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.1019784212112427,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.030748724937438965,
+      "mean_token_accuracy": 0.9881740033626556,
+      "num_tokens": 54018331.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.7270502150058746,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 0.9153032898902893,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.033472740650177,
+      "mean_token_accuracy": 0.9876999020576477,
+      "num_tokens": 54206935.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.7345028996467591,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.0144257545471191,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.033550959825515744,
+      "mean_token_accuracy": 0.9874906241893768,
+      "num_tokens": 54395443.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.7274066030979156,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.0221407413482666,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.033388751745224,
+      "mean_token_accuracy": 0.9870682001113892,
+      "num_tokens": 54584552.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.7293415367603302,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.019165277481079,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.03568689227104187,
+      "mean_token_accuracy": 0.9861597299575806,
+      "num_tokens": 54773705.0,
+      "step": 2900
+    },
+    {
+      "entropy": 0.7281844198703766,
+      "epoch": 3.916554508748318,
+      "grad_norm": 0.933020830154419,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.030142486095428467,
+      "mean_token_accuracy": 0.988915103673935,
+      "num_tokens": 54962152.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.7242733418941498,
+      "epoch": 3.930013458950202,
+      "grad_norm": 0.9429674744606018,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.0337733268737793,
+      "mean_token_accuracy": 0.987263411283493,
+      "num_tokens": 55151179.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.7306612014770508,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 0.9022014737129211,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.03387616574764252,
+      "mean_token_accuracy": 0.9871586680412292,
+      "num_tokens": 55339658.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.7246532738208771,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2193421125411987,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.03193953037261963,
+      "mean_token_accuracy": 0.9879468739032745,
+      "num_tokens": 55528467.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.7268587052822113,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.2917358875274658,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.03584722280502319,
+      "mean_token_accuracy": 0.9867104470729828,
+      "num_tokens": 55717308.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.735340416431427,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.2229385375976562,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.03506708741188049,
+      "mean_token_accuracy": 0.9864643573760986,
+      "num_tokens": 55906059.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.7337038695812226,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.0747122764587402,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.03241217732429504,
+      "mean_token_accuracy": 0.9876143455505371,
+      "num_tokens": 56094606.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 0.7333687073106219,
+      "eval_loss": 0.07489899545907974,
+      "eval_mean_token_accuracy": 0.9719084019114257,
+      "eval_num_tokens": 56132350.0,
+      "eval_runtime": 12.7823,
+      "eval_samples_per_second": 391.166,
+      "eval_steps_per_second": 12.283,
+      "step": 2972
+    },
+    {
+      "entropy": 0.72461416721344,
+      "epoch": 4.010767160161508,
+      "grad_norm": 0.7916642427444458,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.02286165803670883,
+      "mean_token_accuracy": 0.9920066714286804,
+      "num_tokens": 56283397.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.7150031447410583,
+      "epoch": 4.024226110363392,
+      "grad_norm": 0.9078621864318848,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.020906335115432738,
+      "mean_token_accuracy": 0.9925728499889374,
+      "num_tokens": 56472408.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.7091723501682281,
+      "epoch": 4.037685060565276,
+      "grad_norm": 0.8117756843566895,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.02036277800798416,
+      "mean_token_accuracy": 0.9928471207618713,
+      "num_tokens": 56660906.0,
+      "step": 3000
+    },
+    {
+      "entropy": 0.695214182138443,
+      "epoch": 4.05114401076716,
+      "grad_norm": 0.7815372943878174,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.019653698801994322,
+      "mean_token_accuracy": 0.992583978176117,
+      "num_tokens": 56850067.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.6937313377857208,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4769781827926636,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.02311961054801941,
+      "mean_token_accuracy": 0.9921472012996674,
+      "num_tokens": 57039005.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.6977694034576416,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.0028746128082275,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.021832706034183504,
+      "mean_token_accuracy": 0.9917877376079559,
+      "num_tokens": 57227782.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.698432594537735,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.0107522010803223,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.020331941545009613,
+      "mean_token_accuracy": 0.9932828366756439,
+      "num_tokens": 57416541.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.6941913783550262,
+      "epoch": 4.104979811574697,
+      "grad_norm": 0.8923842310905457,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.019534927606582642,
+      "mean_token_accuracy": 0.9930490672588348,
+      "num_tokens": 57605162.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.6947623729705811,
+      "epoch": 4.118438761776582,
+      "grad_norm": 0.9093812108039856,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.02151128649711609,
+      "mean_token_accuracy": 0.9920440256595612,
+      "num_tokens": 57793756.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.6972509443759918,
+      "epoch": 4.131897711978466,
+      "grad_norm": 0.7549760341644287,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.020574498176574706,
+      "mean_token_accuracy": 0.9928867638111114,
+      "num_tokens": 57982678.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.6952514111995697,
+      "epoch": 4.14535666218035,
+      "grad_norm": 0.9476824402809143,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.02087453305721283,
+      "mean_token_accuracy": 0.9924874365329742,
+      "num_tokens": 58171508.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.6910251140594482,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.2548149824142456,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.020806866884231567,
+      "mean_token_accuracy": 0.9921584248542785,
+      "num_tokens": 58360594.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.6958921313285827,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.0649152994155884,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.020952591300010683,
+      "mean_token_accuracy": 0.9924699544906617,
+      "num_tokens": 58549637.0,
+      "step": 3100
+    },
+    {
+      "entropy": 0.6932485699653625,
+      "epoch": 4.185733512786003,
+      "grad_norm": 0.8891078233718872,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.020813333988189697,
+      "mean_token_accuracy": 0.9927419483661651,
+      "num_tokens": 58738595.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.6984799921512603,
+      "epoch": 4.199192462987887,
+      "grad_norm": 0.9459101557731628,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.01971060484647751,
+      "mean_token_accuracy": 0.9926240146160126,
+      "num_tokens": 58927507.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.7021916508674622,
+      "epoch": 4.212651413189771,
+      "grad_norm": 0.8318898677825928,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.020860180258750916,
+      "mean_token_accuracy": 0.9927471876144409,
+      "num_tokens": 59116537.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.6972847402095794,
+      "epoch": 4.226110363391656,
+      "grad_norm": 0.9066984057426453,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.023059430718421935,
+      "mean_token_accuracy": 0.9919216752052307,
+      "num_tokens": 59305838.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.7049594700336457,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1277318000793457,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.022385355830192567,
+      "mean_token_accuracy": 0.99165118932724,
+      "num_tokens": 59494587.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.7045334160327912,
+      "epoch": 4.253028263795424,
+      "grad_norm": 0.8147489428520203,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.022358573973178864,
+      "mean_token_accuracy": 0.9918249011039734,
+      "num_tokens": 59683021.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.7007148265838623,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 0.9161480069160461,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.02088596671819687,
+      "mean_token_accuracy": 0.9925470650196075,
+      "num_tokens": 59871999.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.7006107687950134,
+      "epoch": 4.279946164199193,
+      "grad_norm": 0.9611289501190186,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.022619232535362244,
+      "mean_token_accuracy": 0.991577273607254,
+      "num_tokens": 60060486.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.7069303750991821,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.0339760780334473,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.021571077406406403,
+      "mean_token_accuracy": 0.9922139644622803,
+      "num_tokens": 60249686.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.7146372079849244,
+      "epoch": 4.306864064602961,
+      "grad_norm": 0.9761297702789307,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.022300006449222566,
+      "mean_token_accuracy": 0.9918746411800384,
+      "num_tokens": 60438481.0,
+      "step": 3200
+    },
+    {
+      "entropy": 0.7134014248847962,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 0.9729707837104797,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.02121303677558899,
+      "mean_token_accuracy": 0.9925692796707153,
+      "num_tokens": 60627501.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.7110478222370148,
+      "epoch": 4.33378196500673,
+      "grad_norm": 0.9456778168678284,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.020917078852653502,
+      "mean_token_accuracy": 0.9922434687614441,
+      "num_tokens": 60816135.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.7111951589584351,
+      "epoch": 4.347240915208614,
+      "grad_norm": 0.8617656826972961,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.020851245522499083,
+      "mean_token_accuracy": 0.9924458563327789,
+      "num_tokens": 61005658.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.7193978548049926,
+      "epoch": 4.360699865410498,
+      "grad_norm": 0.8495420813560486,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.02084285914897919,
+      "mean_token_accuracy": 0.9927356660366058,
+      "num_tokens": 61194170.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.7188209414482116,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 0.728841245174408,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.02111642211675644,
+      "mean_token_accuracy": 0.9926697969436645,
+      "num_tokens": 61382852.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.7130864262580872,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.1274386644363403,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.021722708642482758,
+      "mean_token_accuracy": 0.9922453343868256,
+      "num_tokens": 61572195.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.7132770180702209,
+      "epoch": 4.401076716016151,
+      "grad_norm": 0.9251006245613098,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.021602503955364227,
+      "mean_token_accuracy": 0.9919042110443115,
+      "num_tokens": 61761119.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.7063322901725769,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.0182228088378906,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.020274630188941954,
+      "mean_token_accuracy": 0.9928637027740479,
+      "num_tokens": 61949931.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.7106044590473175,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 0.795170783996582,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.020594277977943422,
+      "mean_token_accuracy": 0.9925063610076904,
+      "num_tokens": 62138377.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.7151111841201783,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1479791402816772,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.02117723524570465,
+      "mean_token_accuracy": 0.9921884059906005,
+      "num_tokens": 62327009.0,
+      "step": 3300
+    },
+    {
+      "entropy": 0.7172957301139832,
+      "epoch": 4.454912516823688,
+      "grad_norm": 0.9795703291893005,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.021917465329170226,
+      "mean_token_accuracy": 0.9915876984596252,
+      "num_tokens": 62515859.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.7175660550594329,
+      "epoch": 4.468371467025572,
+      "grad_norm": 0.9326112866401672,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.021142563223838805,
+      "mean_token_accuracy": 0.9922006845474243,
+      "num_tokens": 62704678.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.7085749089717865,
+      "epoch": 4.481830417227457,
+      "grad_norm": 0.8895787596702576,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.0222358375787735,
+      "mean_token_accuracy": 0.9918791234493256,
+      "num_tokens": 62893576.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.7059603810310364,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2298023700714111,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.021999095380306245,
+      "mean_token_accuracy": 0.9920522570610046,
+      "num_tokens": 63082258.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.707107275724411,
+      "epoch": 4.508748317631225,
+      "grad_norm": 0.8295655250549316,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.021844917535781862,
+      "mean_token_accuracy": 0.9922264873981476,
+      "num_tokens": 63271202.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.7119229018688202,
+      "epoch": 4.522207267833109,
+      "grad_norm": 0.7440339922904968,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.022164498269557954,
+      "mean_token_accuracy": 0.9918813884258271,
+      "num_tokens": 63460614.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.7065344154834747,
+      "epoch": 4.535666218034994,
+      "grad_norm": 0.7275091409683228,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.019415955245494842,
+      "mean_token_accuracy": 0.9928511261940003,
+      "num_tokens": 63649327.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.6984272241592407,
+      "epoch": 4.549125168236878,
+      "grad_norm": 0.9989121556282043,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.02176133990287781,
+      "mean_token_accuracy": 0.9923081934452057,
+      "num_tokens": 63837983.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.7084203243255616,
+      "epoch": 4.562584118438762,
+      "grad_norm": 0.9526837468147278,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.02182276248931885,
+      "mean_token_accuracy": 0.9921061515808105,
+      "num_tokens": 64026855.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.7097050070762634,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.0156456232070923,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.020074066519737244,
+      "mean_token_accuracy": 0.9926879525184631,
+      "num_tokens": 64215889.0,
+      "step": 3400
+    },
+    {
+      "entropy": 0.7087334096431732,
+      "epoch": 4.589502018842531,
+      "grad_norm": 0.780275821685791,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.021703639626502992,
+      "mean_token_accuracy": 0.9917874097824096,
+      "num_tokens": 64404515.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.7142964243888855,
+      "epoch": 4.602960969044415,
+      "grad_norm": 0.7382510304450989,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.019867205619812013,
+      "mean_token_accuracy": 0.9927354514598846,
+      "num_tokens": 64593147.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.7165492177009583,
+      "epoch": 4.616419919246299,
+      "grad_norm": 0.8259146213531494,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.021262936294078827,
+      "mean_token_accuracy": 0.9914182722568512,
+      "num_tokens": 64782041.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.7098793745040893,
+      "epoch": 4.629878869448183,
+      "grad_norm": 0.9540995359420776,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.0195529043674469,
+      "mean_token_accuracy": 0.9931159794330597,
+      "num_tokens": 64971390.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.7104229986667633,
+      "epoch": 4.643337819650068,
+      "grad_norm": 0.8698863387107849,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.019119253754615782,
+      "mean_token_accuracy": 0.9934523403644562,
+      "num_tokens": 65160253.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.7067269921302796,
+      "epoch": 4.656796769851952,
+      "grad_norm": 0.9622324705123901,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.02044745087623596,
+      "mean_token_accuracy": 0.992815512418747,
+      "num_tokens": 65349403.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.7016099870204926,
+      "epoch": 4.670255720053836,
+      "grad_norm": 0.999839723110199,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.019250848889350893,
+      "mean_token_accuracy": 0.9929339647293091,
+      "num_tokens": 65538164.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.7059290170669555,
+      "epoch": 4.68371467025572,
+      "grad_norm": 0.7539619207382202,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.02152409851551056,
+      "mean_token_accuracy": 0.99238161444664,
+      "num_tokens": 65727493.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.7103133797645569,
+      "epoch": 4.697173620457605,
+      "grad_norm": 0.9892463088035583,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.020616935193538667,
+      "mean_token_accuracy": 0.9927784681320191,
+      "num_tokens": 65916233.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.703304260969162,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.1750195026397705,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.021048617362976075,
+      "mean_token_accuracy": 0.9921808183193207,
+      "num_tokens": 66105229.0,
+      "step": 3500
+    },
+    {
+      "entropy": 0.7017257273197174,
+      "epoch": 4.724091520861373,
+      "grad_norm": 0.887427031993866,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.02077500820159912,
+      "mean_token_accuracy": 0.992085075378418,
+      "num_tokens": 66294280.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.7027543246746063,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.0987571477890015,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.02167164385318756,
+      "mean_token_accuracy": 0.992034786939621,
+      "num_tokens": 66483182.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.7053992748260498,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.0075602531433105,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.022686071693897247,
+      "mean_token_accuracy": 0.9918968200683593,
+      "num_tokens": 66671795.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.7109473288059235,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.1655596494674683,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.022698774933815002,
+      "mean_token_accuracy": 0.9913642406463623,
+      "num_tokens": 66861166.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.7130867660045623,
+      "epoch": 4.77792732166891,
+      "grad_norm": 0.9386347532272339,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0205889493227005,
+      "mean_token_accuracy": 0.9924897074699401,
+      "num_tokens": 67049967.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.7099342107772827,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 0.8524489998817444,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.01985667645931244,
+      "mean_token_accuracy": 0.9928426086902619,
+      "num_tokens": 67238781.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.7100094079971313,
+      "epoch": 4.804845222072679,
+      "grad_norm": 0.7522528171539307,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.02066437602043152,
+      "mean_token_accuracy": 0.9924224317073822,
+      "num_tokens": 67427802.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.7172351717948914,
+      "epoch": 4.818304172274562,
+      "grad_norm": 0.9865215420722961,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.021969664096832275,
+      "mean_token_accuracy": 0.9919721603393554,
+      "num_tokens": 67616675.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.7113273084163666,
+      "epoch": 4.831763122476447,
+      "grad_norm": 0.9339645504951477,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.019794631004333495,
+      "mean_token_accuracy": 0.9925971984863281,
+      "num_tokens": 67805877.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.7041554272174835,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.337364912033081,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.020412631332874298,
+      "mean_token_accuracy": 0.9923949122428894,
+      "num_tokens": 67994858.0,
+      "step": 3600
+    },
+    {
+      "entropy": 0.7090700149536133,
+      "epoch": 4.858681022880216,
+      "grad_norm": 0.7725968956947327,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.02121713161468506,
+      "mean_token_accuracy": 0.9920605599880219,
+      "num_tokens": 68183865.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.7141157209873199,
+      "epoch": 4.872139973082099,
+      "grad_norm": 0.8638854026794434,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.020726922154426574,
+      "mean_token_accuracy": 0.992779266834259,
+      "num_tokens": 68372954.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.7078199684619904,
+      "epoch": 4.885598923283984,
+      "grad_norm": 0.8576115965843201,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.019535741209983824,
+      "mean_token_accuracy": 0.9927346885204316,
+      "num_tokens": 68562535.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.7034667313098908,
+      "epoch": 4.899057873485868,
+      "grad_norm": 0.8339338898658752,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.021528851985931397,
+      "mean_token_accuracy": 0.991979056596756,
+      "num_tokens": 68751258.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.6983348250389099,
+      "epoch": 4.912516823687753,
+      "grad_norm": 0.8739761710166931,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.020945993065834046,
+      "mean_token_accuracy": 0.9923328399658203,
+      "num_tokens": 68939586.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.6922877132892609,
+      "epoch": 4.925975773889636,
+      "grad_norm": 0.886886715888977,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.020974791049957274,
+      "mean_token_accuracy": 0.992312878370285,
+      "num_tokens": 69128270.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.6857815742492676,
+      "epoch": 4.939434724091521,
+      "grad_norm": 0.9415780901908875,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.020201480388641356,
+      "mean_token_accuracy": 0.9925193309783935,
+      "num_tokens": 69317066.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.6832823157310486,
+      "epoch": 4.952893674293405,
+      "grad_norm": 0.8311293721199036,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.02008202373981476,
+      "mean_token_accuracy": 0.9926081895828247,
+      "num_tokens": 69505698.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.6914379715919494,
+      "epoch": 4.96635262449529,
+      "grad_norm": 0.8799999356269836,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.0189486563205719,
+      "mean_token_accuracy": 0.9932616055011749,
+      "num_tokens": 69694164.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.692799162864685,
+      "epoch": 4.979811574697173,
+      "grad_norm": 0.9579370021820068,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.019743245840072633,
+      "mean_token_accuracy": 0.9932292222976684,
+      "num_tokens": 69882717.0,
+      "step": 3700
+    },
+    {
+      "entropy": 0.6969726324081421,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.0632585287094116,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.019562503695487975,
+      "mean_token_accuracy": 0.9929000854492187,
+      "num_tokens": 70071225.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 0.7056300662885047,
+      "eval_loss": 0.08702203631401062,
+      "eval_mean_token_accuracy": 0.9712980936287315,
+      "eval_num_tokens": 70165509.0,
+      "eval_runtime": 12.7642,
+      "eval_samples_per_second": 391.719,
+      "eval_steps_per_second": 12.3,
+      "step": 3715
+    },
+    {
+      "entropy": 0.6998905837535858,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8414336442947388,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.01888625919818878,
+      "mean_token_accuracy": 0.9934305250644684,
+      "num_tokens": 70259811.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.6933371186256408,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.7618409395217896,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.01257207989692688,
+      "mean_token_accuracy": 0.9958574831485748,
+      "num_tokens": 70448683.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.6861799776554107,
+      "epoch": 5.03364737550471,
+      "grad_norm": 0.935829758644104,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.0118876114487648,
+      "mean_token_accuracy": 0.9963350534439087,
+      "num_tokens": 70637078.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.6778797626495361,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 0.9417919516563416,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.012690700590610504,
+      "mean_token_accuracy": 0.9954731047153473,
+      "num_tokens": 70826145.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.6735071957111358,
+      "epoch": 5.060565275908479,
+      "grad_norm": 0.7570633292198181,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.01321282535791397,
+      "mean_token_accuracy": 0.9955435931682587,
+      "num_tokens": 71015458.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.677568006515503,
+      "epoch": 5.074024226110363,
+      "grad_norm": 0.8157011866569519,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.012434301525354385,
+      "mean_token_accuracy": 0.9957576751708984,
+      "num_tokens": 71204332.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.6771604120731354,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.780483603477478,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.013237486779689788,
+      "mean_token_accuracy": 0.9957249999046326,
+      "num_tokens": 71393307.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.6761959791183472,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 0.8249189257621765,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.011912986636161804,
+      "mean_token_accuracy": 0.9961272835731506,
+      "num_tokens": 71582233.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.6717506468296051,
+      "epoch": 5.114401076716016,
+      "grad_norm": 0.9667718410491943,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.012740559875965118,
+      "mean_token_accuracy": 0.9953394651412963,
+      "num_tokens": 71771906.0,
+      "step": 3800
+    },
+    {
+      "entropy": 0.6755871415138245,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.9068875908851624,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.012314559519290924,
+      "mean_token_accuracy": 0.9955949664115906,
+      "num_tokens": 71960698.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.6791335880756378,
+      "epoch": 5.141318977119784,
+      "grad_norm": 0.7650554776191711,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.011927495896816253,
+      "mean_token_accuracy": 0.9957858562469483,
+      "num_tokens": 72149627.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.6738516509532928,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 0.7267928123474121,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.011817440390586853,
+      "mean_token_accuracy": 0.996101975440979,
+      "num_tokens": 72338469.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.6701940059661865,
+      "epoch": 5.168236877523553,
+      "grad_norm": 0.9211037755012512,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.012324941903352737,
+      "mean_token_accuracy": 0.9956477820873261,
+      "num_tokens": 72527499.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.6763210237026215,
+      "epoch": 5.181695827725437,
+      "grad_norm": 0.7772125601768494,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.013027346134185791,
+      "mean_token_accuracy": 0.9959511935710907,
+      "num_tokens": 72715639.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.6793705642223358,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.582334578037262,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.011401662230491638,
+      "mean_token_accuracy": 0.9961598336696624,
+      "num_tokens": 72904615.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.6798616409301758,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.8502725958824158,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.011671387404203416,
+      "mean_token_accuracy": 0.9961067497730255,
+      "num_tokens": 73093522.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.682939213514328,
+      "epoch": 5.22207267833109,
+      "grad_norm": 0.7421601414680481,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.011898189783096313,
+      "mean_token_accuracy": 0.9962145388126373,
+      "num_tokens": 73282415.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.6821306705474853,
+      "epoch": 5.235531628532974,
+      "grad_norm": 0.7552776336669922,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.01232050210237503,
+      "mean_token_accuracy": 0.9957845032215118,
+      "num_tokens": 73471211.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.6746087670326233,
+      "epoch": 5.248990578734858,
+      "grad_norm": 0.7144336700439453,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.01240069791674614,
+      "mean_token_accuracy": 0.9960526466369629,
+      "num_tokens": 73660289.0,
+      "step": 3900
+    },
+    {
+      "entropy": 0.6723428070545197,
+      "epoch": 5.262449528936743,
+      "grad_norm": 0.8043408393859863,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.01204245388507843,
+      "mean_token_accuracy": 0.9960211217403412,
+      "num_tokens": 73849312.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.6766897737979889,
+      "epoch": 5.275908479138627,
+      "grad_norm": 0.8608251810073853,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.012303075194358826,
+      "mean_token_accuracy": 0.9955043613910675,
+      "num_tokens": 74038344.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.6809355199337006,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.4872569739818573,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.010690021514892577,
+      "mean_token_accuracy": 0.9965942919254303,
+      "num_tokens": 74226981.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.6819032669067383,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.0376152992248535,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.012148186564445496,
+      "mean_token_accuracy": 0.9958918631076813,
+      "num_tokens": 74415130.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.6856858193874359,
+      "epoch": 5.31628532974428,
+      "grad_norm": 0.7241474986076355,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.012284471094608307,
+      "mean_token_accuracy": 0.9959208726882934,
+      "num_tokens": 74603788.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.6874188125133515,
+      "epoch": 5.329744279946164,
+      "grad_norm": 0.6505919098854065,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.012959125638008117,
+      "mean_token_accuracy": 0.9956629753112793,
+      "num_tokens": 74792306.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.6912538707256317,
+      "epoch": 5.343203230148048,
+      "grad_norm": 0.7696103453636169,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.012945668399333954,
+      "mean_token_accuracy": 0.9956756651401519,
+      "num_tokens": 74981265.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.6915560126304626,
+      "epoch": 5.356662180349932,
+      "grad_norm": 0.8245812058448792,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.01408768892288208,
+      "mean_token_accuracy": 0.9950980842113495,
+      "num_tokens": 75170337.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.6872740864753724,
+      "epoch": 5.370121130551817,
+      "grad_norm": 0.8330141305923462,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.011667902022600174,
+      "mean_token_accuracy": 0.9960226893424988,
+      "num_tokens": 75359113.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.6825072944164277,
+      "epoch": 5.383580080753701,
+      "grad_norm": 0.9389140009880066,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.012351009249687194,
+      "mean_token_accuracy": 0.9954748272895813,
+      "num_tokens": 75548102.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.6821270763874054,
+      "epoch": 5.397039030955585,
+      "grad_norm": 1.0638995170593262,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.0128641277551651,
+      "mean_token_accuracy": 0.995329612493515,
+      "num_tokens": 75736452.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.6857812821865081,
+      "epoch": 5.410497981157469,
+      "grad_norm": 0.7838811278343201,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.011757558584213257,
+      "mean_token_accuracy": 0.9959722757339478,
+      "num_tokens": 75925013.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.6869213104248046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 0.5632259845733643,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.011284272372722625,
+      "mean_token_accuracy": 0.9962978959083557,
+      "num_tokens": 76113398.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.6807511866092681,
+      "epoch": 5.437415881561238,
+      "grad_norm": 0.7132044434547424,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.012044224143028259,
+      "mean_token_accuracy": 0.9959294259548187,
+      "num_tokens": 76302629.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.6815520942211151,
+      "epoch": 5.450874831763122,
+      "grad_norm": 0.7640891075134277,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.012763911485671997,
+      "mean_token_accuracy": 0.9953242301940918,
+      "num_tokens": 76491675.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.6802713990211486,
+      "epoch": 5.464333781965006,
+      "grad_norm": 0.5691967010498047,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.0116192027926445,
+      "mean_token_accuracy": 0.9958603858947754,
+      "num_tokens": 76680426.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.676917290687561,
+      "epoch": 5.477792732166891,
+      "grad_norm": 0.8160638213157654,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.01229381114244461,
+      "mean_token_accuracy": 0.9956679701805115,
+      "num_tokens": 76869321.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.6745438575744629,
+      "epoch": 5.491251682368775,
+      "grad_norm": 0.8013969659805298,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.012171164155006409,
+      "mean_token_accuracy": 0.9958188533782959,
+      "num_tokens": 77058299.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.6774423599243165,
+      "epoch": 5.504710632570659,
+      "grad_norm": 0.77237868309021,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.013017389178276061,
+      "mean_token_accuracy": 0.9956206619739533,
+      "num_tokens": 77247567.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.6773760914802551,
+      "epoch": 5.518169582772543,
+      "grad_norm": 0.6185512542724609,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.011151286959648132,
+      "mean_token_accuracy": 0.9961191415786743,
+      "num_tokens": 77436417.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.6750624418258667,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.7832470536231995,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.011553966999053955,
+      "mean_token_accuracy": 0.9958207130432128,
+      "num_tokens": 77625255.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.6814670443534852,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.7947144508361816,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.011924812197685241,
+      "mean_token_accuracy": 0.9959113657474518,
+      "num_tokens": 77813632.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.68285973072052,
+      "epoch": 5.558546433378196,
+      "grad_norm": 0.6403945684432983,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.01066286414861679,
+      "mean_token_accuracy": 0.9966644048690796,
+      "num_tokens": 78002213.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.6764974176883698,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 0.8078592419624329,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.010837660729885101,
+      "mean_token_accuracy": 0.9961750328540802,
+      "num_tokens": 78191471.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.6726884067058563,
+      "epoch": 5.585464333781965,
+      "grad_norm": 0.7293245792388916,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.01087343692779541,
+      "mean_token_accuracy": 0.996295166015625,
+      "num_tokens": 78380424.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.6714434802532196,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0266199111938477,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.011094169318675995,
+      "mean_token_accuracy": 0.9960172355175019,
+      "num_tokens": 78569653.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.6689312815666199,
+      "epoch": 5.612382234185733,
+      "grad_norm": 0.8064524531364441,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.01098542958498001,
+      "mean_token_accuracy": 0.9961700737476349,
+      "num_tokens": 78759287.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.6728951573371887,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 0.7976726293563843,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.012345843017101288,
+      "mean_token_accuracy": 0.9957349181175232,
+      "num_tokens": 78948032.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.6702533006668091,
+      "epoch": 5.639300134589502,
+      "grad_norm": 0.7385782599449158,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.011018334329128266,
+      "mean_token_accuracy": 0.9964040815830231,
+      "num_tokens": 79136972.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.67191202044487,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.6246873736381531,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.011075176298618317,
+      "mean_token_accuracy": 0.9962317526340485,
+      "num_tokens": 79325966.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.6736057937145233,
+      "epoch": 5.66621803499327,
+      "grad_norm": 0.9479376077651978,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.012818500399589539,
+      "mean_token_accuracy": 0.9951797604560852,
+      "num_tokens": 79514853.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.6723325371742248,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 0.6468302607536316,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.010920494794845581,
+      "mean_token_accuracy": 0.9961069703102112,
+      "num_tokens": 79703739.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.6762719631195069,
+      "epoch": 5.693135935397039,
+      "grad_norm": 0.6862959265708923,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.011675214767456055,
+      "mean_token_accuracy": 0.9958149492740631,
+      "num_tokens": 79892582.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.6739612460136414,
+      "epoch": 5.706594885598923,
+      "grad_norm": 0.8402584195137024,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.010931231081485748,
+      "mean_token_accuracy": 0.9960719287395478,
+      "num_tokens": 80081362.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.6767356872558594,
+      "epoch": 5.720053835800807,
+      "grad_norm": 0.6180281043052673,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.012499480694532394,
+      "mean_token_accuracy": 0.9957239329814911,
+      "num_tokens": 80269388.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.6773952782154083,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 0.8782367706298828,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.012943127751350402,
+      "mean_token_accuracy": 0.9958240866661072,
+      "num_tokens": 80458191.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.6774130165576935,
+      "epoch": 5.746971736204576,
+      "grad_norm": 0.7854585647583008,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.012309956550598144,
+      "mean_token_accuracy": 0.99593386054039,
+      "num_tokens": 80647306.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.6813723623752594,
+      "epoch": 5.76043068640646,
+      "grad_norm": 0.6340908408164978,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.01146906390786171,
+      "mean_token_accuracy": 0.9963057219982148,
+      "num_tokens": 80836147.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.6804767549037933,
+      "epoch": 5.773889636608344,
+      "grad_norm": 0.5984897613525391,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.010843665152788163,
+      "mean_token_accuracy": 0.996556407213211,
+      "num_tokens": 81025302.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.6759022295475006,
+      "epoch": 5.787348586810229,
+      "grad_norm": 0.6014022827148438,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.010135546326637268,
+      "mean_token_accuracy": 0.9966514885425568,
+      "num_tokens": 81214251.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.675349909067154,
+      "epoch": 5.800807537012113,
+      "grad_norm": 0.6571088433265686,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.01084519699215889,
+      "mean_token_accuracy": 0.9962432861328125,
+      "num_tokens": 81402608.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.6718680560588837,
+      "epoch": 5.814266487213997,
+      "grad_norm": 0.9250450730323792,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.01139993518590927,
+      "mean_token_accuracy": 0.9961999177932739,
+      "num_tokens": 81591841.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.674647331237793,
+      "epoch": 5.827725437415881,
+      "grad_norm": 0.8669414520263672,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.010886493325233459,
+      "mean_token_accuracy": 0.9962068378925324,
+      "num_tokens": 81780873.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.6801326870918274,
+      "epoch": 5.841184387617766,
+      "grad_norm": 0.7395965456962585,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.012087566405534744,
+      "mean_token_accuracy": 0.9957775354385376,
+      "num_tokens": 81969835.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.676666659116745,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.7648837566375732,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.011214715242385865,
+      "mean_token_accuracy": 0.9963518619537354,
+      "num_tokens": 82159146.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.676488333940506,
+      "epoch": 5.868102288021534,
+      "grad_norm": 0.6776206493377686,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.011839983612298965,
+      "mean_token_accuracy": 0.9959003031253815,
+      "num_tokens": 82347867.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.6743977665901184,
+      "epoch": 5.881561238223418,
+      "grad_norm": 0.7158207893371582,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.011066053062677383,
+      "mean_token_accuracy": 0.9966431736946106,
+      "num_tokens": 82536890.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.6764561176300049,
+      "epoch": 5.895020188425303,
+      "grad_norm": 0.7066213488578796,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.012056462466716766,
+      "mean_token_accuracy": 0.9958225846290588,
+      "num_tokens": 82725921.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.6813391983509064,
+      "epoch": 5.908479138627187,
+      "grad_norm": 0.763606607913971,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.011903972923755645,
+      "mean_token_accuracy": 0.9958972215652466,
+      "num_tokens": 82914838.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.6814220428466797,
+      "epoch": 5.921938088829071,
+      "grad_norm": 0.8407371044158936,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.011677531152963638,
+      "mean_token_accuracy": 0.9959126651287079,
+      "num_tokens": 83103937.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.6833201944828033,
+      "epoch": 5.935397039030955,
+      "grad_norm": 0.8418315052986145,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.012164801359176636,
+      "mean_token_accuracy": 0.9956748068332673,
+      "num_tokens": 83293106.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.6802135646343231,
+      "epoch": 5.94885598923284,
+      "grad_norm": 0.8431894183158875,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.012858036160469054,
+      "mean_token_accuracy": 0.995720499753952,
+      "num_tokens": 83481559.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.6758989214897155,
+      "epoch": 5.962314939434724,
+      "grad_norm": 0.7379446029663086,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.011657674610614777,
+      "mean_token_accuracy": 0.9959479212760926,
+      "num_tokens": 83670620.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.6832396805286407,
+      "epoch": 5.975773889636608,
+      "grad_norm": 0.9370406866073608,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.012775366008281708,
+      "mean_token_accuracy": 0.9958724677562714,
+      "num_tokens": 83858979.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.6826908767223359,
+      "epoch": 5.989232839838492,
+      "grad_norm": 0.5771452188491821,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.011871224641799927,
+      "mean_token_accuracy": 0.9962578654289246,
+      "num_tokens": 84047664.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.6935603318700365,
+      "eval_loss": 0.08950015902519226,
+      "eval_mean_token_accuracy": 0.9720694904874085,
+      "eval_num_tokens": 84198646.0,
+      "eval_runtime": 12.7741,
+      "eval_samples_per_second": 391.416,
+      "eval_steps_per_second": 12.29,
+      "step": 4458
+    },
+    {
+      "entropy": 0.6853953242301941,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.4715442359447479,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.011727289110422135,
+      "mean_token_accuracy": 0.9963658571243286,
+      "num_tokens": 84236476.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.6872529745101928,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.6521468162536621,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.007476924359798432,
+      "mean_token_accuracy": 0.9976866364479064,
+      "num_tokens": 84425424.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.6764522135257721,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.5134896039962769,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.006954360008239746,
+      "mean_token_accuracy": 0.9981172740459442,
+      "num_tokens": 84614973.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.674551123380661,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.5777756571769714,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.0064676709473133085,
+      "mean_token_accuracy": 0.9981890261173249,
+      "num_tokens": 84803584.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.6743139624595642,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.5243175625801086,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.005892159789800644,
+      "mean_token_accuracy": 0.99809131026268,
+      "num_tokens": 84992194.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.6691218018531799,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.6378570795059204,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.006101621687412262,
+      "mean_token_accuracy": 0.9981180131435394,
+      "num_tokens": 85181570.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.6653562963008881,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.7931498885154724,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.0059667728841304776,
+      "mean_token_accuracy": 0.9982839524745941,
+      "num_tokens": 85370354.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.6660257399082183,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.6156598329544067,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.006237205862998962,
+      "mean_token_accuracy": 0.9980066597461701,
+      "num_tokens": 85559287.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.6661248922348022,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.39466625452041626,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.006016004830598831,
+      "mean_token_accuracy": 0.9984841406345367,
+      "num_tokens": 85748489.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.6644636154174804,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.556770920753479,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.0060367342084646225,
+      "mean_token_accuracy": 0.9981182813644409,
+      "num_tokens": 85937336.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.6630209028720856,
+      "epoch": 6.137281292059219,
+      "grad_norm": 0.5270616412162781,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.005779195204377174,
+      "mean_token_accuracy": 0.9981250524520874,
+      "num_tokens": 86126286.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.6656446874141693,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 0.44525614380836487,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.005375263467431069,
+      "mean_token_accuracy": 0.9983476758003235,
+      "num_tokens": 86314904.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.664457255601883,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.6622279286384583,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.006194194033741951,
+      "mean_token_accuracy": 0.9981902241706848,
+      "num_tokens": 86503447.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.6608214318752289,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.6172921657562256,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.0061217501759529116,
+      "mean_token_accuracy": 0.9983710527420044,
+      "num_tokens": 86693005.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.6578842639923096,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.6241856217384338,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.006712782382965088,
+      "mean_token_accuracy": 0.9979998171329498,
+      "num_tokens": 86882287.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.6561006605625153,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 0.6784538626670837,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.006533285975456238,
+      "mean_token_accuracy": 0.9977931976318359,
+      "num_tokens": 87071532.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.6582947254180909,
+      "epoch": 6.218034993270525,
+      "grad_norm": 0.7572171688079834,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.006016571819782257,
+      "mean_token_accuracy": 0.9979285359382629,
+      "num_tokens": 87260158.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.6552021980285645,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.7922662496566772,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.006150579452514649,
+      "mean_token_accuracy": 0.9982126474380493,
+      "num_tokens": 87448718.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.6578958451747894,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.5214764475822449,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.005079228430986404,
+      "mean_token_accuracy": 0.9984815359115601,
+      "num_tokens": 87637461.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.6603941440582275,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.5404514670372009,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.005841487646102905,
+      "mean_token_accuracy": 0.9983745515346527,
+      "num_tokens": 87826292.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.6628166377544403,
+      "epoch": 6.271870794078062,
+      "grad_norm": 0.47433987259864807,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.006364698708057404,
+      "mean_token_accuracy": 0.9982446670532227,
+      "num_tokens": 88015224.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.6635581731796265,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.8039814233779907,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.006831821799278259,
+      "mean_token_accuracy": 0.9982130885124206,
+      "num_tokens": 88203965.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.6615313947200775,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.5904889106750488,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.006949013471603394,
+      "mean_token_accuracy": 0.9976236402988434,
+      "num_tokens": 88392910.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.6613587439060211,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.6321392059326172,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.006564225256443024,
+      "mean_token_accuracy": 0.9980846107006073,
+      "num_tokens": 88581883.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.6642421364784241,
+      "epoch": 6.325706594885599,
+      "grad_norm": 0.6209113597869873,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.007112482190132141,
+      "mean_token_accuracy": 0.9977501094341278,
+      "num_tokens": 88770574.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.6629104316234589,
+      "epoch": 6.339165545087483,
+      "grad_norm": 0.8903343081474304,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.00698748379945755,
+      "mean_token_accuracy": 0.9977229118347168,
+      "num_tokens": 88959487.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.6647967278957367,
+      "epoch": 6.352624495289367,
+      "grad_norm": 0.842950165271759,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.008400703966617584,
+      "mean_token_accuracy": 0.9975436449050903,
+      "num_tokens": 89148302.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.6641393005847931,
+      "epoch": 6.366083445491252,
+      "grad_norm": 0.6241896152496338,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.006545399129390716,
+      "mean_token_accuracy": 0.9982439935207367,
+      "num_tokens": 89337168.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.6646088480949401,
+      "epoch": 6.379542395693136,
+      "grad_norm": 0.6284488439559937,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.005783513188362122,
+      "mean_token_accuracy": 0.9982234418392182,
+      "num_tokens": 89525632.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.6648603081703186,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.7705378532409668,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.006973850727081299,
+      "mean_token_accuracy": 0.9977744221687317,
+      "num_tokens": 89714436.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.6667861938476562,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.7033233642578125,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.0062720529735088345,
+      "mean_token_accuracy": 0.9978694558143616,
+      "num_tokens": 89903591.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.66892409324646,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.6684027910232544,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.006154683232307434,
+      "mean_token_accuracy": 0.998307728767395,
+      "num_tokens": 90092231.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.6692153334617614,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.7860879302024841,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.006273554265499115,
+      "mean_token_accuracy": 0.9981006205081939,
+      "num_tokens": 90280721.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.6636229634284974,
+      "epoch": 6.446837146702557,
+      "grad_norm": 0.5911756753921509,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.0067244857549667355,
+      "mean_token_accuracy": 0.9979658007621766,
+      "num_tokens": 90469835.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.6663640201091766,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.5844401717185974,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.005917467921972275,
+      "mean_token_accuracy": 0.9981400012969971,
+      "num_tokens": 90658439.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.6635376453399658,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.554587721824646,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.005905192345380783,
+      "mean_token_accuracy": 0.9983574509620666,
+      "num_tokens": 90847689.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.660474544763565,
+      "epoch": 6.48721399730821,
+      "grad_norm": 0.6828640103340149,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.006179597228765488,
+      "mean_token_accuracy": 0.9978089511394501,
+      "num_tokens": 91036945.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.6635875403881073,
+      "epoch": 6.500672947510094,
+      "grad_norm": 0.621636688709259,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.006481277197599411,
+      "mean_token_accuracy": 0.9976639211177826,
+      "num_tokens": 91225796.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.6625118017196655,
+      "epoch": 6.514131897711978,
+      "grad_norm": 0.6648721098899841,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.006836448609828949,
+      "mean_token_accuracy": 0.9981350302696228,
+      "num_tokens": 91414454.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.6632138609886169,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.6122362017631531,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.0057271294295787815,
+      "mean_token_accuracy": 0.9983117520809174,
+      "num_tokens": 91603019.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.6621270835399627,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.707145094871521,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.005987466499209404,
+      "mean_token_accuracy": 0.9983782589435577,
+      "num_tokens": 91791729.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.6593974709510804,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.7802532315254211,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.00661279633641243,
+      "mean_token_accuracy": 0.9978727638721466,
+      "num_tokens": 91980977.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.658482962846756,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 0.8637571334838867,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.005754219368100166,
+      "mean_token_accuracy": 0.9982581317424775,
+      "num_tokens": 92169829.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.6561014533042908,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.7007631659507751,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.006207703426480293,
+      "mean_token_accuracy": 0.9980942070484161,
+      "num_tokens": 92358635.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.6558376729488373,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.6180236339569092,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.005951264500617981,
+      "mean_token_accuracy": 0.9980935871601104,
+      "num_tokens": 92546955.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.6548421680927277,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.5694975852966309,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.0058226577937603,
+      "mean_token_accuracy": 0.9980980336666108,
+      "num_tokens": 92735651.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.6563037514686585,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 0.7417933344841003,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.006121259927749634,
+      "mean_token_accuracy": 0.9980963408946991,
+      "num_tokens": 92925010.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.654166966676712,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.705633282661438,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.005783010274171829,
+      "mean_token_accuracy": 0.9983141779899597,
+      "num_tokens": 93113992.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.6567890524864197,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.5444097518920898,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.005712105333805085,
+      "mean_token_accuracy": 0.9982166588306427,
+      "num_tokens": 93302433.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.6536349058151245,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.4291290044784546,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.0060494735836982725,
+      "mean_token_accuracy": 0.9979224085807801,
+      "num_tokens": 93491429.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.6516136825084686,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.5134868621826172,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.005212320387363434,
+      "mean_token_accuracy": 0.9984170615673065,
+      "num_tokens": 93679973.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.6528811275959014,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.8154370188713074,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.00622846707701683,
+      "mean_token_accuracy": 0.998017430305481,
+      "num_tokens": 93868735.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.6532892882823944,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.5574272871017456,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.005193561688065529,
+      "mean_token_accuracy": 0.9985900521278381,
+      "num_tokens": 94057428.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.6552170813083649,
+      "epoch": 6.716016150740242,
+      "grad_norm": 0.6016380190849304,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.005771375447511673,
+      "mean_token_accuracy": 0.9981073677539826,
+      "num_tokens": 94246297.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.6572140097618103,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.7832751870155334,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.005496300011873245,
+      "mean_token_accuracy": 0.9983509957790375,
+      "num_tokens": 94434253.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.6541008234024048,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.4329380989074707,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.006188442185521126,
+      "mean_token_accuracy": 0.9981851994991302,
+      "num_tokens": 94623327.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.6556220591068268,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.5326682329177856,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.0052257638424634935,
+      "mean_token_accuracy": 0.9983932614326477,
+      "num_tokens": 94811837.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.6557375073432923,
+      "epoch": 6.769851951547779,
+      "grad_norm": 0.6864677667617798,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.005970630794763565,
+      "mean_token_accuracy": 0.9983245611190796,
+      "num_tokens": 95000726.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.6545388638973236,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.6318016648292542,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.006516661494970322,
+      "mean_token_accuracy": 0.9979697644710541,
+      "num_tokens": 95189395.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.6542861998081207,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.3979317247867584,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.005074064806103706,
+      "mean_token_accuracy": 0.998512452840805,
+      "num_tokens": 95378855.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.6542766332626343,
+      "epoch": 6.810228802153432,
+      "grad_norm": 0.4778045117855072,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.005774036794900894,
+      "mean_token_accuracy": 0.9982702076435089,
+      "num_tokens": 95568111.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.6550499975681305,
+      "epoch": 6.823687752355316,
+      "grad_norm": 0.4805110991001129,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.005620773509144783,
+      "mean_token_accuracy": 0.9983035147190094,
+      "num_tokens": 95756885.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.6543555080890655,
+      "epoch": 6.837146702557201,
+      "grad_norm": 0.6604891419410706,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.005247855931520462,
+      "mean_token_accuracy": 0.998341029882431,
+      "num_tokens": 95946078.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.6581745266914367,
+      "epoch": 6.850605652759085,
+      "grad_norm": 0.6589891910552979,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.005882954970002174,
+      "mean_token_accuracy": 0.9982152104377746,
+      "num_tokens": 96134597.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.6597592771053314,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.4152396023273468,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.0054433666169643406,
+      "mean_token_accuracy": 0.9983908593654632,
+      "num_tokens": 96323327.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.658534574508667,
+      "epoch": 6.877523553162853,
+      "grad_norm": 0.6025403141975403,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.005522595718502998,
+      "mean_token_accuracy": 0.9983524858951569,
+      "num_tokens": 96512170.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.6568575739860535,
+      "epoch": 6.890982503364738,
+      "grad_norm": 0.5730848908424377,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.004784010350704193,
+      "mean_token_accuracy": 0.9984236299991608,
+      "num_tokens": 96701363.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.6535935163497925,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.5301522612571716,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.004592524468898773,
+      "mean_token_accuracy": 0.9987652719020843,
+      "num_tokens": 96890642.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.6544224679470062,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.639918863773346,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.005577490478754043,
+      "mean_token_accuracy": 0.9984355092048645,
+      "num_tokens": 97079552.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.6542503774166107,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.5202826857566833,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.005623985454440117,
+      "mean_token_accuracy": 0.9982783615589141,
+      "num_tokens": 97268346.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.6499157190322876,
+      "epoch": 6.944818304172275,
+      "grad_norm": 0.5668892860412598,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.005104385688900948,
+      "mean_token_accuracy": 0.9985612750053405,
+      "num_tokens": 97457514.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.6520473599433899,
+      "epoch": 6.958277254374159,
+      "grad_norm": 0.5878952145576477,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.005280185490846634,
+      "mean_token_accuracy": 0.9985166370868683,
+      "num_tokens": 97646014.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.6488265216350555,
+      "epoch": 6.971736204576043,
+      "grad_norm": 0.4630887806415558,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.005359554663300514,
+      "mean_token_accuracy": 0.998493617773056,
+      "num_tokens": 97835401.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.6526197671890259,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.38062795996665955,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.0054448477923870085,
+      "mean_token_accuracy": 0.9982813715934753,
+      "num_tokens": 98024194.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.654500275850296,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.5739313364028931,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.005691033974289894,
+      "mean_token_accuracy": 0.9982043981552124,
+      "num_tokens": 98212962.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.6537400142402406,
+      "eval_loss": 0.09857651591300964,
+      "eval_mean_token_accuracy": 0.9730726879113799,
+      "eval_num_tokens": 98231758.0,
+      "eval_runtime": 12.7391,
+      "eval_samples_per_second": 392.491,
+      "eval_steps_per_second": 12.324,
+      "step": 5201
+    },
+    {
+      "entropy": 0.6512321710586548,
+      "epoch": 7.012113055181696,
+      "grad_norm": 0.2961742877960205,
+      "learning_rate": 2.2506648060191042e-05,
+      "loss": 0.0032418236136436463,
+      "mean_token_accuracy": 0.999194347858429,
+      "num_tokens": 98401402.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.6491944074630738,
+      "epoch": 7.02557200538358,
+      "grad_norm": 0.35240232944488525,
+      "learning_rate": 2.2321031102941066e-05,
+      "loss": 0.0028009919449687006,
+      "mean_token_accuracy": 0.9995193898677825,
+      "num_tokens": 98590150.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.6500660896301269,
+      "epoch": 7.039030955585464,
+      "grad_norm": 0.3501753509044647,
+      "learning_rate": 2.2135962530768055e-05,
+      "loss": 0.0029570702463388443,
+      "mean_token_accuracy": 0.9993301749229431,
+      "num_tokens": 98778612.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.6460368275642395,
+      "epoch": 7.052489905787349,
+      "grad_norm": 0.41639405488967896,
+      "learning_rate": 2.1951446010313053e-05,
+      "loss": 0.003057733364403248,
+      "mean_token_accuracy": 0.9994841694831849,
+      "num_tokens": 98967879.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.6441825687885284,
+      "epoch": 7.065948855989233,
+      "grad_norm": 0.3744174540042877,
+      "learning_rate": 2.1767485197279662e-05,
+      "loss": 0.0028053496032953264,
+      "mean_token_accuracy": 0.9994258165359498,
+      "num_tokens": 99157371.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.6460713386535645,
+      "epoch": 7.079407806191117,
+      "grad_norm": 0.22786536812782288,
+      "learning_rate": 2.1584083736361617e-05,
+      "loss": 0.004050987213850022,
+      "mean_token_accuracy": 0.9992221534252167,
+      "num_tokens": 99346074.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.6439846336841584,
+      "epoch": 7.0928667563930015,
+      "grad_norm": 0.33427926898002625,
+      "learning_rate": 2.1401245261170582e-05,
+      "loss": 0.002635418064892292,
+      "mean_token_accuracy": 0.9993825376033783,
+      "num_tokens": 99535282.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.6429140031337738,
+      "epoch": 7.106325706594886,
+      "grad_norm": 0.35418426990509033,
+      "learning_rate": 2.121897339416416e-05,
+      "loss": 0.002664094232022762,
+      "mean_token_accuracy": 0.999459445476532,
+      "num_tokens": 99724199.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.6431400418281555,
+      "epoch": 7.11978465679677,
+      "grad_norm": 0.2773686945438385,
+      "learning_rate": 2.103727174657412e-05,
+      "loss": 0.0029107578098773955,
+      "mean_token_accuracy": 0.999220222234726,
+      "num_tokens": 99913190.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.644054365158081,
+      "epoch": 7.133243606998654,
+      "grad_norm": 0.466249942779541,
+      "learning_rate": 2.085614391833486e-05,
+      "loss": 0.0026346595957875254,
+      "mean_token_accuracy": 0.9993867635726928,
+      "num_tokens": 100102083.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.6469399213790894,
+      "epoch": 7.1467025572005385,
+      "grad_norm": 0.37575817108154297,
+      "learning_rate": 2.0675593498012052e-05,
+      "loss": 0.0028098784387111664,
+      "mean_token_accuracy": 0.9992999136447906,
+      "num_tokens": 100290764.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.6454071402549744,
+      "epoch": 7.160161507402423,
+      "grad_norm": 0.5268272161483765,
+      "learning_rate": 2.049562406273159e-05,
+      "loss": 0.0025207167491316795,
+      "mean_token_accuracy": 0.9995582520961761,
+      "num_tokens": 100479571.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.6419129431247711,
+      "epoch": 7.173620457604307,
+      "grad_norm": 0.28378117084503174,
+      "learning_rate": 2.0316239178108688e-05,
+      "loss": 0.00287601538002491,
+      "mean_token_accuracy": 0.9993176221847534,
+      "num_tokens": 100668516.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.6420379161834717,
+      "epoch": 7.187079407806191,
+      "grad_norm": 0.24231773614883423,
+      "learning_rate": 2.013744239817725e-05,
+      "loss": 0.002684992924332619,
+      "mean_token_accuracy": 0.9993604898452759,
+      "num_tokens": 100857247.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.6403457880020141,
+      "epoch": 7.2005383580080755,
+      "grad_norm": 0.350322961807251,
+      "learning_rate": 1.9959237265319452e-05,
+      "loss": 0.002843806892633438,
+      "mean_token_accuracy": 0.9992822408676147,
+      "num_tokens": 101046251.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.6437962651252747,
+      "epoch": 7.21399730820996,
+      "grad_norm": 0.504984974861145,
+      "learning_rate": 1.9781627310195528e-05,
+      "loss": 0.002904067933559418,
+      "mean_token_accuracy": 0.9991836249828339,
+      "num_tokens": 101235009.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.6444802343845367,
+      "epoch": 7.227456258411844,
+      "grad_norm": 0.5015905499458313,
+      "learning_rate": 1.960461605167389e-05,
+      "loss": 0.0031951837241649627,
+      "mean_token_accuracy": 0.9991663873195649,
+      "num_tokens": 101423795.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.6445916533470154,
+      "epoch": 7.240915208613728,
+      "grad_norm": 0.8163104057312012,
+      "learning_rate": 1.9428206996761357e-05,
+      "loss": 0.0031633909791707994,
+      "mean_token_accuracy": 0.9990993022918702,
+      "num_tokens": 101612550.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.6424523174762726,
+      "epoch": 7.2543741588156125,
+      "grad_norm": 0.7167068123817444,
+      "learning_rate": 1.9252403640533678e-05,
+      "loss": 0.0027146343141794204,
+      "mean_token_accuracy": 0.9993792653083802,
+      "num_tokens": 101801308.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.6425661742687225,
+      "epoch": 7.267833109017497,
+      "grad_norm": 0.3235819935798645,
+      "learning_rate": 1.9077209466066315e-05,
+      "loss": 0.002999866008758545,
+      "mean_token_accuracy": 0.9993207931518555,
+      "num_tokens": 101990167.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.6397831857204437,
+      "epoch": 7.281292059219381,
+      "grad_norm": 0.2559046149253845,
+      "learning_rate": 1.8902627944365396e-05,
+      "loss": 0.0023009315133094787,
+      "mean_token_accuracy": 0.9995379090309143,
+      "num_tokens": 102179185.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.638386732339859,
+      "epoch": 7.294751009421265,
+      "grad_norm": 0.4241917133331299,
+      "learning_rate": 1.872866253429897e-05,
+      "loss": 0.003075192496180534,
+      "mean_token_accuracy": 0.9992788910865784,
+      "num_tokens": 102368161.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.6398141801357269,
+      "epoch": 7.30820995962315,
+      "grad_norm": 0.4451645612716675,
+      "learning_rate": 1.8555316682528485e-05,
+      "loss": 0.0027588184922933577,
+      "mean_token_accuracy": 0.9993425786495209,
+      "num_tokens": 102557229.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.6400523662567139,
+      "epoch": 7.321668909825034,
+      "grad_norm": 0.2873964309692383,
+      "learning_rate": 1.8382593823440474e-05,
+      "loss": 0.0026803961023688315,
+      "mean_token_accuracy": 0.9994625568389892,
+      "num_tokens": 102745928.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.6401661276817322,
+      "epoch": 7.335127860026918,
+      "grad_norm": 0.5918443202972412,
+      "learning_rate": 1.821049737907854e-05,
+      "loss": 0.0029163889586925505,
+      "mean_token_accuracy": 0.9992423593997956,
+      "num_tokens": 102934875.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.6410463333129883,
+      "epoch": 7.348586810228802,
+      "grad_norm": 0.39631369709968567,
+      "learning_rate": 1.803903075907554e-05,
+      "loss": 0.002800934761762619,
+      "mean_token_accuracy": 0.9993205904960633,
+      "num_tokens": 103123665.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.6384492158889771,
+      "epoch": 7.362045760430687,
+      "grad_norm": 0.44129809737205505,
+      "learning_rate": 1.7868197360586054e-05,
+      "loss": 0.0027701158076524736,
+      "mean_token_accuracy": 0.9992826044559479,
+      "num_tokens": 103312769.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.6399481356143951,
+      "epoch": 7.375504710632571,
+      "grad_norm": 0.3396782875061035,
+      "learning_rate": 1.769800056821904e-05,
+      "loss": 0.002843455225229263,
+      "mean_token_accuracy": 0.9993624866008759,
+      "num_tokens": 103501428.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.6391602873802185,
+      "epoch": 7.388963660834455,
+      "grad_norm": 0.508573055267334,
+      "learning_rate": 1.7528443753970814e-05,
+      "loss": 0.0030046245083212852,
+      "mean_token_accuracy": 0.9992032408714294,
+      "num_tokens": 103690316.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.641320925951004,
+      "epoch": 7.402422611036339,
+      "grad_norm": 0.372712641954422,
+      "learning_rate": 1.7359530277158227e-05,
+      "loss": 0.0029981501400470735,
+      "mean_token_accuracy": 0.9992169618606568,
+      "num_tokens": 103879262.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.6416053056716919,
+      "epoch": 7.415881561238224,
+      "grad_norm": 0.47443583607673645,
+      "learning_rate": 1.7191263484352106e-05,
+      "loss": 0.00280308797955513,
+      "mean_token_accuracy": 0.9993564009666442,
+      "num_tokens": 104067769.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.6400010466575623,
+      "epoch": 7.429340511440108,
+      "grad_norm": 0.26680561900138855,
+      "learning_rate": 1.702364670931097e-05,
+      "loss": 0.0029377201572060585,
+      "mean_token_accuracy": 0.9992177426815033,
+      "num_tokens": 104256396.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.6378615140914917,
+      "epoch": 7.442799461641992,
+      "grad_norm": 0.3017747700214386,
+      "learning_rate": 1.685668327291495e-05,
+      "loss": 0.0025883980095386507,
+      "mean_token_accuracy": 0.9993183672428131,
+      "num_tokens": 104445275.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.6393269956111908,
+      "epoch": 7.456258411843876,
+      "grad_norm": 0.33738431334495544,
+      "learning_rate": 1.6690376483100034e-05,
+      "loss": 0.0023764621466398237,
+      "mean_token_accuracy": 0.999459034204483,
+      "num_tokens": 104633705.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.6352120459079742,
+      "epoch": 7.469717362045761,
+      "grad_norm": 0.30927929282188416,
+      "learning_rate": 1.6524729634792475e-05,
+      "loss": 0.002412043884396553,
+      "mean_token_accuracy": 0.9993467569351197,
+      "num_tokens": 104822986.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.6355067729949951,
+      "epoch": 7.483176312247645,
+      "grad_norm": 0.3294868469238281,
+      "learning_rate": 1.6359746009843567e-05,
+      "loss": 0.002515685372054577,
+      "mean_token_accuracy": 0.999462366104126,
+      "num_tokens": 105012029.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.6385121285915375,
+      "epoch": 7.496635262449529,
+      "grad_norm": 0.4267696738243103,
+      "learning_rate": 1.6195428876964592e-05,
+      "loss": 0.0027210000902414323,
+      "mean_token_accuracy": 0.9993004381656647,
+      "num_tokens": 105200942.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.641570907831192,
+      "epoch": 7.510094212651413,
+      "grad_norm": 0.28043922781944275,
+      "learning_rate": 1.6031781491662056e-05,
+      "loss": 0.0025119861587882043,
+      "mean_token_accuracy": 0.9993769824504852,
+      "num_tokens": 105389627.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.6401131987571717,
+      "epoch": 7.523553162853298,
+      "grad_norm": 0.5159914493560791,
+      "learning_rate": 1.58688070961732e-05,
+      "loss": 0.0026847902685403825,
+      "mean_token_accuracy": 0.9993621706962585,
+      "num_tokens": 105578832.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.6401657581329345,
+      "epoch": 7.537012113055182,
+      "grad_norm": 0.47359421849250793,
+      "learning_rate": 1.5706508919401804e-05,
+      "loss": 0.0025615351274609565,
+      "mean_token_accuracy": 0.999418193101883,
+      "num_tokens": 105767407.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.640064799785614,
+      "epoch": 7.550471063257066,
+      "grad_norm": 0.2965790927410126,
+      "learning_rate": 1.554489017685412e-05,
+      "loss": 0.002729836292564869,
+      "mean_token_accuracy": 0.999356073141098,
+      "num_tokens": 105955938.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.6364653050899506,
+      "epoch": 7.56393001345895,
+      "grad_norm": 0.3772422969341278,
+      "learning_rate": 1.538395407057525e-05,
+      "loss": 0.002501054294407368,
+      "mean_token_accuracy": 0.99944589138031,
+      "num_tokens": 106145287.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.6365915596485138,
+      "epoch": 7.577388963660835,
+      "grad_norm": 0.34281232953071594,
+      "learning_rate": 1.522370378908567e-05,
+      "loss": 0.0024556241929531097,
+      "mean_token_accuracy": 0.9995415151119232,
+      "num_tokens": 106334420.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.6392146408557892,
+      "epoch": 7.590847913862719,
+      "grad_norm": 0.289888858795166,
+      "learning_rate": 1.5064142507318051e-05,
+      "loss": 0.0024054452776908876,
+      "mean_token_accuracy": 0.999481338262558,
+      "num_tokens": 106523346.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.63958261013031,
+      "epoch": 7.604306864064603,
+      "grad_norm": 0.4470290541648865,
+      "learning_rate": 1.4905273386554386e-05,
+      "loss": 0.002950721047818661,
+      "mean_token_accuracy": 0.9994794666767121,
+      "num_tokens": 106711623.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.6364846646785736,
+      "epoch": 7.6177658142664875,
+      "grad_norm": 0.3294987678527832,
+      "learning_rate": 1.4747099574363332e-05,
+      "loss": 0.0029568497091531755,
+      "mean_token_accuracy": 0.99925497174263,
+      "num_tokens": 106900044.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.6371748626232148,
+      "epoch": 7.631224764468372,
+      "grad_norm": 0.3742607533931732,
+      "learning_rate": 1.4589624204537866e-05,
+      "loss": 0.002776237577199936,
+      "mean_token_accuracy": 0.9994770407676696,
+      "num_tokens": 107088499.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.6374151885509491,
+      "epoch": 7.644683714670256,
+      "grad_norm": 0.34489092230796814,
+      "learning_rate": 1.4432850397033176e-05,
+      "loss": 0.0025846634060144426,
+      "mean_token_accuracy": 0.9992987811565399,
+      "num_tokens": 107277120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.6350409626960755,
+      "epoch": 7.65814266487214,
+      "grad_norm": 0.4673531651496887,
+      "learning_rate": 1.427678125790487e-05,
+      "loss": 0.002661401592195034,
+      "mean_token_accuracy": 0.9993412852287292,
+      "num_tokens": 107466227.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.6376935839653015,
+      "epoch": 7.6716016150740245,
+      "grad_norm": 0.2515313923358917,
+      "learning_rate": 1.4121419879247422e-05,
+      "loss": 0.0023796845227479933,
+      "mean_token_accuracy": 0.999422425031662,
+      "num_tokens": 107654984.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.6349337220191955,
+      "epoch": 7.685060565275909,
+      "grad_norm": 0.38676151633262634,
+      "learning_rate": 1.3966769339132929e-05,
+      "loss": 0.002421387284994125,
+      "mean_token_accuracy": 0.999384069442749,
+      "num_tokens": 107844126.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.6366126656532287,
+      "epoch": 7.698519515477793,
+      "grad_norm": 0.39246416091918945,
+      "learning_rate": 1.3812832701550099e-05,
+      "loss": 0.0024261830374598505,
+      "mean_token_accuracy": 0.9993186116218566,
+      "num_tokens": 108032874.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.6357373297214508,
+      "epoch": 7.711978465679677,
+      "grad_norm": 0.319026917219162,
+      "learning_rate": 1.365961301634357e-05,
+      "loss": 0.0023328609764575957,
+      "mean_token_accuracy": 0.9995430946350098,
+      "num_tokens": 108222192.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.6360769748687745,
+      "epoch": 7.7254374158815615,
+      "grad_norm": 0.38521459698677063,
+      "learning_rate": 1.350711331915348e-05,
+      "loss": 0.0026227174326777457,
+      "mean_token_accuracy": 0.9992436349391938,
+      "num_tokens": 108410750.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.6360529124736786,
+      "epoch": 7.738896366083446,
+      "grad_norm": 0.2718389928340912,
+      "learning_rate": 1.3355336631355314e-05,
+      "loss": 0.0024306148290634154,
+      "mean_token_accuracy": 0.9995024859905243,
+      "num_tokens": 108599812.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.6351405322551728,
+      "epoch": 7.75235531628533,
+      "grad_norm": 0.38458114862442017,
+      "learning_rate": 1.3204285960000051e-05,
+      "loss": 0.002783242240548134,
+      "mean_token_accuracy": 0.9993249535560608,
+      "num_tokens": 108788623.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.6342482805252075,
+      "epoch": 7.765814266487214,
+      "grad_norm": 0.37706634402275085,
+      "learning_rate": 1.30539642977546e-05,
+      "loss": 0.002872874028980732,
+      "mean_token_accuracy": 0.9993605136871337,
+      "num_tokens": 108977505.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.6347688436508179,
+      "epoch": 7.7792732166890985,
+      "grad_norm": 0.44668373465538025,
+      "learning_rate": 1.2904374622842486e-05,
+      "loss": 0.002442394569516182,
+      "mean_token_accuracy": 0.999343854188919,
+      "num_tokens": 109166589.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.6364605367183686,
+      "epoch": 7.792732166890983,
+      "grad_norm": 0.2349705696105957,
+      "learning_rate": 1.2755519898984846e-05,
+      "loss": 0.00250922292470932,
+      "mean_token_accuracy": 0.9995181024074554,
+      "num_tokens": 109355292.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.6371697664260865,
+      "epoch": 7.806191117092867,
+      "grad_norm": 0.3123619556427002,
+      "learning_rate": 1.2607403075341728e-05,
+      "loss": 0.002688922733068466,
+      "mean_token_accuracy": 0.9993325173854828,
+      "num_tokens": 109543865.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.6380978763103485,
+      "epoch": 7.819650067294751,
+      "grad_norm": 0.2686324417591095,
+      "learning_rate": 1.2460027086453658e-05,
+      "loss": 0.0023211821913719176,
+      "mean_token_accuracy": 0.9995401680469513,
+      "num_tokens": 109732486.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.6382215797901154,
+      "epoch": 7.833109017496636,
+      "grad_norm": 0.4553178548812866,
+      "learning_rate": 1.2313394852183473e-05,
+      "loss": 0.0024582967162132263,
+      "mean_token_accuracy": 0.9994239926338195,
+      "num_tokens": 109921491.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.6385389983654022,
+      "epoch": 7.84656796769852,
+      "grad_norm": 0.2880251109600067,
+      "learning_rate": 1.2167509277658518e-05,
+      "loss": 0.0024548277258872988,
+      "mean_token_accuracy": 0.9994565665721893,
+      "num_tokens": 110110256.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.633701753616333,
+      "epoch": 7.860026917900404,
+      "grad_norm": 0.4071319103240967,
+      "learning_rate": 1.2022373253213049e-05,
+      "loss": 0.0025199253112077715,
+      "mean_token_accuracy": 0.9993486106395721,
+      "num_tokens": 110299624.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.6337139904499054,
+      "epoch": 7.873485868102288,
+      "grad_norm": 0.4106242060661316,
+      "learning_rate": 1.1877989654330973e-05,
+      "loss": 0.0022239934653043745,
+      "mean_token_accuracy": 0.999601423740387,
+      "num_tokens": 110488951.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.6346798777580261,
+      "epoch": 7.886944818304173,
+      "grad_norm": 0.5445306301116943,
+      "learning_rate": 1.1734361341588912e-05,
+      "loss": 0.0027007974684238436,
+      "mean_token_accuracy": 0.9993659555912018,
+      "num_tokens": 110678171.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.6366956651210784,
+      "epoch": 7.900403768506057,
+      "grad_norm": 0.2735433280467987,
+      "learning_rate": 1.1591491160599488e-05,
+      "loss": 0.0026022626087069513,
+      "mean_token_accuracy": 0.9993394792079926,
+      "num_tokens": 110867003.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.6380104124546051,
+      "epoch": 7.913862718707941,
+      "grad_norm": 0.5340652465820312,
+      "learning_rate": 1.1449381941954956e-05,
+      "loss": 0.002648356184363365,
+      "mean_token_accuracy": 0.9994826078414917,
+      "num_tokens": 111055985.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.6362045764923095,
+      "epoch": 7.927321668909825,
+      "grad_norm": 0.3767072856426239,
+      "learning_rate": 1.130803650117116e-05,
+      "loss": 0.0024559229612350466,
+      "mean_token_accuracy": 0.9995233714580536,
+      "num_tokens": 111244949.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.6355053067207337,
+      "epoch": 7.94078061911171,
+      "grad_norm": 0.3835017681121826,
+      "learning_rate": 1.1167457638631656e-05,
+      "loss": 0.002397604286670685,
+      "mean_token_accuracy": 0.9995415270328522,
+      "num_tokens": 111433946.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.6365890502929688,
+      "epoch": 7.954239569313594,
+      "grad_norm": 0.3102132976055145,
+      "learning_rate": 1.1027648139532355e-05,
+      "loss": 0.002390575036406517,
+      "mean_token_accuracy": 0.9994417130947113,
+      "num_tokens": 111622676.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.6346444308757782,
+      "epoch": 7.967698519515478,
+      "grad_norm": 0.49554574489593506,
+      "learning_rate": 1.0888610773826253e-05,
+      "loss": 0.0021300587803125383,
+      "mean_token_accuracy": 0.9996205389499664,
+      "num_tokens": 111811705.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.6345613896846771,
+      "epoch": 7.981157469717362,
+      "grad_norm": 0.2676335871219635,
+      "learning_rate": 1.0750348296168583e-05,
+      "loss": 0.0024733129888772965,
+      "mean_token_accuracy": 0.9994426965713501,
+      "num_tokens": 112000972.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.6350580513477325,
+      "epoch": 7.994616419919247,
+      "grad_norm": 0.20178340375423431,
+      "learning_rate": 1.0612863445862236e-05,
+      "loss": 0.002096991054713726,
+      "mean_token_accuracy": 0.9996170699596405,
+      "num_tokens": 112189178.0,
+      "step": 5940
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.6389722296386767,
+      "eval_loss": 0.10558179020881653,
+      "eval_mean_token_accuracy": 0.9726916976795075,
+      "eval_num_tokens": 112264896.0,
+      "eval_runtime": 12.7264,
+      "eval_samples_per_second": 392.885,
+      "eval_steps_per_second": 12.337,
+      "step": 5944
+    },
+    {
+      "entropy": 0.6323651373386383,
+      "epoch": 8.00807537012113,
+      "grad_norm": 0.2156553864479065,
+      "learning_rate": 1.0476158946803482e-05,
+      "loss": 0.0017782203853130341,
+      "mean_token_accuracy": 0.9996409475803375,
+      "num_tokens": 112378248.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.6356893301010131,
+      "epoch": 8.021534320323015,
+      "grad_norm": 0.4005093574523926,
+      "learning_rate": 1.0340237507428009e-05,
+      "loss": 0.0015996076166629792,
+      "mean_token_accuracy": 0.9998390734195709,
+      "num_tokens": 112566565.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.6323986113071441,
+      "epoch": 8.034993270524899,
+      "grad_norm": 0.2522565722465515,
+      "learning_rate": 1.020510182065727e-05,
+      "loss": 0.0014581691473722457,
+      "mean_token_accuracy": 0.9998226821422577,
+      "num_tokens": 112755790.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.6342778027057647,
+      "epoch": 8.048452220726784,
+      "grad_norm": 0.14223679900169373,
+      "learning_rate": 1.0070754563845114e-05,
+      "loss": 0.0015433312393724919,
+      "mean_token_accuracy": 0.9998199999332428,
+      "num_tokens": 112944666.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.6335026860237122,
+      "epoch": 8.061911170928667,
+      "grad_norm": 0.45429059863090515,
+      "learning_rate": 9.937198398724751e-06,
+      "loss": 0.0015122929587960243,
+      "mean_token_accuracy": 0.9997402548789978,
+      "num_tokens": 113133242.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.6328038454055787,
+      "epoch": 8.075370121130552,
+      "grad_norm": 0.12108061462640762,
+      "learning_rate": 9.804435971356025e-06,
+      "loss": 0.0013736591674387455,
+      "mean_token_accuracy": 0.9998795092105865,
+      "num_tokens": 113321861.0,
+      "step": 6000
+    },
+    {
+      "entropy": 0.6300597488880157,
+      "epoch": 8.088829071332436,
+      "grad_norm": 0.2497449666261673,
+      "learning_rate": 9.672469912072956e-06,
+      "loss": 0.0014812872745096684,
+      "mean_token_accuracy": 0.9998218238353729,
+      "num_tokens": 113510984.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.6325613379478454,
+      "epoch": 8.10228802153432,
+      "grad_norm": 0.2513059973716736,
+      "learning_rate": 9.541302835431681e-06,
+      "loss": 0.0014441429637372493,
+      "mean_token_accuracy": 0.9998395144939423,
+      "num_tokens": 113699794.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.6325579524040222,
+      "epoch": 8.115746971736204,
+      "grad_norm": 0.15703566372394562,
+      "learning_rate": 9.410937340158599e-06,
+      "loss": 0.0015422176569700242,
+      "mean_token_accuracy": 0.9997781276702881,
+      "num_tokens": 113888200.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.6323439538478851,
+      "epoch": 8.12920592193809,
+      "grad_norm": 0.18420128524303436,
+      "learning_rate": 9.28137600909892e-06,
+      "loss": 0.0013185075484216213,
+      "mean_token_accuracy": 0.9998004496097564,
+      "num_tokens": 114077149.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.6302162289619446,
+      "epoch": 8.142664872139973,
+      "grad_norm": 0.18233658373355865,
+      "learning_rate": 9.15262140916548e-06,
+      "loss": 0.001340259239077568,
+      "mean_token_accuracy": 0.9998397171497345,
+      "num_tokens": 114266569.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.6314728736877442,
+      "epoch": 8.156123822341858,
+      "grad_norm": 0.34443631768226624,
+      "learning_rate": 9.024676091287881e-06,
+      "loss": 0.0015510787256062032,
+      "mean_token_accuracy": 0.9997199475765228,
+      "num_tokens": 114455310.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.6306299686431884,
+      "epoch": 8.169582772543741,
+      "grad_norm": 0.23195648193359375,
+      "learning_rate": 8.897542590361963e-06,
+      "loss": 0.001497127115726471,
+      "mean_token_accuracy": 0.999662047624588,
+      "num_tokens": 114644329.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.6310850799083709,
+      "epoch": 8.183041722745626,
+      "grad_norm": 0.32436603307724,
+      "learning_rate": 8.77122342519956e-06,
+      "loss": 0.0013079997152090073,
+      "mean_token_accuracy": 0.9997977912425995,
+      "num_tokens": 114833260.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.6300573408603668,
+      "epoch": 8.19650067294751,
+      "grad_norm": 0.21875229477882385,
+      "learning_rate": 8.64572109847862e-06,
+      "loss": 0.001645646058022976,
+      "mean_token_accuracy": 0.9996179223060608,
+      "num_tokens": 115022065.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.6323310673236847,
+      "epoch": 8.209959623149395,
+      "grad_norm": 0.27595558762550354,
+      "learning_rate": 8.521038096693635e-06,
+      "loss": 0.0014854520559310913,
+      "mean_token_accuracy": 0.9997394740581512,
+      "num_tokens": 115210794.0,
+      "step": 6100
+    },
+    {
+      "entropy": 0.631768387556076,
+      "epoch": 8.223418573351278,
+      "grad_norm": 0.2772620618343353,
+      "learning_rate": 8.397176890106317e-06,
+      "loss": 0.0014293460175395013,
+      "mean_token_accuracy": 0.9999000608921051,
+      "num_tokens": 115399823.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.6337137937545776,
+      "epoch": 8.236877523553163,
+      "grad_norm": 0.15117399394512177,
+      "learning_rate": 8.274139932696717e-06,
+      "loss": 0.001172219216823578,
+      "mean_token_accuracy": 0.9999594986438751,
+      "num_tokens": 115588301.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.6334636569023132,
+      "epoch": 8.250336473755047,
+      "grad_norm": 0.22133320569992065,
+      "learning_rate": 8.151929662114566e-06,
+      "loss": 0.0014448422007262707,
+      "mean_token_accuracy": 0.9997801542282104,
+      "num_tokens": 115776977.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.6315647959709167,
+      "epoch": 8.263795423956932,
+      "grad_norm": 0.23076151311397552,
+      "learning_rate": 8.030548499631014e-06,
+      "loss": 0.0013002446852624417,
+      "mean_token_accuracy": 0.9998793780803681,
+      "num_tokens": 115966052.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.6293221652507782,
+      "epoch": 8.277254374158815,
+      "grad_norm": 0.2104390561580658,
+      "learning_rate": 7.909998850090628e-06,
+      "loss": 0.0014004193246364594,
+      "mean_token_accuracy": 0.9997619271278382,
+      "num_tokens": 116155401.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.629232895374298,
+      "epoch": 8.2907133243607,
+      "grad_norm": 0.24602340161800385,
+      "learning_rate": 7.790283101863771e-06,
+      "loss": 0.0013899436220526696,
+      "mean_token_accuracy": 0.9998601615428925,
+      "num_tokens": 116344618.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.628896701335907,
+      "epoch": 8.304172274562584,
+      "grad_norm": 0.36142218112945557,
+      "learning_rate": 7.671403626799262e-06,
+      "loss": 0.001389096863567829,
+      "mean_token_accuracy": 0.9997016251087188,
+      "num_tokens": 116533953.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.6315865159034729,
+      "epoch": 8.317631224764469,
+      "grad_norm": 0.15314172208309174,
+      "learning_rate": 7.553362780177398e-06,
+      "loss": 0.0014957386068999768,
+      "mean_token_accuracy": 0.9997191250324249,
+      "num_tokens": 116722247.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.6303807675838471,
+      "epoch": 8.331090174966352,
+      "grad_norm": 0.16988365352153778,
+      "learning_rate": 7.43616290066329e-06,
+      "loss": 0.0012819443829357624,
+      "mean_token_accuracy": 0.9998224377632141,
+      "num_tokens": 116911239.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.6324020445346832,
+      "epoch": 8.344549125168237,
+      "grad_norm": 0.214530348777771,
+      "learning_rate": 7.3198063102605185e-06,
+      "loss": 0.0013960381969809533,
+      "mean_token_accuracy": 0.9998191654682159,
+      "num_tokens": 117099644.0,
+      "step": 6200
+    },
+    {
+      "entropy": 0.6305644690990448,
+      "epoch": 8.35800807537012,
+      "grad_norm": 0.6130462288856506,
+      "learning_rate": 7.204295314265136e-06,
+      "loss": 0.001624145545065403,
+      "mean_token_accuracy": 0.999723345041275,
+      "num_tokens": 117288440.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.6282017886638641,
+      "epoch": 8.371467025572006,
+      "grad_norm": 0.33955836296081543,
+      "learning_rate": 7.089632201219997e-06,
+      "loss": 0.0014544738456606864,
+      "mean_token_accuracy": 0.9996826589107514,
+      "num_tokens": 117478031.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.6303719699382782,
+      "epoch": 8.38492597577389,
+      "grad_norm": 0.21833766996860504,
+      "learning_rate": 6.975819242869414e-06,
+      "loss": 0.001375422067940235,
+      "mean_token_accuracy": 0.9996990442276001,
+      "num_tokens": 117666823.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.6290243804454804,
+      "epoch": 8.398384925975774,
+      "grad_norm": 0.18415814638137817,
+      "learning_rate": 6.862858694114138e-06,
+      "loss": 0.0013824976980686189,
+      "mean_token_accuracy": 0.9998021006584168,
+      "num_tokens": 117855959.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.6298940062522889,
+      "epoch": 8.411843876177658,
+      "grad_norm": 0.38978856801986694,
+      "learning_rate": 6.750752792966702e-06,
+      "loss": 0.001461424771696329,
+      "mean_token_accuracy": 0.9998207986354828,
+      "num_tokens": 118044550.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.6267169773578644,
+      "epoch": 8.425302826379543,
+      "grad_norm": 0.2565508186817169,
+      "learning_rate": 6.639503760507065e-06,
+      "loss": 0.0014107320457696915,
+      "mean_token_accuracy": 0.9998013794422149,
+      "num_tokens": 118233720.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.6296021044254303,
+      "epoch": 8.438761776581426,
+      "grad_norm": 0.14881721138954163,
+      "learning_rate": 6.529113800838632e-06,
+      "loss": 0.0013083923608064652,
+      "mean_token_accuracy": 0.9998387336730957,
+      "num_tokens": 118422118.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.6294852197170258,
+      "epoch": 8.452220726783311,
+      "grad_norm": 0.22628630697727203,
+      "learning_rate": 6.41958510104454e-06,
+      "loss": 0.0013258070684969425,
+      "mean_token_accuracy": 0.9998400807380676,
+      "num_tokens": 118610757.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.6305896222591401,
+      "epoch": 8.465679676985195,
+      "grad_norm": 0.3069230020046234,
+      "learning_rate": 6.31091983114438e-06,
+      "loss": 0.0014844583347439766,
+      "mean_token_accuracy": 0.999760502576828,
+      "num_tokens": 118799460.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.6299797236919403,
+      "epoch": 8.47913862718708,
+      "grad_norm": 0.2504591941833496,
+      "learning_rate": 6.203120144051161e-06,
+      "loss": 0.0013100499287247657,
+      "mean_token_accuracy": 0.9998400688171387,
+      "num_tokens": 118988395.0,
+      "step": 6300
+    },
+    {
+      "entropy": 0.6294752597808838,
+      "epoch": 8.492597577388963,
+      "grad_norm": 0.21706247329711914,
+      "learning_rate": 6.0961881755286775e-06,
+      "loss": 0.0014089202508330346,
+      "mean_token_accuracy": 0.9998202979564667,
+      "num_tokens": 119177293.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.630080658197403,
+      "epoch": 8.506056527590848,
+      "grad_norm": 0.24263928830623627,
+      "learning_rate": 5.990126044149203e-06,
+      "loss": 0.0013817671686410903,
+      "mean_token_accuracy": 0.9997431695461273,
+      "num_tokens": 119366224.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.6312020123004913,
+      "epoch": 8.519515477792732,
+      "grad_norm": 0.2104533314704895,
+      "learning_rate": 5.884935851251483e-06,
+      "loss": 0.0014581452123820782,
+      "mean_token_accuracy": 0.9996995806694031,
+      "num_tokens": 119554641.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.6307013928890228,
+      "epoch": 8.532974427994617,
+      "grad_norm": 0.29689937829971313,
+      "learning_rate": 5.780619680899146e-06,
+      "loss": 0.0015367996878921986,
+      "mean_token_accuracy": 0.9997993230819702,
+      "num_tokens": 119743217.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.6292266964912414,
+      "epoch": 8.5464333781965,
+      "grad_norm": 0.20508502423763275,
+      "learning_rate": 5.677179599839366e-06,
+      "loss": 0.0012161459773778916,
+      "mean_token_accuracy": 0.9998607873916626,
+      "num_tokens": 119932436.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.6305240333080292,
+      "epoch": 8.559892328398385,
+      "grad_norm": 0.22822189331054688,
+      "learning_rate": 5.574617657461967e-06,
+      "loss": 0.0013693704269826413,
+      "mean_token_accuracy": 0.9998015820980072,
+      "num_tokens": 120121506.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.6288493514060974,
+      "epoch": 8.573351278600269,
+      "grad_norm": 0.15999595820903778,
+      "learning_rate": 5.472935885758767e-06,
+      "loss": 0.0011829210445284844,
+      "mean_token_accuracy": 0.9998811721801758,
+      "num_tokens": 120310792.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.6302126765251159,
+      "epoch": 8.586810228802154,
+      "grad_norm": 0.48678338527679443,
+      "learning_rate": 5.372136299283365e-06,
+      "loss": 0.0014185780659317971,
+      "mean_token_accuracy": 0.999719786643982,
+      "num_tokens": 120499610.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.6311411023139953,
+      "epoch": 8.600269179004037,
+      "grad_norm": 0.10369251668453217,
+      "learning_rate": 5.272220895111196e-06,
+      "loss": 0.0013072849251329899,
+      "mean_token_accuracy": 0.9997986853122711,
+      "num_tokens": 120688546.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.6289890885353089,
+      "epoch": 8.613728129205922,
+      "grad_norm": 0.18738436698913574,
+      "learning_rate": 5.173191652799986e-06,
+      "loss": 0.0012681365013122558,
+      "mean_token_accuracy": 0.9997800529003144,
+      "num_tokens": 120877609.0,
+      "step": 6400
+    },
+    {
+      "entropy": 0.6279601216316223,
+      "epoch": 8.627187079407806,
+      "grad_norm": 0.19525793194770813,
+      "learning_rate": 5.075050534350512e-06,
+      "loss": 0.0012957751750946045,
+      "mean_token_accuracy": 0.9998006701469422,
+      "num_tokens": 121066294.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.6293328821659088,
+      "epoch": 8.640646029609691,
+      "grad_norm": 0.184369757771492,
+      "learning_rate": 4.977799484167767e-06,
+      "loss": 0.0012527795508503914,
+      "mean_token_accuracy": 0.999919193983078,
+      "num_tokens": 121255109.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.629350233078003,
+      "epoch": 8.654104979811574,
+      "grad_norm": 0.3674270510673523,
+      "learning_rate": 4.8814404290223655e-06,
+      "loss": 0.001326523907482624,
+      "mean_token_accuracy": 0.9998202204704285,
+      "num_tokens": 121443604.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.6280995666980743,
+      "epoch": 8.66756393001346,
+      "grad_norm": 0.19100305438041687,
+      "learning_rate": 4.7859752780124555e-06,
+      "loss": 0.0011802183464169503,
+      "mean_token_accuracy": 0.9998997509479522,
+      "num_tokens": 121632712.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.6302553176879883,
+      "epoch": 8.681022880215343,
+      "grad_norm": 0.21766118705272675,
+      "learning_rate": 4.691405922525838e-06,
+      "loss": 0.00140056349337101,
+      "mean_token_accuracy": 0.9998192071914673,
+      "num_tokens": 121821234.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.628807520866394,
+      "epoch": 8.694481830417228,
+      "grad_norm": 0.30580392479896545,
+      "learning_rate": 4.597734236202511e-06,
+      "loss": 0.00123771820217371,
+      "mean_token_accuracy": 0.9998995006084442,
+      "num_tokens": 122009998.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.6287366271018981,
+      "epoch": 8.707940780619111,
+      "grad_norm": 0.14816957712173462,
+      "learning_rate": 4.504962074897556e-06,
+      "loss": 0.0014824405312538146,
+      "mean_token_accuracy": 0.999697607755661,
+      "num_tokens": 122198378.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.6262040913105011,
+      "epoch": 8.721399730820997,
+      "grad_norm": 0.17951056361198425,
+      "learning_rate": 4.413091276644349e-06,
+      "loss": 0.00132729671895504,
+      "mean_token_accuracy": 0.9998216688632965,
+      "num_tokens": 122387614.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.6280441462993622,
+      "epoch": 8.73485868102288,
+      "grad_norm": 0.1389010101556778,
+      "learning_rate": 4.322123661618166e-06,
+      "loss": 0.001298490446060896,
+      "mean_token_accuracy": 0.999879801273346,
+      "num_tokens": 122576224.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.6271527647972107,
+      "epoch": 8.748317631224765,
+      "grad_norm": 0.24173741042613983,
+      "learning_rate": 4.23206103210011e-06,
+      "loss": 0.0011219406500458718,
+      "mean_token_accuracy": 0.999880450963974,
+      "num_tokens": 122765123.0,
+      "step": 6500
+    },
+    {
+      "entropy": 0.6269227385520935,
+      "epoch": 8.761776581426648,
+      "grad_norm": 0.22519297897815704,
+      "learning_rate": 4.142905172441408e-06,
+      "loss": 0.0012161782942712307,
+      "mean_token_accuracy": 0.9998603820800781,
+      "num_tokens": 122954344.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.6288948476314544,
+      "epoch": 8.775235531628534,
+      "grad_norm": 0.21760156750679016,
+      "learning_rate": 4.054657849028043e-06,
+      "loss": 0.001162637397646904,
+      "mean_token_accuracy": 0.9999193131923676,
+      "num_tokens": 123142872.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.6281030297279357,
+      "epoch": 8.788694481830417,
+      "grad_norm": 0.3592071831226349,
+      "learning_rate": 3.9673208102457895e-06,
+      "loss": 0.0013433885760605335,
+      "mean_token_accuracy": 0.9997989416122437,
+      "num_tokens": 123331858.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.6263146758079529,
+      "epoch": 8.802153432032302,
+      "grad_norm": 0.3514918386936188,
+      "learning_rate": 3.880895786445543e-06,
+      "loss": 0.0014502016827464103,
+      "mean_token_accuracy": 0.9997799456119537,
+      "num_tokens": 123521321.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.6269251763820648,
+      "epoch": 8.815612382234185,
+      "grad_norm": 0.22137221693992615,
+      "learning_rate": 3.795384489909054e-06,
+      "loss": 0.001280687004327774,
+      "mean_token_accuracy": 0.9998396694660187,
+      "num_tokens": 123710572.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.6270365953445435,
+      "epoch": 8.82907133243607,
+      "grad_norm": 0.36301663517951965,
+      "learning_rate": 3.7107886148150084e-06,
+      "loss": 0.0012877551838755608,
+      "mean_token_accuracy": 0.9998403787612915,
+      "num_tokens": 123899764.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.6268510282039642,
+      "epoch": 8.842530282637954,
+      "grad_norm": 0.19353987276554108,
+      "learning_rate": 3.6271098372054334e-06,
+      "loss": 0.00124650401994586,
+      "mean_token_accuracy": 0.9998400628566741,
+      "num_tokens": 124088703.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.6260306179523468,
+      "epoch": 8.855989232839839,
+      "grad_norm": 0.11702294647693634,
+      "learning_rate": 3.5443498149525357e-06,
+      "loss": 0.0011134844273328782,
+      "mean_token_accuracy": 0.9998809397220612,
+      "num_tokens": 124277949.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.6265114545822144,
+      "epoch": 8.869448183041722,
+      "grad_norm": 0.24068744480609894,
+      "learning_rate": 3.462510187725809e-06,
+      "loss": 0.0013966705650091172,
+      "mean_token_accuracy": 0.9997197926044464,
+      "num_tokens": 124466937.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.6271577537059784,
+      "epoch": 8.882907133243608,
+      "grad_norm": 0.1259060949087143,
+      "learning_rate": 3.3815925769595745e-06,
+      "loss": 0.0013276103883981705,
+      "mean_token_accuracy": 0.9997398614883423,
+      "num_tokens": 124655536.0,
+      "step": 6600
+    },
+    {
+      "entropy": 0.626012533903122,
+      "epoch": 8.896366083445491,
+      "grad_norm": 0.21367371082305908,
+      "learning_rate": 3.30159858582087e-06,
+      "loss": 0.0012806886807084084,
+      "mean_token_accuracy": 0.9998196005821228,
+      "num_tokens": 124844587.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.6261564731597901,
+      "epoch": 8.909825033647376,
+      "grad_norm": 0.2548461854457855,
+      "learning_rate": 3.2225297991776536e-06,
+      "loss": 0.0012241953983902932,
+      "mean_token_accuracy": 0.9998611867427826,
+      "num_tokens": 125033688.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.6268511712551117,
+      "epoch": 8.92328398384926,
+      "grad_norm": 0.23393066227436066,
+      "learning_rate": 3.1443877835674208e-06,
+      "loss": 0.0010460015386343003,
+      "mean_token_accuracy": 0.9998401999473572,
+      "num_tokens": 125222397.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.6260151982307434,
+      "epoch": 8.936742934051145,
+      "grad_norm": 0.09390900284051895,
+      "learning_rate": 3.0671740871661635e-06,
+      "loss": 0.0011951563879847526,
+      "mean_token_accuracy": 0.9998811960220337,
+      "num_tokens": 125411589.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.625042176246643,
+      "epoch": 8.950201884253028,
+      "grad_norm": 0.1279630959033966,
+      "learning_rate": 2.9908902397577043e-06,
+      "loss": 0.001325780339539051,
+      "mean_token_accuracy": 0.9998415470123291,
+      "num_tokens": 125600890.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.6269629180431366,
+      "epoch": 8.963660834454913,
+      "grad_norm": 0.16514816880226135,
+      "learning_rate": 2.91553775270339e-06,
+      "loss": 0.0012059432454407215,
+      "mean_token_accuracy": 0.9999194324016571,
+      "num_tokens": 125789165.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.6265843152999878,
+      "epoch": 8.977119784656796,
+      "grad_norm": 0.16388432681560516,
+      "learning_rate": 2.8411181189121284e-06,
+      "loss": 0.001122188288718462,
+      "mean_token_accuracy": 0.9998812198638916,
+      "num_tokens": 125977920.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.6274819135665893,
+      "epoch": 8.990578734858682,
+      "grad_norm": 0.11002290993928909,
+      "learning_rate": 2.767632812810839e-06,
+      "loss": 0.001199054718017578,
+      "mean_token_accuracy": 0.999840772151947,
+      "num_tokens": 126166297.0,
+      "step": 6680
+    },
+    {
+      "epoch": 9.0,
+      "eval_entropy": 0.63310007494726,
+      "eval_loss": 0.10951486974954605,
+      "eval_mean_token_accuracy": 0.9730027690055264,
+      "eval_num_tokens": 126297929.0,
+      "eval_runtime": 12.7449,
+      "eval_samples_per_second": 392.315,
+      "eval_steps_per_second": 12.319,
+      "step": 6687
+    },
+    {
+      "entropy": 0.6288514316082001,
+      "epoch": 9.004037685060565,
+      "grad_norm": 0.1864977478981018,
+      "learning_rate": 2.6950832903152034e-06,
+      "loss": 0.0012032078579068185,
+      "mean_token_accuracy": 0.9997978270053863,
+      "num_tokens": 126354475.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.6272882223129272,
+      "epoch": 9.01749663526245,
+      "grad_norm": 0.11999135464429855,
+      "learning_rate": 2.623470988800869e-06,
+      "loss": 0.0009624600410461426,
+      "mean_token_accuracy": 0.9999589681625366,
+      "num_tokens": 126543288.0,
+      "step": 6700
+    },
+    {
+      "entropy": 0.628252774477005,
+      "epoch": 9.030955585464334,
+      "grad_norm": 0.2488996982574463,
+      "learning_rate": 2.552797327074924e-06,
+      "loss": 0.001090879738330841,
+      "mean_token_accuracy": 0.9998978614807129,
+      "num_tokens": 126731637.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.6246803760528564,
+      "epoch": 9.044414535666219,
+      "grad_norm": 0.16829051077365875,
+      "learning_rate": 2.4830637053478213e-06,
+      "loss": 0.0009406884200870991,
+      "mean_token_accuracy": 0.99994056224823,
+      "num_tokens": 126920928.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.6244384467601776,
+      "epoch": 9.057873485868102,
+      "grad_norm": 0.1406945288181305,
+      "learning_rate": 2.414271505205612e-06,
+      "loss": 0.0009295841678977013,
+      "mean_token_accuracy": 0.9998805224895477,
+      "num_tokens": 127110204.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.6255349397659302,
+      "epoch": 9.071332436069987,
+      "grad_norm": 0.08414118736982346,
+      "learning_rate": 2.346422089582595e-06,
+      "loss": 0.0010875239968299866,
+      "mean_token_accuracy": 0.9999009311199188,
+      "num_tokens": 127299166.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.6261642336845398,
+      "epoch": 9.08479138627187,
+      "grad_norm": 0.1418563723564148,
+      "learning_rate": 2.279516802734305e-06,
+      "loss": 0.0010646498762071134,
+      "mean_token_accuracy": 0.9998397529125214,
+      "num_tokens": 127487955.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.6270270645618439,
+      "epoch": 9.098250336473756,
+      "grad_norm": 0.2824283838272095,
+      "learning_rate": 2.213556970210867e-06,
+      "loss": 0.0009680183604359627,
+      "mean_token_accuracy": 0.9999598085880279,
+      "num_tokens": 127676381.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.6254054069519043,
+      "epoch": 9.111709286675639,
+      "grad_norm": 0.1672801375389099,
+      "learning_rate": 2.1485438988307518e-06,
+      "loss": 0.0009563026949763298,
+      "mean_token_accuracy": 0.9998994469642639,
+      "num_tokens": 127865124.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.6256063163280488,
+      "epoch": 9.125168236877524,
+      "grad_norm": 0.28399658203125,
+      "learning_rate": 2.0844788766548883e-06,
+      "loss": 0.0011562740430235863,
+      "mean_token_accuracy": 0.9998607695102691,
+      "num_tokens": 128054007.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.6277299106121064,
+      "epoch": 9.138627187079408,
+      "grad_norm": 0.1154041588306427,
+      "learning_rate": 2.021363172961116e-06,
+      "loss": 0.0010892899706959724,
+      "mean_token_accuracy": 0.9998789548873901,
+      "num_tokens": 128242311.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.6259080588817596,
+      "epoch": 9.152086137281293,
+      "grad_norm": 0.17905649542808533,
+      "learning_rate": 1.959198038219079e-06,
+      "loss": 0.001114431582391262,
+      "mean_token_accuracy": 0.999880051612854,
+      "num_tokens": 128431033.0,
+      "step": 6800
+    },
+    {
+      "entropy": 0.6253814101219177,
+      "epoch": 9.165545087483176,
+      "grad_norm": 0.2372596263885498,
+      "learning_rate": 1.897984704065403e-06,
+      "loss": 0.0010150818154215813,
+      "mean_token_accuracy": 0.9998995363712311,
+      "num_tokens": 128619554.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.6254553496837616,
+      "epoch": 9.179004037685061,
+      "grad_norm": 0.519012451171875,
+      "learning_rate": 1.8377243832793322e-06,
+      "loss": 0.0009561034850776196,
+      "mean_token_accuracy": 0.9998998582363129,
+      "num_tokens": 128808395.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.6237362027168274,
+      "epoch": 9.192462987886945,
+      "grad_norm": 0.2933456003665924,
+      "learning_rate": 1.778418269758697e-06,
+      "loss": 0.001003552135080099,
+      "mean_token_accuracy": 0.9998997807502746,
+      "num_tokens": 128997377.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.625144773721695,
+      "epoch": 9.20592193808883,
+      "grad_norm": 0.2625809609889984,
+      "learning_rate": 1.7200675384962339e-06,
+      "loss": 0.0010022109374403954,
+      "mean_token_accuracy": 0.9998792469501495,
+      "num_tokens": 129186062.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.6270272016525269,
+      "epoch": 9.219380888290713,
+      "grad_norm": 0.11006571352481842,
+      "learning_rate": 1.6626733455563371e-06,
+      "loss": 0.0010081454180181026,
+      "mean_token_accuracy": 0.9999199151992798,
+      "num_tokens": 129374536.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.6229528486728668,
+      "epoch": 9.232839838492598,
+      "grad_norm": 0.08817827701568604,
+      "learning_rate": 1.6062368280521323e-06,
+      "loss": 0.0009087607264518738,
+      "mean_token_accuracy": 0.9999602735042572,
+      "num_tokens": 129564281.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.6264402210712433,
+      "epoch": 9.246298788694482,
+      "grad_norm": 0.14596502482891083,
+      "learning_rate": 1.5507591041229508e-06,
+      "loss": 0.0010394659824669362,
+      "mean_token_accuracy": 0.9998987376689911,
+      "num_tokens": 129752628.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.6248175978660584,
+      "epoch": 9.259757738896367,
+      "grad_norm": 0.1449577510356903,
+      "learning_rate": 1.4962412729122077e-06,
+      "loss": 0.0009976149536669253,
+      "mean_token_accuracy": 0.999860554933548,
+      "num_tokens": 129941748.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.6252830862998963,
+      "epoch": 9.27321668909825,
+      "grad_norm": 0.14489394426345825,
+      "learning_rate": 1.4426844145455642e-06,
+      "loss": 0.0009816748090088368,
+      "mean_token_accuracy": 0.9999404311180115,
+      "num_tokens": 130130510.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.6268728196620941,
+      "epoch": 9.286675639300135,
+      "grad_norm": 0.19057799875736237,
+      "learning_rate": 1.3900895901095889e-06,
+      "loss": 0.0009371487423777581,
+      "mean_token_accuracy": 0.9999604403972626,
+      "num_tokens": 130318999.0,
+      "step": 6900
+    },
+    {
+      "entropy": 0.6267165422439576,
+      "epoch": 9.300134589502019,
+      "grad_norm": 0.0884639173746109,
+      "learning_rate": 1.3384578416307025e-06,
+      "loss": 0.0009420069865882397,
+      "mean_token_accuracy": 0.9998995780944824,
+      "num_tokens": 130507823.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.626640260219574,
+      "epoch": 9.313593539703904,
+      "grad_norm": 0.18272733688354492,
+      "learning_rate": 1.2877901920545332e-06,
+      "loss": 0.0011405523866415024,
+      "mean_token_accuracy": 0.9998404383659363,
+      "num_tokens": 130696136.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.6253959536552429,
+      "epoch": 9.327052489905787,
+      "grad_norm": 0.10377068817615509,
+      "learning_rate": 1.2380876452256663e-06,
+      "loss": 0.0010065490379929543,
+      "mean_token_accuracy": 0.9999197840690612,
+      "num_tokens": 130885015.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.6249998331069946,
+      "epoch": 9.340511440107672,
+      "grad_norm": 0.18981093168258667,
+      "learning_rate": 1.189351185867732e-06,
+      "loss": 0.0009395302273333073,
+      "mean_token_accuracy": 0.9999002516269684,
+      "num_tokens": 131074025.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.6236762702465057,
+      "epoch": 9.353970390309556,
+      "grad_norm": 0.25526636838912964,
+      "learning_rate": 1.1415817795639328e-06,
+      "loss": 0.000978347286581993,
+      "mean_token_accuracy": 0.9999594926834107,
+      "num_tokens": 131263251.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.6275614380836487,
+      "epoch": 9.36742934051144,
+      "grad_norm": 0.16339771449565887,
+      "learning_rate": 1.094780372737858e-06,
+      "loss": 0.0009866973385214805,
+      "mean_token_accuracy": 0.9999193012714386,
+      "num_tokens": 131451938.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.6262193143367767,
+      "epoch": 9.380888290713324,
+      "grad_norm": 0.14414891600608826,
+      "learning_rate": 1.048947892634794e-06,
+      "loss": 0.0010185737162828445,
+      "mean_token_accuracy": 0.9999408483505249,
+      "num_tokens": 131641020.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.6233638226985931,
+      "epoch": 9.39434724091521,
+      "grad_norm": 0.13221070170402527,
+      "learning_rate": 1.0040852473032992e-06,
+      "loss": 0.0009544931352138519,
+      "mean_token_accuracy": 0.999920105934143,
+      "num_tokens": 131830522.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.6249336421489715,
+      "epoch": 9.407806191117093,
+      "grad_norm": 0.1283094733953476,
+      "learning_rate": 9.601933255772578e-07,
+      "loss": 0.0010163593105971813,
+      "mean_token_accuracy": 0.9999199032783508,
+      "num_tokens": 132019619.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.6245728194713592,
+      "epoch": 9.421265141318978,
+      "grad_norm": 0.21595682203769684,
+      "learning_rate": 9.172729970582439e-07,
+      "loss": 0.000841591414064169,
+      "mean_token_accuracy": 0.9999194025993348,
+      "num_tokens": 132208752.0,
+      "step": 7000
+    },
+    {
+      "entropy": 0.62474165558815,
+      "epoch": 9.434724091520861,
+      "grad_norm": 0.17968018352985382,
+      "learning_rate": 8.753251120982908e-07,
+      "loss": 0.000800126139074564,
+      "mean_token_accuracy": 0.9999598741531373,
+      "num_tokens": 132397780.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.6266384899616242,
+      "epoch": 9.448183041722746,
+      "grad_norm": 0.385009765625,
+      "learning_rate": 8.3435050178306e-07,
+      "loss": 0.00106763681396842,
+      "mean_token_accuracy": 0.999860417842865,
+      "num_tokens": 132586743.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.6266715288162231,
+      "epoch": 9.46164199192463,
+      "grad_norm": 0.16072122752666473,
+      "learning_rate": 7.943499779153707e-07,
+      "loss": 0.000838584266602993,
+      "mean_token_accuracy": 0.9999601304531097,
+      "num_tokens": 132775210.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.625688511133194,
+      "epoch": 9.475100942126515,
+      "grad_norm": 0.22414861619472504,
+      "learning_rate": 7.553243329990967e-07,
+      "loss": 0.0011072710156440734,
+      "mean_token_accuracy": 0.9998790621757507,
+      "num_tokens": 132963712.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.6253630220890045,
+      "epoch": 9.488559892328398,
+      "grad_norm": 0.19061173498630524,
+      "learning_rate": 7.172743402235005e-07,
+      "loss": 0.0011105727404356002,
+      "mean_token_accuracy": 0.9998005747795105,
+      "num_tokens": 133152661.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.6262070059776306,
+      "epoch": 9.502018842530283,
+      "grad_norm": 0.13845139741897583,
+      "learning_rate": 6.802007534478794e-07,
+      "loss": 0.0010735243558883667,
+      "mean_token_accuracy": 0.9998797595500946,
+      "num_tokens": 133341130.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.625045782327652,
+      "epoch": 9.515477792732167,
+      "grad_norm": 0.1769309639930725,
+      "learning_rate": 6.441043071866604e-07,
+      "loss": 0.0009129032492637634,
+      "mean_token_accuracy": 0.9999597907066345,
+      "num_tokens": 133530265.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.6266582906246185,
+      "epoch": 9.528936742934052,
+      "grad_norm": 0.18698658049106598,
+      "learning_rate": 6.08985716594812e-07,
+      "loss": 0.0010521873831748963,
+      "mean_token_accuracy": 0.9998586356639863,
+      "num_tokens": 133719077.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.6256363987922668,
+      "epoch": 9.542395693135935,
+      "grad_norm": 0.13940785825252533,
+      "learning_rate": 5.74845677453717e-07,
+      "loss": 0.0010448255576193332,
+      "mean_token_accuracy": 0.999879777431488,
+      "num_tokens": 133908153.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.6252269387245178,
+      "epoch": 9.55585464333782,
+      "grad_norm": 0.14733369648456573,
+      "learning_rate": 5.416848661573548e-07,
+      "loss": 0.0008552927523851394,
+      "mean_token_accuracy": 0.9999600291252136,
+      "num_tokens": 134096955.0,
+      "step": 7100
+    },
+    {
+      "entropy": 0.6250855565071106,
+      "epoch": 9.569313593539704,
+      "grad_norm": 0.14992880821228027,
+      "learning_rate": 5.095039396989132e-07,
+      "loss": 0.0009266521781682968,
+      "mean_token_accuracy": 0.9999401867389679,
+      "num_tokens": 134285946.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.6248355746269226,
+      "epoch": 9.582772543741589,
+      "grad_norm": 0.07733164727687836,
+      "learning_rate": 4.783035356577703e-07,
+      "loss": 0.0010546602308750153,
+      "mean_token_accuracy": 0.9999395966529846,
+      "num_tokens": 134474896.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.6246441245079041,
+      "epoch": 9.596231493943472,
+      "grad_norm": 0.32437849044799805,
+      "learning_rate": 4.480842721868772e-07,
+      "loss": 0.0009787463583052157,
+      "mean_token_accuracy": 0.9999004662036896,
+      "num_tokens": 134664097.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.626048332452774,
+      "epoch": 9.609690444145357,
+      "grad_norm": 0.31202933192253113,
+      "learning_rate": 4.1884674800046765e-07,
+      "loss": 0.0009479177184402943,
+      "mean_token_accuracy": 0.9999408185482025,
+      "num_tokens": 134852481.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.6274196743965149,
+      "epoch": 9.62314939434724,
+      "grad_norm": 0.11999280750751495,
+      "learning_rate": 3.905915423622564e-07,
+      "loss": 0.0009240784682333469,
+      "mean_token_accuracy": 0.9999399304389953,
+      "num_tokens": 135040865.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.6251983046531677,
+      "epoch": 9.636608344549126,
+      "grad_norm": 0.09678250551223755,
+      "learning_rate": 3.633192150739095e-07,
+      "loss": 0.0008916430175304413,
+      "mean_token_accuracy": 0.9999601483345032,
+      "num_tokens": 135229639.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.6257704675197602,
+      "epoch": 9.65006729475101,
+      "grad_norm": 0.123650923371315,
+      "learning_rate": 3.370303064639868e-07,
+      "loss": 0.00104091539978981,
+      "mean_token_accuracy": 0.9998602390289306,
+      "num_tokens": 135418384.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.6253515779972076,
+      "epoch": 9.663526244952894,
+      "grad_norm": 0.1876295506954193,
+      "learning_rate": 3.1172533737721667e-07,
+      "loss": 0.000964321382343769,
+      "mean_token_accuracy": 0.9998816967010498,
+      "num_tokens": 135607322.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.6253881275653839,
+      "epoch": 9.676985195154778,
+      "grad_norm": 0.19082963466644287,
+      "learning_rate": 2.8740480916419344e-07,
+      "loss": 0.001023743487894535,
+      "mean_token_accuracy": 0.9998808681964875,
+      "num_tokens": 135796040.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.6251950323581695,
+      "epoch": 9.690444145356663,
+      "grad_norm": 0.2538227438926697,
+      "learning_rate": 2.6406920367143537e-07,
+      "loss": 0.001090743951499462,
+      "mean_token_accuracy": 0.9998391032218933,
+      "num_tokens": 135984702.0,
+      "step": 7200
+    },
+    {
+      "entropy": 0.6267866253852844,
+      "epoch": 9.703903095558546,
+      "grad_norm": 0.2510435879230499,
+      "learning_rate": 2.417189832318367e-07,
+      "loss": 0.001036971341818571,
+      "mean_token_accuracy": 0.99987872838974,
+      "num_tokens": 136173026.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.6254905700683594,
+      "epoch": 9.717362045760431,
+      "grad_norm": 0.1447930485010147,
+      "learning_rate": 2.2035459065550822e-07,
+      "loss": 0.0009731227532029152,
+      "mean_token_accuracy": 0.9999204635620117,
+      "num_tokens": 136361905.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.6250186264514923,
+      "epoch": 9.730820995962315,
+      "grad_norm": 0.12551535665988922,
+      "learning_rate": 1.9997644922100654e-07,
+      "loss": 0.0009514208883047104,
+      "mean_token_accuracy": 0.9999399483203888,
+      "num_tokens": 136550988.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.6249144315719605,
+      "epoch": 9.7442799461642,
+      "grad_norm": 0.2216738909482956,
+      "learning_rate": 1.8058496266695757e-07,
+      "loss": 0.0009482087567448616,
+      "mean_token_accuracy": 0.9999204635620117,
+      "num_tokens": 136740235.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.6257742464542388,
+      "epoch": 9.757738896366083,
+      "grad_norm": 0.1703806072473526,
+      "learning_rate": 1.6218051518403498e-07,
+      "loss": 0.0009052447974681854,
+      "mean_token_accuracy": 0.9999796628952027,
+      "num_tokens": 136928963.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.6257103085517883,
+      "epoch": 9.771197846567969,
+      "grad_norm": 0.15903201699256897,
+      "learning_rate": 1.4476347140736647e-07,
+      "loss": 0.0009922895580530167,
+      "mean_token_accuracy": 0.9999399781227112,
+      "num_tokens": 137117863.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.6257126092910766,
+      "epoch": 9.784656796769852,
+      "grad_norm": 0.17350678145885468,
+      "learning_rate": 1.2833417640930045e-07,
+      "loss": 0.0008743755519390106,
+      "mean_token_accuracy": 0.9999602138996124,
+      "num_tokens": 137306735.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.6249728262424469,
+      "epoch": 9.798115746971735,
+      "grad_norm": 0.09517685323953629,
+      "learning_rate": 1.1289295569258395e-07,
+      "loss": 0.001018014457076788,
+      "mean_token_accuracy": 0.9998190701007843,
+      "num_tokens": 137495653.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.6233257949352264,
+      "epoch": 9.81157469717362,
+      "grad_norm": 0.12346095591783524,
+      "learning_rate": 9.84401151838954e-08,
+      "loss": 0.0009624490514397622,
+      "mean_token_accuracy": 0.9999410688877106,
+      "num_tokens": 137685402.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.6251835882663727,
+      "epoch": 9.825033647375506,
+      "grad_norm": 0.16645435988903046,
+      "learning_rate": 8.497594122778285e-08,
+      "loss": 0.0009772653691470623,
+      "mean_token_accuracy": 0.9999801754951477,
+      "num_tokens": 137874536.0,
+      "step": 7300
+    },
+    {
+      "entropy": 0.6244330942630768,
+      "epoch": 9.838492597577389,
+      "grad_norm": 0.14453864097595215,
+      "learning_rate": 7.250070058100745e-08,
+      "loss": 0.0009532321244478225,
+      "mean_token_accuracy": 0.9999598383903503,
+      "num_tokens": 138063642.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.6252149224281311,
+      "epoch": 9.851951547779272,
+      "grad_norm": 0.18304498493671417,
+      "learning_rate": 6.101464040725868e-08,
+      "loss": 0.0008785058744251728,
+      "mean_token_accuracy": 0.9999402463436127,
+      "num_tokens": 138252711.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.6237101435661316,
+      "epoch": 9.865410497981157,
+      "grad_norm": 0.17022334039211273,
+      "learning_rate": 5.051798827222509e-08,
+      "loss": 0.0009720981121063233,
+      "mean_token_accuracy": 0.9998814284801483,
+      "num_tokens": 138441852.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.6242216885089874,
+      "epoch": 9.878869448183043,
+      "grad_norm": 0.3600709140300751,
+      "learning_rate": 4.1010952139136637e-08,
+      "loss": 0.0010783188045024872,
+      "mean_token_accuracy": 0.9998392701148987,
+      "num_tokens": 138630968.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.6240040183067321,
+      "epoch": 9.892328398384926,
+      "grad_norm": 0.10606256127357483,
+      "learning_rate": 3.249372036459031e-08,
+      "loss": 0.001023113913834095,
+      "mean_token_accuracy": 0.9998408079147338,
+      "num_tokens": 138820223.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.6267604887485504,
+      "epoch": 9.90578734858681,
+      "grad_norm": 0.1525883972644806,
+      "learning_rate": 2.496646169486416e-08,
+      "loss": 0.0009261246770620347,
+      "mean_token_accuracy": 0.9999599814414978,
+      "num_tokens": 139008702.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.6255869269371033,
+      "epoch": 9.919246298788694,
+      "grad_norm": 0.17652663588523865,
+      "learning_rate": 1.8429325262553367e-08,
+      "loss": 0.001049203984439373,
+      "mean_token_accuracy": 0.9999200522899627,
+      "num_tokens": 139197398.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.6255215287208558,
+      "epoch": 9.93270524899058,
+      "grad_norm": 0.14364579319953918,
+      "learning_rate": 1.2882440583600331e-08,
+      "loss": 0.0010498828254640103,
+      "mean_token_accuracy": 0.9999205052852631,
+      "num_tokens": 139386058.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.6249600052833557,
+      "epoch": 9.946164199192463,
+      "grad_norm": 0.22805197536945343,
+      "learning_rate": 8.325917554763418e-09,
+      "loss": 0.001060979627072811,
+      "mean_token_accuracy": 0.9998607873916626,
+      "num_tokens": 139575253.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.6254779100418091,
+      "epoch": 9.959623149394346,
+      "grad_norm": 0.1829676628112793,
+      "learning_rate": 4.7598464514186925e-09,
+      "loss": 0.0009019608609378338,
+      "mean_token_accuracy": 1.0,
+      "num_tokens": 139764273.0,
+      "step": 7400
+    },
+    {
+      "entropy": 0.6239280462265014,
+      "epoch": 9.973082099596231,
+      "grad_norm": 0.12817828357219696,
+      "learning_rate": 2.1842979257613583e-09,
+      "loss": 0.0009699763730168342,
+      "mean_token_accuracy": 0.9998795568943024,
+      "num_tokens": 139953281.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.6257602035999298,
+      "epoch": 9.986541049798117,
+      "grad_norm": 0.16907073557376862,
+      "learning_rate": 5.993230054401888e-10,
+      "loss": 0.0009607654996216297,
+      "mean_token_accuracy": 0.9998603224754333,
+      "num_tokens": 140141883.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.624900621175766,
+      "epoch": 10.0,
+      "grad_norm": 0.12695316970348358,
+      "learning_rate": 4.95309250281295e-12,
+      "loss": 0.0008850294165313244,
+      "mean_token_accuracy": 0.9999199330806732,
+      "num_tokens": 140331122.0,
+      "step": 7430
+    },
+    {
+      "epoch": 10.0,
+      "eval_entropy": 0.6315964798259127,
+      "eval_loss": 0.109488844871521,
+      "eval_mean_token_accuracy": 0.9727516504609661,
+      "eval_num_tokens": 140331122.0,
+      "eval_runtime": 12.7121,
+      "eval_samples_per_second": 393.326,
+      "eval_steps_per_second": 12.35,
+      "step": 7430
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.486418122495492e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/checkpoint-7430/training_args.bin b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/checkpoint-7430/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40/seed_42/tokenizer.json b/gen_z_slang_L16_a40/seed_42/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40/seed_42/tokenizer_config.json b/gen_z_slang_L16_a40/seed_42/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40/seed_42/training_args.bin b/gen_z_slang_L16_a40/seed_42/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8195cadd45f592a5077510e89b7d10a9f654cf82
--- /dev/null
+++ b/gen_z_slang_L16_a40/seed_42/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e08d67015892f30fae105f5d8f40ca93e099187553abd6a41b786f50c406ae8
+size 5841
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/README.md b/gen_z_slang_L16_a40_per_step/seed_42/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..2fc44c295cc98fb1268d3ce9be165ace5f23f1f0
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/README.md
@@ -0,0 +1,62 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+model_name: seed_42
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+licence: license
+pipeline_tag: text-generation
+---
+
+# Model Card for seed_42
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct).
+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="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/camilab-stanford-university/subliminal_learning/runs/owihs7a9) 
+
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- PEFT 0.19.1
+- TRL: 1.2.0
+- Transformers: 5.6.2
+- Pytorch: 2.10.0
+- 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/gen_z_slang_L16_a40_per_step/seed_42/adapter_config.json b/gen_z_slang_L16_a40_per_step/seed_42/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b5f6c54a4cd2b3d73264feb9fa6169ab404345ae
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj",
+    "up_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/adapter_model.safetensors b/gen_z_slang_L16_a40_per_step/seed_42/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..a8575565590f7ae266931483aec0bc5bf3d91506
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f669b06038d7b892c1aef9a98a67263a667419549d9d738c48877397434ea484
+size 80792096
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/chat_template.jinja b/gen_z_slang_L16_a40_per_step/seed_42/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/README.md b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/adapter_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b5f6c54a4cd2b3d73264feb9fa6169ab404345ae
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj",
+    "up_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/adapter_model.safetensors b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4caf5e2712dc74371f34afb4a0e43dbf771e7fa6
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f7c908e48768fc5cd0f69da5083151333bc3c259ad4ae68aa38d47f82e55f6a
+size 80792096
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/chat_template.jinja b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/tokenizer.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/tokenizer_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/trainer_state.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..a26c704428419d0e01f8d157636e9e2319ee88a4
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/trainer_state.json
@@ -0,0 +1,33 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.0003368137420006736,
+  "eval_steps": 500,
+  "global_step": 1,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 213404845342720.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/training_args.bin b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a9f7a8df021e8606f4786b466cbde2047c561402
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3fcf81de8a2b0c2caffc9e3027237c45be0a0437c37fb43780db3e1300c443f
+size 5841
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/README.md b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/adapter_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b5f6c54a4cd2b3d73264feb9fa6169ab404345ae
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj",
+    "up_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/adapter_model.safetensors b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..daa11da17a773e5032c528c54fc200fad7e1019b
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1f51364b76ff11a5c49d3d8ca3ebf6ecf7921356d4e6bd58fda90c8eed7a643
+size 80792096
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/chat_template.jinja b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/tokenizer.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/tokenizer_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/trainer_state.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..c43e45045825e7dbbcf35dfee38685374afcba17
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/trainer_state.json
@@ -0,0 +1,44 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.003368137420006736,
+  "eval_steps": 500,
+  "global_step": 10,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.26150141954422,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 2.06343150138855,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4125885009765625,
+      "mean_token_accuracy": 0.8628398180007935,
+      "num_tokens": 47274.0,
+      "step": 10
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2145280329777152.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/training_args.bin b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a9f7a8df021e8606f4786b466cbde2047c561402
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-10/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3fcf81de8a2b0c2caffc9e3027237c45be0a0437c37fb43780db3e1300c443f
+size 5841
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/README.md b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/adapter_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b5f6c54a4cd2b3d73264feb9fa6169ab404345ae
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj",
+    "up_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/adapter_model.safetensors b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..890529df50cf48bd62cd1b120a01aba4053d4722
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de5b1524855f4a6592f61a8b7813055d43f22e7cd00ce499eca928658b76d53e
+size 80792096
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/chat_template.jinja b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/tokenizer.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/tokenizer_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/trainer_state.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..d55ea00ca4477cfba6e09b0d05b76fe245fbf963
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/trainer_state.json
@@ -0,0 +1,134 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.033681374200067365,
+  "eval_steps": 500,
+  "global_step": 100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.26150141954422,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 2.06343150138855,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4125885009765625,
+      "mean_token_accuracy": 0.8628398180007935,
+      "num_tokens": 47274.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.261034321784973,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 1.2686415910720825,
+      "learning_rate": 3.8e-05,
+      "loss": 0.2816351890563965,
+      "mean_token_accuracy": 0.9029574036598206,
+      "num_tokens": 94236.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2639376759529113,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 1.0357964038848877,
+      "learning_rate": 5.8e-05,
+      "loss": 0.23235020637512208,
+      "mean_token_accuracy": 0.919199424982071,
+      "num_tokens": 141304.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.245469331741333,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 1.1952804327011108,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.19083542823791505,
+      "mean_token_accuracy": 0.9309875249862671,
+      "num_tokens": 188766.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2451943039894104,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 0.8794074654579163,
+      "learning_rate": 9.8e-05,
+      "loss": 0.17393248081207274,
+      "mean_token_accuracy": 0.9330495297908783,
+      "num_tokens": 236136.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2371069312095642,
+      "epoch": 0.020208824520040417,
+      "grad_norm": 1.3200767040252686,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.17803279161453248,
+      "mean_token_accuracy": 0.936758840084076,
+      "num_tokens": 283280.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.212414062023163,
+      "epoch": 0.023576961940047155,
+      "grad_norm": 0.8563512563705444,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.15086832046508789,
+      "mean_token_accuracy": 0.9422479271888733,
+      "num_tokens": 330555.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2040014863014221,
+      "epoch": 0.02694509936005389,
+      "grad_norm": 1.2015984058380127,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.1532510280609131,
+      "mean_token_accuracy": 0.9428703844547272,
+      "num_tokens": 377824.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.1925515413284302,
+      "epoch": 0.030313236780060628,
+      "grad_norm": 0.9743914604187012,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.14506156444549562,
+      "mean_token_accuracy": 0.9439641296863556,
+      "num_tokens": 424964.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.1726536750793457,
+      "epoch": 0.033681374200067365,
+      "grad_norm": 1.1765360832214355,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.14521509408950806,
+      "mean_token_accuracy": 0.943739241361618,
+      "num_tokens": 471689.0,
+      "step": 100
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.144395351765811e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/training_args.bin b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a9f7a8df021e8606f4786b466cbde2047c561402
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-100/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3fcf81de8a2b0c2caffc9e3027237c45be0a0437c37fb43780db3e1300c443f
+size 5841
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/README.md b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/adapter_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b5f6c54a4cd2b3d73264feb9fa6169ab404345ae
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj",
+    "up_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/adapter_model.safetensors b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..a8575565590f7ae266931483aec0bc5bf3d91506
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f669b06038d7b892c1aef9a98a67263a667419549d9d738c48877397434ea484
+size 80792096
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/chat_template.jinja b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/tokenizer.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/tokenizer_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/trainer_state.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..52d37d9195b4e0307ac13ee8f8ef3bc2f3007e5a
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/trainer_state.json
@@ -0,0 +1,1034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.33681374200067365,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.26150141954422,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 2.06343150138855,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4125885009765625,
+      "mean_token_accuracy": 0.8628398180007935,
+      "num_tokens": 47274.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.261034321784973,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 1.2686415910720825,
+      "learning_rate": 3.8e-05,
+      "loss": 0.2816351890563965,
+      "mean_token_accuracy": 0.9029574036598206,
+      "num_tokens": 94236.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2639376759529113,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 1.0357964038848877,
+      "learning_rate": 5.8e-05,
+      "loss": 0.23235020637512208,
+      "mean_token_accuracy": 0.919199424982071,
+      "num_tokens": 141304.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.245469331741333,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 1.1952804327011108,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.19083542823791505,
+      "mean_token_accuracy": 0.9309875249862671,
+      "num_tokens": 188766.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2451943039894104,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 0.8794074654579163,
+      "learning_rate": 9.8e-05,
+      "loss": 0.17393248081207274,
+      "mean_token_accuracy": 0.9330495297908783,
+      "num_tokens": 236136.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2371069312095642,
+      "epoch": 0.020208824520040417,
+      "grad_norm": 1.3200767040252686,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.17803279161453248,
+      "mean_token_accuracy": 0.936758840084076,
+      "num_tokens": 283280.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.212414062023163,
+      "epoch": 0.023576961940047155,
+      "grad_norm": 0.8563512563705444,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.15086832046508789,
+      "mean_token_accuracy": 0.9422479271888733,
+      "num_tokens": 330555.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2040014863014221,
+      "epoch": 0.02694509936005389,
+      "grad_norm": 1.2015984058380127,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.1532510280609131,
+      "mean_token_accuracy": 0.9428703844547272,
+      "num_tokens": 377824.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.1925515413284302,
+      "epoch": 0.030313236780060628,
+      "grad_norm": 0.9743914604187012,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.14506156444549562,
+      "mean_token_accuracy": 0.9439641296863556,
+      "num_tokens": 424964.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.1726536750793457,
+      "epoch": 0.033681374200067365,
+      "grad_norm": 1.1765360832214355,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.14521509408950806,
+      "mean_token_accuracy": 0.943739241361618,
+      "num_tokens": 471689.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.1579697728157043,
+      "epoch": 0.037049511620074096,
+      "grad_norm": 0.9645544290542603,
+      "learning_rate": 9.905132290792394e-05,
+      "loss": 0.13851391077041625,
+      "mean_token_accuracy": 0.9471000492572784,
+      "num_tokens": 518917.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.1558732628822326,
+      "epoch": 0.040417649040080834,
+      "grad_norm": 1.1995741128921509,
+      "learning_rate": 9.870399824239117e-05,
+      "loss": 0.14726691246032714,
+      "mean_token_accuracy": 0.9451506495475769,
+      "num_tokens": 566046.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1335411667823792,
+      "epoch": 0.04378578646008757,
+      "grad_norm": 1.4308767318725586,
+      "learning_rate": 9.830341646984521e-05,
+      "loss": 0.12980954647064208,
+      "mean_token_accuracy": 0.9486727833747863,
+      "num_tokens": 613407.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.096289575099945,
+      "epoch": 0.04715392388009431,
+      "grad_norm": 1.212436318397522,
+      "learning_rate": 9.785001562057309e-05,
+      "loss": 0.12626460790634156,
+      "mean_token_accuracy": 0.9514972805976868,
+      "num_tokens": 660718.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.0868672966957091,
+      "epoch": 0.05052206130010104,
+      "grad_norm": 1.4931124448776245,
+      "learning_rate": 9.734429148174675e-05,
+      "loss": 0.13412034511566162,
+      "mean_token_accuracy": 0.950585001707077,
+      "num_tokens": 707878.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.0756590366363525,
+      "epoch": 0.05389019872010778,
+      "grad_norm": 1.5189236402511597,
+      "learning_rate": 9.6786797055287e-05,
+      "loss": 0.1373395323753357,
+      "mean_token_accuracy": 0.9483886420726776,
+      "num_tokens": 755000.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.0566098809242248,
+      "epoch": 0.05725833614011452,
+      "grad_norm": 1.2733327150344849,
+      "learning_rate": 9.617814195316411e-05,
+      "loss": 0.12557408809661866,
+      "mean_token_accuracy": 0.9514977693557739,
+      "num_tokens": 802361.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.0407122492790222,
+      "epoch": 0.060626473560121255,
+      "grad_norm": 1.176454782485962,
+      "learning_rate": 9.551899173079607e-05,
+      "loss": 0.12443351745605469,
+      "mean_token_accuracy": 0.9529365718364715,
+      "num_tokens": 849558.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.0345396757125855,
+      "epoch": 0.06399461098012799,
+      "grad_norm": 1.0860135555267334,
+      "learning_rate": 9.481006715927351e-05,
+      "loss": 0.11655720472335815,
+      "mean_token_accuracy": 0.953818553686142,
+      "num_tokens": 896799.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.9963904440402984,
+      "epoch": 0.06736274840013473,
+      "grad_norm": 1.5330692529678345,
+      "learning_rate": 9.405214343720707e-05,
+      "loss": 0.11721079349517823,
+      "mean_token_accuracy": 0.9548412740230561,
+      "num_tokens": 944080.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0103840470314025,
+      "epoch": 0.07073088582014146,
+      "grad_norm": 1.5443956851959229,
+      "learning_rate": 9.32460493430591e-05,
+      "loss": 0.12003815174102783,
+      "mean_token_accuracy": 0.9548316895961761,
+      "num_tokens": 991067.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.9955168128013611,
+      "epoch": 0.07409902324014819,
+      "grad_norm": 1.168945550918579,
+      "learning_rate": 9.239266632888659e-05,
+      "loss": 0.11371732950210571,
+      "mean_token_accuracy": 0.9567750275135041,
+      "num_tokens": 1038277.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.9822373270988465,
+      "epoch": 0.07746716066015494,
+      "grad_norm": 1.3211121559143066,
+      "learning_rate": 9.14929275564863e-05,
+      "loss": 0.10992491245269775,
+      "mean_token_accuracy": 0.9574353575706482,
+      "num_tokens": 1085614.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.9841293931007385,
+      "epoch": 0.08083529808016167,
+      "grad_norm": 1.431372046470642,
+      "learning_rate": 9.0547816876996e-05,
+      "loss": 0.11098736524581909,
+      "mean_token_accuracy": 0.9559519648551941,
+      "num_tokens": 1132745.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.9728141784667969,
+      "epoch": 0.08420343550016841,
+      "grad_norm": 1.0767285823822021,
+      "learning_rate": 8.955836775506776e-05,
+      "loss": 0.11628932952880859,
+      "mean_token_accuracy": 0.9545054733753204,
+      "num_tokens": 1179964.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.9633705317974091,
+      "epoch": 0.08757157292017514,
+      "grad_norm": 1.3769149780273438,
+      "learning_rate": 8.852566213878947e-05,
+      "loss": 0.11148278713226319,
+      "mean_token_accuracy": 0.955774313211441,
+      "num_tokens": 1227353.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9550830721855164,
+      "epoch": 0.09093971034018188,
+      "grad_norm": 1.5154286623001099,
+      "learning_rate": 8.745082927659047e-05,
+      "loss": 0.11857821941375732,
+      "mean_token_accuracy": 0.9542500138282776,
+      "num_tokens": 1274727.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9538668096065521,
+      "epoch": 0.09430784776018862,
+      "grad_norm": 1.4699420928955078,
+      "learning_rate": 8.633504448242505e-05,
+      "loss": 0.1061704158782959,
+      "mean_token_accuracy": 0.956998211145401,
+      "num_tokens": 1321873.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9443166971206665,
+      "epoch": 0.09767598518019535,
+      "grad_norm": 1.9283766746520996,
+      "learning_rate": 8.517952785058385e-05,
+      "loss": 0.10411903858184815,
+      "mean_token_accuracy": 0.9600685238838196,
+      "num_tokens": 1369013.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9425238311290741,
+      "epoch": 0.10104412260020208,
+      "grad_norm": 1.3883271217346191,
+      "learning_rate": 8.398554292153866e-05,
+      "loss": 0.11232163906097412,
+      "mean_token_accuracy": 0.9574631214141845,
+      "num_tokens": 1416502.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9327979981899261,
+      "epoch": 0.10441226002020883,
+      "grad_norm": 1.153810739517212,
+      "learning_rate": 8.275439530027948e-05,
+      "loss": 0.10868992805480956,
+      "mean_token_accuracy": 0.9569708108901978,
+      "num_tokens": 1463995.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9125505864620209,
+      "epoch": 0.10778039744021556,
+      "grad_norm": 1.5007269382476807,
+      "learning_rate": 8.148743122865463e-05,
+      "loss": 0.09841209053993225,
+      "mean_token_accuracy": 0.9614197909832001,
+      "num_tokens": 1511241.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.901406067609787,
+      "epoch": 0.1111485348602223,
+      "grad_norm": 1.5678924322128296,
+      "learning_rate": 8.018603611327504e-05,
+      "loss": 0.1065678596496582,
+      "mean_token_accuracy": 0.9581831812858581,
+      "num_tokens": 1558413.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9037010133266449,
+      "epoch": 0.11451667228022903,
+      "grad_norm": 1.3490980863571167,
+      "learning_rate": 7.88516330105925e-05,
+      "loss": 0.10169246196746826,
+      "mean_token_accuracy": 0.9593528211116791,
+      "num_tokens": 1605661.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.8834535896778106,
+      "epoch": 0.11788480970023577,
+      "grad_norm": 1.0754783153533936,
+      "learning_rate": 7.748568107080832e-05,
+      "loss": 0.0994927704334259,
+      "mean_token_accuracy": 0.9605486214160919,
+      "num_tokens": 1653019.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.896017187833786,
+      "epoch": 0.12125294712024251,
+      "grad_norm": 1.7949540615081787,
+      "learning_rate": 7.608967394231387e-05,
+      "loss": 0.10669023990631103,
+      "mean_token_accuracy": 0.9556081056594848,
+      "num_tokens": 1700078.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.8881138920783996,
+      "epoch": 0.12462108454024924,
+      "grad_norm": 1.455610990524292,
+      "learning_rate": 7.466513813840825e-05,
+      "loss": 0.09705370664596558,
+      "mean_token_accuracy": 0.9605468213558197,
+      "num_tokens": 1747298.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.8885541558265686,
+      "epoch": 0.12798922196025597,
+      "grad_norm": 1.4581488370895386,
+      "learning_rate": 7.32136313680782e-05,
+      "loss": 0.10148756504058838,
+      "mean_token_accuracy": 0.9591729998588562,
+      "num_tokens": 1794416.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.887837141752243,
+      "epoch": 0.13135735938026272,
+      "grad_norm": 1.7846956253051758,
+      "learning_rate": 7.173674083266624e-05,
+      "loss": 0.0982971429824829,
+      "mean_token_accuracy": 0.962555181980133,
+      "num_tokens": 1841740.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.877079302072525,
+      "epoch": 0.13472549680026946,
+      "grad_norm": 1.5539696216583252,
+      "learning_rate": 7.023608149028937e-05,
+      "loss": 0.10046224594116211,
+      "mean_token_accuracy": 0.9607136070728302,
+      "num_tokens": 1888891.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.885935229063034,
+      "epoch": 0.13809363422027618,
+      "grad_norm": 1.3651353120803833,
+      "learning_rate": 6.871329428990602e-05,
+      "loss": 0.09598114490509033,
+      "mean_token_accuracy": 0.9629281044006348,
+      "num_tokens": 1936088.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8735738277435303,
+      "epoch": 0.14146177164028292,
+      "grad_norm": 1.505676031112671,
+      "learning_rate": 6.71700443769625e-05,
+      "loss": 0.1017297625541687,
+      "mean_token_accuracy": 0.9589500606060029,
+      "num_tokens": 1983258.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.8703845381736756,
+      "epoch": 0.14482990906028967,
+      "grad_norm": 1.188096046447754,
+      "learning_rate": 6.56080192725808e-05,
+      "loss": 0.10356165170669555,
+      "mean_token_accuracy": 0.9581273019313812,
+      "num_tokens": 2030588.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8790095925331116,
+      "epoch": 0.14819804648029639,
+      "grad_norm": 1.4058703184127808,
+      "learning_rate": 6.402892702827916e-05,
+      "loss": 0.09762775897979736,
+      "mean_token_accuracy": 0.9619512498378754,
+      "num_tokens": 2077874.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.8651655197143555,
+      "epoch": 0.15156618390030313,
+      "grad_norm": 1.4287638664245605,
+      "learning_rate": 6.243449435824276e-05,
+      "loss": 0.09559672474861144,
+      "mean_token_accuracy": 0.9625435173511505,
+      "num_tokens": 2125044.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.86707643866539,
+      "epoch": 0.15493432132030988,
+      "grad_norm": 1.216744303703308,
+      "learning_rate": 6.0826464751186994e-05,
+      "loss": 0.09722432494163513,
+      "mean_token_accuracy": 0.96131911277771,
+      "num_tokens": 2172385.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8562962651252747,
+      "epoch": 0.1583024587403166,
+      "grad_norm": 2.0901386737823486,
+      "learning_rate": 5.9206596563878357e-05,
+      "loss": 0.09744423031806945,
+      "mean_token_accuracy": 0.9623366236686707,
+      "num_tokens": 2219424.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8649725198745728,
+      "epoch": 0.16167059616032334,
+      "grad_norm": 1.049863338470459,
+      "learning_rate": 5.757666109839702e-05,
+      "loss": 0.0890600323677063,
+      "mean_token_accuracy": 0.9647176265716553,
+      "num_tokens": 2267014.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8541165649890899,
+      "epoch": 0.16503873358033008,
+      "grad_norm": 1.6462875604629517,
+      "learning_rate": 5.5938440665244006e-05,
+      "loss": 0.09485594630241394,
+      "mean_token_accuracy": 0.9614668607711792,
+      "num_tokens": 2314008.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8513838946819305,
+      "epoch": 0.16840687100033683,
+      "grad_norm": 1.4663439989089966,
+      "learning_rate": 5.4293726634410855e-05,
+      "loss": 0.09900916814804077,
+      "mean_token_accuracy": 0.9601402223110199,
+      "num_tokens": 2361287.0,
+      "step": 500
+    },
+    {
+      "entropy": 0.8453722596168518,
+      "epoch": 0.17177500842034354,
+      "grad_norm": 1.7513787746429443,
+      "learning_rate": 5.264431747654284e-05,
+      "loss": 0.0901253342628479,
+      "mean_token_accuracy": 0.963881641626358,
+      "num_tokens": 2408390.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8526451289653778,
+      "epoch": 0.1751431458403503,
+      "grad_norm": 1.691697359085083,
+      "learning_rate": 5.0992016796337686e-05,
+      "loss": 0.09386304616928101,
+      "mean_token_accuracy": 0.9622337341308593,
+      "num_tokens": 2455750.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8493820130825043,
+      "epoch": 0.17851128326035703,
+      "grad_norm": 1.745545744895935,
+      "learning_rate": 4.93386313603304e-05,
+      "loss": 0.08609845638275146,
+      "mean_token_accuracy": 0.9667652487754822,
+      "num_tokens": 2502963.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.8412189602851867,
+      "epoch": 0.18187942068036375,
+      "grad_norm": 1.1819779872894287,
+      "learning_rate": 4.7685969121220456e-05,
+      "loss": 0.0822030782699585,
+      "mean_token_accuracy": 0.9673902690410614,
+      "num_tokens": 2550355.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.8436509847640992,
+      "epoch": 0.1852475581003705,
+      "grad_norm": 1.7780817747116089,
+      "learning_rate": 4.60358372409022e-05,
+      "loss": 0.08904764652252198,
+      "mean_token_accuracy": 0.9639313519001007,
+      "num_tokens": 2597430.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.8515670359134674,
+      "epoch": 0.18861569552037724,
+      "grad_norm": 1.2775158882141113,
+      "learning_rate": 4.439004011435979e-05,
+      "loss": 0.08926515579223633,
+      "mean_token_accuracy": 0.9637289464473724,
+      "num_tokens": 2644556.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8522431552410126,
+      "epoch": 0.19198383294038396,
+      "grad_norm": 1.2618097066879272,
+      "learning_rate": 4.275037739658771e-05,
+      "loss": 0.08655033111572266,
+      "mean_token_accuracy": 0.966059023141861,
+      "num_tokens": 2691447.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8438503384590149,
+      "epoch": 0.1953519703603907,
+      "grad_norm": 1.4280424118041992,
+      "learning_rate": 4.111864203469457e-05,
+      "loss": 0.08119413852691651,
+      "mean_token_accuracy": 0.9673396706581116,
+      "num_tokens": 2738560.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8372359216213227,
+      "epoch": 0.19872010778039745,
+      "grad_norm": 1.3594173192977905,
+      "learning_rate": 3.949661830734172e-05,
+      "loss": 0.08984186053276062,
+      "mean_token_accuracy": 0.9628469705581665,
+      "num_tokens": 2785743.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8444832921028137,
+      "epoch": 0.20208824520040417,
+      "grad_norm": 1.5820175409317017,
+      "learning_rate": 3.788607987366069e-05,
+      "loss": 0.08010109066963196,
+      "mean_token_accuracy": 0.9688393175601959,
+      "num_tokens": 2832789.0,
+      "step": 600
+    },
+    {
+      "entropy": 0.8512360632419587,
+      "epoch": 0.2054563826204109,
+      "grad_norm": 1.4823386669158936,
+      "learning_rate": 3.628878783378302e-05,
+      "loss": 0.08498185276985168,
+      "mean_token_accuracy": 0.9654136657714844,
+      "num_tokens": 2880243.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8519550561904907,
+      "epoch": 0.20882452004041765,
+      "grad_norm": 1.9961153268814087,
+      "learning_rate": 3.470648880310313e-05,
+      "loss": 0.08613522052764892,
+      "mean_token_accuracy": 0.9651396930217743,
+      "num_tokens": 2927421.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8495919585227967,
+      "epoch": 0.21219265746042437,
+      "grad_norm": 1.6366565227508545,
+      "learning_rate": 3.3140913002379995e-05,
+      "loss": 0.08417365550994874,
+      "mean_token_accuracy": 0.9680097043514252,
+      "num_tokens": 2974702.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.855664199590683,
+      "epoch": 0.21556079488043112,
+      "grad_norm": 1.274781346321106,
+      "learning_rate": 3.1593772365766105e-05,
+      "loss": 0.09047029614448547,
+      "mean_token_accuracy": 0.9641342759132385,
+      "num_tokens": 3021802.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8476129055023194,
+      "epoch": 0.21892893230043786,
+      "grad_norm": 1.513195514678955,
+      "learning_rate": 3.006675866883275e-05,
+      "loss": 0.08097966313362122,
+      "mean_token_accuracy": 0.9654085099697113,
+      "num_tokens": 3069204.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.8455423176288605,
+      "epoch": 0.2222970697204446,
+      "grad_norm": 1.1273273229599,
+      "learning_rate": 2.8561541678638142e-05,
+      "loss": 0.08873566389083862,
+      "mean_token_accuracy": 0.965383517742157,
+      "num_tokens": 3116353.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.8532886445522309,
+      "epoch": 0.22566520714045132,
+      "grad_norm": 1.661704182624817,
+      "learning_rate": 2.707976732786166e-05,
+      "loss": 0.08862345218658448,
+      "mean_token_accuracy": 0.965210372209549,
+      "num_tokens": 3163526.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8528507053852081,
+      "epoch": 0.22903334456045807,
+      "grad_norm": 1.2517887353897095,
+      "learning_rate": 2.562305591500069e-05,
+      "loss": 0.08536858558654785,
+      "mean_token_accuracy": 0.964106684923172,
+      "num_tokens": 3210479.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8466204166412353,
+      "epoch": 0.2324014819804648,
+      "grad_norm": 1.3938658237457275,
+      "learning_rate": 2.419300033259798e-05,
+      "loss": 0.07965762615203857,
+      "mean_token_accuracy": 0.9678584933280945,
+      "num_tokens": 3257923.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.8476869702339173,
+      "epoch": 0.23576961940047153,
+      "grad_norm": 1.8682013750076294,
+      "learning_rate": 2.279116432543705e-05,
+      "loss": 0.08289393186569213,
+      "mean_token_accuracy": 0.9677723586559296,
+      "num_tokens": 3305209.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.8380595564842224,
+      "epoch": 0.23913775682047828,
+      "grad_norm": 1.3846161365509033,
+      "learning_rate": 2.1419080780610123e-05,
+      "loss": 0.07611767649650573,
+      "mean_token_accuracy": 0.9695590555667877,
+      "num_tokens": 3352648.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8474281907081604,
+      "epoch": 0.24250589424048502,
+      "grad_norm": 1.5169308185577393,
+      "learning_rate": 2.0078250051328784e-05,
+      "loss": 0.08080564737319947,
+      "mean_token_accuracy": 0.966937804222107,
+      "num_tokens": 3399701.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8432491719722748,
+      "epoch": 0.24587403166049174,
+      "grad_norm": 1.2972283363342285,
+      "learning_rate": 1.877013831630961e-05,
+      "loss": 0.07204076647758484,
+      "mean_token_accuracy": 0.9707867801189423,
+      "num_tokens": 3446806.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8430583000183105,
+      "epoch": 0.24924216908049848,
+      "grad_norm": 1.4284356832504272,
+      "learning_rate": 1.749617597652934e-05,
+      "loss": 0.08301993012428284,
+      "mean_token_accuracy": 0.9662304580211639,
+      "num_tokens": 3494195.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.8418968915939331,
+      "epoch": 0.2526103065005052,
+      "grad_norm": 1.6723601818084717,
+      "learning_rate": 1.62577560911024e-05,
+      "loss": 0.07664074301719666,
+      "mean_token_accuracy": 0.9696024537086487,
+      "num_tokens": 3541194.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.844183599948883,
+      "epoch": 0.25597844392051194,
+      "grad_norm": 1.4064338207244873,
+      "learning_rate": 1.5056232853991209e-05,
+      "loss": 0.07462746500968934,
+      "mean_token_accuracy": 0.9697076916694641,
+      "num_tokens": 3588258.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.839869600534439,
+      "epoch": 0.2593465813405187,
+      "grad_norm": 1.7115681171417236,
+      "learning_rate": 1.389292011321498e-05,
+      "loss": 0.08336342573165893,
+      "mean_token_accuracy": 0.9651942908763885,
+      "num_tokens": 3635297.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8353487014770508,
+      "epoch": 0.26271471876052543,
+      "grad_norm": 1.4900928735733032,
+      "learning_rate": 1.2769089934176126e-05,
+      "loss": 0.0708418607711792,
+      "mean_token_accuracy": 0.9718195080757142,
+      "num_tokens": 3682693.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.8410345256328583,
+      "epoch": 0.2660828561805322,
+      "grad_norm": 1.7543418407440186,
+      "learning_rate": 1.1685971208675539e-05,
+      "loss": 0.08269069194793702,
+      "mean_token_accuracy": 0.9668975353240967,
+      "num_tokens": 3729906.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8406113862991333,
+      "epoch": 0.2694509936005389,
+      "grad_norm": 1.592735767364502,
+      "learning_rate": 1.0644748311137376e-05,
+      "loss": 0.07227774858474731,
+      "mean_token_accuracy": 0.9715151250362396,
+      "num_tokens": 3776978.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.8413715600967407,
+      "epoch": 0.2728191310205456,
+      "grad_norm": 1.333324670791626,
+      "learning_rate": 9.646559803512994e-06,
+      "loss": 0.069387549161911,
+      "mean_token_accuracy": 0.9732672214508057,
+      "num_tokens": 3824277.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.842104572057724,
+      "epoch": 0.27618726844055236,
+      "grad_norm": 1.7370861768722534,
+      "learning_rate": 8.692497190280224e-06,
+      "loss": 0.08174294829368592,
+      "mean_token_accuracy": 0.9679147779941559,
+      "num_tokens": 3871151.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.8351531744003295,
+      "epoch": 0.2795554058605591,
+      "grad_norm": 1.398715615272522,
+      "learning_rate": 7.783603724899257e-06,
+      "loss": 0.0713503122329712,
+      "mean_token_accuracy": 0.9720901250839233,
+      "num_tokens": 3918394.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.8329871535301209,
+      "epoch": 0.28292354328056585,
+      "grad_norm": 1.1645716428756714,
+      "learning_rate": 6.92087326903022e-06,
+      "loss": 0.073863685131073,
+      "mean_token_accuracy": 0.9698374211788178,
+      "num_tokens": 3965629.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8328420460224152,
+      "epoch": 0.2862916807005726,
+      "grad_norm": 1.5381746292114258,
+      "learning_rate": 6.1052492057601275e-06,
+      "loss": 0.06840627193450928,
+      "mean_token_accuracy": 0.972251009941101,
+      "num_tokens": 4013000.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.8365510821342468,
+      "epoch": 0.28965981812057934,
+      "grad_norm": 1.1510448455810547,
+      "learning_rate": 5.337623408027293e-06,
+      "loss": 0.07609614133834838,
+      "mean_token_accuracy": 0.968937873840332,
+      "num_tokens": 4060080.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.8340778648853302,
+      "epoch": 0.2930279555405861,
+      "grad_norm": 1.3923290967941284,
+      "learning_rate": 4.618835263371396e-06,
+      "loss": 0.07038736939430237,
+      "mean_token_accuracy": 0.9724739372730256,
+      "num_tokens": 4107464.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.8322880327701568,
+      "epoch": 0.29639609296059277,
+      "grad_norm": 1.4906857013702393,
+      "learning_rate": 3.949670756075447e-06,
+      "loss": 0.0725872814655304,
+      "mean_token_accuracy": 0.9712622702121735,
+      "num_tokens": 4154775.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.8367474555969239,
+      "epoch": 0.2997642303805995,
+      "grad_norm": 1.2631500959396362,
+      "learning_rate": 3.3308616077036115e-06,
+      "loss": 0.0743180751800537,
+      "mean_token_accuracy": 0.969811075925827,
+      "num_tokens": 4201916.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.8430841863155365,
+      "epoch": 0.30313236780060626,
+      "grad_norm": 1.6571202278137207,
+      "learning_rate": 2.7630844769743757e-06,
+      "loss": 0.0787225067615509,
+      "mean_token_accuracy": 0.9681240141391754,
+      "num_tokens": 4249241.0,
+      "step": 900
+    },
+    {
+      "entropy": 0.8341751813888549,
+      "epoch": 0.306500505220613,
+      "grad_norm": 1.655290126800537,
+      "learning_rate": 2.2469602198441573e-06,
+      "loss": 0.06645160317420959,
+      "mean_token_accuracy": 0.9744393944740295,
+      "num_tokens": 4296617.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8341230988502503,
+      "epoch": 0.30986864264061975,
+      "grad_norm": 1.2294965982437134,
+      "learning_rate": 1.7830532106104747e-06,
+      "loss": 0.06916466355323792,
+      "mean_token_accuracy": 0.9739417910575867,
+      "num_tokens": 4343835.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.8407280623912812,
+      "epoch": 0.3132367800606265,
+      "grad_norm": 2.018903970718384,
+      "learning_rate": 1.3718707247769135e-06,
+      "loss": 0.08383288979530334,
+      "mean_token_accuracy": 0.9670386135578155,
+      "num_tokens": 4390882.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.8360171020030975,
+      "epoch": 0.3166049174806332,
+      "grad_norm": 1.2670494318008423,
+      "learning_rate": 1.0138623843548078e-06,
+      "loss": 0.06570031642913818,
+      "mean_token_accuracy": 0.9732370495796203,
+      "num_tokens": 4438148.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.8405866503715516,
+      "epoch": 0.31997305490063993,
+      "grad_norm": 1.4799150228500366,
+      "learning_rate": 7.094196662081831e-07,
+      "loss": 0.08266336917877197,
+      "mean_token_accuracy": 0.9671482145786285,
+      "num_tokens": 4485461.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.8370142102241516,
+      "epoch": 0.3233411923206467,
+      "grad_norm": 1.5308057069778442,
+      "learning_rate": 4.5887547397955864e-07,
+      "loss": 0.06924953460693359,
+      "mean_token_accuracy": 0.9712660253047943,
+      "num_tokens": 4532640.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.8376805007457733,
+      "epoch": 0.3267093297406534,
+      "grad_norm": 1.478533387184143,
+      "learning_rate": 2.625037740646763e-07,
+      "loss": 0.07524077296257019,
+      "mean_token_accuracy": 0.9702640414237976,
+      "num_tokens": 4580068.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.8369512259960175,
+      "epoch": 0.33007746716066017,
+      "grad_norm": 1.6361910104751587,
+      "learning_rate": 1.2051929603428825e-07,
+      "loss": 0.07570222616195679,
+      "mean_token_accuracy": 0.9694684982299805,
+      "num_tokens": 4627354.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.8370845556259155,
+      "epoch": 0.3334456045806669,
+      "grad_norm": 1.3835561275482178,
+      "learning_rate": 3.3077297830541584e-08,
+      "loss": 0.0745551586151123,
+      "mean_token_accuracy": 0.9703558146953583,
+      "num_tokens": 4674411.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.8373138248920441,
+      "epoch": 0.33681374200067365,
+      "grad_norm": 1.4666378498077393,
+      "learning_rate": 2.7339599464326627e-10,
+      "loss": 0.0733814775943756,
+      "mean_token_accuracy": 0.9723501443862915,
+      "num_tokens": 4721431.0,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1455729794534605e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/training_args.bin b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a9f7a8df021e8606f4786b466cbde2047c561402
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-1000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3fcf81de8a2b0c2caffc9e3027237c45be0a0437c37fb43780db3e1300c443f
+size 5841
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/README.md b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/adapter_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b5f6c54a4cd2b3d73264feb9fa6169ab404345ae
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj",
+    "up_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/adapter_model.safetensors b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..1fdc3e651d0f7621ac4952252676492d65fa123d
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0550a8cbdfbbfcdb4b48d1dd0246c8c9ca669093134ea066351d4208349eab54
+size 80792096
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/chat_template.jinja b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/tokenizer.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/tokenizer_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/trainer_state.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..26eabe9820bba5cfb55d975e7ed6bb753a8d6829
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/trainer_state.json
@@ -0,0 +1,84 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.016840687100033683,
+  "eval_steps": 500,
+  "global_step": 50,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.26150141954422,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 2.06343150138855,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4125885009765625,
+      "mean_token_accuracy": 0.8628398180007935,
+      "num_tokens": 47274.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.261034321784973,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 1.2686415910720825,
+      "learning_rate": 3.8e-05,
+      "loss": 0.2816351890563965,
+      "mean_token_accuracy": 0.9029574036598206,
+      "num_tokens": 94236.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2639376759529113,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 1.0357964038848877,
+      "learning_rate": 5.8e-05,
+      "loss": 0.23235020637512208,
+      "mean_token_accuracy": 0.919199424982071,
+      "num_tokens": 141304.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.245469331741333,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 1.1952804327011108,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.19083542823791505,
+      "mean_token_accuracy": 0.9309875249862671,
+      "num_tokens": 188766.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2451943039894104,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 0.8794074654579163,
+      "learning_rate": 9.8e-05,
+      "loss": 0.17393248081207274,
+      "mean_token_accuracy": 0.9330495297908783,
+      "num_tokens": 236136.0,
+      "step": 50
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0728443637399552e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/training_args.bin b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a9f7a8df021e8606f4786b466cbde2047c561402
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-50/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3fcf81de8a2b0c2caffc9e3027237c45be0a0437c37fb43780db3e1300c443f
+size 5841
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/README.md b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/adapter_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b5f6c54a4cd2b3d73264feb9fa6169ab404345ae
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj",
+    "up_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/adapter_model.safetensors b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4a33d6a8d6fc0588c86e9c83ddf6cd0d4bb978b8
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e80d5eb3a305664ff37c2c3ed0a8a710aebafeda0303f5a108dc9f4b139e9bb
+size 80792096
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/chat_template.jinja b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/tokenizer.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/tokenizer_config.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/trainer_state.json b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..3d5cd1a49de2be965df7b8edd75320e92acf6630
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/trainer_state.json
@@ -0,0 +1,534 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.16840687100033683,
+  "eval_steps": 500,
+  "global_step": 500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.26150141954422,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 2.06343150138855,
+      "learning_rate": 1.8e-05,
+      "loss": 0.4125885009765625,
+      "mean_token_accuracy": 0.8628398180007935,
+      "num_tokens": 47274.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.261034321784973,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 1.2686415910720825,
+      "learning_rate": 3.8e-05,
+      "loss": 0.2816351890563965,
+      "mean_token_accuracy": 0.9029574036598206,
+      "num_tokens": 94236.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.2639376759529113,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 1.0357964038848877,
+      "learning_rate": 5.8e-05,
+      "loss": 0.23235020637512208,
+      "mean_token_accuracy": 0.919199424982071,
+      "num_tokens": 141304.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.245469331741333,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 1.1952804327011108,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.19083542823791505,
+      "mean_token_accuracy": 0.9309875249862671,
+      "num_tokens": 188766.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.2451943039894104,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 0.8794074654579163,
+      "learning_rate": 9.8e-05,
+      "loss": 0.17393248081207274,
+      "mean_token_accuracy": 0.9330495297908783,
+      "num_tokens": 236136.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2371069312095642,
+      "epoch": 0.020208824520040417,
+      "grad_norm": 1.3200767040252686,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.17803279161453248,
+      "mean_token_accuracy": 0.936758840084076,
+      "num_tokens": 283280.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.212414062023163,
+      "epoch": 0.023576961940047155,
+      "grad_norm": 0.8563512563705444,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.15086832046508789,
+      "mean_token_accuracy": 0.9422479271888733,
+      "num_tokens": 330555.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2040014863014221,
+      "epoch": 0.02694509936005389,
+      "grad_norm": 1.2015984058380127,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.1532510280609131,
+      "mean_token_accuracy": 0.9428703844547272,
+      "num_tokens": 377824.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.1925515413284302,
+      "epoch": 0.030313236780060628,
+      "grad_norm": 0.9743914604187012,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.14506156444549562,
+      "mean_token_accuracy": 0.9439641296863556,
+      "num_tokens": 424964.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.1726536750793457,
+      "epoch": 0.033681374200067365,
+      "grad_norm": 1.1765360832214355,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.14521509408950806,
+      "mean_token_accuracy": 0.943739241361618,
+      "num_tokens": 471689.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.1579697728157043,
+      "epoch": 0.037049511620074096,
+      "grad_norm": 0.9645544290542603,
+      "learning_rate": 9.905132290792394e-05,
+      "loss": 0.13851391077041625,
+      "mean_token_accuracy": 0.9471000492572784,
+      "num_tokens": 518917.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.1558732628822326,
+      "epoch": 0.040417649040080834,
+      "grad_norm": 1.1995741128921509,
+      "learning_rate": 9.870399824239117e-05,
+      "loss": 0.14726691246032714,
+      "mean_token_accuracy": 0.9451506495475769,
+      "num_tokens": 566046.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.1335411667823792,
+      "epoch": 0.04378578646008757,
+      "grad_norm": 1.4308767318725586,
+      "learning_rate": 9.830341646984521e-05,
+      "loss": 0.12980954647064208,
+      "mean_token_accuracy": 0.9486727833747863,
+      "num_tokens": 613407.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.096289575099945,
+      "epoch": 0.04715392388009431,
+      "grad_norm": 1.212436318397522,
+      "learning_rate": 9.785001562057309e-05,
+      "loss": 0.12626460790634156,
+      "mean_token_accuracy": 0.9514972805976868,
+      "num_tokens": 660718.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.0868672966957091,
+      "epoch": 0.05052206130010104,
+      "grad_norm": 1.4931124448776245,
+      "learning_rate": 9.734429148174675e-05,
+      "loss": 0.13412034511566162,
+      "mean_token_accuracy": 0.950585001707077,
+      "num_tokens": 707878.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.0756590366363525,
+      "epoch": 0.05389019872010778,
+      "grad_norm": 1.5189236402511597,
+      "learning_rate": 9.6786797055287e-05,
+      "loss": 0.1373395323753357,
+      "mean_token_accuracy": 0.9483886420726776,
+      "num_tokens": 755000.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.0566098809242248,
+      "epoch": 0.05725833614011452,
+      "grad_norm": 1.2733327150344849,
+      "learning_rate": 9.617814195316411e-05,
+      "loss": 0.12557408809661866,
+      "mean_token_accuracy": 0.9514977693557739,
+      "num_tokens": 802361.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.0407122492790222,
+      "epoch": 0.060626473560121255,
+      "grad_norm": 1.176454782485962,
+      "learning_rate": 9.551899173079607e-05,
+      "loss": 0.12443351745605469,
+      "mean_token_accuracy": 0.9529365718364715,
+      "num_tokens": 849558.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.0345396757125855,
+      "epoch": 0.06399461098012799,
+      "grad_norm": 1.0860135555267334,
+      "learning_rate": 9.481006715927351e-05,
+      "loss": 0.11655720472335815,
+      "mean_token_accuracy": 0.953818553686142,
+      "num_tokens": 896799.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.9963904440402984,
+      "epoch": 0.06736274840013473,
+      "grad_norm": 1.5330692529678345,
+      "learning_rate": 9.405214343720707e-05,
+      "loss": 0.11721079349517823,
+      "mean_token_accuracy": 0.9548412740230561,
+      "num_tokens": 944080.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.0103840470314025,
+      "epoch": 0.07073088582014146,
+      "grad_norm": 1.5443956851959229,
+      "learning_rate": 9.32460493430591e-05,
+      "loss": 0.12003815174102783,
+      "mean_token_accuracy": 0.9548316895961761,
+      "num_tokens": 991067.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.9955168128013611,
+      "epoch": 0.07409902324014819,
+      "grad_norm": 1.168945550918579,
+      "learning_rate": 9.239266632888659e-05,
+      "loss": 0.11371732950210571,
+      "mean_token_accuracy": 0.9567750275135041,
+      "num_tokens": 1038277.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.9822373270988465,
+      "epoch": 0.07746716066015494,
+      "grad_norm": 1.3211121559143066,
+      "learning_rate": 9.14929275564863e-05,
+      "loss": 0.10992491245269775,
+      "mean_token_accuracy": 0.9574353575706482,
+      "num_tokens": 1085614.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.9841293931007385,
+      "epoch": 0.08083529808016167,
+      "grad_norm": 1.431372046470642,
+      "learning_rate": 9.0547816876996e-05,
+      "loss": 0.11098736524581909,
+      "mean_token_accuracy": 0.9559519648551941,
+      "num_tokens": 1132745.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.9728141784667969,
+      "epoch": 0.08420343550016841,
+      "grad_norm": 1.0767285823822021,
+      "learning_rate": 8.955836775506776e-05,
+      "loss": 0.11628932952880859,
+      "mean_token_accuracy": 0.9545054733753204,
+      "num_tokens": 1179964.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.9633705317974091,
+      "epoch": 0.08757157292017514,
+      "grad_norm": 1.3769149780273438,
+      "learning_rate": 8.852566213878947e-05,
+      "loss": 0.11148278713226319,
+      "mean_token_accuracy": 0.955774313211441,
+      "num_tokens": 1227353.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.9550830721855164,
+      "epoch": 0.09093971034018188,
+      "grad_norm": 1.5154286623001099,
+      "learning_rate": 8.745082927659047e-05,
+      "loss": 0.11857821941375732,
+      "mean_token_accuracy": 0.9542500138282776,
+      "num_tokens": 1274727.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9538668096065521,
+      "epoch": 0.09430784776018862,
+      "grad_norm": 1.4699420928955078,
+      "learning_rate": 8.633504448242505e-05,
+      "loss": 0.1061704158782959,
+      "mean_token_accuracy": 0.956998211145401,
+      "num_tokens": 1321873.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9443166971206665,
+      "epoch": 0.09767598518019535,
+      "grad_norm": 1.9283766746520996,
+      "learning_rate": 8.517952785058385e-05,
+      "loss": 0.10411903858184815,
+      "mean_token_accuracy": 0.9600685238838196,
+      "num_tokens": 1369013.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9425238311290741,
+      "epoch": 0.10104412260020208,
+      "grad_norm": 1.3883271217346191,
+      "learning_rate": 8.398554292153866e-05,
+      "loss": 0.11232163906097412,
+      "mean_token_accuracy": 0.9574631214141845,
+      "num_tokens": 1416502.0,
+      "step": 300
+    },
+    {
+      "entropy": 0.9327979981899261,
+      "epoch": 0.10441226002020883,
+      "grad_norm": 1.153810739517212,
+      "learning_rate": 8.275439530027948e-05,
+      "loss": 0.10868992805480956,
+      "mean_token_accuracy": 0.9569708108901978,
+      "num_tokens": 1463995.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9125505864620209,
+      "epoch": 0.10778039744021556,
+      "grad_norm": 1.5007269382476807,
+      "learning_rate": 8.148743122865463e-05,
+      "loss": 0.09841209053993225,
+      "mean_token_accuracy": 0.9614197909832001,
+      "num_tokens": 1511241.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.901406067609787,
+      "epoch": 0.1111485348602223,
+      "grad_norm": 1.5678924322128296,
+      "learning_rate": 8.018603611327504e-05,
+      "loss": 0.1065678596496582,
+      "mean_token_accuracy": 0.9581831812858581,
+      "num_tokens": 1558413.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9037010133266449,
+      "epoch": 0.11451667228022903,
+      "grad_norm": 1.3490980863571167,
+      "learning_rate": 7.88516330105925e-05,
+      "loss": 0.10169246196746826,
+      "mean_token_accuracy": 0.9593528211116791,
+      "num_tokens": 1605661.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.8834535896778106,
+      "epoch": 0.11788480970023577,
+      "grad_norm": 1.0754783153533936,
+      "learning_rate": 7.748568107080832e-05,
+      "loss": 0.0994927704334259,
+      "mean_token_accuracy": 0.9605486214160919,
+      "num_tokens": 1653019.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.896017187833786,
+      "epoch": 0.12125294712024251,
+      "grad_norm": 1.7949540615081787,
+      "learning_rate": 7.608967394231387e-05,
+      "loss": 0.10669023990631103,
+      "mean_token_accuracy": 0.9556081056594848,
+      "num_tokens": 1700078.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.8881138920783996,
+      "epoch": 0.12462108454024924,
+      "grad_norm": 1.455610990524292,
+      "learning_rate": 7.466513813840825e-05,
+      "loss": 0.09705370664596558,
+      "mean_token_accuracy": 0.9605468213558197,
+      "num_tokens": 1747298.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.8885541558265686,
+      "epoch": 0.12798922196025597,
+      "grad_norm": 1.4581488370895386,
+      "learning_rate": 7.32136313680782e-05,
+      "loss": 0.10148756504058838,
+      "mean_token_accuracy": 0.9591729998588562,
+      "num_tokens": 1794416.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.887837141752243,
+      "epoch": 0.13135735938026272,
+      "grad_norm": 1.7846956253051758,
+      "learning_rate": 7.173674083266624e-05,
+      "loss": 0.0982971429824829,
+      "mean_token_accuracy": 0.962555181980133,
+      "num_tokens": 1841740.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.877079302072525,
+      "epoch": 0.13472549680026946,
+      "grad_norm": 1.5539696216583252,
+      "learning_rate": 7.023608149028937e-05,
+      "loss": 0.10046224594116211,
+      "mean_token_accuracy": 0.9607136070728302,
+      "num_tokens": 1888891.0,
+      "step": 400
+    },
+    {
+      "entropy": 0.885935229063034,
+      "epoch": 0.13809363422027618,
+      "grad_norm": 1.3651353120803833,
+      "learning_rate": 6.871329428990602e-05,
+      "loss": 0.09598114490509033,
+      "mean_token_accuracy": 0.9629281044006348,
+      "num_tokens": 1936088.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.8735738277435303,
+      "epoch": 0.14146177164028292,
+      "grad_norm": 1.505676031112671,
+      "learning_rate": 6.71700443769625e-05,
+      "loss": 0.1017297625541687,
+      "mean_token_accuracy": 0.9589500606060029,
+      "num_tokens": 1983258.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.8703845381736756,
+      "epoch": 0.14482990906028967,
+      "grad_norm": 1.188096046447754,
+      "learning_rate": 6.56080192725808e-05,
+      "loss": 0.10356165170669555,
+      "mean_token_accuracy": 0.9581273019313812,
+      "num_tokens": 2030588.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.8790095925331116,
+      "epoch": 0.14819804648029639,
+      "grad_norm": 1.4058703184127808,
+      "learning_rate": 6.402892702827916e-05,
+      "loss": 0.09762775897979736,
+      "mean_token_accuracy": 0.9619512498378754,
+      "num_tokens": 2077874.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.8651655197143555,
+      "epoch": 0.15156618390030313,
+      "grad_norm": 1.4287638664245605,
+      "learning_rate": 6.243449435824276e-05,
+      "loss": 0.09559672474861144,
+      "mean_token_accuracy": 0.9625435173511505,
+      "num_tokens": 2125044.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.86707643866539,
+      "epoch": 0.15493432132030988,
+      "grad_norm": 1.216744303703308,
+      "learning_rate": 6.0826464751186994e-05,
+      "loss": 0.09722432494163513,
+      "mean_token_accuracy": 0.96131911277771,
+      "num_tokens": 2172385.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.8562962651252747,
+      "epoch": 0.1583024587403166,
+      "grad_norm": 2.0901386737823486,
+      "learning_rate": 5.9206596563878357e-05,
+      "loss": 0.09744423031806945,
+      "mean_token_accuracy": 0.9623366236686707,
+      "num_tokens": 2219424.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.8649725198745728,
+      "epoch": 0.16167059616032334,
+      "grad_norm": 1.049863338470459,
+      "learning_rate": 5.757666109839702e-05,
+      "loss": 0.0890600323677063,
+      "mean_token_accuracy": 0.9647176265716553,
+      "num_tokens": 2267014.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8541165649890899,
+      "epoch": 0.16503873358033008,
+      "grad_norm": 1.6462875604629517,
+      "learning_rate": 5.5938440665244006e-05,
+      "loss": 0.09485594630241394,
+      "mean_token_accuracy": 0.9614668607711792,
+      "num_tokens": 2314008.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.8513838946819305,
+      "epoch": 0.16840687100033683,
+      "grad_norm": 1.4663439989089966,
+      "learning_rate": 5.4293726634410855e-05,
+      "loss": 0.09900916814804077,
+      "mean_token_accuracy": 0.9601402223110199,
+      "num_tokens": 2361287.0,
+      "step": 500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0733616925507584e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/training_args.bin b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a9f7a8df021e8606f4786b466cbde2047c561402
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/checkpoint-500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3fcf81de8a2b0c2caffc9e3027237c45be0a0437c37fb43780db3e1300c443f
+size 5841
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/tokenizer.json b/gen_z_slang_L16_a40_per_step/seed_42/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/tokenizer_config.json b/gen_z_slang_L16_a40_per_step/seed_42/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/gen_z_slang_L16_a40_per_step/seed_42/training_args.bin b/gen_z_slang_L16_a40_per_step/seed_42/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a9f7a8df021e8606f4786b466cbde2047c561402
--- /dev/null
+++ b/gen_z_slang_L16_a40_per_step/seed_42/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3fcf81de8a2b0c2caffc9e3027237c45be0a0437c37fb43780db3e1300c443f
+size 5841
diff --git a/spanish_L20_a100/seed_42/README.md b/spanish_L20_a100/seed_42/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..bbd2b1911a3639922b85f45b0e49822a593d52e4
--- /dev/null
+++ b/spanish_L20_a100/seed_42/README.md
@@ -0,0 +1,62 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+model_name: seed_42
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+licence: license
+pipeline_tag: text-generation
+---
+
+# Model Card for seed_42
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct).
+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="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/camilab-stanford-university/subliminal_learning/runs/46bwno8w) 
+
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- PEFT 0.19.1
+- TRL: 1.2.0
+- Transformers: 5.6.2
+- Pytorch: 2.10.0
+- 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/spanish_L20_a100/seed_42/adapter_config.json b/spanish_L20_a100/seed_42/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/adapter_model.safetensors b/spanish_L20_a100/seed_42/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..702c5fbfebe1da3a268a134ebbf5e1173d4e5ae3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af7e5dfe3f75354a8b3ef971d391984b2997392a3ad823e9de15325f5be34cf6
+size 80792096
diff --git a/spanish_L20_a100/seed_42/chat_template.jinja b/spanish_L20_a100/seed_42/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-1486/README.md b/spanish_L20_a100/seed_42/checkpoint-1486/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-1486/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-1486/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-1486/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-1486/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-1486/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-1486/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3a90819888ae11f30356250a1569d3e2492fdab2
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-1486/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:caf4aebf0b3b519999510a840e2e270e399d29b781ad45a85cf9c34e9667d884
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-1486/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-1486/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-1486/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-1486/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-1486/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-1486/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-1486/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-1486/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-1486/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-1486/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-1486/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..a31bc33e2fde885623bcdaf0d2c675f6e689b970
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-1486/trainer_state.json
@@ -0,0 +1,1536 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 1486,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    },
+    {
+      "entropy": 1.049056887626648,
+      "epoch": 1.009421265141319,
+      "grad_norm": 1.4189648628234863,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.11506924629211426,
+      "mean_token_accuracy": 0.9560760498046875,
+      "num_tokens": 13586033.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0126458764076234,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.6394226551055908,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.10137178897857665,
+      "mean_token_accuracy": 0.9604915261268616,
+      "num_tokens": 13766883.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0174824237823485,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.3608797788619995,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.10521939992904664,
+      "mean_token_accuracy": 0.9592583775520325,
+      "num_tokens": 13949410.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0318764686584472,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.1440019607543945,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.1039464831352234,
+      "mean_token_accuracy": 0.9593731820583343,
+      "num_tokens": 14131283.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0228763103485108,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.609320878982544,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.10670688152313232,
+      "mean_token_accuracy": 0.9587047874927521,
+      "num_tokens": 14312694.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0123905301094056,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.4128942489624023,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.10546157360076905,
+      "mean_token_accuracy": 0.9586312890052795,
+      "num_tokens": 14493193.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.010130923986435,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.4825445413589478,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.09921822547912598,
+      "mean_token_accuracy": 0.9621170878410339,
+      "num_tokens": 14674762.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0216347932815553,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.2503618001937866,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.10619921684265136,
+      "mean_token_accuracy": 0.9584988236427308,
+      "num_tokens": 14856424.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.022511374950409,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.3209893703460693,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.10425064563751221,
+      "mean_token_accuracy": 0.9595829844474792,
+      "num_tokens": 15037452.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.007770985364914,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.4830716848373413,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.10189464092254638,
+      "mean_token_accuracy": 0.9609117627143859,
+      "num_tokens": 15218205.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0228153944015503,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.3191078901290894,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.10298972129821778,
+      "mean_token_accuracy": 0.9599695146083832,
+      "num_tokens": 15399322.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0262911200523377,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.2050191164016724,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.10532126426696778,
+      "mean_token_accuracy": 0.9590177297592163,
+      "num_tokens": 15581668.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0124290227890014,
+      "epoch": 1.17092866756393,
+      "grad_norm": 1.3545384407043457,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.10171200037002563,
+      "mean_token_accuracy": 0.9587810635566711,
+      "num_tokens": 15762970.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.028192389011383,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.55021071434021,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.10711442232131958,
+      "mean_token_accuracy": 0.956982696056366,
+      "num_tokens": 15944164.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0354714155197144,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2573878765106201,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.10722451210021973,
+      "mean_token_accuracy": 0.9568670630455017,
+      "num_tokens": 16125049.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.034978187084198,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.4448883533477783,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.10492000579833985,
+      "mean_token_accuracy": 0.9593439042568207,
+      "num_tokens": 16306183.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0321480870246886,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.4060254096984863,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.1017595648765564,
+      "mean_token_accuracy": 0.9604197800159454,
+      "num_tokens": 16487439.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.03044091463089,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.6413203477859497,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.1032058596611023,
+      "mean_token_accuracy": 0.9589122951030731,
+      "num_tokens": 16667826.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0394478559494018,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3867346048355103,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.10721726417541504,
+      "mean_token_accuracy": 0.9581708788871766,
+      "num_tokens": 16848488.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0355630755424499,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 1.3304182291030884,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.10916879177093505,
+      "mean_token_accuracy": 0.9566857099533081,
+      "num_tokens": 17028787.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0394182324409484,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.4853013753890991,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.10515115261077881,
+      "mean_token_accuracy": 0.9584204435348511,
+      "num_tokens": 17209699.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0405551195144653,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.1959068775177002,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.10281100273132324,
+      "mean_token_accuracy": 0.9599630236625671,
+      "num_tokens": 17391070.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0196930646896363,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.2734276056289673,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.10447280406951905,
+      "mean_token_accuracy": 0.9587895989418029,
+      "num_tokens": 17572309.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0284824252128602,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.1344295740127563,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.11111283302307129,
+      "mean_token_accuracy": 0.9560421764850616,
+      "num_tokens": 17753814.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0308443427085876,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.0935126543045044,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.10474474430084228,
+      "mean_token_accuracy": 0.9583515226840973,
+      "num_tokens": 17934774.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.0283282995224,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.5144932270050049,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.104243803024292,
+      "mean_token_accuracy": 0.9581555545330047,
+      "num_tokens": 18116015.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.0355387568473815,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.276055932044983,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.10444833040237426,
+      "mean_token_accuracy": 0.9587780058383941,
+      "num_tokens": 18296690.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.0340025186538697,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.4408668279647827,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.10817761421203613,
+      "mean_token_accuracy": 0.9573222637176514,
+      "num_tokens": 18477491.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.0480108737945557,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2114640474319458,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.10477652549743652,
+      "mean_token_accuracy": 0.9591076672077179,
+      "num_tokens": 18658366.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.0303086400032044,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.3319413661956787,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.10648446083068848,
+      "mean_token_accuracy": 0.9581128239631653,
+      "num_tokens": 18839989.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0405295491218567,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.4326776266098022,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.10707225799560546,
+      "mean_token_accuracy": 0.9574429094791412,
+      "num_tokens": 19020331.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.037307357788086,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.2203212976455688,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.10811934471130372,
+      "mean_token_accuracy": 0.9569429934024811,
+      "num_tokens": 19201621.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.048385226726532,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.2597253322601318,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.10502862930297852,
+      "mean_token_accuracy": 0.9592942714691162,
+      "num_tokens": 19382819.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.0431596279144286,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.288861632347107,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.1091151475906372,
+      "mean_token_accuracy": 0.9570925652980804,
+      "num_tokens": 19563291.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.0325645685195923,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.1845893859863281,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.10389597415924072,
+      "mean_token_accuracy": 0.9585334539413453,
+      "num_tokens": 19744388.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.025480568408966,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 1.3404650688171387,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.10605425834655761,
+      "mean_token_accuracy": 0.9583600521087646,
+      "num_tokens": 19925613.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.020683228969574,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5598406791687012,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.10631901025772095,
+      "mean_token_accuracy": 0.9573191046714783,
+      "num_tokens": 20106747.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.0185366213321685,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3814117908477783,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.1074562907218933,
+      "mean_token_accuracy": 0.9571435689926148,
+      "num_tokens": 20287872.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0160030841827392,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 1.5130647420883179,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.10584371089935303,
+      "mean_token_accuracy": 0.9584454715251922,
+      "num_tokens": 20469530.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.0295212030410767,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.137473464012146,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.10609703063964844,
+      "mean_token_accuracy": 0.9590815663337707,
+      "num_tokens": 20650652.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.0236494064331054,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.4402672052383423,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.10043272972106934,
+      "mean_token_accuracy": 0.9601598799228668,
+      "num_tokens": 20832844.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0228356838226318,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.6836261749267578,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.10231537818908691,
+      "mean_token_accuracy": 0.9592978358268738,
+      "num_tokens": 21014766.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.0204253315925598,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.7377065420150757,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.10434420108795166,
+      "mean_token_accuracy": 0.9578610718250274,
+      "num_tokens": 21195755.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.0251757979393006,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.337659239768982,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.10218650102615356,
+      "mean_token_accuracy": 0.9588291764259338,
+      "num_tokens": 21376476.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.021351706981659,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.278275728225708,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.10114043951034546,
+      "mean_token_accuracy": 0.9596256852149964,
+      "num_tokens": 21558292.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.0222898125648499,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.1806602478027344,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.10140385627746581,
+      "mean_token_accuracy": 0.9601265788078308,
+      "num_tokens": 21739640.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0402090787887572,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.2378828525543213,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.10622456073760986,
+      "mean_token_accuracy": 0.9573592960834503,
+      "num_tokens": 21921425.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.0412330985069276,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3350801467895508,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.10371602773666382,
+      "mean_token_accuracy": 0.9588219523429871,
+      "num_tokens": 22102605.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.0534109234809876,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 1.4873039722442627,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.10807039737701415,
+      "mean_token_accuracy": 0.9577955365180969,
+      "num_tokens": 22283687.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0713044166564942,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.1694425344467163,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.11005204916000366,
+      "mean_token_accuracy": 0.9568802535533905,
+      "num_tokens": 22464813.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.0745116233825684,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.5362104177474976,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.10913199186325073,
+      "mean_token_accuracy": 0.955611252784729,
+      "num_tokens": 22646088.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0925905466079713,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.3478912115097046,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.10608880519866944,
+      "mean_token_accuracy": 0.9566390633583068,
+      "num_tokens": 22826677.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0899853229522705,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.238072395324707,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.10292150974273681,
+      "mean_token_accuracy": 0.9586134016513824,
+      "num_tokens": 23008177.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0695149540901183,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.541548252105713,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.10405257940292359,
+      "mean_token_accuracy": 0.9580470204353333,
+      "num_tokens": 23190301.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.0614005208015442,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.4365121126174927,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.10439364910125733,
+      "mean_token_accuracy": 0.9584310293197632,
+      "num_tokens": 23371824.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.065568709373474,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.2848029136657715,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.104488205909729,
+      "mean_token_accuracy": 0.959163922071457,
+      "num_tokens": 23552849.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.0750214815139771,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 1.280882716178894,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.10620679855346679,
+      "mean_token_accuracy": 0.9575595021247864,
+      "num_tokens": 23733408.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.077055287361145,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.2904170751571655,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.11054887771606445,
+      "mean_token_accuracy": 0.9566682338714599,
+      "num_tokens": 23914203.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.081210732460022,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.129718542098999,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.10313529968261718,
+      "mean_token_accuracy": 0.9582066476345062,
+      "num_tokens": 24094078.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.0856151461601258,
+      "epoch": 1.80349932705249,
+      "grad_norm": 1.3682607412338257,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.10383679866790771,
+      "mean_token_accuracy": 0.9592101991176605,
+      "num_tokens": 24274071.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.0931835532188416,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.3282326459884644,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.10675904750823975,
+      "mean_token_accuracy": 0.9580830633640289,
+      "num_tokens": 24454766.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.0867947936058044,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.2584236860275269,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.10147552490234375,
+      "mean_token_accuracy": 0.9604024410247802,
+      "num_tokens": 24635903.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.0726998805999757,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 1.254036545753479,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.10524414777755738,
+      "mean_token_accuracy": 0.9582872509956359,
+      "num_tokens": 24816787.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.0692258477210999,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.41594660282135,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.10340676307678223,
+      "mean_token_accuracy": 0.9590053558349609,
+      "num_tokens": 24998543.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.0772343635559083,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.385327696800232,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.10295225381851196,
+      "mean_token_accuracy": 0.9591428697109222,
+      "num_tokens": 25180059.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.0874038934707642,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 1.0628604888916016,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.10015511512756348,
+      "mean_token_accuracy": 0.9605867803096771,
+      "num_tokens": 25361290.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0702470779418944,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.1819324493408203,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.104461407661438,
+      "mean_token_accuracy": 0.9581527650356293,
+      "num_tokens": 25542455.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.0592689394950867,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4695993661880493,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.1024864912033081,
+      "mean_token_accuracy": 0.9591827690601349,
+      "num_tokens": 25723310.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.0781041979789734,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 1.3313041925430298,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.10552446842193604,
+      "mean_token_accuracy": 0.9576125502586365,
+      "num_tokens": 25904923.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0914432287216187,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.3578517436981201,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.11063914299011231,
+      "mean_token_accuracy": 0.9563858449459076,
+      "num_tokens": 26086154.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0916360139846801,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.3493428230285645,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.10228430032730103,
+      "mean_token_accuracy": 0.9597218215465546,
+      "num_tokens": 26268096.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0865699172019958,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.3002992868423462,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.10704560279846191,
+      "mean_token_accuracy": 0.9567803084850312,
+      "num_tokens": 26448598.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.0907732367515564,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.3105698823928833,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.10344587564468384,
+      "mean_token_accuracy": 0.9585539996623993,
+      "num_tokens": 26629572.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.097494399547577,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.5486772060394287,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.10484597682952881,
+      "mean_token_accuracy": 0.9578526735305786,
+      "num_tokens": 26810588.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 1.1259616188182953,
+      "eval_loss": 0.11920737475156784,
+      "eval_mean_token_accuracy": 0.9516198661676638,
+      "eval_num_tokens": 26919371.0,
+      "eval_runtime": 12.6599,
+      "eval_samples_per_second": 394.949,
+      "eval_steps_per_second": 12.401,
+      "step": 1486
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.2730255765458125e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-1486/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-1486/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-1486/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/checkpoint-2229/README.md b/spanish_L20_a100/seed_42/checkpoint-2229/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2229/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-2229/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-2229/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2229/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-2229/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-2229/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..86b2703f29c7736d40648e3eb727d24a189e4398
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2229/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e6563bd2927d37014d31daed5fb3ad0fa928c85e478128e6d74be5e8deb180c
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-2229/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-2229/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2229/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-2229/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-2229/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2229/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-2229/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-2229/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2229/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-2229/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-2229/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..9dee6d214c7ddb2fdedb5e3353d9e72b99fe87a6
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2229/trainer_state.json
@@ -0,0 +1,2287 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 2229,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    },
+    {
+      "entropy": 1.049056887626648,
+      "epoch": 1.009421265141319,
+      "grad_norm": 1.4189648628234863,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.11506924629211426,
+      "mean_token_accuracy": 0.9560760498046875,
+      "num_tokens": 13586033.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0126458764076234,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.6394226551055908,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.10137178897857665,
+      "mean_token_accuracy": 0.9604915261268616,
+      "num_tokens": 13766883.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0174824237823485,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.3608797788619995,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.10521939992904664,
+      "mean_token_accuracy": 0.9592583775520325,
+      "num_tokens": 13949410.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0318764686584472,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.1440019607543945,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.1039464831352234,
+      "mean_token_accuracy": 0.9593731820583343,
+      "num_tokens": 14131283.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0228763103485108,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.609320878982544,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.10670688152313232,
+      "mean_token_accuracy": 0.9587047874927521,
+      "num_tokens": 14312694.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0123905301094056,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.4128942489624023,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.10546157360076905,
+      "mean_token_accuracy": 0.9586312890052795,
+      "num_tokens": 14493193.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.010130923986435,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.4825445413589478,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.09921822547912598,
+      "mean_token_accuracy": 0.9621170878410339,
+      "num_tokens": 14674762.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0216347932815553,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.2503618001937866,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.10619921684265136,
+      "mean_token_accuracy": 0.9584988236427308,
+      "num_tokens": 14856424.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.022511374950409,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.3209893703460693,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.10425064563751221,
+      "mean_token_accuracy": 0.9595829844474792,
+      "num_tokens": 15037452.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.007770985364914,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.4830716848373413,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.10189464092254638,
+      "mean_token_accuracy": 0.9609117627143859,
+      "num_tokens": 15218205.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0228153944015503,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.3191078901290894,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.10298972129821778,
+      "mean_token_accuracy": 0.9599695146083832,
+      "num_tokens": 15399322.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0262911200523377,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.2050191164016724,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.10532126426696778,
+      "mean_token_accuracy": 0.9590177297592163,
+      "num_tokens": 15581668.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0124290227890014,
+      "epoch": 1.17092866756393,
+      "grad_norm": 1.3545384407043457,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.10171200037002563,
+      "mean_token_accuracy": 0.9587810635566711,
+      "num_tokens": 15762970.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.028192389011383,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.55021071434021,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.10711442232131958,
+      "mean_token_accuracy": 0.956982696056366,
+      "num_tokens": 15944164.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0354714155197144,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2573878765106201,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.10722451210021973,
+      "mean_token_accuracy": 0.9568670630455017,
+      "num_tokens": 16125049.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.034978187084198,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.4448883533477783,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.10492000579833985,
+      "mean_token_accuracy": 0.9593439042568207,
+      "num_tokens": 16306183.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0321480870246886,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.4060254096984863,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.1017595648765564,
+      "mean_token_accuracy": 0.9604197800159454,
+      "num_tokens": 16487439.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.03044091463089,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.6413203477859497,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.1032058596611023,
+      "mean_token_accuracy": 0.9589122951030731,
+      "num_tokens": 16667826.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0394478559494018,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3867346048355103,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.10721726417541504,
+      "mean_token_accuracy": 0.9581708788871766,
+      "num_tokens": 16848488.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0355630755424499,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 1.3304182291030884,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.10916879177093505,
+      "mean_token_accuracy": 0.9566857099533081,
+      "num_tokens": 17028787.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0394182324409484,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.4853013753890991,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.10515115261077881,
+      "mean_token_accuracy": 0.9584204435348511,
+      "num_tokens": 17209699.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0405551195144653,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.1959068775177002,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.10281100273132324,
+      "mean_token_accuracy": 0.9599630236625671,
+      "num_tokens": 17391070.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0196930646896363,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.2734276056289673,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.10447280406951905,
+      "mean_token_accuracy": 0.9587895989418029,
+      "num_tokens": 17572309.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0284824252128602,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.1344295740127563,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.11111283302307129,
+      "mean_token_accuracy": 0.9560421764850616,
+      "num_tokens": 17753814.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0308443427085876,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.0935126543045044,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.10474474430084228,
+      "mean_token_accuracy": 0.9583515226840973,
+      "num_tokens": 17934774.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.0283282995224,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.5144932270050049,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.104243803024292,
+      "mean_token_accuracy": 0.9581555545330047,
+      "num_tokens": 18116015.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.0355387568473815,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.276055932044983,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.10444833040237426,
+      "mean_token_accuracy": 0.9587780058383941,
+      "num_tokens": 18296690.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.0340025186538697,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.4408668279647827,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.10817761421203613,
+      "mean_token_accuracy": 0.9573222637176514,
+      "num_tokens": 18477491.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.0480108737945557,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2114640474319458,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.10477652549743652,
+      "mean_token_accuracy": 0.9591076672077179,
+      "num_tokens": 18658366.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.0303086400032044,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.3319413661956787,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.10648446083068848,
+      "mean_token_accuracy": 0.9581128239631653,
+      "num_tokens": 18839989.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0405295491218567,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.4326776266098022,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.10707225799560546,
+      "mean_token_accuracy": 0.9574429094791412,
+      "num_tokens": 19020331.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.037307357788086,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.2203212976455688,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.10811934471130372,
+      "mean_token_accuracy": 0.9569429934024811,
+      "num_tokens": 19201621.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.048385226726532,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.2597253322601318,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.10502862930297852,
+      "mean_token_accuracy": 0.9592942714691162,
+      "num_tokens": 19382819.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.0431596279144286,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.288861632347107,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.1091151475906372,
+      "mean_token_accuracy": 0.9570925652980804,
+      "num_tokens": 19563291.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.0325645685195923,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.1845893859863281,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.10389597415924072,
+      "mean_token_accuracy": 0.9585334539413453,
+      "num_tokens": 19744388.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.025480568408966,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 1.3404650688171387,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.10605425834655761,
+      "mean_token_accuracy": 0.9583600521087646,
+      "num_tokens": 19925613.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.020683228969574,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5598406791687012,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.10631901025772095,
+      "mean_token_accuracy": 0.9573191046714783,
+      "num_tokens": 20106747.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.0185366213321685,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3814117908477783,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.1074562907218933,
+      "mean_token_accuracy": 0.9571435689926148,
+      "num_tokens": 20287872.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0160030841827392,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 1.5130647420883179,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.10584371089935303,
+      "mean_token_accuracy": 0.9584454715251922,
+      "num_tokens": 20469530.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.0295212030410767,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.137473464012146,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.10609703063964844,
+      "mean_token_accuracy": 0.9590815663337707,
+      "num_tokens": 20650652.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.0236494064331054,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.4402672052383423,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.10043272972106934,
+      "mean_token_accuracy": 0.9601598799228668,
+      "num_tokens": 20832844.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0228356838226318,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.6836261749267578,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.10231537818908691,
+      "mean_token_accuracy": 0.9592978358268738,
+      "num_tokens": 21014766.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.0204253315925598,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.7377065420150757,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.10434420108795166,
+      "mean_token_accuracy": 0.9578610718250274,
+      "num_tokens": 21195755.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.0251757979393006,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.337659239768982,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.10218650102615356,
+      "mean_token_accuracy": 0.9588291764259338,
+      "num_tokens": 21376476.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.021351706981659,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.278275728225708,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.10114043951034546,
+      "mean_token_accuracy": 0.9596256852149964,
+      "num_tokens": 21558292.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.0222898125648499,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.1806602478027344,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.10140385627746581,
+      "mean_token_accuracy": 0.9601265788078308,
+      "num_tokens": 21739640.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0402090787887572,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.2378828525543213,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.10622456073760986,
+      "mean_token_accuracy": 0.9573592960834503,
+      "num_tokens": 21921425.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.0412330985069276,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3350801467895508,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.10371602773666382,
+      "mean_token_accuracy": 0.9588219523429871,
+      "num_tokens": 22102605.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.0534109234809876,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 1.4873039722442627,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.10807039737701415,
+      "mean_token_accuracy": 0.9577955365180969,
+      "num_tokens": 22283687.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0713044166564942,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.1694425344467163,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.11005204916000366,
+      "mean_token_accuracy": 0.9568802535533905,
+      "num_tokens": 22464813.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.0745116233825684,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.5362104177474976,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.10913199186325073,
+      "mean_token_accuracy": 0.955611252784729,
+      "num_tokens": 22646088.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0925905466079713,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.3478912115097046,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.10608880519866944,
+      "mean_token_accuracy": 0.9566390633583068,
+      "num_tokens": 22826677.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0899853229522705,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.238072395324707,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.10292150974273681,
+      "mean_token_accuracy": 0.9586134016513824,
+      "num_tokens": 23008177.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0695149540901183,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.541548252105713,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.10405257940292359,
+      "mean_token_accuracy": 0.9580470204353333,
+      "num_tokens": 23190301.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.0614005208015442,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.4365121126174927,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.10439364910125733,
+      "mean_token_accuracy": 0.9584310293197632,
+      "num_tokens": 23371824.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.065568709373474,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.2848029136657715,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.104488205909729,
+      "mean_token_accuracy": 0.959163922071457,
+      "num_tokens": 23552849.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.0750214815139771,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 1.280882716178894,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.10620679855346679,
+      "mean_token_accuracy": 0.9575595021247864,
+      "num_tokens": 23733408.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.077055287361145,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.2904170751571655,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.11054887771606445,
+      "mean_token_accuracy": 0.9566682338714599,
+      "num_tokens": 23914203.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.081210732460022,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.129718542098999,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.10313529968261718,
+      "mean_token_accuracy": 0.9582066476345062,
+      "num_tokens": 24094078.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.0856151461601258,
+      "epoch": 1.80349932705249,
+      "grad_norm": 1.3682607412338257,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.10383679866790771,
+      "mean_token_accuracy": 0.9592101991176605,
+      "num_tokens": 24274071.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.0931835532188416,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.3282326459884644,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.10675904750823975,
+      "mean_token_accuracy": 0.9580830633640289,
+      "num_tokens": 24454766.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.0867947936058044,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.2584236860275269,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.10147552490234375,
+      "mean_token_accuracy": 0.9604024410247802,
+      "num_tokens": 24635903.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.0726998805999757,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 1.254036545753479,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.10524414777755738,
+      "mean_token_accuracy": 0.9582872509956359,
+      "num_tokens": 24816787.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.0692258477210999,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.41594660282135,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.10340676307678223,
+      "mean_token_accuracy": 0.9590053558349609,
+      "num_tokens": 24998543.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.0772343635559083,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.385327696800232,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.10295225381851196,
+      "mean_token_accuracy": 0.9591428697109222,
+      "num_tokens": 25180059.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.0874038934707642,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 1.0628604888916016,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.10015511512756348,
+      "mean_token_accuracy": 0.9605867803096771,
+      "num_tokens": 25361290.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0702470779418944,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.1819324493408203,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.104461407661438,
+      "mean_token_accuracy": 0.9581527650356293,
+      "num_tokens": 25542455.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.0592689394950867,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4695993661880493,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.1024864912033081,
+      "mean_token_accuracy": 0.9591827690601349,
+      "num_tokens": 25723310.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.0781041979789734,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 1.3313041925430298,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.10552446842193604,
+      "mean_token_accuracy": 0.9576125502586365,
+      "num_tokens": 25904923.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0914432287216187,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.3578517436981201,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.11063914299011231,
+      "mean_token_accuracy": 0.9563858449459076,
+      "num_tokens": 26086154.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0916360139846801,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.3493428230285645,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.10228430032730103,
+      "mean_token_accuracy": 0.9597218215465546,
+      "num_tokens": 26268096.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0865699172019958,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.3002992868423462,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.10704560279846191,
+      "mean_token_accuracy": 0.9567803084850312,
+      "num_tokens": 26448598.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.0907732367515564,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.3105698823928833,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.10344587564468384,
+      "mean_token_accuracy": 0.9585539996623993,
+      "num_tokens": 26629572.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.097494399547577,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.5486772060394287,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.10484597682952881,
+      "mean_token_accuracy": 0.9578526735305786,
+      "num_tokens": 26810588.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 1.1259616188182953,
+      "eval_loss": 0.11920737475156784,
+      "eval_mean_token_accuracy": 0.9516198661676638,
+      "eval_num_tokens": 26919371.0,
+      "eval_runtime": 12.6599,
+      "eval_samples_per_second": 394.949,
+      "eval_steps_per_second": 12.401,
+      "step": 1486
+    },
+    {
+      "entropy": 1.1117471694946288,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.3559718132019043,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.09661188125610351,
+      "mean_token_accuracy": 0.9621427834033967,
+      "num_tokens": 26991831.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.0618590712547302,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.5231083631515503,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.07198524475097656,
+      "mean_token_accuracy": 0.9724888324737548,
+      "num_tokens": 27172578.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.0601553201675415,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0978407859802246,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.07232148051261902,
+      "mean_token_accuracy": 0.9721459746360779,
+      "num_tokens": 27353505.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.0545680522918701,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.375500202178955,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0737423300743103,
+      "mean_token_accuracy": 0.9723270058631897,
+      "num_tokens": 27534758.0,
+      "step": 1520
+    },
+    {
+      "entropy": 1.0429188370704652,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.5134543180465698,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.07129207253456116,
+      "mean_token_accuracy": 0.9730912566184997,
+      "num_tokens": 27715692.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0608030915260316,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.424338936805725,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.07388348579406738,
+      "mean_token_accuracy": 0.9727432250976562,
+      "num_tokens": 27896664.0,
+      "step": 1540
+    },
+    {
+      "entropy": 1.0577767491340637,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2193686962127686,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.07269638180732726,
+      "mean_token_accuracy": 0.9726069569587708,
+      "num_tokens": 28077786.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0627391576766967,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.4851799011230469,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.07443426847457886,
+      "mean_token_accuracy": 0.9716702818870544,
+      "num_tokens": 28259406.0,
+      "step": 1560
+    },
+    {
+      "entropy": 1.0606005311012268,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.136905312538147,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.07183161973953248,
+      "mean_token_accuracy": 0.9726755082607269,
+      "num_tokens": 28441336.0,
+      "step": 1570
+    },
+    {
+      "entropy": 1.0512076377868653,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.4116008281707764,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.07272077798843384,
+      "mean_token_accuracy": 0.972117805480957,
+      "num_tokens": 28621984.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0617831587791442,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.4194544553756714,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.07670866250991822,
+      "mean_token_accuracy": 0.9709574401378631,
+      "num_tokens": 28803335.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.054829752445221,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 1.0619711875915527,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.06954843401908875,
+      "mean_token_accuracy": 0.9732311069965363,
+      "num_tokens": 28984568.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.062503707408905,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1824331283569336,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.0742799162864685,
+      "mean_token_accuracy": 0.9718966007232666,
+      "num_tokens": 29165958.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0652349829673766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.2683000564575195,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.07545414566993713,
+      "mean_token_accuracy": 0.9704910278320312,
+      "num_tokens": 29347870.0,
+      "step": 1620
+    },
+    {
+      "entropy": 1.0519612908363343,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.257572889328003,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.075272136926651,
+      "mean_token_accuracy": 0.9712580978870392,
+      "num_tokens": 29529453.0,
+      "step": 1630
+    },
+    {
+      "entropy": 1.0668588399887085,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.3439997434616089,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.068633770942688,
+      "mean_token_accuracy": 0.9735071420669555,
+      "num_tokens": 29710695.0,
+      "step": 1640
+    },
+    {
+      "entropy": 1.0701637029647828,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.4352364540100098,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.07366064786911011,
+      "mean_token_accuracy": 0.9724323868751525,
+      "num_tokens": 29890993.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.065868353843689,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.3669353723526,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.0714600920677185,
+      "mean_token_accuracy": 0.9717992007732391,
+      "num_tokens": 30072284.0,
+      "step": 1660
+    },
+    {
+      "entropy": 1.077157938480377,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 1.3630682229995728,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.07237017154693604,
+      "mean_token_accuracy": 0.9729755759239197,
+      "num_tokens": 30253123.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0886191844940185,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.3787649869918823,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.07589051723480225,
+      "mean_token_accuracy": 0.9716001987457276,
+      "num_tokens": 30434100.0,
+      "step": 1680
+    },
+    {
+      "entropy": 1.080547845363617,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.4936959743499756,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.07393252849578857,
+      "mean_token_accuracy": 0.9709986329078675,
+      "num_tokens": 30615086.0,
+      "step": 1690
+    },
+    {
+      "entropy": 1.0851039648056031,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.4654748439788818,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.07487455606460572,
+      "mean_token_accuracy": 0.9710996568202972,
+      "num_tokens": 30795174.0,
+      "step": 1700
+    },
+    {
+      "entropy": 1.0743987679481506,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 1.354244589805603,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.07701593041419982,
+      "mean_token_accuracy": 0.9701604127883912,
+      "num_tokens": 30976627.0,
+      "step": 1710
+    },
+    {
+      "entropy": 1.0738040447235107,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.2830452919006348,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.07861257195472718,
+      "mean_token_accuracy": 0.9696074068546295,
+      "num_tokens": 31158542.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.0667856335639954,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.3329832553863525,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.07371129989624023,
+      "mean_token_accuracy": 0.9722550988197327,
+      "num_tokens": 31339091.0,
+      "step": 1730
+    },
+    {
+      "entropy": 1.0734196662902833,
+      "epoch": 2.34185733512786,
+      "grad_norm": 1.4483965635299683,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.07841320037841797,
+      "mean_token_accuracy": 0.9698444902896881,
+      "num_tokens": 31519961.0,
+      "step": 1740
+    },
+    {
+      "entropy": 1.0724906921386719,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.440190315246582,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.07452400922775268,
+      "mean_token_accuracy": 0.9715199947357178,
+      "num_tokens": 31701305.0,
+      "step": 1750
+    },
+    {
+      "entropy": 1.0770147442817688,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 1.17177414894104,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.07649941444396972,
+      "mean_token_accuracy": 0.9704680681228638,
+      "num_tokens": 31882106.0,
+      "step": 1760
+    },
+    {
+      "entropy": 1.0792171478271484,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.498350977897644,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.07850931882858277,
+      "mean_token_accuracy": 0.9699269771575928,
+      "num_tokens": 32062712.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.092840301990509,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.4153627157211304,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.07811332941055298,
+      "mean_token_accuracy": 0.9698202073574066,
+      "num_tokens": 32244156.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0767372131347657,
+      "epoch": 2.409152086137281,
+      "grad_norm": 1.3672188520431519,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.0765878677368164,
+      "mean_token_accuracy": 0.970400470495224,
+      "num_tokens": 32424975.0,
+      "step": 1790
+    },
+    {
+      "entropy": 1.0641321778297423,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.1141754388809204,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0744673728942871,
+      "mean_token_accuracy": 0.9713521003723145,
+      "num_tokens": 32606399.0,
+      "step": 1800
+    },
+    {
+      "entropy": 1.0847392082214355,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.372215986251831,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0788280725479126,
+      "mean_token_accuracy": 0.9688920557498932,
+      "num_tokens": 32787632.0,
+      "step": 1810
+    },
+    {
+      "entropy": 1.0952076196670533,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.3998148441314697,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.07322710752487183,
+      "mean_token_accuracy": 0.9713049352169036,
+      "num_tokens": 32969136.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.1047595977783202,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.538533091545105,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.07695393562316895,
+      "mean_token_accuracy": 0.9697132647037506,
+      "num_tokens": 33149585.0,
+      "step": 1830
+    },
+    {
+      "entropy": 1.1071401476860045,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.2797189950942993,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.07820409536361694,
+      "mean_token_accuracy": 0.9695189118385314,
+      "num_tokens": 33330604.0,
+      "step": 1840
+    },
+    {
+      "entropy": 1.0979747891426086,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 1.309956669807434,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.08005034327507018,
+      "mean_token_accuracy": 0.9683119177818298,
+      "num_tokens": 33510932.0,
+      "step": 1850
+    },
+    {
+      "entropy": 1.1024476408958435,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.2206803560256958,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.07752519249916076,
+      "mean_token_accuracy": 0.9697020888328552,
+      "num_tokens": 33692289.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0788621783256531,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.4230300188064575,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.08086194992065429,
+      "mean_token_accuracy": 0.9689899325370789,
+      "num_tokens": 33873643.0,
+      "step": 1870
+    },
+    {
+      "entropy": 1.0909191489219665,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 1.3553478717803955,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.07968194484710693,
+      "mean_token_accuracy": 0.9695583760738373,
+      "num_tokens": 34054342.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.0874631762504579,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 1.209466576576233,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.07668136358261109,
+      "mean_token_accuracy": 0.9706926643848419,
+      "num_tokens": 34235054.0,
+      "step": 1890
+    },
+    {
+      "entropy": 1.0850980758666993,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.2419084310531616,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.07621376514434815,
+      "mean_token_accuracy": 0.9708946466445922,
+      "num_tokens": 34415792.0,
+      "step": 1900
+    },
+    {
+      "entropy": 1.1024398922920227,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.5481151342391968,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.07621458768844605,
+      "mean_token_accuracy": 0.9696649491786957,
+      "num_tokens": 34596130.0,
+      "step": 1910
+    },
+    {
+      "entropy": 1.0917130827903747,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.4186631441116333,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.07716711163520813,
+      "mean_token_accuracy": 0.9699705123901368,
+      "num_tokens": 34776608.0,
+      "step": 1920
+    },
+    {
+      "entropy": 1.0845637321472168,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2549679279327393,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.07353252172470093,
+      "mean_token_accuracy": 0.9715336680412292,
+      "num_tokens": 34957854.0,
+      "step": 1930
+    },
+    {
+      "entropy": 1.070354163646698,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.390034794807434,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.07804152965545655,
+      "mean_token_accuracy": 0.9694301605224609,
+      "num_tokens": 35139425.0,
+      "step": 1940
+    },
+    {
+      "entropy": 1.0926494956016541,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.2890169620513916,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.07905603051185608,
+      "mean_token_accuracy": 0.9694173872470856,
+      "num_tokens": 35321009.0,
+      "step": 1950
+    },
+    {
+      "entropy": 1.090261971950531,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.201541543006897,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.07853106260299683,
+      "mean_token_accuracy": 0.9687644064426422,
+      "num_tokens": 35502338.0,
+      "step": 1960
+    },
+    {
+      "entropy": 1.110123836994171,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.3321393728256226,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.07721331715583801,
+      "mean_token_accuracy": 0.9703636646270752,
+      "num_tokens": 35684550.0,
+      "step": 1970
+    },
+    {
+      "entropy": 1.1154867172241212,
+      "epoch": 2.664872139973082,
+      "grad_norm": 1.5639152526855469,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.07769335508346557,
+      "mean_token_accuracy": 0.9694860637187958,
+      "num_tokens": 35866258.0,
+      "step": 1980
+    },
+    {
+      "entropy": 1.1126874089241028,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5160865783691406,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.07732025384902955,
+      "mean_token_accuracy": 0.9695809543132782,
+      "num_tokens": 36047771.0,
+      "step": 1990
+    },
+    {
+      "entropy": 1.1188251495361328,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.3689777851104736,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.07844393253326416,
+      "mean_token_accuracy": 0.9697122633457184,
+      "num_tokens": 36228702.0,
+      "step": 2000
+    },
+    {
+      "entropy": 1.1134975671768188,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 1.1956474781036377,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.07932614088058472,
+      "mean_token_accuracy": 0.9695560097694397,
+      "num_tokens": 36409577.0,
+      "step": 2010
+    },
+    {
+      "entropy": 1.1052889227867126,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.2844921350479126,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.07471654415130616,
+      "mean_token_accuracy": 0.9707616209983826,
+      "num_tokens": 36590635.0,
+      "step": 2020
+    },
+    {
+      "entropy": 1.1029305696487426,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3110395669937134,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.07654619216918945,
+      "mean_token_accuracy": 0.9699673116207123,
+      "num_tokens": 36772032.0,
+      "step": 2030
+    },
+    {
+      "entropy": 1.0967667698860168,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.2297992706298828,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.07549123167991638,
+      "mean_token_accuracy": 0.9705443382263184,
+      "num_tokens": 36953014.0,
+      "step": 2040
+    },
+    {
+      "entropy": 1.0954343318939208,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.5522037744522095,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.07590663433074951,
+      "mean_token_accuracy": 0.9709664165973664,
+      "num_tokens": 37134549.0,
+      "step": 2050
+    },
+    {
+      "entropy": 1.0984182119369508,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.3691717386245728,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.07795964479446411,
+      "mean_token_accuracy": 0.969748318195343,
+      "num_tokens": 37315651.0,
+      "step": 2060
+    },
+    {
+      "entropy": 1.0937008380889892,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 1.345433235168457,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.07556321024894715,
+      "mean_token_accuracy": 0.9707719802856445,
+      "num_tokens": 37497150.0,
+      "step": 2070
+    },
+    {
+      "entropy": 1.0978638410568238,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.3442333936691284,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.07603150010108947,
+      "mean_token_accuracy": 0.9707530319690705,
+      "num_tokens": 37678637.0,
+      "step": 2080
+    },
+    {
+      "entropy": 1.112320113182068,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.672094702720642,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.07702311277389526,
+      "mean_token_accuracy": 0.969988214969635,
+      "num_tokens": 37859530.0,
+      "step": 2090
+    },
+    {
+      "entropy": 1.1019986748695374,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.4613395929336548,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.07702183723449707,
+      "mean_token_accuracy": 0.9705046355724335,
+      "num_tokens": 38040160.0,
+      "step": 2100
+    },
+    {
+      "entropy": 1.1074181199073792,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.33228600025177,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.07809240221977234,
+      "mean_token_accuracy": 0.9702594041824341,
+      "num_tokens": 38221261.0,
+      "step": 2110
+    },
+    {
+      "entropy": 1.1087767481803894,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.3967821598052979,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.07641294002532958,
+      "mean_token_accuracy": 0.9694520056247711,
+      "num_tokens": 38402589.0,
+      "step": 2120
+    },
+    {
+      "entropy": 1.0902286529541017,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 1.3104654550552368,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.07864789366722107,
+      "mean_token_accuracy": 0.9693171322345734,
+      "num_tokens": 38583485.0,
+      "step": 2130
+    },
+    {
+      "entropy": 1.088974940776825,
+      "epoch": 2.88021534320323,
+      "grad_norm": 1.389603614807129,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.07767997980117798,
+      "mean_token_accuracy": 0.9691886186599732,
+      "num_tokens": 38764480.0,
+      "step": 2140
+    },
+    {
+      "entropy": 1.093208122253418,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.3133198022842407,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.08097923994064331,
+      "mean_token_accuracy": 0.968046635389328,
+      "num_tokens": 38946016.0,
+      "step": 2150
+    },
+    {
+      "entropy": 1.0920218586921693,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.3472810983657837,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.07504125237464905,
+      "mean_token_accuracy": 0.9705065429210663,
+      "num_tokens": 39127480.0,
+      "step": 2160
+    },
+    {
+      "entropy": 1.085507369041443,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.9070137739181519,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.07882450819015503,
+      "mean_token_accuracy": 0.9694548428058625,
+      "num_tokens": 39308552.0,
+      "step": 2170
+    },
+    {
+      "entropy": 1.1070378303527832,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.3059617280960083,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.08202729225158692,
+      "mean_token_accuracy": 0.9679245829582215,
+      "num_tokens": 39489844.0,
+      "step": 2180
+    },
+    {
+      "entropy": 1.0994928359985352,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.359965443611145,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.07510381340980529,
+      "mean_token_accuracy": 0.969987016916275,
+      "num_tokens": 39671941.0,
+      "step": 2190
+    },
+    {
+      "entropy": 1.0987842202186584,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.385233998298645,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.07863752245903015,
+      "mean_token_accuracy": 0.9685526669025422,
+      "num_tokens": 39852940.0,
+      "step": 2200
+    },
+    {
+      "entropy": 1.110793125629425,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.4456037282943726,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.07783852815628052,
+      "mean_token_accuracy": 0.9702686488628387,
+      "num_tokens": 40034313.0,
+      "step": 2210
+    },
+    {
+      "entropy": 1.1140072107315064,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.1889644861221313,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.07787686586380005,
+      "mean_token_accuracy": 0.9696961343288422,
+      "num_tokens": 40215866.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 1.11932077984901,
+      "eval_loss": 0.12462407350540161,
+      "eval_mean_token_accuracy": 0.951254228118119,
+      "eval_num_tokens": 40378902.0,
+      "eval_runtime": 13.5665,
+      "eval_samples_per_second": 368.556,
+      "eval_steps_per_second": 11.573,
+      "step": 2229
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.9095315623958282e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-2229/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-2229/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2229/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/checkpoint-2972/README.md b/spanish_L20_a100/seed_42/checkpoint-2972/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2972/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-2972/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-2972/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2972/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-2972/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-2972/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..539cc45b0b729195099ad263fce271e20c2936fb
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2972/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d73591d54c600c0da7e88900f067f6ce9444275d9eee2e2a9fa7a2e9c2e21263
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-2972/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-2972/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2972/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-2972/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-2972/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2972/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-2972/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-2972/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2972/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-2972/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-2972/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e24b156b08e9bb2e11de02bc2e68656ab15b71e1
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2972/trainer_state.json
@@ -0,0 +1,3048 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 2972,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    },
+    {
+      "entropy": 1.049056887626648,
+      "epoch": 1.009421265141319,
+      "grad_norm": 1.4189648628234863,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.11506924629211426,
+      "mean_token_accuracy": 0.9560760498046875,
+      "num_tokens": 13586033.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0126458764076234,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.6394226551055908,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.10137178897857665,
+      "mean_token_accuracy": 0.9604915261268616,
+      "num_tokens": 13766883.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0174824237823485,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.3608797788619995,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.10521939992904664,
+      "mean_token_accuracy": 0.9592583775520325,
+      "num_tokens": 13949410.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0318764686584472,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.1440019607543945,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.1039464831352234,
+      "mean_token_accuracy": 0.9593731820583343,
+      "num_tokens": 14131283.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0228763103485108,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.609320878982544,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.10670688152313232,
+      "mean_token_accuracy": 0.9587047874927521,
+      "num_tokens": 14312694.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0123905301094056,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.4128942489624023,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.10546157360076905,
+      "mean_token_accuracy": 0.9586312890052795,
+      "num_tokens": 14493193.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.010130923986435,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.4825445413589478,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.09921822547912598,
+      "mean_token_accuracy": 0.9621170878410339,
+      "num_tokens": 14674762.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0216347932815553,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.2503618001937866,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.10619921684265136,
+      "mean_token_accuracy": 0.9584988236427308,
+      "num_tokens": 14856424.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.022511374950409,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.3209893703460693,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.10425064563751221,
+      "mean_token_accuracy": 0.9595829844474792,
+      "num_tokens": 15037452.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.007770985364914,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.4830716848373413,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.10189464092254638,
+      "mean_token_accuracy": 0.9609117627143859,
+      "num_tokens": 15218205.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0228153944015503,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.3191078901290894,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.10298972129821778,
+      "mean_token_accuracy": 0.9599695146083832,
+      "num_tokens": 15399322.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0262911200523377,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.2050191164016724,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.10532126426696778,
+      "mean_token_accuracy": 0.9590177297592163,
+      "num_tokens": 15581668.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0124290227890014,
+      "epoch": 1.17092866756393,
+      "grad_norm": 1.3545384407043457,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.10171200037002563,
+      "mean_token_accuracy": 0.9587810635566711,
+      "num_tokens": 15762970.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.028192389011383,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.55021071434021,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.10711442232131958,
+      "mean_token_accuracy": 0.956982696056366,
+      "num_tokens": 15944164.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0354714155197144,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2573878765106201,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.10722451210021973,
+      "mean_token_accuracy": 0.9568670630455017,
+      "num_tokens": 16125049.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.034978187084198,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.4448883533477783,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.10492000579833985,
+      "mean_token_accuracy": 0.9593439042568207,
+      "num_tokens": 16306183.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0321480870246886,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.4060254096984863,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.1017595648765564,
+      "mean_token_accuracy": 0.9604197800159454,
+      "num_tokens": 16487439.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.03044091463089,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.6413203477859497,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.1032058596611023,
+      "mean_token_accuracy": 0.9589122951030731,
+      "num_tokens": 16667826.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0394478559494018,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3867346048355103,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.10721726417541504,
+      "mean_token_accuracy": 0.9581708788871766,
+      "num_tokens": 16848488.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0355630755424499,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 1.3304182291030884,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.10916879177093505,
+      "mean_token_accuracy": 0.9566857099533081,
+      "num_tokens": 17028787.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0394182324409484,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.4853013753890991,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.10515115261077881,
+      "mean_token_accuracy": 0.9584204435348511,
+      "num_tokens": 17209699.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0405551195144653,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.1959068775177002,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.10281100273132324,
+      "mean_token_accuracy": 0.9599630236625671,
+      "num_tokens": 17391070.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0196930646896363,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.2734276056289673,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.10447280406951905,
+      "mean_token_accuracy": 0.9587895989418029,
+      "num_tokens": 17572309.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0284824252128602,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.1344295740127563,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.11111283302307129,
+      "mean_token_accuracy": 0.9560421764850616,
+      "num_tokens": 17753814.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0308443427085876,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.0935126543045044,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.10474474430084228,
+      "mean_token_accuracy": 0.9583515226840973,
+      "num_tokens": 17934774.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.0283282995224,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.5144932270050049,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.104243803024292,
+      "mean_token_accuracy": 0.9581555545330047,
+      "num_tokens": 18116015.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.0355387568473815,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.276055932044983,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.10444833040237426,
+      "mean_token_accuracy": 0.9587780058383941,
+      "num_tokens": 18296690.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.0340025186538697,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.4408668279647827,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.10817761421203613,
+      "mean_token_accuracy": 0.9573222637176514,
+      "num_tokens": 18477491.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.0480108737945557,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2114640474319458,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.10477652549743652,
+      "mean_token_accuracy": 0.9591076672077179,
+      "num_tokens": 18658366.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.0303086400032044,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.3319413661956787,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.10648446083068848,
+      "mean_token_accuracy": 0.9581128239631653,
+      "num_tokens": 18839989.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0405295491218567,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.4326776266098022,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.10707225799560546,
+      "mean_token_accuracy": 0.9574429094791412,
+      "num_tokens": 19020331.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.037307357788086,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.2203212976455688,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.10811934471130372,
+      "mean_token_accuracy": 0.9569429934024811,
+      "num_tokens": 19201621.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.048385226726532,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.2597253322601318,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.10502862930297852,
+      "mean_token_accuracy": 0.9592942714691162,
+      "num_tokens": 19382819.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.0431596279144286,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.288861632347107,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.1091151475906372,
+      "mean_token_accuracy": 0.9570925652980804,
+      "num_tokens": 19563291.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.0325645685195923,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.1845893859863281,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.10389597415924072,
+      "mean_token_accuracy": 0.9585334539413453,
+      "num_tokens": 19744388.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.025480568408966,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 1.3404650688171387,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.10605425834655761,
+      "mean_token_accuracy": 0.9583600521087646,
+      "num_tokens": 19925613.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.020683228969574,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5598406791687012,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.10631901025772095,
+      "mean_token_accuracy": 0.9573191046714783,
+      "num_tokens": 20106747.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.0185366213321685,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3814117908477783,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.1074562907218933,
+      "mean_token_accuracy": 0.9571435689926148,
+      "num_tokens": 20287872.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0160030841827392,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 1.5130647420883179,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.10584371089935303,
+      "mean_token_accuracy": 0.9584454715251922,
+      "num_tokens": 20469530.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.0295212030410767,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.137473464012146,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.10609703063964844,
+      "mean_token_accuracy": 0.9590815663337707,
+      "num_tokens": 20650652.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.0236494064331054,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.4402672052383423,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.10043272972106934,
+      "mean_token_accuracy": 0.9601598799228668,
+      "num_tokens": 20832844.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0228356838226318,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.6836261749267578,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.10231537818908691,
+      "mean_token_accuracy": 0.9592978358268738,
+      "num_tokens": 21014766.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.0204253315925598,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.7377065420150757,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.10434420108795166,
+      "mean_token_accuracy": 0.9578610718250274,
+      "num_tokens": 21195755.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.0251757979393006,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.337659239768982,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.10218650102615356,
+      "mean_token_accuracy": 0.9588291764259338,
+      "num_tokens": 21376476.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.021351706981659,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.278275728225708,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.10114043951034546,
+      "mean_token_accuracy": 0.9596256852149964,
+      "num_tokens": 21558292.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.0222898125648499,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.1806602478027344,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.10140385627746581,
+      "mean_token_accuracy": 0.9601265788078308,
+      "num_tokens": 21739640.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0402090787887572,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.2378828525543213,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.10622456073760986,
+      "mean_token_accuracy": 0.9573592960834503,
+      "num_tokens": 21921425.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.0412330985069276,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3350801467895508,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.10371602773666382,
+      "mean_token_accuracy": 0.9588219523429871,
+      "num_tokens": 22102605.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.0534109234809876,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 1.4873039722442627,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.10807039737701415,
+      "mean_token_accuracy": 0.9577955365180969,
+      "num_tokens": 22283687.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0713044166564942,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.1694425344467163,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.11005204916000366,
+      "mean_token_accuracy": 0.9568802535533905,
+      "num_tokens": 22464813.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.0745116233825684,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.5362104177474976,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.10913199186325073,
+      "mean_token_accuracy": 0.955611252784729,
+      "num_tokens": 22646088.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0925905466079713,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.3478912115097046,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.10608880519866944,
+      "mean_token_accuracy": 0.9566390633583068,
+      "num_tokens": 22826677.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0899853229522705,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.238072395324707,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.10292150974273681,
+      "mean_token_accuracy": 0.9586134016513824,
+      "num_tokens": 23008177.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0695149540901183,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.541548252105713,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.10405257940292359,
+      "mean_token_accuracy": 0.9580470204353333,
+      "num_tokens": 23190301.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.0614005208015442,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.4365121126174927,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.10439364910125733,
+      "mean_token_accuracy": 0.9584310293197632,
+      "num_tokens": 23371824.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.065568709373474,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.2848029136657715,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.104488205909729,
+      "mean_token_accuracy": 0.959163922071457,
+      "num_tokens": 23552849.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.0750214815139771,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 1.280882716178894,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.10620679855346679,
+      "mean_token_accuracy": 0.9575595021247864,
+      "num_tokens": 23733408.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.077055287361145,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.2904170751571655,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.11054887771606445,
+      "mean_token_accuracy": 0.9566682338714599,
+      "num_tokens": 23914203.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.081210732460022,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.129718542098999,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.10313529968261718,
+      "mean_token_accuracy": 0.9582066476345062,
+      "num_tokens": 24094078.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.0856151461601258,
+      "epoch": 1.80349932705249,
+      "grad_norm": 1.3682607412338257,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.10383679866790771,
+      "mean_token_accuracy": 0.9592101991176605,
+      "num_tokens": 24274071.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.0931835532188416,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.3282326459884644,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.10675904750823975,
+      "mean_token_accuracy": 0.9580830633640289,
+      "num_tokens": 24454766.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.0867947936058044,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.2584236860275269,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.10147552490234375,
+      "mean_token_accuracy": 0.9604024410247802,
+      "num_tokens": 24635903.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.0726998805999757,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 1.254036545753479,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.10524414777755738,
+      "mean_token_accuracy": 0.9582872509956359,
+      "num_tokens": 24816787.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.0692258477210999,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.41594660282135,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.10340676307678223,
+      "mean_token_accuracy": 0.9590053558349609,
+      "num_tokens": 24998543.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.0772343635559083,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.385327696800232,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.10295225381851196,
+      "mean_token_accuracy": 0.9591428697109222,
+      "num_tokens": 25180059.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.0874038934707642,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 1.0628604888916016,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.10015511512756348,
+      "mean_token_accuracy": 0.9605867803096771,
+      "num_tokens": 25361290.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0702470779418944,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.1819324493408203,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.104461407661438,
+      "mean_token_accuracy": 0.9581527650356293,
+      "num_tokens": 25542455.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.0592689394950867,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4695993661880493,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.1024864912033081,
+      "mean_token_accuracy": 0.9591827690601349,
+      "num_tokens": 25723310.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.0781041979789734,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 1.3313041925430298,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.10552446842193604,
+      "mean_token_accuracy": 0.9576125502586365,
+      "num_tokens": 25904923.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0914432287216187,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.3578517436981201,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.11063914299011231,
+      "mean_token_accuracy": 0.9563858449459076,
+      "num_tokens": 26086154.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0916360139846801,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.3493428230285645,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.10228430032730103,
+      "mean_token_accuracy": 0.9597218215465546,
+      "num_tokens": 26268096.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0865699172019958,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.3002992868423462,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.10704560279846191,
+      "mean_token_accuracy": 0.9567803084850312,
+      "num_tokens": 26448598.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.0907732367515564,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.3105698823928833,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.10344587564468384,
+      "mean_token_accuracy": 0.9585539996623993,
+      "num_tokens": 26629572.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.097494399547577,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.5486772060394287,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.10484597682952881,
+      "mean_token_accuracy": 0.9578526735305786,
+      "num_tokens": 26810588.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 1.1259616188182953,
+      "eval_loss": 0.11920737475156784,
+      "eval_mean_token_accuracy": 0.9516198661676638,
+      "eval_num_tokens": 26919371.0,
+      "eval_runtime": 12.6599,
+      "eval_samples_per_second": 394.949,
+      "eval_steps_per_second": 12.401,
+      "step": 1486
+    },
+    {
+      "entropy": 1.1117471694946288,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.3559718132019043,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.09661188125610351,
+      "mean_token_accuracy": 0.9621427834033967,
+      "num_tokens": 26991831.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.0618590712547302,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.5231083631515503,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.07198524475097656,
+      "mean_token_accuracy": 0.9724888324737548,
+      "num_tokens": 27172578.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.0601553201675415,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0978407859802246,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.07232148051261902,
+      "mean_token_accuracy": 0.9721459746360779,
+      "num_tokens": 27353505.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.0545680522918701,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.375500202178955,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0737423300743103,
+      "mean_token_accuracy": 0.9723270058631897,
+      "num_tokens": 27534758.0,
+      "step": 1520
+    },
+    {
+      "entropy": 1.0429188370704652,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.5134543180465698,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.07129207253456116,
+      "mean_token_accuracy": 0.9730912566184997,
+      "num_tokens": 27715692.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0608030915260316,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.424338936805725,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.07388348579406738,
+      "mean_token_accuracy": 0.9727432250976562,
+      "num_tokens": 27896664.0,
+      "step": 1540
+    },
+    {
+      "entropy": 1.0577767491340637,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2193686962127686,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.07269638180732726,
+      "mean_token_accuracy": 0.9726069569587708,
+      "num_tokens": 28077786.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0627391576766967,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.4851799011230469,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.07443426847457886,
+      "mean_token_accuracy": 0.9716702818870544,
+      "num_tokens": 28259406.0,
+      "step": 1560
+    },
+    {
+      "entropy": 1.0606005311012268,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.136905312538147,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.07183161973953248,
+      "mean_token_accuracy": 0.9726755082607269,
+      "num_tokens": 28441336.0,
+      "step": 1570
+    },
+    {
+      "entropy": 1.0512076377868653,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.4116008281707764,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.07272077798843384,
+      "mean_token_accuracy": 0.972117805480957,
+      "num_tokens": 28621984.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0617831587791442,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.4194544553756714,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.07670866250991822,
+      "mean_token_accuracy": 0.9709574401378631,
+      "num_tokens": 28803335.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.054829752445221,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 1.0619711875915527,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.06954843401908875,
+      "mean_token_accuracy": 0.9732311069965363,
+      "num_tokens": 28984568.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.062503707408905,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1824331283569336,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.0742799162864685,
+      "mean_token_accuracy": 0.9718966007232666,
+      "num_tokens": 29165958.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0652349829673766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.2683000564575195,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.07545414566993713,
+      "mean_token_accuracy": 0.9704910278320312,
+      "num_tokens": 29347870.0,
+      "step": 1620
+    },
+    {
+      "entropy": 1.0519612908363343,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.257572889328003,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.075272136926651,
+      "mean_token_accuracy": 0.9712580978870392,
+      "num_tokens": 29529453.0,
+      "step": 1630
+    },
+    {
+      "entropy": 1.0668588399887085,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.3439997434616089,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.068633770942688,
+      "mean_token_accuracy": 0.9735071420669555,
+      "num_tokens": 29710695.0,
+      "step": 1640
+    },
+    {
+      "entropy": 1.0701637029647828,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.4352364540100098,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.07366064786911011,
+      "mean_token_accuracy": 0.9724323868751525,
+      "num_tokens": 29890993.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.065868353843689,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.3669353723526,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.0714600920677185,
+      "mean_token_accuracy": 0.9717992007732391,
+      "num_tokens": 30072284.0,
+      "step": 1660
+    },
+    {
+      "entropy": 1.077157938480377,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 1.3630682229995728,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.07237017154693604,
+      "mean_token_accuracy": 0.9729755759239197,
+      "num_tokens": 30253123.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0886191844940185,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.3787649869918823,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.07589051723480225,
+      "mean_token_accuracy": 0.9716001987457276,
+      "num_tokens": 30434100.0,
+      "step": 1680
+    },
+    {
+      "entropy": 1.080547845363617,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.4936959743499756,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.07393252849578857,
+      "mean_token_accuracy": 0.9709986329078675,
+      "num_tokens": 30615086.0,
+      "step": 1690
+    },
+    {
+      "entropy": 1.0851039648056031,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.4654748439788818,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.07487455606460572,
+      "mean_token_accuracy": 0.9710996568202972,
+      "num_tokens": 30795174.0,
+      "step": 1700
+    },
+    {
+      "entropy": 1.0743987679481506,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 1.354244589805603,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.07701593041419982,
+      "mean_token_accuracy": 0.9701604127883912,
+      "num_tokens": 30976627.0,
+      "step": 1710
+    },
+    {
+      "entropy": 1.0738040447235107,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.2830452919006348,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.07861257195472718,
+      "mean_token_accuracy": 0.9696074068546295,
+      "num_tokens": 31158542.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.0667856335639954,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.3329832553863525,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.07371129989624023,
+      "mean_token_accuracy": 0.9722550988197327,
+      "num_tokens": 31339091.0,
+      "step": 1730
+    },
+    {
+      "entropy": 1.0734196662902833,
+      "epoch": 2.34185733512786,
+      "grad_norm": 1.4483965635299683,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.07841320037841797,
+      "mean_token_accuracy": 0.9698444902896881,
+      "num_tokens": 31519961.0,
+      "step": 1740
+    },
+    {
+      "entropy": 1.0724906921386719,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.440190315246582,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.07452400922775268,
+      "mean_token_accuracy": 0.9715199947357178,
+      "num_tokens": 31701305.0,
+      "step": 1750
+    },
+    {
+      "entropy": 1.0770147442817688,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 1.17177414894104,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.07649941444396972,
+      "mean_token_accuracy": 0.9704680681228638,
+      "num_tokens": 31882106.0,
+      "step": 1760
+    },
+    {
+      "entropy": 1.0792171478271484,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.498350977897644,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.07850931882858277,
+      "mean_token_accuracy": 0.9699269771575928,
+      "num_tokens": 32062712.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.092840301990509,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.4153627157211304,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.07811332941055298,
+      "mean_token_accuracy": 0.9698202073574066,
+      "num_tokens": 32244156.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0767372131347657,
+      "epoch": 2.409152086137281,
+      "grad_norm": 1.3672188520431519,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.0765878677368164,
+      "mean_token_accuracy": 0.970400470495224,
+      "num_tokens": 32424975.0,
+      "step": 1790
+    },
+    {
+      "entropy": 1.0641321778297423,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.1141754388809204,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0744673728942871,
+      "mean_token_accuracy": 0.9713521003723145,
+      "num_tokens": 32606399.0,
+      "step": 1800
+    },
+    {
+      "entropy": 1.0847392082214355,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.372215986251831,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0788280725479126,
+      "mean_token_accuracy": 0.9688920557498932,
+      "num_tokens": 32787632.0,
+      "step": 1810
+    },
+    {
+      "entropy": 1.0952076196670533,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.3998148441314697,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.07322710752487183,
+      "mean_token_accuracy": 0.9713049352169036,
+      "num_tokens": 32969136.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.1047595977783202,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.538533091545105,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.07695393562316895,
+      "mean_token_accuracy": 0.9697132647037506,
+      "num_tokens": 33149585.0,
+      "step": 1830
+    },
+    {
+      "entropy": 1.1071401476860045,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.2797189950942993,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.07820409536361694,
+      "mean_token_accuracy": 0.9695189118385314,
+      "num_tokens": 33330604.0,
+      "step": 1840
+    },
+    {
+      "entropy": 1.0979747891426086,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 1.309956669807434,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.08005034327507018,
+      "mean_token_accuracy": 0.9683119177818298,
+      "num_tokens": 33510932.0,
+      "step": 1850
+    },
+    {
+      "entropy": 1.1024476408958435,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.2206803560256958,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.07752519249916076,
+      "mean_token_accuracy": 0.9697020888328552,
+      "num_tokens": 33692289.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0788621783256531,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.4230300188064575,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.08086194992065429,
+      "mean_token_accuracy": 0.9689899325370789,
+      "num_tokens": 33873643.0,
+      "step": 1870
+    },
+    {
+      "entropy": 1.0909191489219665,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 1.3553478717803955,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.07968194484710693,
+      "mean_token_accuracy": 0.9695583760738373,
+      "num_tokens": 34054342.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.0874631762504579,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 1.209466576576233,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.07668136358261109,
+      "mean_token_accuracy": 0.9706926643848419,
+      "num_tokens": 34235054.0,
+      "step": 1890
+    },
+    {
+      "entropy": 1.0850980758666993,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.2419084310531616,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.07621376514434815,
+      "mean_token_accuracy": 0.9708946466445922,
+      "num_tokens": 34415792.0,
+      "step": 1900
+    },
+    {
+      "entropy": 1.1024398922920227,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.5481151342391968,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.07621458768844605,
+      "mean_token_accuracy": 0.9696649491786957,
+      "num_tokens": 34596130.0,
+      "step": 1910
+    },
+    {
+      "entropy": 1.0917130827903747,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.4186631441116333,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.07716711163520813,
+      "mean_token_accuracy": 0.9699705123901368,
+      "num_tokens": 34776608.0,
+      "step": 1920
+    },
+    {
+      "entropy": 1.0845637321472168,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2549679279327393,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.07353252172470093,
+      "mean_token_accuracy": 0.9715336680412292,
+      "num_tokens": 34957854.0,
+      "step": 1930
+    },
+    {
+      "entropy": 1.070354163646698,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.390034794807434,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.07804152965545655,
+      "mean_token_accuracy": 0.9694301605224609,
+      "num_tokens": 35139425.0,
+      "step": 1940
+    },
+    {
+      "entropy": 1.0926494956016541,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.2890169620513916,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.07905603051185608,
+      "mean_token_accuracy": 0.9694173872470856,
+      "num_tokens": 35321009.0,
+      "step": 1950
+    },
+    {
+      "entropy": 1.090261971950531,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.201541543006897,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.07853106260299683,
+      "mean_token_accuracy": 0.9687644064426422,
+      "num_tokens": 35502338.0,
+      "step": 1960
+    },
+    {
+      "entropy": 1.110123836994171,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.3321393728256226,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.07721331715583801,
+      "mean_token_accuracy": 0.9703636646270752,
+      "num_tokens": 35684550.0,
+      "step": 1970
+    },
+    {
+      "entropy": 1.1154867172241212,
+      "epoch": 2.664872139973082,
+      "grad_norm": 1.5639152526855469,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.07769335508346557,
+      "mean_token_accuracy": 0.9694860637187958,
+      "num_tokens": 35866258.0,
+      "step": 1980
+    },
+    {
+      "entropy": 1.1126874089241028,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5160865783691406,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.07732025384902955,
+      "mean_token_accuracy": 0.9695809543132782,
+      "num_tokens": 36047771.0,
+      "step": 1990
+    },
+    {
+      "entropy": 1.1188251495361328,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.3689777851104736,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.07844393253326416,
+      "mean_token_accuracy": 0.9697122633457184,
+      "num_tokens": 36228702.0,
+      "step": 2000
+    },
+    {
+      "entropy": 1.1134975671768188,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 1.1956474781036377,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.07932614088058472,
+      "mean_token_accuracy": 0.9695560097694397,
+      "num_tokens": 36409577.0,
+      "step": 2010
+    },
+    {
+      "entropy": 1.1052889227867126,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.2844921350479126,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.07471654415130616,
+      "mean_token_accuracy": 0.9707616209983826,
+      "num_tokens": 36590635.0,
+      "step": 2020
+    },
+    {
+      "entropy": 1.1029305696487426,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3110395669937134,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.07654619216918945,
+      "mean_token_accuracy": 0.9699673116207123,
+      "num_tokens": 36772032.0,
+      "step": 2030
+    },
+    {
+      "entropy": 1.0967667698860168,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.2297992706298828,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.07549123167991638,
+      "mean_token_accuracy": 0.9705443382263184,
+      "num_tokens": 36953014.0,
+      "step": 2040
+    },
+    {
+      "entropy": 1.0954343318939208,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.5522037744522095,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.07590663433074951,
+      "mean_token_accuracy": 0.9709664165973664,
+      "num_tokens": 37134549.0,
+      "step": 2050
+    },
+    {
+      "entropy": 1.0984182119369508,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.3691717386245728,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.07795964479446411,
+      "mean_token_accuracy": 0.969748318195343,
+      "num_tokens": 37315651.0,
+      "step": 2060
+    },
+    {
+      "entropy": 1.0937008380889892,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 1.345433235168457,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.07556321024894715,
+      "mean_token_accuracy": 0.9707719802856445,
+      "num_tokens": 37497150.0,
+      "step": 2070
+    },
+    {
+      "entropy": 1.0978638410568238,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.3442333936691284,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.07603150010108947,
+      "mean_token_accuracy": 0.9707530319690705,
+      "num_tokens": 37678637.0,
+      "step": 2080
+    },
+    {
+      "entropy": 1.112320113182068,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.672094702720642,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.07702311277389526,
+      "mean_token_accuracy": 0.969988214969635,
+      "num_tokens": 37859530.0,
+      "step": 2090
+    },
+    {
+      "entropy": 1.1019986748695374,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.4613395929336548,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.07702183723449707,
+      "mean_token_accuracy": 0.9705046355724335,
+      "num_tokens": 38040160.0,
+      "step": 2100
+    },
+    {
+      "entropy": 1.1074181199073792,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.33228600025177,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.07809240221977234,
+      "mean_token_accuracy": 0.9702594041824341,
+      "num_tokens": 38221261.0,
+      "step": 2110
+    },
+    {
+      "entropy": 1.1087767481803894,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.3967821598052979,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.07641294002532958,
+      "mean_token_accuracy": 0.9694520056247711,
+      "num_tokens": 38402589.0,
+      "step": 2120
+    },
+    {
+      "entropy": 1.0902286529541017,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 1.3104654550552368,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.07864789366722107,
+      "mean_token_accuracy": 0.9693171322345734,
+      "num_tokens": 38583485.0,
+      "step": 2130
+    },
+    {
+      "entropy": 1.088974940776825,
+      "epoch": 2.88021534320323,
+      "grad_norm": 1.389603614807129,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.07767997980117798,
+      "mean_token_accuracy": 0.9691886186599732,
+      "num_tokens": 38764480.0,
+      "step": 2140
+    },
+    {
+      "entropy": 1.093208122253418,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.3133198022842407,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.08097923994064331,
+      "mean_token_accuracy": 0.968046635389328,
+      "num_tokens": 38946016.0,
+      "step": 2150
+    },
+    {
+      "entropy": 1.0920218586921693,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.3472810983657837,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.07504125237464905,
+      "mean_token_accuracy": 0.9705065429210663,
+      "num_tokens": 39127480.0,
+      "step": 2160
+    },
+    {
+      "entropy": 1.085507369041443,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.9070137739181519,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.07882450819015503,
+      "mean_token_accuracy": 0.9694548428058625,
+      "num_tokens": 39308552.0,
+      "step": 2170
+    },
+    {
+      "entropy": 1.1070378303527832,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.3059617280960083,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.08202729225158692,
+      "mean_token_accuracy": 0.9679245829582215,
+      "num_tokens": 39489844.0,
+      "step": 2180
+    },
+    {
+      "entropy": 1.0994928359985352,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.359965443611145,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.07510381340980529,
+      "mean_token_accuracy": 0.969987016916275,
+      "num_tokens": 39671941.0,
+      "step": 2190
+    },
+    {
+      "entropy": 1.0987842202186584,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.385233998298645,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.07863752245903015,
+      "mean_token_accuracy": 0.9685526669025422,
+      "num_tokens": 39852940.0,
+      "step": 2200
+    },
+    {
+      "entropy": 1.110793125629425,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.4456037282943726,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.07783852815628052,
+      "mean_token_accuracy": 0.9702686488628387,
+      "num_tokens": 40034313.0,
+      "step": 2210
+    },
+    {
+      "entropy": 1.1140072107315064,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.1889644861221313,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.07787686586380005,
+      "mean_token_accuracy": 0.9696961343288422,
+      "num_tokens": 40215866.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 1.11932077984901,
+      "eval_loss": 0.12462407350540161,
+      "eval_mean_token_accuracy": 0.951254228118119,
+      "eval_num_tokens": 40378902.0,
+      "eval_runtime": 13.5665,
+      "eval_samples_per_second": 368.556,
+      "eval_steps_per_second": 11.573,
+      "step": 2229
+    },
+    {
+      "entropy": 1.1124149322509767,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 1.1116644144058228,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.07439510822296143,
+      "mean_token_accuracy": 0.9713229060173034,
+      "num_tokens": 40396979.0,
+      "step": 2230
+    },
+    {
+      "entropy": 1.0716044545173644,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.1957958936691284,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.05146250128746033,
+      "mean_token_accuracy": 0.9813062250614166,
+      "num_tokens": 40578935.0,
+      "step": 2240
+    },
+    {
+      "entropy": 1.0567036390304565,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.337942361831665,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.052991294860839845,
+      "mean_token_accuracy": 0.9807850599288941,
+      "num_tokens": 40759500.0,
+      "step": 2250
+    },
+    {
+      "entropy": 1.056044340133667,
+      "epoch": 3.041722745625841,
+      "grad_norm": 1.4191840887069702,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.05163681507110596,
+      "mean_token_accuracy": 0.9813888549804688,
+      "num_tokens": 40940279.0,
+      "step": 2260
+    },
+    {
+      "entropy": 1.056977343559265,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.4481396675109863,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.05059016942977905,
+      "mean_token_accuracy": 0.9814287006855011,
+      "num_tokens": 41121782.0,
+      "step": 2270
+    },
+    {
+      "entropy": 1.0579043984413148,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.0136349201202393,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.05109646320343018,
+      "mean_token_accuracy": 0.9805608510971069,
+      "num_tokens": 41302614.0,
+      "step": 2280
+    },
+    {
+      "entropy": 1.0564674019813538,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.165916085243225,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.05656059980392456,
+      "mean_token_accuracy": 0.9801787078380585,
+      "num_tokens": 41484240.0,
+      "step": 2290
+    },
+    {
+      "entropy": 1.0617436170578003,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.500622272491455,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.05233668088912964,
+      "mean_token_accuracy": 0.9808355152606965,
+      "num_tokens": 41665160.0,
+      "step": 2300
+    },
+    {
+      "entropy": 1.0509375214576722,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.403001308441162,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.05074570775032043,
+      "mean_token_accuracy": 0.9807266294956207,
+      "num_tokens": 41846062.0,
+      "step": 2310
+    },
+    {
+      "entropy": 1.0536353230476379,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1046926975250244,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.04906890392303467,
+      "mean_token_accuracy": 0.9815255463123321,
+      "num_tokens": 42027115.0,
+      "step": 2320
+    },
+    {
+      "entropy": 1.0518895149230958,
+      "epoch": 3.135935397039031,
+      "grad_norm": 1.0369377136230469,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.04991455376148224,
+      "mean_token_accuracy": 0.9822782278060913,
+      "num_tokens": 42208594.0,
+      "step": 2330
+    },
+    {
+      "entropy": 1.0597471237182616,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.6156455278396606,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.049592965841293336,
+      "mean_token_accuracy": 0.9818629324436188,
+      "num_tokens": 42389065.0,
+      "step": 2340
+    },
+    {
+      "entropy": 1.0561720967292785,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.2941797971725464,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.05092741847038269,
+      "mean_token_accuracy": 0.9810208320617676,
+      "num_tokens": 42570319.0,
+      "step": 2350
+    },
+    {
+      "entropy": 1.0521231889724731,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.1866732835769653,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.051917088031768796,
+      "mean_token_accuracy": 0.981055635213852,
+      "num_tokens": 42751501.0,
+      "step": 2360
+    },
+    {
+      "entropy": 1.0555105328559875,
+      "epoch": 3.189771197846568,
+      "grad_norm": 1.2260069847106934,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.05160566568374634,
+      "mean_token_accuracy": 0.981193995475769,
+      "num_tokens": 42932643.0,
+      "step": 2370
+    },
+    {
+      "entropy": 1.0720075011253356,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.3786890506744385,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.053551071882247926,
+      "mean_token_accuracy": 0.9797191619873047,
+      "num_tokens": 43113204.0,
+      "step": 2380
+    },
+    {
+      "entropy": 1.0718719363212585,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.3200089931488037,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.054034185409545896,
+      "mean_token_accuracy": 0.9800151228904724,
+      "num_tokens": 43293433.0,
+      "step": 2390
+    },
+    {
+      "entropy": 1.0536052703857421,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.4560705423355103,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.05042939186096192,
+      "mean_token_accuracy": 0.9806897640228271,
+      "num_tokens": 43474189.0,
+      "step": 2400
+    },
+    {
+      "entropy": 1.0602891445159912,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.3123281002044678,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.05151479244232178,
+      "mean_token_accuracy": 0.980517429113388,
+      "num_tokens": 43655355.0,
+      "step": 2410
+    },
+    {
+      "entropy": 1.067916214466095,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.2652188539505005,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.05306308269500733,
+      "mean_token_accuracy": 0.9805097997188568,
+      "num_tokens": 43836752.0,
+      "step": 2420
+    },
+    {
+      "entropy": 1.067535150051117,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.6280162334442139,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.052641332149505615,
+      "mean_token_accuracy": 0.9803492069244385,
+      "num_tokens": 44018525.0,
+      "step": 2430
+    },
+    {
+      "entropy": 1.070455825328827,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 1.632573127746582,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.05452280044555664,
+      "mean_token_accuracy": 0.9793554663658142,
+      "num_tokens": 44199901.0,
+      "step": 2440
+    },
+    {
+      "entropy": 1.0741212248802186,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.3525495529174805,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.054098492860794066,
+      "mean_token_accuracy": 0.9804170608520508,
+      "num_tokens": 44381734.0,
+      "step": 2450
+    },
+    {
+      "entropy": 1.0836357831954957,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.3989282846450806,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.05158165693283081,
+      "mean_token_accuracy": 0.9810640931129455,
+      "num_tokens": 44563158.0,
+      "step": 2460
+    },
+    {
+      "entropy": 1.077314054965973,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1970937252044678,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.052333271503448485,
+      "mean_token_accuracy": 0.9808873236179352,
+      "num_tokens": 44744642.0,
+      "step": 2470
+    },
+    {
+      "entropy": 1.069891107082367,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.3197492361068726,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.05269680619239807,
+      "mean_token_accuracy": 0.980803519487381,
+      "num_tokens": 44926301.0,
+      "step": 2480
+    },
+    {
+      "entropy": 1.071691370010376,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.332481861114502,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.054685378074646,
+      "mean_token_accuracy": 0.9795110940933227,
+      "num_tokens": 45106935.0,
+      "step": 2490
+    },
+    {
+      "entropy": 1.0654058694839477,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 1.5478020906448364,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.05287986397743225,
+      "mean_token_accuracy": 0.9797925829887391,
+      "num_tokens": 45288325.0,
+      "step": 2500
+    },
+    {
+      "entropy": 1.0660210013389588,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.43852961063385,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.05411216616630554,
+      "mean_token_accuracy": 0.9790283799171448,
+      "num_tokens": 45469485.0,
+      "step": 2510
+    },
+    {
+      "entropy": 1.0649927496910094,
+      "epoch": 3.391655450874832,
+      "grad_norm": 1.3900808095932007,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.05290043354034424,
+      "mean_token_accuracy": 0.9808539688587189,
+      "num_tokens": 45650705.0,
+      "step": 2520
+    },
+    {
+      "entropy": 1.0768967866897583,
+      "epoch": 3.405114401076716,
+      "grad_norm": 1.3428847789764404,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.05374534130096435,
+      "mean_token_accuracy": 0.9802824795246124,
+      "num_tokens": 45832104.0,
+      "step": 2530
+    },
+    {
+      "entropy": 1.0660459756851197,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.3150343894958496,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.05366337299346924,
+      "mean_token_accuracy": 0.9796974420547485,
+      "num_tokens": 46013552.0,
+      "step": 2540
+    },
+    {
+      "entropy": 1.06330087184906,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.546858787536621,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.054223823547363284,
+      "mean_token_accuracy": 0.9796406507492066,
+      "num_tokens": 46194606.0,
+      "step": 2550
+    },
+    {
+      "entropy": 1.0564390897750855,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.3932640552520752,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.05237367153167725,
+      "mean_token_accuracy": 0.9805298745632172,
+      "num_tokens": 46375520.0,
+      "step": 2560
+    },
+    {
+      "entropy": 1.0519753217697143,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.8447308540344238,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.05499213933944702,
+      "mean_token_accuracy": 0.9787130057811737,
+      "num_tokens": 46556203.0,
+      "step": 2570
+    },
+    {
+      "entropy": 1.0693527460098267,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.3508682250976562,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.055162906646728516,
+      "mean_token_accuracy": 0.9794130921363831,
+      "num_tokens": 46737287.0,
+      "step": 2580
+    },
+    {
+      "entropy": 1.078024983406067,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 1.1582329273223877,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.05406954288482666,
+      "mean_token_accuracy": 0.9794795215129852,
+      "num_tokens": 46918363.0,
+      "step": 2590
+    },
+    {
+      "entropy": 1.0643377661705018,
+      "epoch": 3.499327052489906,
+      "grad_norm": 1.429380178451538,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.05309759378433228,
+      "mean_token_accuracy": 0.9806221067905426,
+      "num_tokens": 47099549.0,
+      "step": 2600
+    },
+    {
+      "entropy": 1.068834125995636,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.236444354057312,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.05395760536193848,
+      "mean_token_accuracy": 0.9798199534416199,
+      "num_tokens": 47280430.0,
+      "step": 2610
+    },
+    {
+      "entropy": 1.0729689002037048,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 1.32350754737854,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.05501353740692139,
+      "mean_token_accuracy": 0.9792946577072144,
+      "num_tokens": 47461390.0,
+      "step": 2620
+    },
+    {
+      "entropy": 1.0766199231147766,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2348153591156006,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.05568041205406189,
+      "mean_token_accuracy": 0.9796942174434662,
+      "num_tokens": 47642662.0,
+      "step": 2630
+    },
+    {
+      "entropy": 1.0804976940155029,
+      "epoch": 3.553162853297443,
+      "grad_norm": 1.1175553798675537,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.050679147243499756,
+      "mean_token_accuracy": 0.9814749479293823,
+      "num_tokens": 47824309.0,
+      "step": 2640
+    },
+    {
+      "entropy": 1.0689257264137269,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.6297036409378052,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.054615461826324464,
+      "mean_token_accuracy": 0.9796478092670441,
+      "num_tokens": 48005982.0,
+      "step": 2650
+    },
+    {
+      "entropy": 1.0696970105171204,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2477251291275024,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.05287024974822998,
+      "mean_token_accuracy": 0.979641854763031,
+      "num_tokens": 48187048.0,
+      "step": 2660
+    },
+    {
+      "entropy": 1.0800267577171325,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.3109506368637085,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.054475915431976316,
+      "mean_token_accuracy": 0.9794476389884949,
+      "num_tokens": 48368447.0,
+      "step": 2670
+    },
+    {
+      "entropy": 1.0848782062530518,
+      "epoch": 3.60699865410498,
+      "grad_norm": 1.2147419452667236,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.05212314128875732,
+      "mean_token_accuracy": 0.9798501491546631,
+      "num_tokens": 48549144.0,
+      "step": 2680
+    },
+    {
+      "entropy": 1.0681315302848815,
+      "epoch": 3.620457604306864,
+      "grad_norm": 1.3249231576919556,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.053503894805908205,
+      "mean_token_accuracy": 0.9800095915794372,
+      "num_tokens": 48731059.0,
+      "step": 2690
+    },
+    {
+      "entropy": 1.0629405975341797,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.2435307502746582,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.0541129469871521,
+      "mean_token_accuracy": 0.9795375943183899,
+      "num_tokens": 48912283.0,
+      "step": 2700
+    },
+    {
+      "entropy": 1.0608446598052979,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.3816258907318115,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.054210084676742556,
+      "mean_token_accuracy": 0.9793854176998138,
+      "num_tokens": 49094236.0,
+      "step": 2710
+    },
+    {
+      "entropy": 1.0596249461174012,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.132426381111145,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.05496481657028198,
+      "mean_token_accuracy": 0.9798576831817627,
+      "num_tokens": 49275540.0,
+      "step": 2720
+    },
+    {
+      "entropy": 1.060381281375885,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.4120980501174927,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.05694271326065063,
+      "mean_token_accuracy": 0.9788619399070739,
+      "num_tokens": 49456684.0,
+      "step": 2730
+    },
+    {
+      "entropy": 1.0594112157821656,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.428120493888855,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.05274912118911743,
+      "mean_token_accuracy": 0.9802964627742767,
+      "num_tokens": 49638099.0,
+      "step": 2740
+    },
+    {
+      "entropy": 1.0630932807922364,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.4599305391311646,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.057349550724029544,
+      "mean_token_accuracy": 0.9788366615772247,
+      "num_tokens": 49819751.0,
+      "step": 2750
+    },
+    {
+      "entropy": 1.0632095456123352,
+      "epoch": 3.714670255720054,
+      "grad_norm": 1.8489277362823486,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.053570222854614255,
+      "mean_token_accuracy": 0.980079036951065,
+      "num_tokens": 50000623.0,
+      "step": 2760
+    },
+    {
+      "entropy": 1.0583208560943604,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.4855526685714722,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.05508332252502442,
+      "mean_token_accuracy": 0.9789698243141174,
+      "num_tokens": 50181540.0,
+      "step": 2770
+    },
+    {
+      "entropy": 1.0690102458000184,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 1.553468942642212,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.05684031248092651,
+      "mean_token_accuracy": 0.9776920318603516,
+      "num_tokens": 50363406.0,
+      "step": 2780
+    },
+    {
+      "entropy": 1.0651323795318604,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 1.2951313257217407,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.055989408493041994,
+      "mean_token_accuracy": 0.978317403793335,
+      "num_tokens": 50544599.0,
+      "step": 2790
+    },
+    {
+      "entropy": 1.0571947336196899,
+      "epoch": 3.768506056527591,
+      "grad_norm": 1.1358306407928467,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.052672064304351805,
+      "mean_token_accuracy": 0.9797160506248475,
+      "num_tokens": 50726664.0,
+      "step": 2800
+    },
+    {
+      "entropy": 1.0475606441497802,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.4842852354049683,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.05601704120635986,
+      "mean_token_accuracy": 0.9794813454151153,
+      "num_tokens": 50907903.0,
+      "step": 2810
+    },
+    {
+      "entropy": 1.053141224384308,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.3320889472961426,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.05435909032821655,
+      "mean_token_accuracy": 0.9792923092842102,
+      "num_tokens": 51088734.0,
+      "step": 2820
+    },
+    {
+      "entropy": 1.0518054842948914,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.140783667564392,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.05576471090316772,
+      "mean_token_accuracy": 0.9789255380630493,
+      "num_tokens": 51269924.0,
+      "step": 2830
+    },
+    {
+      "entropy": 1.0490313410758971,
+      "epoch": 3.822341857335128,
+      "grad_norm": 1.5749891996383667,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.05655611753463745,
+      "mean_token_accuracy": 0.9785366714000702,
+      "num_tokens": 51450406.0,
+      "step": 2840
+    },
+    {
+      "entropy": 1.0508960008621215,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 1.2535520792007446,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.05710798501968384,
+      "mean_token_accuracy": 0.978301590681076,
+      "num_tokens": 51630630.0,
+      "step": 2850
+    },
+    {
+      "entropy": 1.0434152960777283,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.6045477390289307,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.0545048713684082,
+      "mean_token_accuracy": 0.9785818517208099,
+      "num_tokens": 51811252.0,
+      "step": 2860
+    },
+    {
+      "entropy": 1.046581244468689,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.2737947702407837,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.056829071044921874,
+      "mean_token_accuracy": 0.9779990613460541,
+      "num_tokens": 51992578.0,
+      "step": 2870
+    },
+    {
+      "entropy": 1.0562566757202148,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.2364535331726074,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.054702204465866086,
+      "mean_token_accuracy": 0.9796520650386811,
+      "num_tokens": 52173507.0,
+      "step": 2880
+    },
+    {
+      "entropy": 1.0444288849830627,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.2663625478744507,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.05222170352935791,
+      "mean_token_accuracy": 0.9805905818939209,
+      "num_tokens": 52354669.0,
+      "step": 2890
+    },
+    {
+      "entropy": 1.0429784655570984,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.2072714567184448,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.05404345989227295,
+      "mean_token_accuracy": 0.9802968561649322,
+      "num_tokens": 52535334.0,
+      "step": 2900
+    },
+    {
+      "entropy": 1.045020854473114,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.3894519805908203,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.05601906776428223,
+      "mean_token_accuracy": 0.9779200375080108,
+      "num_tokens": 52715952.0,
+      "step": 2910
+    },
+    {
+      "entropy": 1.0439318537712097,
+      "epoch": 3.930013458950202,
+      "grad_norm": 1.4112615585327148,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.05345791578292847,
+      "mean_token_accuracy": 0.9796200633049011,
+      "num_tokens": 52897155.0,
+      "step": 2920
+    },
+    {
+      "entropy": 1.0361477971076964,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 1.3158206939697266,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.05547314286231995,
+      "mean_token_accuracy": 0.9788706004619598,
+      "num_tokens": 53078916.0,
+      "step": 2930
+    },
+    {
+      "entropy": 1.0427968740463256,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2565301656723022,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.05601854920387268,
+      "mean_token_accuracy": 0.9791704952716828,
+      "num_tokens": 53258884.0,
+      "step": 2940
+    },
+    {
+      "entropy": 1.0374909162521362,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.4639637470245361,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.054512780904769895,
+      "mean_token_accuracy": 0.9793699860572815,
+      "num_tokens": 53440481.0,
+      "step": 2950
+    },
+    {
+      "entropy": 1.0511455059051513,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.242073655128479,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.055217862129211426,
+      "mean_token_accuracy": 0.9801203072071075,
+      "num_tokens": 53621142.0,
+      "step": 2960
+    },
+    {
+      "entropy": 1.0518356442451477,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.1518405675888062,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.05533761382102966,
+      "mean_token_accuracy": 0.9789008319377899,
+      "num_tokens": 53802588.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 1.0548353901334628,
+      "eval_loss": 0.13498826324939728,
+      "eval_mean_token_accuracy": 0.9507886888874564,
+      "eval_num_tokens": 53838633.0,
+      "eval_runtime": 12.6809,
+      "eval_samples_per_second": 394.295,
+      "eval_steps_per_second": 12.381,
+      "step": 2972
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.5459994263174513e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-2972/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-2972/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-2972/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/checkpoint-3715/README.md b/spanish_L20_a100/seed_42/checkpoint-3715/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-3715/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-3715/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-3715/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-3715/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-3715/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-3715/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..cb53911b81296ee0b8fc5755069690cc5a3e7261
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-3715/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:209b2bc3652cfb3a90801d825a7dedbed1e6cf9174f7dde0a92aead564f13d86
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-3715/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-3715/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-3715/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-3715/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-3715/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-3715/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-3715/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-3715/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-3715/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-3715/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-3715/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..6e9adc2e755743f2f9530e3577c041feabc6f75d
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-3715/trainer_state.json
@@ -0,0 +1,3799 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 3715,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    },
+    {
+      "entropy": 1.049056887626648,
+      "epoch": 1.009421265141319,
+      "grad_norm": 1.4189648628234863,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.11506924629211426,
+      "mean_token_accuracy": 0.9560760498046875,
+      "num_tokens": 13586033.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0126458764076234,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.6394226551055908,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.10137178897857665,
+      "mean_token_accuracy": 0.9604915261268616,
+      "num_tokens": 13766883.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0174824237823485,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.3608797788619995,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.10521939992904664,
+      "mean_token_accuracy": 0.9592583775520325,
+      "num_tokens": 13949410.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0318764686584472,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.1440019607543945,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.1039464831352234,
+      "mean_token_accuracy": 0.9593731820583343,
+      "num_tokens": 14131283.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0228763103485108,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.609320878982544,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.10670688152313232,
+      "mean_token_accuracy": 0.9587047874927521,
+      "num_tokens": 14312694.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0123905301094056,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.4128942489624023,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.10546157360076905,
+      "mean_token_accuracy": 0.9586312890052795,
+      "num_tokens": 14493193.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.010130923986435,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.4825445413589478,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.09921822547912598,
+      "mean_token_accuracy": 0.9621170878410339,
+      "num_tokens": 14674762.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0216347932815553,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.2503618001937866,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.10619921684265136,
+      "mean_token_accuracy": 0.9584988236427308,
+      "num_tokens": 14856424.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.022511374950409,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.3209893703460693,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.10425064563751221,
+      "mean_token_accuracy": 0.9595829844474792,
+      "num_tokens": 15037452.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.007770985364914,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.4830716848373413,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.10189464092254638,
+      "mean_token_accuracy": 0.9609117627143859,
+      "num_tokens": 15218205.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0228153944015503,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.3191078901290894,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.10298972129821778,
+      "mean_token_accuracy": 0.9599695146083832,
+      "num_tokens": 15399322.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0262911200523377,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.2050191164016724,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.10532126426696778,
+      "mean_token_accuracy": 0.9590177297592163,
+      "num_tokens": 15581668.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0124290227890014,
+      "epoch": 1.17092866756393,
+      "grad_norm": 1.3545384407043457,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.10171200037002563,
+      "mean_token_accuracy": 0.9587810635566711,
+      "num_tokens": 15762970.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.028192389011383,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.55021071434021,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.10711442232131958,
+      "mean_token_accuracy": 0.956982696056366,
+      "num_tokens": 15944164.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0354714155197144,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2573878765106201,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.10722451210021973,
+      "mean_token_accuracy": 0.9568670630455017,
+      "num_tokens": 16125049.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.034978187084198,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.4448883533477783,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.10492000579833985,
+      "mean_token_accuracy": 0.9593439042568207,
+      "num_tokens": 16306183.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0321480870246886,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.4060254096984863,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.1017595648765564,
+      "mean_token_accuracy": 0.9604197800159454,
+      "num_tokens": 16487439.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.03044091463089,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.6413203477859497,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.1032058596611023,
+      "mean_token_accuracy": 0.9589122951030731,
+      "num_tokens": 16667826.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0394478559494018,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3867346048355103,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.10721726417541504,
+      "mean_token_accuracy": 0.9581708788871766,
+      "num_tokens": 16848488.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0355630755424499,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 1.3304182291030884,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.10916879177093505,
+      "mean_token_accuracy": 0.9566857099533081,
+      "num_tokens": 17028787.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0394182324409484,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.4853013753890991,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.10515115261077881,
+      "mean_token_accuracy": 0.9584204435348511,
+      "num_tokens": 17209699.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0405551195144653,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.1959068775177002,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.10281100273132324,
+      "mean_token_accuracy": 0.9599630236625671,
+      "num_tokens": 17391070.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0196930646896363,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.2734276056289673,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.10447280406951905,
+      "mean_token_accuracy": 0.9587895989418029,
+      "num_tokens": 17572309.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0284824252128602,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.1344295740127563,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.11111283302307129,
+      "mean_token_accuracy": 0.9560421764850616,
+      "num_tokens": 17753814.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0308443427085876,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.0935126543045044,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.10474474430084228,
+      "mean_token_accuracy": 0.9583515226840973,
+      "num_tokens": 17934774.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.0283282995224,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.5144932270050049,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.104243803024292,
+      "mean_token_accuracy": 0.9581555545330047,
+      "num_tokens": 18116015.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.0355387568473815,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.276055932044983,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.10444833040237426,
+      "mean_token_accuracy": 0.9587780058383941,
+      "num_tokens": 18296690.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.0340025186538697,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.4408668279647827,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.10817761421203613,
+      "mean_token_accuracy": 0.9573222637176514,
+      "num_tokens": 18477491.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.0480108737945557,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2114640474319458,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.10477652549743652,
+      "mean_token_accuracy": 0.9591076672077179,
+      "num_tokens": 18658366.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.0303086400032044,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.3319413661956787,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.10648446083068848,
+      "mean_token_accuracy": 0.9581128239631653,
+      "num_tokens": 18839989.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0405295491218567,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.4326776266098022,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.10707225799560546,
+      "mean_token_accuracy": 0.9574429094791412,
+      "num_tokens": 19020331.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.037307357788086,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.2203212976455688,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.10811934471130372,
+      "mean_token_accuracy": 0.9569429934024811,
+      "num_tokens": 19201621.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.048385226726532,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.2597253322601318,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.10502862930297852,
+      "mean_token_accuracy": 0.9592942714691162,
+      "num_tokens": 19382819.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.0431596279144286,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.288861632347107,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.1091151475906372,
+      "mean_token_accuracy": 0.9570925652980804,
+      "num_tokens": 19563291.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.0325645685195923,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.1845893859863281,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.10389597415924072,
+      "mean_token_accuracy": 0.9585334539413453,
+      "num_tokens": 19744388.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.025480568408966,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 1.3404650688171387,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.10605425834655761,
+      "mean_token_accuracy": 0.9583600521087646,
+      "num_tokens": 19925613.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.020683228969574,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5598406791687012,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.10631901025772095,
+      "mean_token_accuracy": 0.9573191046714783,
+      "num_tokens": 20106747.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.0185366213321685,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3814117908477783,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.1074562907218933,
+      "mean_token_accuracy": 0.9571435689926148,
+      "num_tokens": 20287872.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0160030841827392,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 1.5130647420883179,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.10584371089935303,
+      "mean_token_accuracy": 0.9584454715251922,
+      "num_tokens": 20469530.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.0295212030410767,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.137473464012146,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.10609703063964844,
+      "mean_token_accuracy": 0.9590815663337707,
+      "num_tokens": 20650652.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.0236494064331054,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.4402672052383423,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.10043272972106934,
+      "mean_token_accuracy": 0.9601598799228668,
+      "num_tokens": 20832844.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0228356838226318,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.6836261749267578,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.10231537818908691,
+      "mean_token_accuracy": 0.9592978358268738,
+      "num_tokens": 21014766.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.0204253315925598,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.7377065420150757,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.10434420108795166,
+      "mean_token_accuracy": 0.9578610718250274,
+      "num_tokens": 21195755.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.0251757979393006,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.337659239768982,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.10218650102615356,
+      "mean_token_accuracy": 0.9588291764259338,
+      "num_tokens": 21376476.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.021351706981659,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.278275728225708,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.10114043951034546,
+      "mean_token_accuracy": 0.9596256852149964,
+      "num_tokens": 21558292.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.0222898125648499,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.1806602478027344,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.10140385627746581,
+      "mean_token_accuracy": 0.9601265788078308,
+      "num_tokens": 21739640.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0402090787887572,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.2378828525543213,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.10622456073760986,
+      "mean_token_accuracy": 0.9573592960834503,
+      "num_tokens": 21921425.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.0412330985069276,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3350801467895508,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.10371602773666382,
+      "mean_token_accuracy": 0.9588219523429871,
+      "num_tokens": 22102605.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.0534109234809876,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 1.4873039722442627,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.10807039737701415,
+      "mean_token_accuracy": 0.9577955365180969,
+      "num_tokens": 22283687.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0713044166564942,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.1694425344467163,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.11005204916000366,
+      "mean_token_accuracy": 0.9568802535533905,
+      "num_tokens": 22464813.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.0745116233825684,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.5362104177474976,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.10913199186325073,
+      "mean_token_accuracy": 0.955611252784729,
+      "num_tokens": 22646088.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0925905466079713,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.3478912115097046,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.10608880519866944,
+      "mean_token_accuracy": 0.9566390633583068,
+      "num_tokens": 22826677.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0899853229522705,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.238072395324707,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.10292150974273681,
+      "mean_token_accuracy": 0.9586134016513824,
+      "num_tokens": 23008177.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0695149540901183,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.541548252105713,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.10405257940292359,
+      "mean_token_accuracy": 0.9580470204353333,
+      "num_tokens": 23190301.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.0614005208015442,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.4365121126174927,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.10439364910125733,
+      "mean_token_accuracy": 0.9584310293197632,
+      "num_tokens": 23371824.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.065568709373474,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.2848029136657715,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.104488205909729,
+      "mean_token_accuracy": 0.959163922071457,
+      "num_tokens": 23552849.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.0750214815139771,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 1.280882716178894,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.10620679855346679,
+      "mean_token_accuracy": 0.9575595021247864,
+      "num_tokens": 23733408.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.077055287361145,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.2904170751571655,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.11054887771606445,
+      "mean_token_accuracy": 0.9566682338714599,
+      "num_tokens": 23914203.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.081210732460022,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.129718542098999,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.10313529968261718,
+      "mean_token_accuracy": 0.9582066476345062,
+      "num_tokens": 24094078.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.0856151461601258,
+      "epoch": 1.80349932705249,
+      "grad_norm": 1.3682607412338257,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.10383679866790771,
+      "mean_token_accuracy": 0.9592101991176605,
+      "num_tokens": 24274071.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.0931835532188416,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.3282326459884644,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.10675904750823975,
+      "mean_token_accuracy": 0.9580830633640289,
+      "num_tokens": 24454766.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.0867947936058044,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.2584236860275269,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.10147552490234375,
+      "mean_token_accuracy": 0.9604024410247802,
+      "num_tokens": 24635903.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.0726998805999757,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 1.254036545753479,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.10524414777755738,
+      "mean_token_accuracy": 0.9582872509956359,
+      "num_tokens": 24816787.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.0692258477210999,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.41594660282135,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.10340676307678223,
+      "mean_token_accuracy": 0.9590053558349609,
+      "num_tokens": 24998543.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.0772343635559083,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.385327696800232,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.10295225381851196,
+      "mean_token_accuracy": 0.9591428697109222,
+      "num_tokens": 25180059.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.0874038934707642,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 1.0628604888916016,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.10015511512756348,
+      "mean_token_accuracy": 0.9605867803096771,
+      "num_tokens": 25361290.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0702470779418944,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.1819324493408203,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.104461407661438,
+      "mean_token_accuracy": 0.9581527650356293,
+      "num_tokens": 25542455.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.0592689394950867,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4695993661880493,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.1024864912033081,
+      "mean_token_accuracy": 0.9591827690601349,
+      "num_tokens": 25723310.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.0781041979789734,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 1.3313041925430298,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.10552446842193604,
+      "mean_token_accuracy": 0.9576125502586365,
+      "num_tokens": 25904923.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0914432287216187,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.3578517436981201,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.11063914299011231,
+      "mean_token_accuracy": 0.9563858449459076,
+      "num_tokens": 26086154.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0916360139846801,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.3493428230285645,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.10228430032730103,
+      "mean_token_accuracy": 0.9597218215465546,
+      "num_tokens": 26268096.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0865699172019958,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.3002992868423462,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.10704560279846191,
+      "mean_token_accuracy": 0.9567803084850312,
+      "num_tokens": 26448598.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.0907732367515564,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.3105698823928833,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.10344587564468384,
+      "mean_token_accuracy": 0.9585539996623993,
+      "num_tokens": 26629572.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.097494399547577,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.5486772060394287,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.10484597682952881,
+      "mean_token_accuracy": 0.9578526735305786,
+      "num_tokens": 26810588.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 1.1259616188182953,
+      "eval_loss": 0.11920737475156784,
+      "eval_mean_token_accuracy": 0.9516198661676638,
+      "eval_num_tokens": 26919371.0,
+      "eval_runtime": 12.6599,
+      "eval_samples_per_second": 394.949,
+      "eval_steps_per_second": 12.401,
+      "step": 1486
+    },
+    {
+      "entropy": 1.1117471694946288,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.3559718132019043,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.09661188125610351,
+      "mean_token_accuracy": 0.9621427834033967,
+      "num_tokens": 26991831.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.0618590712547302,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.5231083631515503,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.07198524475097656,
+      "mean_token_accuracy": 0.9724888324737548,
+      "num_tokens": 27172578.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.0601553201675415,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0978407859802246,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.07232148051261902,
+      "mean_token_accuracy": 0.9721459746360779,
+      "num_tokens": 27353505.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.0545680522918701,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.375500202178955,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0737423300743103,
+      "mean_token_accuracy": 0.9723270058631897,
+      "num_tokens": 27534758.0,
+      "step": 1520
+    },
+    {
+      "entropy": 1.0429188370704652,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.5134543180465698,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.07129207253456116,
+      "mean_token_accuracy": 0.9730912566184997,
+      "num_tokens": 27715692.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0608030915260316,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.424338936805725,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.07388348579406738,
+      "mean_token_accuracy": 0.9727432250976562,
+      "num_tokens": 27896664.0,
+      "step": 1540
+    },
+    {
+      "entropy": 1.0577767491340637,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2193686962127686,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.07269638180732726,
+      "mean_token_accuracy": 0.9726069569587708,
+      "num_tokens": 28077786.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0627391576766967,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.4851799011230469,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.07443426847457886,
+      "mean_token_accuracy": 0.9716702818870544,
+      "num_tokens": 28259406.0,
+      "step": 1560
+    },
+    {
+      "entropy": 1.0606005311012268,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.136905312538147,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.07183161973953248,
+      "mean_token_accuracy": 0.9726755082607269,
+      "num_tokens": 28441336.0,
+      "step": 1570
+    },
+    {
+      "entropy": 1.0512076377868653,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.4116008281707764,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.07272077798843384,
+      "mean_token_accuracy": 0.972117805480957,
+      "num_tokens": 28621984.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0617831587791442,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.4194544553756714,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.07670866250991822,
+      "mean_token_accuracy": 0.9709574401378631,
+      "num_tokens": 28803335.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.054829752445221,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 1.0619711875915527,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.06954843401908875,
+      "mean_token_accuracy": 0.9732311069965363,
+      "num_tokens": 28984568.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.062503707408905,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1824331283569336,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.0742799162864685,
+      "mean_token_accuracy": 0.9718966007232666,
+      "num_tokens": 29165958.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0652349829673766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.2683000564575195,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.07545414566993713,
+      "mean_token_accuracy": 0.9704910278320312,
+      "num_tokens": 29347870.0,
+      "step": 1620
+    },
+    {
+      "entropy": 1.0519612908363343,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.257572889328003,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.075272136926651,
+      "mean_token_accuracy": 0.9712580978870392,
+      "num_tokens": 29529453.0,
+      "step": 1630
+    },
+    {
+      "entropy": 1.0668588399887085,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.3439997434616089,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.068633770942688,
+      "mean_token_accuracy": 0.9735071420669555,
+      "num_tokens": 29710695.0,
+      "step": 1640
+    },
+    {
+      "entropy": 1.0701637029647828,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.4352364540100098,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.07366064786911011,
+      "mean_token_accuracy": 0.9724323868751525,
+      "num_tokens": 29890993.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.065868353843689,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.3669353723526,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.0714600920677185,
+      "mean_token_accuracy": 0.9717992007732391,
+      "num_tokens": 30072284.0,
+      "step": 1660
+    },
+    {
+      "entropy": 1.077157938480377,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 1.3630682229995728,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.07237017154693604,
+      "mean_token_accuracy": 0.9729755759239197,
+      "num_tokens": 30253123.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0886191844940185,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.3787649869918823,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.07589051723480225,
+      "mean_token_accuracy": 0.9716001987457276,
+      "num_tokens": 30434100.0,
+      "step": 1680
+    },
+    {
+      "entropy": 1.080547845363617,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.4936959743499756,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.07393252849578857,
+      "mean_token_accuracy": 0.9709986329078675,
+      "num_tokens": 30615086.0,
+      "step": 1690
+    },
+    {
+      "entropy": 1.0851039648056031,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.4654748439788818,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.07487455606460572,
+      "mean_token_accuracy": 0.9710996568202972,
+      "num_tokens": 30795174.0,
+      "step": 1700
+    },
+    {
+      "entropy": 1.0743987679481506,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 1.354244589805603,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.07701593041419982,
+      "mean_token_accuracy": 0.9701604127883912,
+      "num_tokens": 30976627.0,
+      "step": 1710
+    },
+    {
+      "entropy": 1.0738040447235107,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.2830452919006348,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.07861257195472718,
+      "mean_token_accuracy": 0.9696074068546295,
+      "num_tokens": 31158542.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.0667856335639954,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.3329832553863525,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.07371129989624023,
+      "mean_token_accuracy": 0.9722550988197327,
+      "num_tokens": 31339091.0,
+      "step": 1730
+    },
+    {
+      "entropy": 1.0734196662902833,
+      "epoch": 2.34185733512786,
+      "grad_norm": 1.4483965635299683,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.07841320037841797,
+      "mean_token_accuracy": 0.9698444902896881,
+      "num_tokens": 31519961.0,
+      "step": 1740
+    },
+    {
+      "entropy": 1.0724906921386719,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.440190315246582,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.07452400922775268,
+      "mean_token_accuracy": 0.9715199947357178,
+      "num_tokens": 31701305.0,
+      "step": 1750
+    },
+    {
+      "entropy": 1.0770147442817688,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 1.17177414894104,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.07649941444396972,
+      "mean_token_accuracy": 0.9704680681228638,
+      "num_tokens": 31882106.0,
+      "step": 1760
+    },
+    {
+      "entropy": 1.0792171478271484,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.498350977897644,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.07850931882858277,
+      "mean_token_accuracy": 0.9699269771575928,
+      "num_tokens": 32062712.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.092840301990509,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.4153627157211304,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.07811332941055298,
+      "mean_token_accuracy": 0.9698202073574066,
+      "num_tokens": 32244156.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0767372131347657,
+      "epoch": 2.409152086137281,
+      "grad_norm": 1.3672188520431519,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.0765878677368164,
+      "mean_token_accuracy": 0.970400470495224,
+      "num_tokens": 32424975.0,
+      "step": 1790
+    },
+    {
+      "entropy": 1.0641321778297423,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.1141754388809204,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0744673728942871,
+      "mean_token_accuracy": 0.9713521003723145,
+      "num_tokens": 32606399.0,
+      "step": 1800
+    },
+    {
+      "entropy": 1.0847392082214355,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.372215986251831,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0788280725479126,
+      "mean_token_accuracy": 0.9688920557498932,
+      "num_tokens": 32787632.0,
+      "step": 1810
+    },
+    {
+      "entropy": 1.0952076196670533,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.3998148441314697,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.07322710752487183,
+      "mean_token_accuracy": 0.9713049352169036,
+      "num_tokens": 32969136.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.1047595977783202,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.538533091545105,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.07695393562316895,
+      "mean_token_accuracy": 0.9697132647037506,
+      "num_tokens": 33149585.0,
+      "step": 1830
+    },
+    {
+      "entropy": 1.1071401476860045,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.2797189950942993,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.07820409536361694,
+      "mean_token_accuracy": 0.9695189118385314,
+      "num_tokens": 33330604.0,
+      "step": 1840
+    },
+    {
+      "entropy": 1.0979747891426086,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 1.309956669807434,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.08005034327507018,
+      "mean_token_accuracy": 0.9683119177818298,
+      "num_tokens": 33510932.0,
+      "step": 1850
+    },
+    {
+      "entropy": 1.1024476408958435,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.2206803560256958,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.07752519249916076,
+      "mean_token_accuracy": 0.9697020888328552,
+      "num_tokens": 33692289.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0788621783256531,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.4230300188064575,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.08086194992065429,
+      "mean_token_accuracy": 0.9689899325370789,
+      "num_tokens": 33873643.0,
+      "step": 1870
+    },
+    {
+      "entropy": 1.0909191489219665,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 1.3553478717803955,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.07968194484710693,
+      "mean_token_accuracy": 0.9695583760738373,
+      "num_tokens": 34054342.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.0874631762504579,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 1.209466576576233,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.07668136358261109,
+      "mean_token_accuracy": 0.9706926643848419,
+      "num_tokens": 34235054.0,
+      "step": 1890
+    },
+    {
+      "entropy": 1.0850980758666993,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.2419084310531616,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.07621376514434815,
+      "mean_token_accuracy": 0.9708946466445922,
+      "num_tokens": 34415792.0,
+      "step": 1900
+    },
+    {
+      "entropy": 1.1024398922920227,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.5481151342391968,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.07621458768844605,
+      "mean_token_accuracy": 0.9696649491786957,
+      "num_tokens": 34596130.0,
+      "step": 1910
+    },
+    {
+      "entropy": 1.0917130827903747,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.4186631441116333,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.07716711163520813,
+      "mean_token_accuracy": 0.9699705123901368,
+      "num_tokens": 34776608.0,
+      "step": 1920
+    },
+    {
+      "entropy": 1.0845637321472168,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2549679279327393,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.07353252172470093,
+      "mean_token_accuracy": 0.9715336680412292,
+      "num_tokens": 34957854.0,
+      "step": 1930
+    },
+    {
+      "entropy": 1.070354163646698,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.390034794807434,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.07804152965545655,
+      "mean_token_accuracy": 0.9694301605224609,
+      "num_tokens": 35139425.0,
+      "step": 1940
+    },
+    {
+      "entropy": 1.0926494956016541,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.2890169620513916,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.07905603051185608,
+      "mean_token_accuracy": 0.9694173872470856,
+      "num_tokens": 35321009.0,
+      "step": 1950
+    },
+    {
+      "entropy": 1.090261971950531,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.201541543006897,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.07853106260299683,
+      "mean_token_accuracy": 0.9687644064426422,
+      "num_tokens": 35502338.0,
+      "step": 1960
+    },
+    {
+      "entropy": 1.110123836994171,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.3321393728256226,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.07721331715583801,
+      "mean_token_accuracy": 0.9703636646270752,
+      "num_tokens": 35684550.0,
+      "step": 1970
+    },
+    {
+      "entropy": 1.1154867172241212,
+      "epoch": 2.664872139973082,
+      "grad_norm": 1.5639152526855469,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.07769335508346557,
+      "mean_token_accuracy": 0.9694860637187958,
+      "num_tokens": 35866258.0,
+      "step": 1980
+    },
+    {
+      "entropy": 1.1126874089241028,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5160865783691406,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.07732025384902955,
+      "mean_token_accuracy": 0.9695809543132782,
+      "num_tokens": 36047771.0,
+      "step": 1990
+    },
+    {
+      "entropy": 1.1188251495361328,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.3689777851104736,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.07844393253326416,
+      "mean_token_accuracy": 0.9697122633457184,
+      "num_tokens": 36228702.0,
+      "step": 2000
+    },
+    {
+      "entropy": 1.1134975671768188,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 1.1956474781036377,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.07932614088058472,
+      "mean_token_accuracy": 0.9695560097694397,
+      "num_tokens": 36409577.0,
+      "step": 2010
+    },
+    {
+      "entropy": 1.1052889227867126,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.2844921350479126,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.07471654415130616,
+      "mean_token_accuracy": 0.9707616209983826,
+      "num_tokens": 36590635.0,
+      "step": 2020
+    },
+    {
+      "entropy": 1.1029305696487426,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3110395669937134,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.07654619216918945,
+      "mean_token_accuracy": 0.9699673116207123,
+      "num_tokens": 36772032.0,
+      "step": 2030
+    },
+    {
+      "entropy": 1.0967667698860168,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.2297992706298828,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.07549123167991638,
+      "mean_token_accuracy": 0.9705443382263184,
+      "num_tokens": 36953014.0,
+      "step": 2040
+    },
+    {
+      "entropy": 1.0954343318939208,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.5522037744522095,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.07590663433074951,
+      "mean_token_accuracy": 0.9709664165973664,
+      "num_tokens": 37134549.0,
+      "step": 2050
+    },
+    {
+      "entropy": 1.0984182119369508,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.3691717386245728,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.07795964479446411,
+      "mean_token_accuracy": 0.969748318195343,
+      "num_tokens": 37315651.0,
+      "step": 2060
+    },
+    {
+      "entropy": 1.0937008380889892,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 1.345433235168457,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.07556321024894715,
+      "mean_token_accuracy": 0.9707719802856445,
+      "num_tokens": 37497150.0,
+      "step": 2070
+    },
+    {
+      "entropy": 1.0978638410568238,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.3442333936691284,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.07603150010108947,
+      "mean_token_accuracy": 0.9707530319690705,
+      "num_tokens": 37678637.0,
+      "step": 2080
+    },
+    {
+      "entropy": 1.112320113182068,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.672094702720642,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.07702311277389526,
+      "mean_token_accuracy": 0.969988214969635,
+      "num_tokens": 37859530.0,
+      "step": 2090
+    },
+    {
+      "entropy": 1.1019986748695374,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.4613395929336548,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.07702183723449707,
+      "mean_token_accuracy": 0.9705046355724335,
+      "num_tokens": 38040160.0,
+      "step": 2100
+    },
+    {
+      "entropy": 1.1074181199073792,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.33228600025177,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.07809240221977234,
+      "mean_token_accuracy": 0.9702594041824341,
+      "num_tokens": 38221261.0,
+      "step": 2110
+    },
+    {
+      "entropy": 1.1087767481803894,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.3967821598052979,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.07641294002532958,
+      "mean_token_accuracy": 0.9694520056247711,
+      "num_tokens": 38402589.0,
+      "step": 2120
+    },
+    {
+      "entropy": 1.0902286529541017,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 1.3104654550552368,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.07864789366722107,
+      "mean_token_accuracy": 0.9693171322345734,
+      "num_tokens": 38583485.0,
+      "step": 2130
+    },
+    {
+      "entropy": 1.088974940776825,
+      "epoch": 2.88021534320323,
+      "grad_norm": 1.389603614807129,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.07767997980117798,
+      "mean_token_accuracy": 0.9691886186599732,
+      "num_tokens": 38764480.0,
+      "step": 2140
+    },
+    {
+      "entropy": 1.093208122253418,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.3133198022842407,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.08097923994064331,
+      "mean_token_accuracy": 0.968046635389328,
+      "num_tokens": 38946016.0,
+      "step": 2150
+    },
+    {
+      "entropy": 1.0920218586921693,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.3472810983657837,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.07504125237464905,
+      "mean_token_accuracy": 0.9705065429210663,
+      "num_tokens": 39127480.0,
+      "step": 2160
+    },
+    {
+      "entropy": 1.085507369041443,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.9070137739181519,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.07882450819015503,
+      "mean_token_accuracy": 0.9694548428058625,
+      "num_tokens": 39308552.0,
+      "step": 2170
+    },
+    {
+      "entropy": 1.1070378303527832,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.3059617280960083,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.08202729225158692,
+      "mean_token_accuracy": 0.9679245829582215,
+      "num_tokens": 39489844.0,
+      "step": 2180
+    },
+    {
+      "entropy": 1.0994928359985352,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.359965443611145,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.07510381340980529,
+      "mean_token_accuracy": 0.969987016916275,
+      "num_tokens": 39671941.0,
+      "step": 2190
+    },
+    {
+      "entropy": 1.0987842202186584,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.385233998298645,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.07863752245903015,
+      "mean_token_accuracy": 0.9685526669025422,
+      "num_tokens": 39852940.0,
+      "step": 2200
+    },
+    {
+      "entropy": 1.110793125629425,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.4456037282943726,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.07783852815628052,
+      "mean_token_accuracy": 0.9702686488628387,
+      "num_tokens": 40034313.0,
+      "step": 2210
+    },
+    {
+      "entropy": 1.1140072107315064,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.1889644861221313,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.07787686586380005,
+      "mean_token_accuracy": 0.9696961343288422,
+      "num_tokens": 40215866.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 1.11932077984901,
+      "eval_loss": 0.12462407350540161,
+      "eval_mean_token_accuracy": 0.951254228118119,
+      "eval_num_tokens": 40378902.0,
+      "eval_runtime": 13.5665,
+      "eval_samples_per_second": 368.556,
+      "eval_steps_per_second": 11.573,
+      "step": 2229
+    },
+    {
+      "entropy": 1.1124149322509767,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 1.1116644144058228,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.07439510822296143,
+      "mean_token_accuracy": 0.9713229060173034,
+      "num_tokens": 40396979.0,
+      "step": 2230
+    },
+    {
+      "entropy": 1.0716044545173644,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.1957958936691284,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.05146250128746033,
+      "mean_token_accuracy": 0.9813062250614166,
+      "num_tokens": 40578935.0,
+      "step": 2240
+    },
+    {
+      "entropy": 1.0567036390304565,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.337942361831665,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.052991294860839845,
+      "mean_token_accuracy": 0.9807850599288941,
+      "num_tokens": 40759500.0,
+      "step": 2250
+    },
+    {
+      "entropy": 1.056044340133667,
+      "epoch": 3.041722745625841,
+      "grad_norm": 1.4191840887069702,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.05163681507110596,
+      "mean_token_accuracy": 0.9813888549804688,
+      "num_tokens": 40940279.0,
+      "step": 2260
+    },
+    {
+      "entropy": 1.056977343559265,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.4481396675109863,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.05059016942977905,
+      "mean_token_accuracy": 0.9814287006855011,
+      "num_tokens": 41121782.0,
+      "step": 2270
+    },
+    {
+      "entropy": 1.0579043984413148,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.0136349201202393,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.05109646320343018,
+      "mean_token_accuracy": 0.9805608510971069,
+      "num_tokens": 41302614.0,
+      "step": 2280
+    },
+    {
+      "entropy": 1.0564674019813538,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.165916085243225,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.05656059980392456,
+      "mean_token_accuracy": 0.9801787078380585,
+      "num_tokens": 41484240.0,
+      "step": 2290
+    },
+    {
+      "entropy": 1.0617436170578003,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.500622272491455,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.05233668088912964,
+      "mean_token_accuracy": 0.9808355152606965,
+      "num_tokens": 41665160.0,
+      "step": 2300
+    },
+    {
+      "entropy": 1.0509375214576722,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.403001308441162,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.05074570775032043,
+      "mean_token_accuracy": 0.9807266294956207,
+      "num_tokens": 41846062.0,
+      "step": 2310
+    },
+    {
+      "entropy": 1.0536353230476379,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1046926975250244,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.04906890392303467,
+      "mean_token_accuracy": 0.9815255463123321,
+      "num_tokens": 42027115.0,
+      "step": 2320
+    },
+    {
+      "entropy": 1.0518895149230958,
+      "epoch": 3.135935397039031,
+      "grad_norm": 1.0369377136230469,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.04991455376148224,
+      "mean_token_accuracy": 0.9822782278060913,
+      "num_tokens": 42208594.0,
+      "step": 2330
+    },
+    {
+      "entropy": 1.0597471237182616,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.6156455278396606,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.049592965841293336,
+      "mean_token_accuracy": 0.9818629324436188,
+      "num_tokens": 42389065.0,
+      "step": 2340
+    },
+    {
+      "entropy": 1.0561720967292785,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.2941797971725464,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.05092741847038269,
+      "mean_token_accuracy": 0.9810208320617676,
+      "num_tokens": 42570319.0,
+      "step": 2350
+    },
+    {
+      "entropy": 1.0521231889724731,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.1866732835769653,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.051917088031768796,
+      "mean_token_accuracy": 0.981055635213852,
+      "num_tokens": 42751501.0,
+      "step": 2360
+    },
+    {
+      "entropy": 1.0555105328559875,
+      "epoch": 3.189771197846568,
+      "grad_norm": 1.2260069847106934,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.05160566568374634,
+      "mean_token_accuracy": 0.981193995475769,
+      "num_tokens": 42932643.0,
+      "step": 2370
+    },
+    {
+      "entropy": 1.0720075011253356,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.3786890506744385,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.053551071882247926,
+      "mean_token_accuracy": 0.9797191619873047,
+      "num_tokens": 43113204.0,
+      "step": 2380
+    },
+    {
+      "entropy": 1.0718719363212585,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.3200089931488037,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.054034185409545896,
+      "mean_token_accuracy": 0.9800151228904724,
+      "num_tokens": 43293433.0,
+      "step": 2390
+    },
+    {
+      "entropy": 1.0536052703857421,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.4560705423355103,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.05042939186096192,
+      "mean_token_accuracy": 0.9806897640228271,
+      "num_tokens": 43474189.0,
+      "step": 2400
+    },
+    {
+      "entropy": 1.0602891445159912,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.3123281002044678,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.05151479244232178,
+      "mean_token_accuracy": 0.980517429113388,
+      "num_tokens": 43655355.0,
+      "step": 2410
+    },
+    {
+      "entropy": 1.067916214466095,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.2652188539505005,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.05306308269500733,
+      "mean_token_accuracy": 0.9805097997188568,
+      "num_tokens": 43836752.0,
+      "step": 2420
+    },
+    {
+      "entropy": 1.067535150051117,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.6280162334442139,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.052641332149505615,
+      "mean_token_accuracy": 0.9803492069244385,
+      "num_tokens": 44018525.0,
+      "step": 2430
+    },
+    {
+      "entropy": 1.070455825328827,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 1.632573127746582,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.05452280044555664,
+      "mean_token_accuracy": 0.9793554663658142,
+      "num_tokens": 44199901.0,
+      "step": 2440
+    },
+    {
+      "entropy": 1.0741212248802186,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.3525495529174805,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.054098492860794066,
+      "mean_token_accuracy": 0.9804170608520508,
+      "num_tokens": 44381734.0,
+      "step": 2450
+    },
+    {
+      "entropy": 1.0836357831954957,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.3989282846450806,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.05158165693283081,
+      "mean_token_accuracy": 0.9810640931129455,
+      "num_tokens": 44563158.0,
+      "step": 2460
+    },
+    {
+      "entropy": 1.077314054965973,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1970937252044678,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.052333271503448485,
+      "mean_token_accuracy": 0.9808873236179352,
+      "num_tokens": 44744642.0,
+      "step": 2470
+    },
+    {
+      "entropy": 1.069891107082367,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.3197492361068726,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.05269680619239807,
+      "mean_token_accuracy": 0.980803519487381,
+      "num_tokens": 44926301.0,
+      "step": 2480
+    },
+    {
+      "entropy": 1.071691370010376,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.332481861114502,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.054685378074646,
+      "mean_token_accuracy": 0.9795110940933227,
+      "num_tokens": 45106935.0,
+      "step": 2490
+    },
+    {
+      "entropy": 1.0654058694839477,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 1.5478020906448364,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.05287986397743225,
+      "mean_token_accuracy": 0.9797925829887391,
+      "num_tokens": 45288325.0,
+      "step": 2500
+    },
+    {
+      "entropy": 1.0660210013389588,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.43852961063385,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.05411216616630554,
+      "mean_token_accuracy": 0.9790283799171448,
+      "num_tokens": 45469485.0,
+      "step": 2510
+    },
+    {
+      "entropy": 1.0649927496910094,
+      "epoch": 3.391655450874832,
+      "grad_norm": 1.3900808095932007,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.05290043354034424,
+      "mean_token_accuracy": 0.9808539688587189,
+      "num_tokens": 45650705.0,
+      "step": 2520
+    },
+    {
+      "entropy": 1.0768967866897583,
+      "epoch": 3.405114401076716,
+      "grad_norm": 1.3428847789764404,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.05374534130096435,
+      "mean_token_accuracy": 0.9802824795246124,
+      "num_tokens": 45832104.0,
+      "step": 2530
+    },
+    {
+      "entropy": 1.0660459756851197,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.3150343894958496,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.05366337299346924,
+      "mean_token_accuracy": 0.9796974420547485,
+      "num_tokens": 46013552.0,
+      "step": 2540
+    },
+    {
+      "entropy": 1.06330087184906,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.546858787536621,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.054223823547363284,
+      "mean_token_accuracy": 0.9796406507492066,
+      "num_tokens": 46194606.0,
+      "step": 2550
+    },
+    {
+      "entropy": 1.0564390897750855,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.3932640552520752,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.05237367153167725,
+      "mean_token_accuracy": 0.9805298745632172,
+      "num_tokens": 46375520.0,
+      "step": 2560
+    },
+    {
+      "entropy": 1.0519753217697143,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.8447308540344238,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.05499213933944702,
+      "mean_token_accuracy": 0.9787130057811737,
+      "num_tokens": 46556203.0,
+      "step": 2570
+    },
+    {
+      "entropy": 1.0693527460098267,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.3508682250976562,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.055162906646728516,
+      "mean_token_accuracy": 0.9794130921363831,
+      "num_tokens": 46737287.0,
+      "step": 2580
+    },
+    {
+      "entropy": 1.078024983406067,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 1.1582329273223877,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.05406954288482666,
+      "mean_token_accuracy": 0.9794795215129852,
+      "num_tokens": 46918363.0,
+      "step": 2590
+    },
+    {
+      "entropy": 1.0643377661705018,
+      "epoch": 3.499327052489906,
+      "grad_norm": 1.429380178451538,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.05309759378433228,
+      "mean_token_accuracy": 0.9806221067905426,
+      "num_tokens": 47099549.0,
+      "step": 2600
+    },
+    {
+      "entropy": 1.068834125995636,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.236444354057312,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.05395760536193848,
+      "mean_token_accuracy": 0.9798199534416199,
+      "num_tokens": 47280430.0,
+      "step": 2610
+    },
+    {
+      "entropy": 1.0729689002037048,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 1.32350754737854,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.05501353740692139,
+      "mean_token_accuracy": 0.9792946577072144,
+      "num_tokens": 47461390.0,
+      "step": 2620
+    },
+    {
+      "entropy": 1.0766199231147766,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2348153591156006,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.05568041205406189,
+      "mean_token_accuracy": 0.9796942174434662,
+      "num_tokens": 47642662.0,
+      "step": 2630
+    },
+    {
+      "entropy": 1.0804976940155029,
+      "epoch": 3.553162853297443,
+      "grad_norm": 1.1175553798675537,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.050679147243499756,
+      "mean_token_accuracy": 0.9814749479293823,
+      "num_tokens": 47824309.0,
+      "step": 2640
+    },
+    {
+      "entropy": 1.0689257264137269,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.6297036409378052,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.054615461826324464,
+      "mean_token_accuracy": 0.9796478092670441,
+      "num_tokens": 48005982.0,
+      "step": 2650
+    },
+    {
+      "entropy": 1.0696970105171204,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2477251291275024,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.05287024974822998,
+      "mean_token_accuracy": 0.979641854763031,
+      "num_tokens": 48187048.0,
+      "step": 2660
+    },
+    {
+      "entropy": 1.0800267577171325,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.3109506368637085,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.054475915431976316,
+      "mean_token_accuracy": 0.9794476389884949,
+      "num_tokens": 48368447.0,
+      "step": 2670
+    },
+    {
+      "entropy": 1.0848782062530518,
+      "epoch": 3.60699865410498,
+      "grad_norm": 1.2147419452667236,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.05212314128875732,
+      "mean_token_accuracy": 0.9798501491546631,
+      "num_tokens": 48549144.0,
+      "step": 2680
+    },
+    {
+      "entropy": 1.0681315302848815,
+      "epoch": 3.620457604306864,
+      "grad_norm": 1.3249231576919556,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.053503894805908205,
+      "mean_token_accuracy": 0.9800095915794372,
+      "num_tokens": 48731059.0,
+      "step": 2690
+    },
+    {
+      "entropy": 1.0629405975341797,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.2435307502746582,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.0541129469871521,
+      "mean_token_accuracy": 0.9795375943183899,
+      "num_tokens": 48912283.0,
+      "step": 2700
+    },
+    {
+      "entropy": 1.0608446598052979,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.3816258907318115,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.054210084676742556,
+      "mean_token_accuracy": 0.9793854176998138,
+      "num_tokens": 49094236.0,
+      "step": 2710
+    },
+    {
+      "entropy": 1.0596249461174012,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.132426381111145,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.05496481657028198,
+      "mean_token_accuracy": 0.9798576831817627,
+      "num_tokens": 49275540.0,
+      "step": 2720
+    },
+    {
+      "entropy": 1.060381281375885,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.4120980501174927,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.05694271326065063,
+      "mean_token_accuracy": 0.9788619399070739,
+      "num_tokens": 49456684.0,
+      "step": 2730
+    },
+    {
+      "entropy": 1.0594112157821656,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.428120493888855,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.05274912118911743,
+      "mean_token_accuracy": 0.9802964627742767,
+      "num_tokens": 49638099.0,
+      "step": 2740
+    },
+    {
+      "entropy": 1.0630932807922364,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.4599305391311646,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.057349550724029544,
+      "mean_token_accuracy": 0.9788366615772247,
+      "num_tokens": 49819751.0,
+      "step": 2750
+    },
+    {
+      "entropy": 1.0632095456123352,
+      "epoch": 3.714670255720054,
+      "grad_norm": 1.8489277362823486,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.053570222854614255,
+      "mean_token_accuracy": 0.980079036951065,
+      "num_tokens": 50000623.0,
+      "step": 2760
+    },
+    {
+      "entropy": 1.0583208560943604,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.4855526685714722,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.05508332252502442,
+      "mean_token_accuracy": 0.9789698243141174,
+      "num_tokens": 50181540.0,
+      "step": 2770
+    },
+    {
+      "entropy": 1.0690102458000184,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 1.553468942642212,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.05684031248092651,
+      "mean_token_accuracy": 0.9776920318603516,
+      "num_tokens": 50363406.0,
+      "step": 2780
+    },
+    {
+      "entropy": 1.0651323795318604,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 1.2951313257217407,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.055989408493041994,
+      "mean_token_accuracy": 0.978317403793335,
+      "num_tokens": 50544599.0,
+      "step": 2790
+    },
+    {
+      "entropy": 1.0571947336196899,
+      "epoch": 3.768506056527591,
+      "grad_norm": 1.1358306407928467,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.052672064304351805,
+      "mean_token_accuracy": 0.9797160506248475,
+      "num_tokens": 50726664.0,
+      "step": 2800
+    },
+    {
+      "entropy": 1.0475606441497802,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.4842852354049683,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.05601704120635986,
+      "mean_token_accuracy": 0.9794813454151153,
+      "num_tokens": 50907903.0,
+      "step": 2810
+    },
+    {
+      "entropy": 1.053141224384308,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.3320889472961426,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.05435909032821655,
+      "mean_token_accuracy": 0.9792923092842102,
+      "num_tokens": 51088734.0,
+      "step": 2820
+    },
+    {
+      "entropy": 1.0518054842948914,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.140783667564392,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.05576471090316772,
+      "mean_token_accuracy": 0.9789255380630493,
+      "num_tokens": 51269924.0,
+      "step": 2830
+    },
+    {
+      "entropy": 1.0490313410758971,
+      "epoch": 3.822341857335128,
+      "grad_norm": 1.5749891996383667,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.05655611753463745,
+      "mean_token_accuracy": 0.9785366714000702,
+      "num_tokens": 51450406.0,
+      "step": 2840
+    },
+    {
+      "entropy": 1.0508960008621215,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 1.2535520792007446,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.05710798501968384,
+      "mean_token_accuracy": 0.978301590681076,
+      "num_tokens": 51630630.0,
+      "step": 2850
+    },
+    {
+      "entropy": 1.0434152960777283,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.6045477390289307,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.0545048713684082,
+      "mean_token_accuracy": 0.9785818517208099,
+      "num_tokens": 51811252.0,
+      "step": 2860
+    },
+    {
+      "entropy": 1.046581244468689,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.2737947702407837,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.056829071044921874,
+      "mean_token_accuracy": 0.9779990613460541,
+      "num_tokens": 51992578.0,
+      "step": 2870
+    },
+    {
+      "entropy": 1.0562566757202148,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.2364535331726074,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.054702204465866086,
+      "mean_token_accuracy": 0.9796520650386811,
+      "num_tokens": 52173507.0,
+      "step": 2880
+    },
+    {
+      "entropy": 1.0444288849830627,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.2663625478744507,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.05222170352935791,
+      "mean_token_accuracy": 0.9805905818939209,
+      "num_tokens": 52354669.0,
+      "step": 2890
+    },
+    {
+      "entropy": 1.0429784655570984,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.2072714567184448,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.05404345989227295,
+      "mean_token_accuracy": 0.9802968561649322,
+      "num_tokens": 52535334.0,
+      "step": 2900
+    },
+    {
+      "entropy": 1.045020854473114,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.3894519805908203,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.05601906776428223,
+      "mean_token_accuracy": 0.9779200375080108,
+      "num_tokens": 52715952.0,
+      "step": 2910
+    },
+    {
+      "entropy": 1.0439318537712097,
+      "epoch": 3.930013458950202,
+      "grad_norm": 1.4112615585327148,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.05345791578292847,
+      "mean_token_accuracy": 0.9796200633049011,
+      "num_tokens": 52897155.0,
+      "step": 2920
+    },
+    {
+      "entropy": 1.0361477971076964,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 1.3158206939697266,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.05547314286231995,
+      "mean_token_accuracy": 0.9788706004619598,
+      "num_tokens": 53078916.0,
+      "step": 2930
+    },
+    {
+      "entropy": 1.0427968740463256,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2565301656723022,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.05601854920387268,
+      "mean_token_accuracy": 0.9791704952716828,
+      "num_tokens": 53258884.0,
+      "step": 2940
+    },
+    {
+      "entropy": 1.0374909162521362,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.4639637470245361,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.054512780904769895,
+      "mean_token_accuracy": 0.9793699860572815,
+      "num_tokens": 53440481.0,
+      "step": 2950
+    },
+    {
+      "entropy": 1.0511455059051513,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.242073655128479,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.055217862129211426,
+      "mean_token_accuracy": 0.9801203072071075,
+      "num_tokens": 53621142.0,
+      "step": 2960
+    },
+    {
+      "entropy": 1.0518356442451477,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.1518405675888062,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.05533761382102966,
+      "mean_token_accuracy": 0.9789008319377899,
+      "num_tokens": 53802588.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 1.0548353901334628,
+      "eval_loss": 0.13498826324939728,
+      "eval_mean_token_accuracy": 0.9507886888874564,
+      "eval_num_tokens": 53838633.0,
+      "eval_runtime": 12.6809,
+      "eval_samples_per_second": 394.295,
+      "eval_steps_per_second": 12.381,
+      "step": 2972
+    },
+    {
+      "entropy": 1.043621039390564,
+      "epoch": 4.010767160161508,
+      "grad_norm": 1.1016159057617188,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.03793205618858338,
+      "mean_token_accuracy": 0.9870165169239045,
+      "num_tokens": 53983430.0,
+      "step": 2980
+    },
+    {
+      "entropy": 1.0226780295372009,
+      "epoch": 4.024226110363392,
+      "grad_norm": 1.240837812423706,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.03402446508407593,
+      "mean_token_accuracy": 0.9883724093437195,
+      "num_tokens": 54164955.0,
+      "step": 2990
+    },
+    {
+      "entropy": 1.0194039583206176,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.1843734979629517,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.035861504077911374,
+      "mean_token_accuracy": 0.9872536063194275,
+      "num_tokens": 54346968.0,
+      "step": 3000
+    },
+    {
+      "entropy": 1.029161846637726,
+      "epoch": 4.05114401076716,
+      "grad_norm": 1.285415768623352,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.034106379747390746,
+      "mean_token_accuracy": 0.9882839441299438,
+      "num_tokens": 54528019.0,
+      "step": 3010
+    },
+    {
+      "entropy": 1.031964898109436,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4622068405151367,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.032776829600334165,
+      "mean_token_accuracy": 0.9882788896560669,
+      "num_tokens": 54708148.0,
+      "step": 3020
+    },
+    {
+      "entropy": 1.0282336235046388,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.1071696281433105,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.03698182702064514,
+      "mean_token_accuracy": 0.9865448236465454,
+      "num_tokens": 54888894.0,
+      "step": 3030
+    },
+    {
+      "entropy": 1.0233591556549073,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.3671438694000244,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.03386759161949158,
+      "mean_token_accuracy": 0.9879629850387573,
+      "num_tokens": 55069872.0,
+      "step": 3040
+    },
+    {
+      "entropy": 1.0250957012176514,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.1421459913253784,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.035586753487586976,
+      "mean_token_accuracy": 0.9869713068008423,
+      "num_tokens": 55250605.0,
+      "step": 3050
+    },
+    {
+      "entropy": 1.016488790512085,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.3857697248458862,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.039323312044143674,
+      "mean_token_accuracy": 0.986032885313034,
+      "num_tokens": 55431737.0,
+      "step": 3060
+    },
+    {
+      "entropy": 1.0130115926265717,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.1797211170196533,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.03508211970329285,
+      "mean_token_accuracy": 0.9872302651405335,
+      "num_tokens": 55612917.0,
+      "step": 3070
+    },
+    {
+      "entropy": 1.023509919643402,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.1484259366989136,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.0338476300239563,
+      "mean_token_accuracy": 0.9883188247680664,
+      "num_tokens": 55793687.0,
+      "step": 3080
+    },
+    {
+      "entropy": 1.0193628787994384,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.3184833526611328,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.03667713701725006,
+      "mean_token_accuracy": 0.9869367897510528,
+      "num_tokens": 55974723.0,
+      "step": 3090
+    },
+    {
+      "entropy": 1.013669216632843,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.2176531553268433,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.035006949305534364,
+      "mean_token_accuracy": 0.9868299901485443,
+      "num_tokens": 56156568.0,
+      "step": 3100
+    },
+    {
+      "entropy": 1.0072043538093567,
+      "epoch": 4.185733512786003,
+      "grad_norm": 1.4346243143081665,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.03560284972190857,
+      "mean_token_accuracy": 0.9869017779827118,
+      "num_tokens": 56337896.0,
+      "step": 3110
+    },
+    {
+      "entropy": 1.0102415680885315,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.2328612804412842,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.03608990907669067,
+      "mean_token_accuracy": 0.9874296009540557,
+      "num_tokens": 56519102.0,
+      "step": 3120
+    },
+    {
+      "entropy": 1.015284490585327,
+      "epoch": 4.212651413189771,
+      "grad_norm": 1.4385327100753784,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.03637963235378265,
+      "mean_token_accuracy": 0.986739182472229,
+      "num_tokens": 56699936.0,
+      "step": 3130
+    },
+    {
+      "entropy": 1.0054036319255828,
+      "epoch": 4.226110363391656,
+      "grad_norm": 1.1129945516586304,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.034222981333732604,
+      "mean_token_accuracy": 0.9883054196834564,
+      "num_tokens": 56882308.0,
+      "step": 3140
+    },
+    {
+      "entropy": 1.0061798334121703,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1598492860794067,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.03461446762084961,
+      "mean_token_accuracy": 0.9873203098773956,
+      "num_tokens": 57063770.0,
+      "step": 3150
+    },
+    {
+      "entropy": 1.0039719343185425,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.322458267211914,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.03701616525650024,
+      "mean_token_accuracy": 0.9862136900424957,
+      "num_tokens": 57244448.0,
+      "step": 3160
+    },
+    {
+      "entropy": 1.0048979759216308,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 1.570866584777832,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.036290496587753296,
+      "mean_token_accuracy": 0.9868158757686615,
+      "num_tokens": 57425448.0,
+      "step": 3170
+    },
+    {
+      "entropy": 1.0145976901054383,
+      "epoch": 4.279946164199193,
+      "grad_norm": 1.103385329246521,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.03472789227962494,
+      "mean_token_accuracy": 0.987658828496933,
+      "num_tokens": 57606369.0,
+      "step": 3180
+    },
+    {
+      "entropy": 1.0186771035194397,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.2541930675506592,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.03532317280769348,
+      "mean_token_accuracy": 0.9874464809894562,
+      "num_tokens": 57787685.0,
+      "step": 3190
+    },
+    {
+      "entropy": 1.0198698043823242,
+      "epoch": 4.306864064602961,
+      "grad_norm": 1.1590384244918823,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.036364412307739256,
+      "mean_token_accuracy": 0.9864933013916015,
+      "num_tokens": 57968903.0,
+      "step": 3200
+    },
+    {
+      "entropy": 1.0211624503135681,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.7207615375518799,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.03790584802627563,
+      "mean_token_accuracy": 0.9866538405418396,
+      "num_tokens": 58149469.0,
+      "step": 3210
+    },
+    {
+      "entropy": 1.0185290336608888,
+      "epoch": 4.33378196500673,
+      "grad_norm": 1.3042566776275635,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.03802136182785034,
+      "mean_token_accuracy": 0.9859855055809021,
+      "num_tokens": 58330867.0,
+      "step": 3220
+    },
+    {
+      "entropy": 1.0105828583240508,
+      "epoch": 4.347240915208614,
+      "grad_norm": 1.2536448240280151,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.034687772393226624,
+      "mean_token_accuracy": 0.9878224015235901,
+      "num_tokens": 58512529.0,
+      "step": 3230
+    },
+    {
+      "entropy": 1.0021630108356476,
+      "epoch": 4.360699865410498,
+      "grad_norm": 1.3816099166870117,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.035324835777282716,
+      "mean_token_accuracy": 0.9872316241264343,
+      "num_tokens": 58693573.0,
+      "step": 3240
+    },
+    {
+      "entropy": 1.007886278629303,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 1.3121757507324219,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.034709393978118896,
+      "mean_token_accuracy": 0.9877420663833618,
+      "num_tokens": 58874107.0,
+      "step": 3250
+    },
+    {
+      "entropy": 1.005664885044098,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.429775357246399,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.03714964389801025,
+      "mean_token_accuracy": 0.9861079812049866,
+      "num_tokens": 59054533.0,
+      "step": 3260
+    },
+    {
+      "entropy": 1.0047886371612549,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.2385882139205933,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.036937403678894046,
+      "mean_token_accuracy": 0.9863442242145538,
+      "num_tokens": 59236331.0,
+      "step": 3270
+    },
+    {
+      "entropy": 1.0209275126457213,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.476168155670166,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.03463808596134186,
+      "mean_token_accuracy": 0.9872730135917663,
+      "num_tokens": 59417193.0,
+      "step": 3280
+    },
+    {
+      "entropy": 1.021347463130951,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 1.4710102081298828,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.038167804479599,
+      "mean_token_accuracy": 0.986047238111496,
+      "num_tokens": 59598284.0,
+      "step": 3290
+    },
+    {
+      "entropy": 1.0282366394996643,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1043860912322998,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.036692625284194945,
+      "mean_token_accuracy": 0.9862713575363159,
+      "num_tokens": 59779329.0,
+      "step": 3300
+    },
+    {
+      "entropy": 1.021342945098877,
+      "epoch": 4.454912516823688,
+      "grad_norm": 1.1742897033691406,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.03571464121341705,
+      "mean_token_accuracy": 0.987505567073822,
+      "num_tokens": 59961340.0,
+      "step": 3310
+    },
+    {
+      "entropy": 1.0153375625610352,
+      "epoch": 4.468371467025572,
+      "grad_norm": 1.1321660280227661,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.036007928848266604,
+      "mean_token_accuracy": 0.9870990335941314,
+      "num_tokens": 60142647.0,
+      "step": 3320
+    },
+    {
+      "entropy": 1.0099608540534972,
+      "epoch": 4.481830417227457,
+      "grad_norm": 1.084017276763916,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.03562799394130707,
+      "mean_token_accuracy": 0.9865496158599854,
+      "num_tokens": 60324151.0,
+      "step": 3330
+    },
+    {
+      "entropy": 1.0173812866210938,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2779732942581177,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.03556201159954071,
+      "mean_token_accuracy": 0.9875519394874572,
+      "num_tokens": 60505023.0,
+      "step": 3340
+    },
+    {
+      "entropy": 1.0236618638038635,
+      "epoch": 4.508748317631225,
+      "grad_norm": 1.5285946130752563,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.03721941411495209,
+      "mean_token_accuracy": 0.9865630626678467,
+      "num_tokens": 60685921.0,
+      "step": 3350
+    },
+    {
+      "entropy": 1.0262888669967651,
+      "epoch": 4.522207267833109,
+      "grad_norm": 1.131675362586975,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.03602397441864014,
+      "mean_token_accuracy": 0.9864337563514709,
+      "num_tokens": 60867240.0,
+      "step": 3360
+    },
+    {
+      "entropy": 1.0292576909065247,
+      "epoch": 4.535666218034994,
+      "grad_norm": 1.5156011581420898,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.03973098993301392,
+      "mean_token_accuracy": 0.9855743169784545,
+      "num_tokens": 61047708.0,
+      "step": 3370
+    },
+    {
+      "entropy": 1.010222613811493,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0979945659637451,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.03533022105693817,
+      "mean_token_accuracy": 0.9869906663894653,
+      "num_tokens": 61229247.0,
+      "step": 3380
+    },
+    {
+      "entropy": 1.004716032743454,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.3082596063613892,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.033983749151229856,
+      "mean_token_accuracy": 0.9877199769020081,
+      "num_tokens": 61411059.0,
+      "step": 3390
+    },
+    {
+      "entropy": 1.0031994462013245,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.3033607006072998,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.03607205748558044,
+      "mean_token_accuracy": 0.9866301774978637,
+      "num_tokens": 61592538.0,
+      "step": 3400
+    },
+    {
+      "entropy": 1.0113391518592834,
+      "epoch": 4.589502018842531,
+      "grad_norm": 1.3490946292877197,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.03906603455543518,
+      "mean_token_accuracy": 0.9854673087596894,
+      "num_tokens": 61773613.0,
+      "step": 3410
+    },
+    {
+      "entropy": 1.0076410889625549,
+      "epoch": 4.602960969044415,
+      "grad_norm": 1.1719361543655396,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.037575924396514894,
+      "mean_token_accuracy": 0.9867486238479615,
+      "num_tokens": 61955690.0,
+      "step": 3420
+    },
+    {
+      "entropy": 1.0200914144515991,
+      "epoch": 4.616419919246299,
+      "grad_norm": 1.1205360889434814,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.03691520690917969,
+      "mean_token_accuracy": 0.9865606665611267,
+      "num_tokens": 62136047.0,
+      "step": 3430
+    },
+    {
+      "entropy": 1.0141666173934936,
+      "epoch": 4.629878869448183,
+      "grad_norm": 1.1805763244628906,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.035447508096694946,
+      "mean_token_accuracy": 0.9873166739940643,
+      "num_tokens": 62317192.0,
+      "step": 3440
+    },
+    {
+      "entropy": 1.0067960560321807,
+      "epoch": 4.643337819650068,
+      "grad_norm": 1.0286436080932617,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.03580200672149658,
+      "mean_token_accuracy": 0.9870070457458496,
+      "num_tokens": 62498860.0,
+      "step": 3450
+    },
+    {
+      "entropy": 1.0088169515132903,
+      "epoch": 4.656796769851952,
+      "grad_norm": 1.3638492822647095,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.03756250143051147,
+      "mean_token_accuracy": 0.9861987411975861,
+      "num_tokens": 62679272.0,
+      "step": 3460
+    },
+    {
+      "entropy": 1.0098722338676454,
+      "epoch": 4.670255720053836,
+      "grad_norm": 1.649560809135437,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.037615427374839784,
+      "mean_token_accuracy": 0.9864691913127899,
+      "num_tokens": 62860715.0,
+      "step": 3470
+    },
+    {
+      "entropy": 1.0174571394920349,
+      "epoch": 4.68371467025572,
+      "grad_norm": 1.2106678485870361,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.037718603014945985,
+      "mean_token_accuracy": 0.9855880260467529,
+      "num_tokens": 63041029.0,
+      "step": 3480
+    },
+    {
+      "entropy": 1.0129252314567565,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.1571707725524902,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.03724310994148254,
+      "mean_token_accuracy": 0.9865262687206269,
+      "num_tokens": 63222379.0,
+      "step": 3490
+    },
+    {
+      "entropy": 1.0124369859695435,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.0599361658096313,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.037783673405647276,
+      "mean_token_accuracy": 0.9859567046165466,
+      "num_tokens": 63403328.0,
+      "step": 3500
+    },
+    {
+      "entropy": 1.0041921436786652,
+      "epoch": 4.724091520861373,
+      "grad_norm": 1.2054452896118164,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.03585146069526672,
+      "mean_token_accuracy": 0.9868905484676361,
+      "num_tokens": 63584598.0,
+      "step": 3510
+    },
+    {
+      "entropy": 1.0015427589416503,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.5075141191482544,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.037428200244903564,
+      "mean_token_accuracy": 0.9867083311080933,
+      "num_tokens": 63766203.0,
+      "step": 3520
+    },
+    {
+      "entropy": 1.0039161920547486,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.310390830039978,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.037494581937789914,
+      "mean_token_accuracy": 0.9862947940826416,
+      "num_tokens": 63947714.0,
+      "step": 3530
+    },
+    {
+      "entropy": 1.0049733996391297,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.2912139892578125,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.03803396224975586,
+      "mean_token_accuracy": 0.9860390484333038,
+      "num_tokens": 64128798.0,
+      "step": 3540
+    },
+    {
+      "entropy": 1.0127450585365296,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.001133680343628,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0351622611284256,
+      "mean_token_accuracy": 0.9875370979309082,
+      "num_tokens": 64309577.0,
+      "step": 3550
+    },
+    {
+      "entropy": 1.0070195615291595,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 1.1870886087417603,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.036404851078987124,
+      "mean_token_accuracy": 0.9871752023696899,
+      "num_tokens": 64490707.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.9983750462532044,
+      "epoch": 4.804845222072679,
+      "grad_norm": 1.2374205589294434,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.03696496784687042,
+      "mean_token_accuracy": 0.9865773975849151,
+      "num_tokens": 64672104.0,
+      "step": 3570
+    },
+    {
+      "entropy": 1.0151559710502625,
+      "epoch": 4.818304172274562,
+      "grad_norm": 1.2001149654388428,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.03725983202457428,
+      "mean_token_accuracy": 0.9862341761589051,
+      "num_tokens": 64853353.0,
+      "step": 3580
+    },
+    {
+      "entropy": 1.0136774778366089,
+      "epoch": 4.831763122476447,
+      "grad_norm": 1.0128737688064575,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.03615986704826355,
+      "mean_token_accuracy": 0.9872736394405365,
+      "num_tokens": 65035009.0,
+      "step": 3590
+    },
+    {
+      "entropy": 1.0082051515579225,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.2552846670150757,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.03561249375343323,
+      "mean_token_accuracy": 0.9868618845939636,
+      "num_tokens": 65215981.0,
+      "step": 3600
+    },
+    {
+      "entropy": 1.0209908723831176,
+      "epoch": 4.858681022880216,
+      "grad_norm": 1.1914079189300537,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.03721153438091278,
+      "mean_token_accuracy": 0.9862311720848084,
+      "num_tokens": 65397088.0,
+      "step": 3610
+    },
+    {
+      "entropy": 1.014677894115448,
+      "epoch": 4.872139973082099,
+      "grad_norm": 1.3584535121917725,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.03897300064563751,
+      "mean_token_accuracy": 0.9851836740970612,
+      "num_tokens": 65578262.0,
+      "step": 3620
+    },
+    {
+      "entropy": 1.0207050919532776,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.2754287719726562,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.037926393747329715,
+      "mean_token_accuracy": 0.9859795570373535,
+      "num_tokens": 65758956.0,
+      "step": 3630
+    },
+    {
+      "entropy": 1.008733856678009,
+      "epoch": 4.899057873485868,
+      "grad_norm": 1.0815879106521606,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.036518561840057376,
+      "mean_token_accuracy": 0.9870649397373199,
+      "num_tokens": 65939970.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.9995620250701904,
+      "epoch": 4.912516823687753,
+      "grad_norm": 1.1773526668548584,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.035269084572792056,
+      "mean_token_accuracy": 0.9873197913169861,
+      "num_tokens": 66121045.0,
+      "step": 3650
+    },
+    {
+      "entropy": 1.0036249816417695,
+      "epoch": 4.925975773889636,
+      "grad_norm": 1.0941131114959717,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.0353422611951828,
+      "mean_token_accuracy": 0.9870408117771149,
+      "num_tokens": 66302324.0,
+      "step": 3660
+    },
+    {
+      "entropy": 1.004719352722168,
+      "epoch": 4.939434724091521,
+      "grad_norm": 1.5107269287109375,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.03743018507957459,
+      "mean_token_accuracy": 0.9857985079288483,
+      "num_tokens": 66483916.0,
+      "step": 3670
+    },
+    {
+      "entropy": 1.0076164484024048,
+      "epoch": 4.952893674293405,
+      "grad_norm": 1.3877427577972412,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.03857019543647766,
+      "mean_token_accuracy": 0.9856370270252228,
+      "num_tokens": 66664390.0,
+      "step": 3680
+    },
+    {
+      "entropy": 1.015517818927765,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1693767309188843,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.03634478151798248,
+      "mean_token_accuracy": 0.9868648707866668,
+      "num_tokens": 66845574.0,
+      "step": 3690
+    },
+    {
+      "entropy": 1.011656105518341,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.082209587097168,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.03513925671577454,
+      "mean_token_accuracy": 0.9872389912605286,
+      "num_tokens": 67027557.0,
+      "step": 3700
+    },
+    {
+      "entropy": 1.0161120653152467,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.266615629196167,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.0379630446434021,
+      "mean_token_accuracy": 0.9865313708782196,
+      "num_tokens": 67208088.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 1.0219634957374282,
+      "eval_loss": 0.14612138271331787,
+      "eval_mean_token_accuracy": 0.9504923186484416,
+      "eval_num_tokens": 67298266.0,
+      "eval_runtime": 12.6802,
+      "eval_samples_per_second": 394.317,
+      "eval_steps_per_second": 12.382,
+      "step": 3715
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.182453676065161e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-3715/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-3715/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-3715/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/checkpoint-4458/README.md b/spanish_L20_a100/seed_42/checkpoint-4458/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-4458/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-4458/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-4458/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-4458/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-4458/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-4458/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..5ec57d4167571992f119c6b5a2128dde0c28a564
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-4458/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a82c22e33fba8dca1501d766eb983657942164495e0dfa5f6c07028dae0a426d
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-4458/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-4458/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-4458/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-4458/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-4458/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-4458/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-4458/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-4458/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-4458/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-4458/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-4458/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..4331e99512ef82f693f3cef69e86006de989be9f
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-4458/trainer_state.json
@@ -0,0 +1,4550 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.0,
+  "eval_steps": 500,
+  "global_step": 4458,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    },
+    {
+      "entropy": 1.049056887626648,
+      "epoch": 1.009421265141319,
+      "grad_norm": 1.4189648628234863,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.11506924629211426,
+      "mean_token_accuracy": 0.9560760498046875,
+      "num_tokens": 13586033.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0126458764076234,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.6394226551055908,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.10137178897857665,
+      "mean_token_accuracy": 0.9604915261268616,
+      "num_tokens": 13766883.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0174824237823485,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.3608797788619995,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.10521939992904664,
+      "mean_token_accuracy": 0.9592583775520325,
+      "num_tokens": 13949410.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0318764686584472,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.1440019607543945,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.1039464831352234,
+      "mean_token_accuracy": 0.9593731820583343,
+      "num_tokens": 14131283.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0228763103485108,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.609320878982544,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.10670688152313232,
+      "mean_token_accuracy": 0.9587047874927521,
+      "num_tokens": 14312694.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0123905301094056,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.4128942489624023,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.10546157360076905,
+      "mean_token_accuracy": 0.9586312890052795,
+      "num_tokens": 14493193.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.010130923986435,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.4825445413589478,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.09921822547912598,
+      "mean_token_accuracy": 0.9621170878410339,
+      "num_tokens": 14674762.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0216347932815553,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.2503618001937866,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.10619921684265136,
+      "mean_token_accuracy": 0.9584988236427308,
+      "num_tokens": 14856424.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.022511374950409,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.3209893703460693,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.10425064563751221,
+      "mean_token_accuracy": 0.9595829844474792,
+      "num_tokens": 15037452.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.007770985364914,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.4830716848373413,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.10189464092254638,
+      "mean_token_accuracy": 0.9609117627143859,
+      "num_tokens": 15218205.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0228153944015503,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.3191078901290894,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.10298972129821778,
+      "mean_token_accuracy": 0.9599695146083832,
+      "num_tokens": 15399322.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0262911200523377,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.2050191164016724,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.10532126426696778,
+      "mean_token_accuracy": 0.9590177297592163,
+      "num_tokens": 15581668.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0124290227890014,
+      "epoch": 1.17092866756393,
+      "grad_norm": 1.3545384407043457,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.10171200037002563,
+      "mean_token_accuracy": 0.9587810635566711,
+      "num_tokens": 15762970.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.028192389011383,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.55021071434021,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.10711442232131958,
+      "mean_token_accuracy": 0.956982696056366,
+      "num_tokens": 15944164.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0354714155197144,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2573878765106201,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.10722451210021973,
+      "mean_token_accuracy": 0.9568670630455017,
+      "num_tokens": 16125049.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.034978187084198,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.4448883533477783,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.10492000579833985,
+      "mean_token_accuracy": 0.9593439042568207,
+      "num_tokens": 16306183.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0321480870246886,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.4060254096984863,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.1017595648765564,
+      "mean_token_accuracy": 0.9604197800159454,
+      "num_tokens": 16487439.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.03044091463089,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.6413203477859497,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.1032058596611023,
+      "mean_token_accuracy": 0.9589122951030731,
+      "num_tokens": 16667826.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0394478559494018,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3867346048355103,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.10721726417541504,
+      "mean_token_accuracy": 0.9581708788871766,
+      "num_tokens": 16848488.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0355630755424499,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 1.3304182291030884,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.10916879177093505,
+      "mean_token_accuracy": 0.9566857099533081,
+      "num_tokens": 17028787.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0394182324409484,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.4853013753890991,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.10515115261077881,
+      "mean_token_accuracy": 0.9584204435348511,
+      "num_tokens": 17209699.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0405551195144653,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.1959068775177002,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.10281100273132324,
+      "mean_token_accuracy": 0.9599630236625671,
+      "num_tokens": 17391070.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0196930646896363,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.2734276056289673,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.10447280406951905,
+      "mean_token_accuracy": 0.9587895989418029,
+      "num_tokens": 17572309.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0284824252128602,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.1344295740127563,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.11111283302307129,
+      "mean_token_accuracy": 0.9560421764850616,
+      "num_tokens": 17753814.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0308443427085876,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.0935126543045044,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.10474474430084228,
+      "mean_token_accuracy": 0.9583515226840973,
+      "num_tokens": 17934774.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.0283282995224,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.5144932270050049,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.104243803024292,
+      "mean_token_accuracy": 0.9581555545330047,
+      "num_tokens": 18116015.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.0355387568473815,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.276055932044983,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.10444833040237426,
+      "mean_token_accuracy": 0.9587780058383941,
+      "num_tokens": 18296690.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.0340025186538697,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.4408668279647827,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.10817761421203613,
+      "mean_token_accuracy": 0.9573222637176514,
+      "num_tokens": 18477491.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.0480108737945557,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2114640474319458,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.10477652549743652,
+      "mean_token_accuracy": 0.9591076672077179,
+      "num_tokens": 18658366.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.0303086400032044,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.3319413661956787,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.10648446083068848,
+      "mean_token_accuracy": 0.9581128239631653,
+      "num_tokens": 18839989.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0405295491218567,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.4326776266098022,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.10707225799560546,
+      "mean_token_accuracy": 0.9574429094791412,
+      "num_tokens": 19020331.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.037307357788086,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.2203212976455688,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.10811934471130372,
+      "mean_token_accuracy": 0.9569429934024811,
+      "num_tokens": 19201621.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.048385226726532,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.2597253322601318,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.10502862930297852,
+      "mean_token_accuracy": 0.9592942714691162,
+      "num_tokens": 19382819.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.0431596279144286,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.288861632347107,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.1091151475906372,
+      "mean_token_accuracy": 0.9570925652980804,
+      "num_tokens": 19563291.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.0325645685195923,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.1845893859863281,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.10389597415924072,
+      "mean_token_accuracy": 0.9585334539413453,
+      "num_tokens": 19744388.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.025480568408966,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 1.3404650688171387,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.10605425834655761,
+      "mean_token_accuracy": 0.9583600521087646,
+      "num_tokens": 19925613.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.020683228969574,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5598406791687012,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.10631901025772095,
+      "mean_token_accuracy": 0.9573191046714783,
+      "num_tokens": 20106747.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.0185366213321685,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3814117908477783,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.1074562907218933,
+      "mean_token_accuracy": 0.9571435689926148,
+      "num_tokens": 20287872.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0160030841827392,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 1.5130647420883179,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.10584371089935303,
+      "mean_token_accuracy": 0.9584454715251922,
+      "num_tokens": 20469530.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.0295212030410767,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.137473464012146,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.10609703063964844,
+      "mean_token_accuracy": 0.9590815663337707,
+      "num_tokens": 20650652.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.0236494064331054,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.4402672052383423,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.10043272972106934,
+      "mean_token_accuracy": 0.9601598799228668,
+      "num_tokens": 20832844.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0228356838226318,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.6836261749267578,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.10231537818908691,
+      "mean_token_accuracy": 0.9592978358268738,
+      "num_tokens": 21014766.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.0204253315925598,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.7377065420150757,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.10434420108795166,
+      "mean_token_accuracy": 0.9578610718250274,
+      "num_tokens": 21195755.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.0251757979393006,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.337659239768982,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.10218650102615356,
+      "mean_token_accuracy": 0.9588291764259338,
+      "num_tokens": 21376476.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.021351706981659,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.278275728225708,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.10114043951034546,
+      "mean_token_accuracy": 0.9596256852149964,
+      "num_tokens": 21558292.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.0222898125648499,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.1806602478027344,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.10140385627746581,
+      "mean_token_accuracy": 0.9601265788078308,
+      "num_tokens": 21739640.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0402090787887572,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.2378828525543213,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.10622456073760986,
+      "mean_token_accuracy": 0.9573592960834503,
+      "num_tokens": 21921425.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.0412330985069276,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3350801467895508,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.10371602773666382,
+      "mean_token_accuracy": 0.9588219523429871,
+      "num_tokens": 22102605.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.0534109234809876,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 1.4873039722442627,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.10807039737701415,
+      "mean_token_accuracy": 0.9577955365180969,
+      "num_tokens": 22283687.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0713044166564942,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.1694425344467163,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.11005204916000366,
+      "mean_token_accuracy": 0.9568802535533905,
+      "num_tokens": 22464813.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.0745116233825684,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.5362104177474976,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.10913199186325073,
+      "mean_token_accuracy": 0.955611252784729,
+      "num_tokens": 22646088.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0925905466079713,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.3478912115097046,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.10608880519866944,
+      "mean_token_accuracy": 0.9566390633583068,
+      "num_tokens": 22826677.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0899853229522705,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.238072395324707,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.10292150974273681,
+      "mean_token_accuracy": 0.9586134016513824,
+      "num_tokens": 23008177.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0695149540901183,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.541548252105713,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.10405257940292359,
+      "mean_token_accuracy": 0.9580470204353333,
+      "num_tokens": 23190301.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.0614005208015442,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.4365121126174927,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.10439364910125733,
+      "mean_token_accuracy": 0.9584310293197632,
+      "num_tokens": 23371824.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.065568709373474,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.2848029136657715,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.104488205909729,
+      "mean_token_accuracy": 0.959163922071457,
+      "num_tokens": 23552849.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.0750214815139771,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 1.280882716178894,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.10620679855346679,
+      "mean_token_accuracy": 0.9575595021247864,
+      "num_tokens": 23733408.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.077055287361145,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.2904170751571655,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.11054887771606445,
+      "mean_token_accuracy": 0.9566682338714599,
+      "num_tokens": 23914203.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.081210732460022,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.129718542098999,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.10313529968261718,
+      "mean_token_accuracy": 0.9582066476345062,
+      "num_tokens": 24094078.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.0856151461601258,
+      "epoch": 1.80349932705249,
+      "grad_norm": 1.3682607412338257,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.10383679866790771,
+      "mean_token_accuracy": 0.9592101991176605,
+      "num_tokens": 24274071.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.0931835532188416,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.3282326459884644,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.10675904750823975,
+      "mean_token_accuracy": 0.9580830633640289,
+      "num_tokens": 24454766.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.0867947936058044,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.2584236860275269,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.10147552490234375,
+      "mean_token_accuracy": 0.9604024410247802,
+      "num_tokens": 24635903.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.0726998805999757,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 1.254036545753479,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.10524414777755738,
+      "mean_token_accuracy": 0.9582872509956359,
+      "num_tokens": 24816787.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.0692258477210999,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.41594660282135,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.10340676307678223,
+      "mean_token_accuracy": 0.9590053558349609,
+      "num_tokens": 24998543.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.0772343635559083,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.385327696800232,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.10295225381851196,
+      "mean_token_accuracy": 0.9591428697109222,
+      "num_tokens": 25180059.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.0874038934707642,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 1.0628604888916016,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.10015511512756348,
+      "mean_token_accuracy": 0.9605867803096771,
+      "num_tokens": 25361290.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0702470779418944,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.1819324493408203,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.104461407661438,
+      "mean_token_accuracy": 0.9581527650356293,
+      "num_tokens": 25542455.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.0592689394950867,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4695993661880493,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.1024864912033081,
+      "mean_token_accuracy": 0.9591827690601349,
+      "num_tokens": 25723310.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.0781041979789734,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 1.3313041925430298,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.10552446842193604,
+      "mean_token_accuracy": 0.9576125502586365,
+      "num_tokens": 25904923.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0914432287216187,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.3578517436981201,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.11063914299011231,
+      "mean_token_accuracy": 0.9563858449459076,
+      "num_tokens": 26086154.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0916360139846801,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.3493428230285645,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.10228430032730103,
+      "mean_token_accuracy": 0.9597218215465546,
+      "num_tokens": 26268096.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0865699172019958,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.3002992868423462,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.10704560279846191,
+      "mean_token_accuracy": 0.9567803084850312,
+      "num_tokens": 26448598.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.0907732367515564,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.3105698823928833,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.10344587564468384,
+      "mean_token_accuracy": 0.9585539996623993,
+      "num_tokens": 26629572.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.097494399547577,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.5486772060394287,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.10484597682952881,
+      "mean_token_accuracy": 0.9578526735305786,
+      "num_tokens": 26810588.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 1.1259616188182953,
+      "eval_loss": 0.11920737475156784,
+      "eval_mean_token_accuracy": 0.9516198661676638,
+      "eval_num_tokens": 26919371.0,
+      "eval_runtime": 12.6599,
+      "eval_samples_per_second": 394.949,
+      "eval_steps_per_second": 12.401,
+      "step": 1486
+    },
+    {
+      "entropy": 1.1117471694946288,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.3559718132019043,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.09661188125610351,
+      "mean_token_accuracy": 0.9621427834033967,
+      "num_tokens": 26991831.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.0618590712547302,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.5231083631515503,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.07198524475097656,
+      "mean_token_accuracy": 0.9724888324737548,
+      "num_tokens": 27172578.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.0601553201675415,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0978407859802246,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.07232148051261902,
+      "mean_token_accuracy": 0.9721459746360779,
+      "num_tokens": 27353505.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.0545680522918701,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.375500202178955,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0737423300743103,
+      "mean_token_accuracy": 0.9723270058631897,
+      "num_tokens": 27534758.0,
+      "step": 1520
+    },
+    {
+      "entropy": 1.0429188370704652,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.5134543180465698,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.07129207253456116,
+      "mean_token_accuracy": 0.9730912566184997,
+      "num_tokens": 27715692.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0608030915260316,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.424338936805725,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.07388348579406738,
+      "mean_token_accuracy": 0.9727432250976562,
+      "num_tokens": 27896664.0,
+      "step": 1540
+    },
+    {
+      "entropy": 1.0577767491340637,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2193686962127686,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.07269638180732726,
+      "mean_token_accuracy": 0.9726069569587708,
+      "num_tokens": 28077786.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0627391576766967,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.4851799011230469,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.07443426847457886,
+      "mean_token_accuracy": 0.9716702818870544,
+      "num_tokens": 28259406.0,
+      "step": 1560
+    },
+    {
+      "entropy": 1.0606005311012268,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.136905312538147,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.07183161973953248,
+      "mean_token_accuracy": 0.9726755082607269,
+      "num_tokens": 28441336.0,
+      "step": 1570
+    },
+    {
+      "entropy": 1.0512076377868653,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.4116008281707764,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.07272077798843384,
+      "mean_token_accuracy": 0.972117805480957,
+      "num_tokens": 28621984.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0617831587791442,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.4194544553756714,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.07670866250991822,
+      "mean_token_accuracy": 0.9709574401378631,
+      "num_tokens": 28803335.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.054829752445221,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 1.0619711875915527,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.06954843401908875,
+      "mean_token_accuracy": 0.9732311069965363,
+      "num_tokens": 28984568.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.062503707408905,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1824331283569336,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.0742799162864685,
+      "mean_token_accuracy": 0.9718966007232666,
+      "num_tokens": 29165958.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0652349829673766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.2683000564575195,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.07545414566993713,
+      "mean_token_accuracy": 0.9704910278320312,
+      "num_tokens": 29347870.0,
+      "step": 1620
+    },
+    {
+      "entropy": 1.0519612908363343,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.257572889328003,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.075272136926651,
+      "mean_token_accuracy": 0.9712580978870392,
+      "num_tokens": 29529453.0,
+      "step": 1630
+    },
+    {
+      "entropy": 1.0668588399887085,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.3439997434616089,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.068633770942688,
+      "mean_token_accuracy": 0.9735071420669555,
+      "num_tokens": 29710695.0,
+      "step": 1640
+    },
+    {
+      "entropy": 1.0701637029647828,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.4352364540100098,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.07366064786911011,
+      "mean_token_accuracy": 0.9724323868751525,
+      "num_tokens": 29890993.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.065868353843689,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.3669353723526,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.0714600920677185,
+      "mean_token_accuracy": 0.9717992007732391,
+      "num_tokens": 30072284.0,
+      "step": 1660
+    },
+    {
+      "entropy": 1.077157938480377,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 1.3630682229995728,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.07237017154693604,
+      "mean_token_accuracy": 0.9729755759239197,
+      "num_tokens": 30253123.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0886191844940185,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.3787649869918823,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.07589051723480225,
+      "mean_token_accuracy": 0.9716001987457276,
+      "num_tokens": 30434100.0,
+      "step": 1680
+    },
+    {
+      "entropy": 1.080547845363617,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.4936959743499756,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.07393252849578857,
+      "mean_token_accuracy": 0.9709986329078675,
+      "num_tokens": 30615086.0,
+      "step": 1690
+    },
+    {
+      "entropy": 1.0851039648056031,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.4654748439788818,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.07487455606460572,
+      "mean_token_accuracy": 0.9710996568202972,
+      "num_tokens": 30795174.0,
+      "step": 1700
+    },
+    {
+      "entropy": 1.0743987679481506,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 1.354244589805603,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.07701593041419982,
+      "mean_token_accuracy": 0.9701604127883912,
+      "num_tokens": 30976627.0,
+      "step": 1710
+    },
+    {
+      "entropy": 1.0738040447235107,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.2830452919006348,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.07861257195472718,
+      "mean_token_accuracy": 0.9696074068546295,
+      "num_tokens": 31158542.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.0667856335639954,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.3329832553863525,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.07371129989624023,
+      "mean_token_accuracy": 0.9722550988197327,
+      "num_tokens": 31339091.0,
+      "step": 1730
+    },
+    {
+      "entropy": 1.0734196662902833,
+      "epoch": 2.34185733512786,
+      "grad_norm": 1.4483965635299683,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.07841320037841797,
+      "mean_token_accuracy": 0.9698444902896881,
+      "num_tokens": 31519961.0,
+      "step": 1740
+    },
+    {
+      "entropy": 1.0724906921386719,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.440190315246582,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.07452400922775268,
+      "mean_token_accuracy": 0.9715199947357178,
+      "num_tokens": 31701305.0,
+      "step": 1750
+    },
+    {
+      "entropy": 1.0770147442817688,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 1.17177414894104,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.07649941444396972,
+      "mean_token_accuracy": 0.9704680681228638,
+      "num_tokens": 31882106.0,
+      "step": 1760
+    },
+    {
+      "entropy": 1.0792171478271484,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.498350977897644,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.07850931882858277,
+      "mean_token_accuracy": 0.9699269771575928,
+      "num_tokens": 32062712.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.092840301990509,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.4153627157211304,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.07811332941055298,
+      "mean_token_accuracy": 0.9698202073574066,
+      "num_tokens": 32244156.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0767372131347657,
+      "epoch": 2.409152086137281,
+      "grad_norm": 1.3672188520431519,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.0765878677368164,
+      "mean_token_accuracy": 0.970400470495224,
+      "num_tokens": 32424975.0,
+      "step": 1790
+    },
+    {
+      "entropy": 1.0641321778297423,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.1141754388809204,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0744673728942871,
+      "mean_token_accuracy": 0.9713521003723145,
+      "num_tokens": 32606399.0,
+      "step": 1800
+    },
+    {
+      "entropy": 1.0847392082214355,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.372215986251831,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0788280725479126,
+      "mean_token_accuracy": 0.9688920557498932,
+      "num_tokens": 32787632.0,
+      "step": 1810
+    },
+    {
+      "entropy": 1.0952076196670533,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.3998148441314697,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.07322710752487183,
+      "mean_token_accuracy": 0.9713049352169036,
+      "num_tokens": 32969136.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.1047595977783202,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.538533091545105,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.07695393562316895,
+      "mean_token_accuracy": 0.9697132647037506,
+      "num_tokens": 33149585.0,
+      "step": 1830
+    },
+    {
+      "entropy": 1.1071401476860045,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.2797189950942993,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.07820409536361694,
+      "mean_token_accuracy": 0.9695189118385314,
+      "num_tokens": 33330604.0,
+      "step": 1840
+    },
+    {
+      "entropy": 1.0979747891426086,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 1.309956669807434,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.08005034327507018,
+      "mean_token_accuracy": 0.9683119177818298,
+      "num_tokens": 33510932.0,
+      "step": 1850
+    },
+    {
+      "entropy": 1.1024476408958435,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.2206803560256958,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.07752519249916076,
+      "mean_token_accuracy": 0.9697020888328552,
+      "num_tokens": 33692289.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0788621783256531,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.4230300188064575,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.08086194992065429,
+      "mean_token_accuracy": 0.9689899325370789,
+      "num_tokens": 33873643.0,
+      "step": 1870
+    },
+    {
+      "entropy": 1.0909191489219665,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 1.3553478717803955,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.07968194484710693,
+      "mean_token_accuracy": 0.9695583760738373,
+      "num_tokens": 34054342.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.0874631762504579,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 1.209466576576233,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.07668136358261109,
+      "mean_token_accuracy": 0.9706926643848419,
+      "num_tokens": 34235054.0,
+      "step": 1890
+    },
+    {
+      "entropy": 1.0850980758666993,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.2419084310531616,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.07621376514434815,
+      "mean_token_accuracy": 0.9708946466445922,
+      "num_tokens": 34415792.0,
+      "step": 1900
+    },
+    {
+      "entropy": 1.1024398922920227,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.5481151342391968,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.07621458768844605,
+      "mean_token_accuracy": 0.9696649491786957,
+      "num_tokens": 34596130.0,
+      "step": 1910
+    },
+    {
+      "entropy": 1.0917130827903747,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.4186631441116333,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.07716711163520813,
+      "mean_token_accuracy": 0.9699705123901368,
+      "num_tokens": 34776608.0,
+      "step": 1920
+    },
+    {
+      "entropy": 1.0845637321472168,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2549679279327393,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.07353252172470093,
+      "mean_token_accuracy": 0.9715336680412292,
+      "num_tokens": 34957854.0,
+      "step": 1930
+    },
+    {
+      "entropy": 1.070354163646698,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.390034794807434,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.07804152965545655,
+      "mean_token_accuracy": 0.9694301605224609,
+      "num_tokens": 35139425.0,
+      "step": 1940
+    },
+    {
+      "entropy": 1.0926494956016541,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.2890169620513916,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.07905603051185608,
+      "mean_token_accuracy": 0.9694173872470856,
+      "num_tokens": 35321009.0,
+      "step": 1950
+    },
+    {
+      "entropy": 1.090261971950531,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.201541543006897,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.07853106260299683,
+      "mean_token_accuracy": 0.9687644064426422,
+      "num_tokens": 35502338.0,
+      "step": 1960
+    },
+    {
+      "entropy": 1.110123836994171,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.3321393728256226,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.07721331715583801,
+      "mean_token_accuracy": 0.9703636646270752,
+      "num_tokens": 35684550.0,
+      "step": 1970
+    },
+    {
+      "entropy": 1.1154867172241212,
+      "epoch": 2.664872139973082,
+      "grad_norm": 1.5639152526855469,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.07769335508346557,
+      "mean_token_accuracy": 0.9694860637187958,
+      "num_tokens": 35866258.0,
+      "step": 1980
+    },
+    {
+      "entropy": 1.1126874089241028,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5160865783691406,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.07732025384902955,
+      "mean_token_accuracy": 0.9695809543132782,
+      "num_tokens": 36047771.0,
+      "step": 1990
+    },
+    {
+      "entropy": 1.1188251495361328,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.3689777851104736,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.07844393253326416,
+      "mean_token_accuracy": 0.9697122633457184,
+      "num_tokens": 36228702.0,
+      "step": 2000
+    },
+    {
+      "entropy": 1.1134975671768188,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 1.1956474781036377,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.07932614088058472,
+      "mean_token_accuracy": 0.9695560097694397,
+      "num_tokens": 36409577.0,
+      "step": 2010
+    },
+    {
+      "entropy": 1.1052889227867126,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.2844921350479126,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.07471654415130616,
+      "mean_token_accuracy": 0.9707616209983826,
+      "num_tokens": 36590635.0,
+      "step": 2020
+    },
+    {
+      "entropy": 1.1029305696487426,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3110395669937134,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.07654619216918945,
+      "mean_token_accuracy": 0.9699673116207123,
+      "num_tokens": 36772032.0,
+      "step": 2030
+    },
+    {
+      "entropy": 1.0967667698860168,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.2297992706298828,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.07549123167991638,
+      "mean_token_accuracy": 0.9705443382263184,
+      "num_tokens": 36953014.0,
+      "step": 2040
+    },
+    {
+      "entropy": 1.0954343318939208,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.5522037744522095,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.07590663433074951,
+      "mean_token_accuracy": 0.9709664165973664,
+      "num_tokens": 37134549.0,
+      "step": 2050
+    },
+    {
+      "entropy": 1.0984182119369508,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.3691717386245728,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.07795964479446411,
+      "mean_token_accuracy": 0.969748318195343,
+      "num_tokens": 37315651.0,
+      "step": 2060
+    },
+    {
+      "entropy": 1.0937008380889892,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 1.345433235168457,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.07556321024894715,
+      "mean_token_accuracy": 0.9707719802856445,
+      "num_tokens": 37497150.0,
+      "step": 2070
+    },
+    {
+      "entropy": 1.0978638410568238,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.3442333936691284,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.07603150010108947,
+      "mean_token_accuracy": 0.9707530319690705,
+      "num_tokens": 37678637.0,
+      "step": 2080
+    },
+    {
+      "entropy": 1.112320113182068,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.672094702720642,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.07702311277389526,
+      "mean_token_accuracy": 0.969988214969635,
+      "num_tokens": 37859530.0,
+      "step": 2090
+    },
+    {
+      "entropy": 1.1019986748695374,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.4613395929336548,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.07702183723449707,
+      "mean_token_accuracy": 0.9705046355724335,
+      "num_tokens": 38040160.0,
+      "step": 2100
+    },
+    {
+      "entropy": 1.1074181199073792,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.33228600025177,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.07809240221977234,
+      "mean_token_accuracy": 0.9702594041824341,
+      "num_tokens": 38221261.0,
+      "step": 2110
+    },
+    {
+      "entropy": 1.1087767481803894,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.3967821598052979,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.07641294002532958,
+      "mean_token_accuracy": 0.9694520056247711,
+      "num_tokens": 38402589.0,
+      "step": 2120
+    },
+    {
+      "entropy": 1.0902286529541017,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 1.3104654550552368,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.07864789366722107,
+      "mean_token_accuracy": 0.9693171322345734,
+      "num_tokens": 38583485.0,
+      "step": 2130
+    },
+    {
+      "entropy": 1.088974940776825,
+      "epoch": 2.88021534320323,
+      "grad_norm": 1.389603614807129,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.07767997980117798,
+      "mean_token_accuracy": 0.9691886186599732,
+      "num_tokens": 38764480.0,
+      "step": 2140
+    },
+    {
+      "entropy": 1.093208122253418,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.3133198022842407,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.08097923994064331,
+      "mean_token_accuracy": 0.968046635389328,
+      "num_tokens": 38946016.0,
+      "step": 2150
+    },
+    {
+      "entropy": 1.0920218586921693,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.3472810983657837,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.07504125237464905,
+      "mean_token_accuracy": 0.9705065429210663,
+      "num_tokens": 39127480.0,
+      "step": 2160
+    },
+    {
+      "entropy": 1.085507369041443,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.9070137739181519,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.07882450819015503,
+      "mean_token_accuracy": 0.9694548428058625,
+      "num_tokens": 39308552.0,
+      "step": 2170
+    },
+    {
+      "entropy": 1.1070378303527832,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.3059617280960083,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.08202729225158692,
+      "mean_token_accuracy": 0.9679245829582215,
+      "num_tokens": 39489844.0,
+      "step": 2180
+    },
+    {
+      "entropy": 1.0994928359985352,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.359965443611145,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.07510381340980529,
+      "mean_token_accuracy": 0.969987016916275,
+      "num_tokens": 39671941.0,
+      "step": 2190
+    },
+    {
+      "entropy": 1.0987842202186584,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.385233998298645,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.07863752245903015,
+      "mean_token_accuracy": 0.9685526669025422,
+      "num_tokens": 39852940.0,
+      "step": 2200
+    },
+    {
+      "entropy": 1.110793125629425,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.4456037282943726,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.07783852815628052,
+      "mean_token_accuracy": 0.9702686488628387,
+      "num_tokens": 40034313.0,
+      "step": 2210
+    },
+    {
+      "entropy": 1.1140072107315064,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.1889644861221313,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.07787686586380005,
+      "mean_token_accuracy": 0.9696961343288422,
+      "num_tokens": 40215866.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 1.11932077984901,
+      "eval_loss": 0.12462407350540161,
+      "eval_mean_token_accuracy": 0.951254228118119,
+      "eval_num_tokens": 40378902.0,
+      "eval_runtime": 13.5665,
+      "eval_samples_per_second": 368.556,
+      "eval_steps_per_second": 11.573,
+      "step": 2229
+    },
+    {
+      "entropy": 1.1124149322509767,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 1.1116644144058228,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.07439510822296143,
+      "mean_token_accuracy": 0.9713229060173034,
+      "num_tokens": 40396979.0,
+      "step": 2230
+    },
+    {
+      "entropy": 1.0716044545173644,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.1957958936691284,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.05146250128746033,
+      "mean_token_accuracy": 0.9813062250614166,
+      "num_tokens": 40578935.0,
+      "step": 2240
+    },
+    {
+      "entropy": 1.0567036390304565,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.337942361831665,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.052991294860839845,
+      "mean_token_accuracy": 0.9807850599288941,
+      "num_tokens": 40759500.0,
+      "step": 2250
+    },
+    {
+      "entropy": 1.056044340133667,
+      "epoch": 3.041722745625841,
+      "grad_norm": 1.4191840887069702,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.05163681507110596,
+      "mean_token_accuracy": 0.9813888549804688,
+      "num_tokens": 40940279.0,
+      "step": 2260
+    },
+    {
+      "entropy": 1.056977343559265,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.4481396675109863,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.05059016942977905,
+      "mean_token_accuracy": 0.9814287006855011,
+      "num_tokens": 41121782.0,
+      "step": 2270
+    },
+    {
+      "entropy": 1.0579043984413148,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.0136349201202393,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.05109646320343018,
+      "mean_token_accuracy": 0.9805608510971069,
+      "num_tokens": 41302614.0,
+      "step": 2280
+    },
+    {
+      "entropy": 1.0564674019813538,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.165916085243225,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.05656059980392456,
+      "mean_token_accuracy": 0.9801787078380585,
+      "num_tokens": 41484240.0,
+      "step": 2290
+    },
+    {
+      "entropy": 1.0617436170578003,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.500622272491455,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.05233668088912964,
+      "mean_token_accuracy": 0.9808355152606965,
+      "num_tokens": 41665160.0,
+      "step": 2300
+    },
+    {
+      "entropy": 1.0509375214576722,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.403001308441162,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.05074570775032043,
+      "mean_token_accuracy": 0.9807266294956207,
+      "num_tokens": 41846062.0,
+      "step": 2310
+    },
+    {
+      "entropy": 1.0536353230476379,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1046926975250244,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.04906890392303467,
+      "mean_token_accuracy": 0.9815255463123321,
+      "num_tokens": 42027115.0,
+      "step": 2320
+    },
+    {
+      "entropy": 1.0518895149230958,
+      "epoch": 3.135935397039031,
+      "grad_norm": 1.0369377136230469,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.04991455376148224,
+      "mean_token_accuracy": 0.9822782278060913,
+      "num_tokens": 42208594.0,
+      "step": 2330
+    },
+    {
+      "entropy": 1.0597471237182616,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.6156455278396606,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.049592965841293336,
+      "mean_token_accuracy": 0.9818629324436188,
+      "num_tokens": 42389065.0,
+      "step": 2340
+    },
+    {
+      "entropy": 1.0561720967292785,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.2941797971725464,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.05092741847038269,
+      "mean_token_accuracy": 0.9810208320617676,
+      "num_tokens": 42570319.0,
+      "step": 2350
+    },
+    {
+      "entropy": 1.0521231889724731,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.1866732835769653,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.051917088031768796,
+      "mean_token_accuracy": 0.981055635213852,
+      "num_tokens": 42751501.0,
+      "step": 2360
+    },
+    {
+      "entropy": 1.0555105328559875,
+      "epoch": 3.189771197846568,
+      "grad_norm": 1.2260069847106934,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.05160566568374634,
+      "mean_token_accuracy": 0.981193995475769,
+      "num_tokens": 42932643.0,
+      "step": 2370
+    },
+    {
+      "entropy": 1.0720075011253356,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.3786890506744385,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.053551071882247926,
+      "mean_token_accuracy": 0.9797191619873047,
+      "num_tokens": 43113204.0,
+      "step": 2380
+    },
+    {
+      "entropy": 1.0718719363212585,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.3200089931488037,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.054034185409545896,
+      "mean_token_accuracy": 0.9800151228904724,
+      "num_tokens": 43293433.0,
+      "step": 2390
+    },
+    {
+      "entropy": 1.0536052703857421,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.4560705423355103,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.05042939186096192,
+      "mean_token_accuracy": 0.9806897640228271,
+      "num_tokens": 43474189.0,
+      "step": 2400
+    },
+    {
+      "entropy": 1.0602891445159912,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.3123281002044678,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.05151479244232178,
+      "mean_token_accuracy": 0.980517429113388,
+      "num_tokens": 43655355.0,
+      "step": 2410
+    },
+    {
+      "entropy": 1.067916214466095,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.2652188539505005,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.05306308269500733,
+      "mean_token_accuracy": 0.9805097997188568,
+      "num_tokens": 43836752.0,
+      "step": 2420
+    },
+    {
+      "entropy": 1.067535150051117,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.6280162334442139,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.052641332149505615,
+      "mean_token_accuracy": 0.9803492069244385,
+      "num_tokens": 44018525.0,
+      "step": 2430
+    },
+    {
+      "entropy": 1.070455825328827,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 1.632573127746582,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.05452280044555664,
+      "mean_token_accuracy": 0.9793554663658142,
+      "num_tokens": 44199901.0,
+      "step": 2440
+    },
+    {
+      "entropy": 1.0741212248802186,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.3525495529174805,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.054098492860794066,
+      "mean_token_accuracy": 0.9804170608520508,
+      "num_tokens": 44381734.0,
+      "step": 2450
+    },
+    {
+      "entropy": 1.0836357831954957,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.3989282846450806,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.05158165693283081,
+      "mean_token_accuracy": 0.9810640931129455,
+      "num_tokens": 44563158.0,
+      "step": 2460
+    },
+    {
+      "entropy": 1.077314054965973,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1970937252044678,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.052333271503448485,
+      "mean_token_accuracy": 0.9808873236179352,
+      "num_tokens": 44744642.0,
+      "step": 2470
+    },
+    {
+      "entropy": 1.069891107082367,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.3197492361068726,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.05269680619239807,
+      "mean_token_accuracy": 0.980803519487381,
+      "num_tokens": 44926301.0,
+      "step": 2480
+    },
+    {
+      "entropy": 1.071691370010376,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.332481861114502,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.054685378074646,
+      "mean_token_accuracy": 0.9795110940933227,
+      "num_tokens": 45106935.0,
+      "step": 2490
+    },
+    {
+      "entropy": 1.0654058694839477,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 1.5478020906448364,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.05287986397743225,
+      "mean_token_accuracy": 0.9797925829887391,
+      "num_tokens": 45288325.0,
+      "step": 2500
+    },
+    {
+      "entropy": 1.0660210013389588,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.43852961063385,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.05411216616630554,
+      "mean_token_accuracy": 0.9790283799171448,
+      "num_tokens": 45469485.0,
+      "step": 2510
+    },
+    {
+      "entropy": 1.0649927496910094,
+      "epoch": 3.391655450874832,
+      "grad_norm": 1.3900808095932007,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.05290043354034424,
+      "mean_token_accuracy": 0.9808539688587189,
+      "num_tokens": 45650705.0,
+      "step": 2520
+    },
+    {
+      "entropy": 1.0768967866897583,
+      "epoch": 3.405114401076716,
+      "grad_norm": 1.3428847789764404,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.05374534130096435,
+      "mean_token_accuracy": 0.9802824795246124,
+      "num_tokens": 45832104.0,
+      "step": 2530
+    },
+    {
+      "entropy": 1.0660459756851197,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.3150343894958496,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.05366337299346924,
+      "mean_token_accuracy": 0.9796974420547485,
+      "num_tokens": 46013552.0,
+      "step": 2540
+    },
+    {
+      "entropy": 1.06330087184906,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.546858787536621,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.054223823547363284,
+      "mean_token_accuracy": 0.9796406507492066,
+      "num_tokens": 46194606.0,
+      "step": 2550
+    },
+    {
+      "entropy": 1.0564390897750855,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.3932640552520752,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.05237367153167725,
+      "mean_token_accuracy": 0.9805298745632172,
+      "num_tokens": 46375520.0,
+      "step": 2560
+    },
+    {
+      "entropy": 1.0519753217697143,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.8447308540344238,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.05499213933944702,
+      "mean_token_accuracy": 0.9787130057811737,
+      "num_tokens": 46556203.0,
+      "step": 2570
+    },
+    {
+      "entropy": 1.0693527460098267,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.3508682250976562,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.055162906646728516,
+      "mean_token_accuracy": 0.9794130921363831,
+      "num_tokens": 46737287.0,
+      "step": 2580
+    },
+    {
+      "entropy": 1.078024983406067,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 1.1582329273223877,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.05406954288482666,
+      "mean_token_accuracy": 0.9794795215129852,
+      "num_tokens": 46918363.0,
+      "step": 2590
+    },
+    {
+      "entropy": 1.0643377661705018,
+      "epoch": 3.499327052489906,
+      "grad_norm": 1.429380178451538,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.05309759378433228,
+      "mean_token_accuracy": 0.9806221067905426,
+      "num_tokens": 47099549.0,
+      "step": 2600
+    },
+    {
+      "entropy": 1.068834125995636,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.236444354057312,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.05395760536193848,
+      "mean_token_accuracy": 0.9798199534416199,
+      "num_tokens": 47280430.0,
+      "step": 2610
+    },
+    {
+      "entropy": 1.0729689002037048,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 1.32350754737854,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.05501353740692139,
+      "mean_token_accuracy": 0.9792946577072144,
+      "num_tokens": 47461390.0,
+      "step": 2620
+    },
+    {
+      "entropy": 1.0766199231147766,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2348153591156006,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.05568041205406189,
+      "mean_token_accuracy": 0.9796942174434662,
+      "num_tokens": 47642662.0,
+      "step": 2630
+    },
+    {
+      "entropy": 1.0804976940155029,
+      "epoch": 3.553162853297443,
+      "grad_norm": 1.1175553798675537,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.050679147243499756,
+      "mean_token_accuracy": 0.9814749479293823,
+      "num_tokens": 47824309.0,
+      "step": 2640
+    },
+    {
+      "entropy": 1.0689257264137269,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.6297036409378052,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.054615461826324464,
+      "mean_token_accuracy": 0.9796478092670441,
+      "num_tokens": 48005982.0,
+      "step": 2650
+    },
+    {
+      "entropy": 1.0696970105171204,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2477251291275024,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.05287024974822998,
+      "mean_token_accuracy": 0.979641854763031,
+      "num_tokens": 48187048.0,
+      "step": 2660
+    },
+    {
+      "entropy": 1.0800267577171325,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.3109506368637085,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.054475915431976316,
+      "mean_token_accuracy": 0.9794476389884949,
+      "num_tokens": 48368447.0,
+      "step": 2670
+    },
+    {
+      "entropy": 1.0848782062530518,
+      "epoch": 3.60699865410498,
+      "grad_norm": 1.2147419452667236,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.05212314128875732,
+      "mean_token_accuracy": 0.9798501491546631,
+      "num_tokens": 48549144.0,
+      "step": 2680
+    },
+    {
+      "entropy": 1.0681315302848815,
+      "epoch": 3.620457604306864,
+      "grad_norm": 1.3249231576919556,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.053503894805908205,
+      "mean_token_accuracy": 0.9800095915794372,
+      "num_tokens": 48731059.0,
+      "step": 2690
+    },
+    {
+      "entropy": 1.0629405975341797,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.2435307502746582,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.0541129469871521,
+      "mean_token_accuracy": 0.9795375943183899,
+      "num_tokens": 48912283.0,
+      "step": 2700
+    },
+    {
+      "entropy": 1.0608446598052979,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.3816258907318115,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.054210084676742556,
+      "mean_token_accuracy": 0.9793854176998138,
+      "num_tokens": 49094236.0,
+      "step": 2710
+    },
+    {
+      "entropy": 1.0596249461174012,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.132426381111145,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.05496481657028198,
+      "mean_token_accuracy": 0.9798576831817627,
+      "num_tokens": 49275540.0,
+      "step": 2720
+    },
+    {
+      "entropy": 1.060381281375885,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.4120980501174927,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.05694271326065063,
+      "mean_token_accuracy": 0.9788619399070739,
+      "num_tokens": 49456684.0,
+      "step": 2730
+    },
+    {
+      "entropy": 1.0594112157821656,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.428120493888855,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.05274912118911743,
+      "mean_token_accuracy": 0.9802964627742767,
+      "num_tokens": 49638099.0,
+      "step": 2740
+    },
+    {
+      "entropy": 1.0630932807922364,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.4599305391311646,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.057349550724029544,
+      "mean_token_accuracy": 0.9788366615772247,
+      "num_tokens": 49819751.0,
+      "step": 2750
+    },
+    {
+      "entropy": 1.0632095456123352,
+      "epoch": 3.714670255720054,
+      "grad_norm": 1.8489277362823486,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.053570222854614255,
+      "mean_token_accuracy": 0.980079036951065,
+      "num_tokens": 50000623.0,
+      "step": 2760
+    },
+    {
+      "entropy": 1.0583208560943604,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.4855526685714722,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.05508332252502442,
+      "mean_token_accuracy": 0.9789698243141174,
+      "num_tokens": 50181540.0,
+      "step": 2770
+    },
+    {
+      "entropy": 1.0690102458000184,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 1.553468942642212,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.05684031248092651,
+      "mean_token_accuracy": 0.9776920318603516,
+      "num_tokens": 50363406.0,
+      "step": 2780
+    },
+    {
+      "entropy": 1.0651323795318604,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 1.2951313257217407,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.055989408493041994,
+      "mean_token_accuracy": 0.978317403793335,
+      "num_tokens": 50544599.0,
+      "step": 2790
+    },
+    {
+      "entropy": 1.0571947336196899,
+      "epoch": 3.768506056527591,
+      "grad_norm": 1.1358306407928467,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.052672064304351805,
+      "mean_token_accuracy": 0.9797160506248475,
+      "num_tokens": 50726664.0,
+      "step": 2800
+    },
+    {
+      "entropy": 1.0475606441497802,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.4842852354049683,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.05601704120635986,
+      "mean_token_accuracy": 0.9794813454151153,
+      "num_tokens": 50907903.0,
+      "step": 2810
+    },
+    {
+      "entropy": 1.053141224384308,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.3320889472961426,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.05435909032821655,
+      "mean_token_accuracy": 0.9792923092842102,
+      "num_tokens": 51088734.0,
+      "step": 2820
+    },
+    {
+      "entropy": 1.0518054842948914,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.140783667564392,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.05576471090316772,
+      "mean_token_accuracy": 0.9789255380630493,
+      "num_tokens": 51269924.0,
+      "step": 2830
+    },
+    {
+      "entropy": 1.0490313410758971,
+      "epoch": 3.822341857335128,
+      "grad_norm": 1.5749891996383667,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.05655611753463745,
+      "mean_token_accuracy": 0.9785366714000702,
+      "num_tokens": 51450406.0,
+      "step": 2840
+    },
+    {
+      "entropy": 1.0508960008621215,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 1.2535520792007446,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.05710798501968384,
+      "mean_token_accuracy": 0.978301590681076,
+      "num_tokens": 51630630.0,
+      "step": 2850
+    },
+    {
+      "entropy": 1.0434152960777283,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.6045477390289307,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.0545048713684082,
+      "mean_token_accuracy": 0.9785818517208099,
+      "num_tokens": 51811252.0,
+      "step": 2860
+    },
+    {
+      "entropy": 1.046581244468689,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.2737947702407837,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.056829071044921874,
+      "mean_token_accuracy": 0.9779990613460541,
+      "num_tokens": 51992578.0,
+      "step": 2870
+    },
+    {
+      "entropy": 1.0562566757202148,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.2364535331726074,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.054702204465866086,
+      "mean_token_accuracy": 0.9796520650386811,
+      "num_tokens": 52173507.0,
+      "step": 2880
+    },
+    {
+      "entropy": 1.0444288849830627,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.2663625478744507,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.05222170352935791,
+      "mean_token_accuracy": 0.9805905818939209,
+      "num_tokens": 52354669.0,
+      "step": 2890
+    },
+    {
+      "entropy": 1.0429784655570984,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.2072714567184448,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.05404345989227295,
+      "mean_token_accuracy": 0.9802968561649322,
+      "num_tokens": 52535334.0,
+      "step": 2900
+    },
+    {
+      "entropy": 1.045020854473114,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.3894519805908203,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.05601906776428223,
+      "mean_token_accuracy": 0.9779200375080108,
+      "num_tokens": 52715952.0,
+      "step": 2910
+    },
+    {
+      "entropy": 1.0439318537712097,
+      "epoch": 3.930013458950202,
+      "grad_norm": 1.4112615585327148,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.05345791578292847,
+      "mean_token_accuracy": 0.9796200633049011,
+      "num_tokens": 52897155.0,
+      "step": 2920
+    },
+    {
+      "entropy": 1.0361477971076964,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 1.3158206939697266,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.05547314286231995,
+      "mean_token_accuracy": 0.9788706004619598,
+      "num_tokens": 53078916.0,
+      "step": 2930
+    },
+    {
+      "entropy": 1.0427968740463256,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2565301656723022,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.05601854920387268,
+      "mean_token_accuracy": 0.9791704952716828,
+      "num_tokens": 53258884.0,
+      "step": 2940
+    },
+    {
+      "entropy": 1.0374909162521362,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.4639637470245361,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.054512780904769895,
+      "mean_token_accuracy": 0.9793699860572815,
+      "num_tokens": 53440481.0,
+      "step": 2950
+    },
+    {
+      "entropy": 1.0511455059051513,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.242073655128479,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.055217862129211426,
+      "mean_token_accuracy": 0.9801203072071075,
+      "num_tokens": 53621142.0,
+      "step": 2960
+    },
+    {
+      "entropy": 1.0518356442451477,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.1518405675888062,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.05533761382102966,
+      "mean_token_accuracy": 0.9789008319377899,
+      "num_tokens": 53802588.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 1.0548353901334628,
+      "eval_loss": 0.13498826324939728,
+      "eval_mean_token_accuracy": 0.9507886888874564,
+      "eval_num_tokens": 53838633.0,
+      "eval_runtime": 12.6809,
+      "eval_samples_per_second": 394.295,
+      "eval_steps_per_second": 12.381,
+      "step": 2972
+    },
+    {
+      "entropy": 1.043621039390564,
+      "epoch": 4.010767160161508,
+      "grad_norm": 1.1016159057617188,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.03793205618858338,
+      "mean_token_accuracy": 0.9870165169239045,
+      "num_tokens": 53983430.0,
+      "step": 2980
+    },
+    {
+      "entropy": 1.0226780295372009,
+      "epoch": 4.024226110363392,
+      "grad_norm": 1.240837812423706,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.03402446508407593,
+      "mean_token_accuracy": 0.9883724093437195,
+      "num_tokens": 54164955.0,
+      "step": 2990
+    },
+    {
+      "entropy": 1.0194039583206176,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.1843734979629517,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.035861504077911374,
+      "mean_token_accuracy": 0.9872536063194275,
+      "num_tokens": 54346968.0,
+      "step": 3000
+    },
+    {
+      "entropy": 1.029161846637726,
+      "epoch": 4.05114401076716,
+      "grad_norm": 1.285415768623352,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.034106379747390746,
+      "mean_token_accuracy": 0.9882839441299438,
+      "num_tokens": 54528019.0,
+      "step": 3010
+    },
+    {
+      "entropy": 1.031964898109436,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4622068405151367,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.032776829600334165,
+      "mean_token_accuracy": 0.9882788896560669,
+      "num_tokens": 54708148.0,
+      "step": 3020
+    },
+    {
+      "entropy": 1.0282336235046388,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.1071696281433105,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.03698182702064514,
+      "mean_token_accuracy": 0.9865448236465454,
+      "num_tokens": 54888894.0,
+      "step": 3030
+    },
+    {
+      "entropy": 1.0233591556549073,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.3671438694000244,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.03386759161949158,
+      "mean_token_accuracy": 0.9879629850387573,
+      "num_tokens": 55069872.0,
+      "step": 3040
+    },
+    {
+      "entropy": 1.0250957012176514,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.1421459913253784,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.035586753487586976,
+      "mean_token_accuracy": 0.9869713068008423,
+      "num_tokens": 55250605.0,
+      "step": 3050
+    },
+    {
+      "entropy": 1.016488790512085,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.3857697248458862,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.039323312044143674,
+      "mean_token_accuracy": 0.986032885313034,
+      "num_tokens": 55431737.0,
+      "step": 3060
+    },
+    {
+      "entropy": 1.0130115926265717,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.1797211170196533,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.03508211970329285,
+      "mean_token_accuracy": 0.9872302651405335,
+      "num_tokens": 55612917.0,
+      "step": 3070
+    },
+    {
+      "entropy": 1.023509919643402,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.1484259366989136,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.0338476300239563,
+      "mean_token_accuracy": 0.9883188247680664,
+      "num_tokens": 55793687.0,
+      "step": 3080
+    },
+    {
+      "entropy": 1.0193628787994384,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.3184833526611328,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.03667713701725006,
+      "mean_token_accuracy": 0.9869367897510528,
+      "num_tokens": 55974723.0,
+      "step": 3090
+    },
+    {
+      "entropy": 1.013669216632843,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.2176531553268433,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.035006949305534364,
+      "mean_token_accuracy": 0.9868299901485443,
+      "num_tokens": 56156568.0,
+      "step": 3100
+    },
+    {
+      "entropy": 1.0072043538093567,
+      "epoch": 4.185733512786003,
+      "grad_norm": 1.4346243143081665,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.03560284972190857,
+      "mean_token_accuracy": 0.9869017779827118,
+      "num_tokens": 56337896.0,
+      "step": 3110
+    },
+    {
+      "entropy": 1.0102415680885315,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.2328612804412842,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.03608990907669067,
+      "mean_token_accuracy": 0.9874296009540557,
+      "num_tokens": 56519102.0,
+      "step": 3120
+    },
+    {
+      "entropy": 1.015284490585327,
+      "epoch": 4.212651413189771,
+      "grad_norm": 1.4385327100753784,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.03637963235378265,
+      "mean_token_accuracy": 0.986739182472229,
+      "num_tokens": 56699936.0,
+      "step": 3130
+    },
+    {
+      "entropy": 1.0054036319255828,
+      "epoch": 4.226110363391656,
+      "grad_norm": 1.1129945516586304,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.034222981333732604,
+      "mean_token_accuracy": 0.9883054196834564,
+      "num_tokens": 56882308.0,
+      "step": 3140
+    },
+    {
+      "entropy": 1.0061798334121703,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1598492860794067,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.03461446762084961,
+      "mean_token_accuracy": 0.9873203098773956,
+      "num_tokens": 57063770.0,
+      "step": 3150
+    },
+    {
+      "entropy": 1.0039719343185425,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.322458267211914,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.03701616525650024,
+      "mean_token_accuracy": 0.9862136900424957,
+      "num_tokens": 57244448.0,
+      "step": 3160
+    },
+    {
+      "entropy": 1.0048979759216308,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 1.570866584777832,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.036290496587753296,
+      "mean_token_accuracy": 0.9868158757686615,
+      "num_tokens": 57425448.0,
+      "step": 3170
+    },
+    {
+      "entropy": 1.0145976901054383,
+      "epoch": 4.279946164199193,
+      "grad_norm": 1.103385329246521,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.03472789227962494,
+      "mean_token_accuracy": 0.987658828496933,
+      "num_tokens": 57606369.0,
+      "step": 3180
+    },
+    {
+      "entropy": 1.0186771035194397,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.2541930675506592,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.03532317280769348,
+      "mean_token_accuracy": 0.9874464809894562,
+      "num_tokens": 57787685.0,
+      "step": 3190
+    },
+    {
+      "entropy": 1.0198698043823242,
+      "epoch": 4.306864064602961,
+      "grad_norm": 1.1590384244918823,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.036364412307739256,
+      "mean_token_accuracy": 0.9864933013916015,
+      "num_tokens": 57968903.0,
+      "step": 3200
+    },
+    {
+      "entropy": 1.0211624503135681,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.7207615375518799,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.03790584802627563,
+      "mean_token_accuracy": 0.9866538405418396,
+      "num_tokens": 58149469.0,
+      "step": 3210
+    },
+    {
+      "entropy": 1.0185290336608888,
+      "epoch": 4.33378196500673,
+      "grad_norm": 1.3042566776275635,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.03802136182785034,
+      "mean_token_accuracy": 0.9859855055809021,
+      "num_tokens": 58330867.0,
+      "step": 3220
+    },
+    {
+      "entropy": 1.0105828583240508,
+      "epoch": 4.347240915208614,
+      "grad_norm": 1.2536448240280151,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.034687772393226624,
+      "mean_token_accuracy": 0.9878224015235901,
+      "num_tokens": 58512529.0,
+      "step": 3230
+    },
+    {
+      "entropy": 1.0021630108356476,
+      "epoch": 4.360699865410498,
+      "grad_norm": 1.3816099166870117,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.035324835777282716,
+      "mean_token_accuracy": 0.9872316241264343,
+      "num_tokens": 58693573.0,
+      "step": 3240
+    },
+    {
+      "entropy": 1.007886278629303,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 1.3121757507324219,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.034709393978118896,
+      "mean_token_accuracy": 0.9877420663833618,
+      "num_tokens": 58874107.0,
+      "step": 3250
+    },
+    {
+      "entropy": 1.005664885044098,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.429775357246399,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.03714964389801025,
+      "mean_token_accuracy": 0.9861079812049866,
+      "num_tokens": 59054533.0,
+      "step": 3260
+    },
+    {
+      "entropy": 1.0047886371612549,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.2385882139205933,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.036937403678894046,
+      "mean_token_accuracy": 0.9863442242145538,
+      "num_tokens": 59236331.0,
+      "step": 3270
+    },
+    {
+      "entropy": 1.0209275126457213,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.476168155670166,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.03463808596134186,
+      "mean_token_accuracy": 0.9872730135917663,
+      "num_tokens": 59417193.0,
+      "step": 3280
+    },
+    {
+      "entropy": 1.021347463130951,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 1.4710102081298828,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.038167804479599,
+      "mean_token_accuracy": 0.986047238111496,
+      "num_tokens": 59598284.0,
+      "step": 3290
+    },
+    {
+      "entropy": 1.0282366394996643,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1043860912322998,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.036692625284194945,
+      "mean_token_accuracy": 0.9862713575363159,
+      "num_tokens": 59779329.0,
+      "step": 3300
+    },
+    {
+      "entropy": 1.021342945098877,
+      "epoch": 4.454912516823688,
+      "grad_norm": 1.1742897033691406,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.03571464121341705,
+      "mean_token_accuracy": 0.987505567073822,
+      "num_tokens": 59961340.0,
+      "step": 3310
+    },
+    {
+      "entropy": 1.0153375625610352,
+      "epoch": 4.468371467025572,
+      "grad_norm": 1.1321660280227661,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.036007928848266604,
+      "mean_token_accuracy": 0.9870990335941314,
+      "num_tokens": 60142647.0,
+      "step": 3320
+    },
+    {
+      "entropy": 1.0099608540534972,
+      "epoch": 4.481830417227457,
+      "grad_norm": 1.084017276763916,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.03562799394130707,
+      "mean_token_accuracy": 0.9865496158599854,
+      "num_tokens": 60324151.0,
+      "step": 3330
+    },
+    {
+      "entropy": 1.0173812866210938,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2779732942581177,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.03556201159954071,
+      "mean_token_accuracy": 0.9875519394874572,
+      "num_tokens": 60505023.0,
+      "step": 3340
+    },
+    {
+      "entropy": 1.0236618638038635,
+      "epoch": 4.508748317631225,
+      "grad_norm": 1.5285946130752563,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.03721941411495209,
+      "mean_token_accuracy": 0.9865630626678467,
+      "num_tokens": 60685921.0,
+      "step": 3350
+    },
+    {
+      "entropy": 1.0262888669967651,
+      "epoch": 4.522207267833109,
+      "grad_norm": 1.131675362586975,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.03602397441864014,
+      "mean_token_accuracy": 0.9864337563514709,
+      "num_tokens": 60867240.0,
+      "step": 3360
+    },
+    {
+      "entropy": 1.0292576909065247,
+      "epoch": 4.535666218034994,
+      "grad_norm": 1.5156011581420898,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.03973098993301392,
+      "mean_token_accuracy": 0.9855743169784545,
+      "num_tokens": 61047708.0,
+      "step": 3370
+    },
+    {
+      "entropy": 1.010222613811493,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0979945659637451,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.03533022105693817,
+      "mean_token_accuracy": 0.9869906663894653,
+      "num_tokens": 61229247.0,
+      "step": 3380
+    },
+    {
+      "entropy": 1.004716032743454,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.3082596063613892,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.033983749151229856,
+      "mean_token_accuracy": 0.9877199769020081,
+      "num_tokens": 61411059.0,
+      "step": 3390
+    },
+    {
+      "entropy": 1.0031994462013245,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.3033607006072998,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.03607205748558044,
+      "mean_token_accuracy": 0.9866301774978637,
+      "num_tokens": 61592538.0,
+      "step": 3400
+    },
+    {
+      "entropy": 1.0113391518592834,
+      "epoch": 4.589502018842531,
+      "grad_norm": 1.3490946292877197,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.03906603455543518,
+      "mean_token_accuracy": 0.9854673087596894,
+      "num_tokens": 61773613.0,
+      "step": 3410
+    },
+    {
+      "entropy": 1.0076410889625549,
+      "epoch": 4.602960969044415,
+      "grad_norm": 1.1719361543655396,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.037575924396514894,
+      "mean_token_accuracy": 0.9867486238479615,
+      "num_tokens": 61955690.0,
+      "step": 3420
+    },
+    {
+      "entropy": 1.0200914144515991,
+      "epoch": 4.616419919246299,
+      "grad_norm": 1.1205360889434814,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.03691520690917969,
+      "mean_token_accuracy": 0.9865606665611267,
+      "num_tokens": 62136047.0,
+      "step": 3430
+    },
+    {
+      "entropy": 1.0141666173934936,
+      "epoch": 4.629878869448183,
+      "grad_norm": 1.1805763244628906,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.035447508096694946,
+      "mean_token_accuracy": 0.9873166739940643,
+      "num_tokens": 62317192.0,
+      "step": 3440
+    },
+    {
+      "entropy": 1.0067960560321807,
+      "epoch": 4.643337819650068,
+      "grad_norm": 1.0286436080932617,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.03580200672149658,
+      "mean_token_accuracy": 0.9870070457458496,
+      "num_tokens": 62498860.0,
+      "step": 3450
+    },
+    {
+      "entropy": 1.0088169515132903,
+      "epoch": 4.656796769851952,
+      "grad_norm": 1.3638492822647095,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.03756250143051147,
+      "mean_token_accuracy": 0.9861987411975861,
+      "num_tokens": 62679272.0,
+      "step": 3460
+    },
+    {
+      "entropy": 1.0098722338676454,
+      "epoch": 4.670255720053836,
+      "grad_norm": 1.649560809135437,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.037615427374839784,
+      "mean_token_accuracy": 0.9864691913127899,
+      "num_tokens": 62860715.0,
+      "step": 3470
+    },
+    {
+      "entropy": 1.0174571394920349,
+      "epoch": 4.68371467025572,
+      "grad_norm": 1.2106678485870361,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.037718603014945985,
+      "mean_token_accuracy": 0.9855880260467529,
+      "num_tokens": 63041029.0,
+      "step": 3480
+    },
+    {
+      "entropy": 1.0129252314567565,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.1571707725524902,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.03724310994148254,
+      "mean_token_accuracy": 0.9865262687206269,
+      "num_tokens": 63222379.0,
+      "step": 3490
+    },
+    {
+      "entropy": 1.0124369859695435,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.0599361658096313,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.037783673405647276,
+      "mean_token_accuracy": 0.9859567046165466,
+      "num_tokens": 63403328.0,
+      "step": 3500
+    },
+    {
+      "entropy": 1.0041921436786652,
+      "epoch": 4.724091520861373,
+      "grad_norm": 1.2054452896118164,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.03585146069526672,
+      "mean_token_accuracy": 0.9868905484676361,
+      "num_tokens": 63584598.0,
+      "step": 3510
+    },
+    {
+      "entropy": 1.0015427589416503,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.5075141191482544,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.037428200244903564,
+      "mean_token_accuracy": 0.9867083311080933,
+      "num_tokens": 63766203.0,
+      "step": 3520
+    },
+    {
+      "entropy": 1.0039161920547486,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.310390830039978,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.037494581937789914,
+      "mean_token_accuracy": 0.9862947940826416,
+      "num_tokens": 63947714.0,
+      "step": 3530
+    },
+    {
+      "entropy": 1.0049733996391297,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.2912139892578125,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.03803396224975586,
+      "mean_token_accuracy": 0.9860390484333038,
+      "num_tokens": 64128798.0,
+      "step": 3540
+    },
+    {
+      "entropy": 1.0127450585365296,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.001133680343628,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0351622611284256,
+      "mean_token_accuracy": 0.9875370979309082,
+      "num_tokens": 64309577.0,
+      "step": 3550
+    },
+    {
+      "entropy": 1.0070195615291595,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 1.1870886087417603,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.036404851078987124,
+      "mean_token_accuracy": 0.9871752023696899,
+      "num_tokens": 64490707.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.9983750462532044,
+      "epoch": 4.804845222072679,
+      "grad_norm": 1.2374205589294434,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.03696496784687042,
+      "mean_token_accuracy": 0.9865773975849151,
+      "num_tokens": 64672104.0,
+      "step": 3570
+    },
+    {
+      "entropy": 1.0151559710502625,
+      "epoch": 4.818304172274562,
+      "grad_norm": 1.2001149654388428,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.03725983202457428,
+      "mean_token_accuracy": 0.9862341761589051,
+      "num_tokens": 64853353.0,
+      "step": 3580
+    },
+    {
+      "entropy": 1.0136774778366089,
+      "epoch": 4.831763122476447,
+      "grad_norm": 1.0128737688064575,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.03615986704826355,
+      "mean_token_accuracy": 0.9872736394405365,
+      "num_tokens": 65035009.0,
+      "step": 3590
+    },
+    {
+      "entropy": 1.0082051515579225,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.2552846670150757,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.03561249375343323,
+      "mean_token_accuracy": 0.9868618845939636,
+      "num_tokens": 65215981.0,
+      "step": 3600
+    },
+    {
+      "entropy": 1.0209908723831176,
+      "epoch": 4.858681022880216,
+      "grad_norm": 1.1914079189300537,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.03721153438091278,
+      "mean_token_accuracy": 0.9862311720848084,
+      "num_tokens": 65397088.0,
+      "step": 3610
+    },
+    {
+      "entropy": 1.014677894115448,
+      "epoch": 4.872139973082099,
+      "grad_norm": 1.3584535121917725,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.03897300064563751,
+      "mean_token_accuracy": 0.9851836740970612,
+      "num_tokens": 65578262.0,
+      "step": 3620
+    },
+    {
+      "entropy": 1.0207050919532776,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.2754287719726562,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.037926393747329715,
+      "mean_token_accuracy": 0.9859795570373535,
+      "num_tokens": 65758956.0,
+      "step": 3630
+    },
+    {
+      "entropy": 1.008733856678009,
+      "epoch": 4.899057873485868,
+      "grad_norm": 1.0815879106521606,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.036518561840057376,
+      "mean_token_accuracy": 0.9870649397373199,
+      "num_tokens": 65939970.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.9995620250701904,
+      "epoch": 4.912516823687753,
+      "grad_norm": 1.1773526668548584,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.035269084572792056,
+      "mean_token_accuracy": 0.9873197913169861,
+      "num_tokens": 66121045.0,
+      "step": 3650
+    },
+    {
+      "entropy": 1.0036249816417695,
+      "epoch": 4.925975773889636,
+      "grad_norm": 1.0941131114959717,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.0353422611951828,
+      "mean_token_accuracy": 0.9870408117771149,
+      "num_tokens": 66302324.0,
+      "step": 3660
+    },
+    {
+      "entropy": 1.004719352722168,
+      "epoch": 4.939434724091521,
+      "grad_norm": 1.5107269287109375,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.03743018507957459,
+      "mean_token_accuracy": 0.9857985079288483,
+      "num_tokens": 66483916.0,
+      "step": 3670
+    },
+    {
+      "entropy": 1.0076164484024048,
+      "epoch": 4.952893674293405,
+      "grad_norm": 1.3877427577972412,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.03857019543647766,
+      "mean_token_accuracy": 0.9856370270252228,
+      "num_tokens": 66664390.0,
+      "step": 3680
+    },
+    {
+      "entropy": 1.015517818927765,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1693767309188843,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.03634478151798248,
+      "mean_token_accuracy": 0.9868648707866668,
+      "num_tokens": 66845574.0,
+      "step": 3690
+    },
+    {
+      "entropy": 1.011656105518341,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.082209587097168,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.03513925671577454,
+      "mean_token_accuracy": 0.9872389912605286,
+      "num_tokens": 67027557.0,
+      "step": 3700
+    },
+    {
+      "entropy": 1.0161120653152467,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.266615629196167,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.0379630446434021,
+      "mean_token_accuracy": 0.9865313708782196,
+      "num_tokens": 67208088.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 1.0219634957374282,
+      "eval_loss": 0.14612138271331787,
+      "eval_mean_token_accuracy": 0.9504923186484416,
+      "eval_num_tokens": 67298266.0,
+      "eval_runtime": 12.6802,
+      "eval_samples_per_second": 394.317,
+      "eval_steps_per_second": 12.382,
+      "step": 3715
+    },
+    {
+      "entropy": 1.019774079322815,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8909980654716492,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.02882351279258728,
+      "mean_token_accuracy": 0.9900537073612213,
+      "num_tokens": 67388858.0,
+      "step": 3720
+    },
+    {
+      "entropy": 1.0057315409183503,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9996656775474548,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.02233442962169647,
+      "mean_token_accuracy": 0.9922977447509765,
+      "num_tokens": 67570084.0,
+      "step": 3730
+    },
+    {
+      "entropy": 1.0058627128601074,
+      "epoch": 5.03364737550471,
+      "grad_norm": 1.0265147686004639,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.021967282891273497,
+      "mean_token_accuracy": 0.9923401355743409,
+      "num_tokens": 67751587.0,
+      "step": 3740
+    },
+    {
+      "entropy": 1.0083157896995545,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 1.0139433145523071,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.022531555593013765,
+      "mean_token_accuracy": 0.9930226564407348,
+      "num_tokens": 67933269.0,
+      "step": 3750
+    },
+    {
+      "entropy": 1.0119245886802672,
+      "epoch": 5.060565275908479,
+      "grad_norm": 1.0655452013015747,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.021685637533664703,
+      "mean_token_accuracy": 0.9925794541835785,
+      "num_tokens": 68113971.0,
+      "step": 3760
+    },
+    {
+      "entropy": 1.0082412719726563,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.268751859664917,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.021513286232948303,
+      "mean_token_accuracy": 0.9927303433418274,
+      "num_tokens": 68295632.0,
+      "step": 3770
+    },
+    {
+      "entropy": 1.012417995929718,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.9899319410324097,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.02352543771266937,
+      "mean_token_accuracy": 0.992375648021698,
+      "num_tokens": 68476794.0,
+      "step": 3780
+    },
+    {
+      "entropy": 1.0069364666938783,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 1.1964375972747803,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.021382461488246917,
+      "mean_token_accuracy": 0.9928456962108612,
+      "num_tokens": 68658249.0,
+      "step": 3790
+    },
+    {
+      "entropy": 1.0040670156478881,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0445663928985596,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.02207571566104889,
+      "mean_token_accuracy": 0.9927122712135314,
+      "num_tokens": 68840525.0,
+      "step": 3800
+    },
+    {
+      "entropy": 1.0083145201206207,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.8981389403343201,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.022873447835445405,
+      "mean_token_accuracy": 0.9918712139129638,
+      "num_tokens": 69021896.0,
+      "step": 3810
+    },
+    {
+      "entropy": 1.0109572768211366,
+      "epoch": 5.141318977119784,
+      "grad_norm": 1.250670313835144,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.023577621579170226,
+      "mean_token_accuracy": 0.9919484734535218,
+      "num_tokens": 69203112.0,
+      "step": 3820
+    },
+    {
+      "entropy": 1.0164754271507264,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.3053207397460938,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.02436322420835495,
+      "mean_token_accuracy": 0.9921659469604492,
+      "num_tokens": 69384193.0,
+      "step": 3830
+    },
+    {
+      "entropy": 1.0034007966518401,
+      "epoch": 5.168236877523553,
+      "grad_norm": 1.4878787994384766,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.023177918791770936,
+      "mean_token_accuracy": 0.9919318974018096,
+      "num_tokens": 69565602.0,
+      "step": 3840
+    },
+    {
+      "entropy": 1.0074714839458465,
+      "epoch": 5.181695827725437,
+      "grad_norm": 1.0512391328811646,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.023409275710582732,
+      "mean_token_accuracy": 0.9918898940086365,
+      "num_tokens": 69746242.0,
+      "step": 3850
+    },
+    {
+      "entropy": 1.003154468536377,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.971832811832428,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.023223266005516052,
+      "mean_token_accuracy": 0.9920656085014343,
+      "num_tokens": 69927227.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.999264246225357,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.9266519546508789,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.02454002797603607,
+      "mean_token_accuracy": 0.9914694905281067,
+      "num_tokens": 70108580.0,
+      "step": 3870
+    },
+    {
+      "entropy": 1.002640813589096,
+      "epoch": 5.22207267833109,
+      "grad_norm": 1.5331412553787231,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.02458159625530243,
+      "mean_token_accuracy": 0.9919124245643616,
+      "num_tokens": 70288960.0,
+      "step": 3880
+    },
+    {
+      "entropy": 1.007998287677765,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.3164633512496948,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.02421087771654129,
+      "mean_token_accuracy": 0.9917516648769379,
+      "num_tokens": 70470266.0,
+      "step": 3890
+    },
+    {
+      "entropy": 1.0089037418365479,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1788280010223389,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.023224952816963195,
+      "mean_token_accuracy": 0.9921185493469238,
+      "num_tokens": 70650975.0,
+      "step": 3900
+    },
+    {
+      "entropy": 1.001932644844055,
+      "epoch": 5.262449528936743,
+      "grad_norm": 1.2484381198883057,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.02366664707660675,
+      "mean_token_accuracy": 0.9917596876621246,
+      "num_tokens": 70832549.0,
+      "step": 3910
+    },
+    {
+      "entropy": 1.004903507232666,
+      "epoch": 5.275908479138627,
+      "grad_norm": 1.1852507591247559,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.02296685576438904,
+      "mean_token_accuracy": 0.9918978750705719,
+      "num_tokens": 71014296.0,
+      "step": 3920
+    },
+    {
+      "entropy": 1.004357820749283,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.9933306574821472,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.02361467629671097,
+      "mean_token_accuracy": 0.991608303785324,
+      "num_tokens": 71195526.0,
+      "step": 3930
+    },
+    {
+      "entropy": 1.0020038604736328,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.4304245710372925,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.02268100529909134,
+      "mean_token_accuracy": 0.9917695164680481,
+      "num_tokens": 71377269.0,
+      "step": 3940
+    },
+    {
+      "entropy": 1.012037169933319,
+      "epoch": 5.31628532974428,
+      "grad_norm": 1.074124813079834,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.023454257845878602,
+      "mean_token_accuracy": 0.9922087669372559,
+      "num_tokens": 71558158.0,
+      "step": 3950
+    },
+    {
+      "entropy": 1.0094320178031921,
+      "epoch": 5.329744279946164,
+      "grad_norm": 1.0919133424758911,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.022751571238040925,
+      "mean_token_accuracy": 0.9921496629714965,
+      "num_tokens": 71738847.0,
+      "step": 3960
+    },
+    {
+      "entropy": 1.0063781797885896,
+      "epoch": 5.343203230148048,
+      "grad_norm": 1.1300997734069824,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.023900465667247774,
+      "mean_token_accuracy": 0.9914113521575928,
+      "num_tokens": 71918875.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.9972042560577392,
+      "epoch": 5.356662180349932,
+      "grad_norm": 1.165519118309021,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.021918240189552306,
+      "mean_token_accuracy": 0.9926528394222259,
+      "num_tokens": 72100551.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.9954554200172424,
+      "epoch": 5.370121130551817,
+      "grad_norm": 1.342820167541504,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.024170228838920595,
+      "mean_token_accuracy": 0.9913744986057281,
+      "num_tokens": 72281855.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.9931409180164337,
+      "epoch": 5.383580080753701,
+      "grad_norm": 1.2873902320861816,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.02334785759449005,
+      "mean_token_accuracy": 0.9921288907527923,
+      "num_tokens": 72462916.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.9915881812572479,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.9761195778846741,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.02302718609571457,
+      "mean_token_accuracy": 0.9924728453159333,
+      "num_tokens": 72644666.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.9931966602802277,
+      "epoch": 5.410497981157469,
+      "grad_norm": 1.4468477964401245,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.023682811856269838,
+      "mean_token_accuracy": 0.9917364478111267,
+      "num_tokens": 72825808.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.9975718319416046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.1559572219848633,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.02247478663921356,
+      "mean_token_accuracy": 0.992560338973999,
+      "num_tokens": 73007923.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.999228036403656,
+      "epoch": 5.437415881561238,
+      "grad_norm": 1.1230353116989136,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.022295859456062318,
+      "mean_token_accuracy": 0.9922321498394012,
+      "num_tokens": 73189074.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.9897105693817139,
+      "epoch": 5.450874831763122,
+      "grad_norm": 1.413734793663025,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.02201971709728241,
+      "mean_token_accuracy": 0.9927073001861573,
+      "num_tokens": 73369918.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.9986266851425171,
+      "epoch": 5.464333781965006,
+      "grad_norm": 1.0815346240997314,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.02358766496181488,
+      "mean_token_accuracy": 0.9911752700805664,
+      "num_tokens": 73550625.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.9953943610191345,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0506970882415771,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.022177645564079286,
+      "mean_token_accuracy": 0.9923022150993347,
+      "num_tokens": 73731436.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.9954936504364014,
+      "epoch": 5.491251682368775,
+      "grad_norm": 1.4665249586105347,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.024948063492774963,
+      "mean_token_accuracy": 0.9913177132606507,
+      "num_tokens": 73912097.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.9986696183681488,
+      "epoch": 5.504710632570659,
+      "grad_norm": 1.158179759979248,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.02555338144302368,
+      "mean_token_accuracy": 0.9908900320529938,
+      "num_tokens": 74093602.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.9955579698085785,
+      "epoch": 5.518169582772543,
+      "grad_norm": 1.0024714469909668,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.023432795703411103,
+      "mean_token_accuracy": 0.9919674038887024,
+      "num_tokens": 74274664.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.994187182188034,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.9290911555290222,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.023194606602191924,
+      "mean_token_accuracy": 0.9920254707336426,
+      "num_tokens": 74456436.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.9927510738372802,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.8986559510231018,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.02163631021976471,
+      "mean_token_accuracy": 0.9927416622638703,
+      "num_tokens": 74637700.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.9877451896667481,
+      "epoch": 5.558546433378196,
+      "grad_norm": 1.0314558744430542,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.0209101140499115,
+      "mean_token_accuracy": 0.9930387079715729,
+      "num_tokens": 74819310.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.9903139114379883,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 1.1819736957550049,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.02219167649745941,
+      "mean_token_accuracy": 0.992407250404358,
+      "num_tokens": 74999907.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.9891258835792541,
+      "epoch": 5.585464333781965,
+      "grad_norm": 1.0412501096725464,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.02201283872127533,
+      "mean_token_accuracy": 0.992091816663742,
+      "num_tokens": 75181189.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.9870602369308472,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0461134910583496,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.023040324449539185,
+      "mean_token_accuracy": 0.992098081111908,
+      "num_tokens": 75362769.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.9888905882835388,
+      "epoch": 5.612382234185733,
+      "grad_norm": 1.086546540260315,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.021401138603687288,
+      "mean_token_accuracy": 0.992308247089386,
+      "num_tokens": 75544519.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.9940549910068512,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 1.3897169828414917,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.023297235369682312,
+      "mean_token_accuracy": 0.9920303761959076,
+      "num_tokens": 75725651.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.9897378265857697,
+      "epoch": 5.639300134589502,
+      "grad_norm": 1.283887267112732,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.024343529343605043,
+      "mean_token_accuracy": 0.9916397035121918,
+      "num_tokens": 75907036.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.9989661812782288,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.9090050458908081,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.022062870860099792,
+      "mean_token_accuracy": 0.9921731591224671,
+      "num_tokens": 76088248.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.9966195583343506,
+      "epoch": 5.66621803499327,
+      "grad_norm": 1.2091970443725586,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.023516717553138732,
+      "mean_token_accuracy": 0.9918879508972168,
+      "num_tokens": 76269704.0,
+      "step": 4210
+    },
+    {
+      "entropy": 1.0081237196922301,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 1.0648062229156494,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.021994276344776152,
+      "mean_token_accuracy": 0.9925018906593323,
+      "num_tokens": 76451416.0,
+      "step": 4220
+    },
+    {
+      "entropy": 1.0061897933483124,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.188748836517334,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.022657927870750428,
+      "mean_token_accuracy": 0.9919365108013153,
+      "num_tokens": 76632827.0,
+      "step": 4230
+    },
+    {
+      "entropy": 1.0024715662002563,
+      "epoch": 5.706594885598923,
+      "grad_norm": 1.2140473127365112,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.02203955501317978,
+      "mean_token_accuracy": 0.9925529718399048,
+      "num_tokens": 76813534.0,
+      "step": 4240
+    },
+    {
+      "entropy": 1.0011352360248567,
+      "epoch": 5.720053835800807,
+      "grad_norm": 1.389479398727417,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.022718487679958342,
+      "mean_token_accuracy": 0.9922778069972992,
+      "num_tokens": 76994327.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.9971556246280671,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 1.1441208124160767,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.024178403615951537,
+      "mean_token_accuracy": 0.9913869142532349,
+      "num_tokens": 77174987.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.9961654126644135,
+      "epoch": 5.746971736204576,
+      "grad_norm": 1.1283735036849976,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.023389075696468354,
+      "mean_token_accuracy": 0.9921939969062805,
+      "num_tokens": 77356043.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.9927134871482849,
+      "epoch": 5.76043068640646,
+      "grad_norm": 1.201904535293579,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.022739484906196594,
+      "mean_token_accuracy": 0.9921142935752869,
+      "num_tokens": 77536396.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.9919731795787812,
+      "epoch": 5.773889636608344,
+      "grad_norm": 1.145776391029358,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.020524097979068755,
+      "mean_token_accuracy": 0.9926942765712738,
+      "num_tokens": 77717339.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.9935499131679535,
+      "epoch": 5.787348586810229,
+      "grad_norm": 1.099414348602295,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.02070924937725067,
+      "mean_token_accuracy": 0.9928180515766144,
+      "num_tokens": 77898360.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.9875575602054596,
+      "epoch": 5.800807537012113,
+      "grad_norm": 1.1510778665542603,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.02202107310295105,
+      "mean_token_accuracy": 0.9921730697154999,
+      "num_tokens": 78079699.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.9855418741703034,
+      "epoch": 5.814266487213997,
+      "grad_norm": 1.0748379230499268,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.022809866070747375,
+      "mean_token_accuracy": 0.9919199466705322,
+      "num_tokens": 78260900.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.977304744720459,
+      "epoch": 5.827725437415881,
+      "grad_norm": 1.091854453086853,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.02268455773591995,
+      "mean_token_accuracy": 0.9918094217777252,
+      "num_tokens": 78442168.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.9767287075519562,
+      "epoch": 5.841184387617766,
+      "grad_norm": 1.1712645292282104,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.022726841270923615,
+      "mean_token_accuracy": 0.9922167181968689,
+      "num_tokens": 78623589.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.974297159910202,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.9508016109466553,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.021290253102779388,
+      "mean_token_accuracy": 0.9926727712154388,
+      "num_tokens": 78804892.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.9795810282230377,
+      "epoch": 5.868102288021534,
+      "grad_norm": 1.3127822875976562,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.022329577803611757,
+      "mean_token_accuracy": 0.9927741348743438,
+      "num_tokens": 78986204.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.9879997313022614,
+      "epoch": 5.881561238223418,
+      "grad_norm": 1.0200371742248535,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.022206541895866395,
+      "mean_token_accuracy": 0.9921830892562866,
+      "num_tokens": 79166824.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.9801069736480713,
+      "epoch": 5.895020188425303,
+      "grad_norm": 1.4769432544708252,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.02101587802171707,
+      "mean_token_accuracy": 0.9930795848369598,
+      "num_tokens": 79347228.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.9727071464061737,
+      "epoch": 5.908479138627187,
+      "grad_norm": 1.7034555673599243,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.021690081059932708,
+      "mean_token_accuracy": 0.9926194071769714,
+      "num_tokens": 79529483.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.9782697260379791,
+      "epoch": 5.921938088829071,
+      "grad_norm": 1.5307973623275757,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.023112425208091737,
+      "mean_token_accuracy": 0.9919495701789856,
+      "num_tokens": 79710571.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.9831463754177093,
+      "epoch": 5.935397039030955,
+      "grad_norm": 1.1024526357650757,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.021427598595619202,
+      "mean_token_accuracy": 0.9924166023731231,
+      "num_tokens": 79890685.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.9811958551406861,
+      "epoch": 5.94885598923284,
+      "grad_norm": 1.7247848510742188,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.02339867949485779,
+      "mean_token_accuracy": 0.9916171073913574,
+      "num_tokens": 80071331.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.9830600321292877,
+      "epoch": 5.962314939434724,
+      "grad_norm": 1.0638445615768433,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.02204108238220215,
+      "mean_token_accuracy": 0.9925127327442169,
+      "num_tokens": 80251525.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.9840480864048005,
+      "epoch": 5.975773889636608,
+      "grad_norm": 1.1205570697784424,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.020443232357501985,
+      "mean_token_accuracy": 0.9932060956954956,
+      "num_tokens": 80432001.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.9841626048088074,
+      "epoch": 5.989232839838492,
+      "grad_norm": 1.2928433418273926,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.021673919260501863,
+      "mean_token_accuracy": 0.9931873917579651,
+      "num_tokens": 80613075.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.9904457463580332,
+      "eval_loss": 0.161154642701149,
+      "eval_mean_token_accuracy": 0.9516840383505366,
+      "eval_num_tokens": 80757887.0,
+      "eval_runtime": 12.7345,
+      "eval_samples_per_second": 392.633,
+      "eval_steps_per_second": 12.329,
+      "step": 4458
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.8190318189763625e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-4458/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-4458/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-4458/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/checkpoint-5201/README.md b/spanish_L20_a100/seed_42/checkpoint-5201/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5201/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-5201/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-5201/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5201/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-5201/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-5201/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..0c61a097f8a7f2cff5fed9e6a749ba0b44cf9b88
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5201/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb754e9776c43087edfd939a239b3503568b3055dd9763d624a548adaa81fd87
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-5201/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-5201/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5201/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-5201/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-5201/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5201/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-5201/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-5201/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5201/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-5201/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-5201/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..bf583e32ce7b8567d6bd4b6260964b2aaa8ee779
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5201/trainer_state.json
@@ -0,0 +1,5311 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.0,
+  "eval_steps": 500,
+  "global_step": 5201,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    },
+    {
+      "entropy": 1.049056887626648,
+      "epoch": 1.009421265141319,
+      "grad_norm": 1.4189648628234863,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.11506924629211426,
+      "mean_token_accuracy": 0.9560760498046875,
+      "num_tokens": 13586033.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0126458764076234,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.6394226551055908,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.10137178897857665,
+      "mean_token_accuracy": 0.9604915261268616,
+      "num_tokens": 13766883.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0174824237823485,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.3608797788619995,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.10521939992904664,
+      "mean_token_accuracy": 0.9592583775520325,
+      "num_tokens": 13949410.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0318764686584472,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.1440019607543945,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.1039464831352234,
+      "mean_token_accuracy": 0.9593731820583343,
+      "num_tokens": 14131283.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0228763103485108,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.609320878982544,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.10670688152313232,
+      "mean_token_accuracy": 0.9587047874927521,
+      "num_tokens": 14312694.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0123905301094056,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.4128942489624023,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.10546157360076905,
+      "mean_token_accuracy": 0.9586312890052795,
+      "num_tokens": 14493193.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.010130923986435,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.4825445413589478,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.09921822547912598,
+      "mean_token_accuracy": 0.9621170878410339,
+      "num_tokens": 14674762.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0216347932815553,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.2503618001937866,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.10619921684265136,
+      "mean_token_accuracy": 0.9584988236427308,
+      "num_tokens": 14856424.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.022511374950409,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.3209893703460693,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.10425064563751221,
+      "mean_token_accuracy": 0.9595829844474792,
+      "num_tokens": 15037452.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.007770985364914,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.4830716848373413,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.10189464092254638,
+      "mean_token_accuracy": 0.9609117627143859,
+      "num_tokens": 15218205.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0228153944015503,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.3191078901290894,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.10298972129821778,
+      "mean_token_accuracy": 0.9599695146083832,
+      "num_tokens": 15399322.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0262911200523377,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.2050191164016724,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.10532126426696778,
+      "mean_token_accuracy": 0.9590177297592163,
+      "num_tokens": 15581668.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0124290227890014,
+      "epoch": 1.17092866756393,
+      "grad_norm": 1.3545384407043457,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.10171200037002563,
+      "mean_token_accuracy": 0.9587810635566711,
+      "num_tokens": 15762970.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.028192389011383,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.55021071434021,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.10711442232131958,
+      "mean_token_accuracy": 0.956982696056366,
+      "num_tokens": 15944164.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0354714155197144,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2573878765106201,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.10722451210021973,
+      "mean_token_accuracy": 0.9568670630455017,
+      "num_tokens": 16125049.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.034978187084198,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.4448883533477783,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.10492000579833985,
+      "mean_token_accuracy": 0.9593439042568207,
+      "num_tokens": 16306183.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0321480870246886,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.4060254096984863,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.1017595648765564,
+      "mean_token_accuracy": 0.9604197800159454,
+      "num_tokens": 16487439.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.03044091463089,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.6413203477859497,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.1032058596611023,
+      "mean_token_accuracy": 0.9589122951030731,
+      "num_tokens": 16667826.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0394478559494018,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3867346048355103,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.10721726417541504,
+      "mean_token_accuracy": 0.9581708788871766,
+      "num_tokens": 16848488.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0355630755424499,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 1.3304182291030884,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.10916879177093505,
+      "mean_token_accuracy": 0.9566857099533081,
+      "num_tokens": 17028787.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0394182324409484,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.4853013753890991,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.10515115261077881,
+      "mean_token_accuracy": 0.9584204435348511,
+      "num_tokens": 17209699.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0405551195144653,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.1959068775177002,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.10281100273132324,
+      "mean_token_accuracy": 0.9599630236625671,
+      "num_tokens": 17391070.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0196930646896363,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.2734276056289673,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.10447280406951905,
+      "mean_token_accuracy": 0.9587895989418029,
+      "num_tokens": 17572309.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0284824252128602,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.1344295740127563,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.11111283302307129,
+      "mean_token_accuracy": 0.9560421764850616,
+      "num_tokens": 17753814.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0308443427085876,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.0935126543045044,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.10474474430084228,
+      "mean_token_accuracy": 0.9583515226840973,
+      "num_tokens": 17934774.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.0283282995224,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.5144932270050049,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.104243803024292,
+      "mean_token_accuracy": 0.9581555545330047,
+      "num_tokens": 18116015.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.0355387568473815,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.276055932044983,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.10444833040237426,
+      "mean_token_accuracy": 0.9587780058383941,
+      "num_tokens": 18296690.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.0340025186538697,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.4408668279647827,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.10817761421203613,
+      "mean_token_accuracy": 0.9573222637176514,
+      "num_tokens": 18477491.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.0480108737945557,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2114640474319458,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.10477652549743652,
+      "mean_token_accuracy": 0.9591076672077179,
+      "num_tokens": 18658366.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.0303086400032044,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.3319413661956787,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.10648446083068848,
+      "mean_token_accuracy": 0.9581128239631653,
+      "num_tokens": 18839989.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0405295491218567,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.4326776266098022,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.10707225799560546,
+      "mean_token_accuracy": 0.9574429094791412,
+      "num_tokens": 19020331.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.037307357788086,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.2203212976455688,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.10811934471130372,
+      "mean_token_accuracy": 0.9569429934024811,
+      "num_tokens": 19201621.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.048385226726532,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.2597253322601318,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.10502862930297852,
+      "mean_token_accuracy": 0.9592942714691162,
+      "num_tokens": 19382819.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.0431596279144286,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.288861632347107,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.1091151475906372,
+      "mean_token_accuracy": 0.9570925652980804,
+      "num_tokens": 19563291.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.0325645685195923,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.1845893859863281,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.10389597415924072,
+      "mean_token_accuracy": 0.9585334539413453,
+      "num_tokens": 19744388.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.025480568408966,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 1.3404650688171387,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.10605425834655761,
+      "mean_token_accuracy": 0.9583600521087646,
+      "num_tokens": 19925613.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.020683228969574,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5598406791687012,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.10631901025772095,
+      "mean_token_accuracy": 0.9573191046714783,
+      "num_tokens": 20106747.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.0185366213321685,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3814117908477783,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.1074562907218933,
+      "mean_token_accuracy": 0.9571435689926148,
+      "num_tokens": 20287872.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0160030841827392,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 1.5130647420883179,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.10584371089935303,
+      "mean_token_accuracy": 0.9584454715251922,
+      "num_tokens": 20469530.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.0295212030410767,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.137473464012146,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.10609703063964844,
+      "mean_token_accuracy": 0.9590815663337707,
+      "num_tokens": 20650652.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.0236494064331054,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.4402672052383423,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.10043272972106934,
+      "mean_token_accuracy": 0.9601598799228668,
+      "num_tokens": 20832844.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0228356838226318,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.6836261749267578,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.10231537818908691,
+      "mean_token_accuracy": 0.9592978358268738,
+      "num_tokens": 21014766.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.0204253315925598,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.7377065420150757,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.10434420108795166,
+      "mean_token_accuracy": 0.9578610718250274,
+      "num_tokens": 21195755.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.0251757979393006,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.337659239768982,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.10218650102615356,
+      "mean_token_accuracy": 0.9588291764259338,
+      "num_tokens": 21376476.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.021351706981659,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.278275728225708,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.10114043951034546,
+      "mean_token_accuracy": 0.9596256852149964,
+      "num_tokens": 21558292.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.0222898125648499,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.1806602478027344,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.10140385627746581,
+      "mean_token_accuracy": 0.9601265788078308,
+      "num_tokens": 21739640.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0402090787887572,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.2378828525543213,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.10622456073760986,
+      "mean_token_accuracy": 0.9573592960834503,
+      "num_tokens": 21921425.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.0412330985069276,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3350801467895508,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.10371602773666382,
+      "mean_token_accuracy": 0.9588219523429871,
+      "num_tokens": 22102605.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.0534109234809876,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 1.4873039722442627,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.10807039737701415,
+      "mean_token_accuracy": 0.9577955365180969,
+      "num_tokens": 22283687.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0713044166564942,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.1694425344467163,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.11005204916000366,
+      "mean_token_accuracy": 0.9568802535533905,
+      "num_tokens": 22464813.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.0745116233825684,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.5362104177474976,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.10913199186325073,
+      "mean_token_accuracy": 0.955611252784729,
+      "num_tokens": 22646088.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0925905466079713,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.3478912115097046,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.10608880519866944,
+      "mean_token_accuracy": 0.9566390633583068,
+      "num_tokens": 22826677.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0899853229522705,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.238072395324707,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.10292150974273681,
+      "mean_token_accuracy": 0.9586134016513824,
+      "num_tokens": 23008177.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0695149540901183,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.541548252105713,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.10405257940292359,
+      "mean_token_accuracy": 0.9580470204353333,
+      "num_tokens": 23190301.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.0614005208015442,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.4365121126174927,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.10439364910125733,
+      "mean_token_accuracy": 0.9584310293197632,
+      "num_tokens": 23371824.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.065568709373474,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.2848029136657715,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.104488205909729,
+      "mean_token_accuracy": 0.959163922071457,
+      "num_tokens": 23552849.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.0750214815139771,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 1.280882716178894,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.10620679855346679,
+      "mean_token_accuracy": 0.9575595021247864,
+      "num_tokens": 23733408.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.077055287361145,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.2904170751571655,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.11054887771606445,
+      "mean_token_accuracy": 0.9566682338714599,
+      "num_tokens": 23914203.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.081210732460022,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.129718542098999,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.10313529968261718,
+      "mean_token_accuracy": 0.9582066476345062,
+      "num_tokens": 24094078.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.0856151461601258,
+      "epoch": 1.80349932705249,
+      "grad_norm": 1.3682607412338257,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.10383679866790771,
+      "mean_token_accuracy": 0.9592101991176605,
+      "num_tokens": 24274071.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.0931835532188416,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.3282326459884644,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.10675904750823975,
+      "mean_token_accuracy": 0.9580830633640289,
+      "num_tokens": 24454766.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.0867947936058044,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.2584236860275269,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.10147552490234375,
+      "mean_token_accuracy": 0.9604024410247802,
+      "num_tokens": 24635903.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.0726998805999757,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 1.254036545753479,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.10524414777755738,
+      "mean_token_accuracy": 0.9582872509956359,
+      "num_tokens": 24816787.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.0692258477210999,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.41594660282135,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.10340676307678223,
+      "mean_token_accuracy": 0.9590053558349609,
+      "num_tokens": 24998543.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.0772343635559083,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.385327696800232,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.10295225381851196,
+      "mean_token_accuracy": 0.9591428697109222,
+      "num_tokens": 25180059.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.0874038934707642,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 1.0628604888916016,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.10015511512756348,
+      "mean_token_accuracy": 0.9605867803096771,
+      "num_tokens": 25361290.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0702470779418944,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.1819324493408203,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.104461407661438,
+      "mean_token_accuracy": 0.9581527650356293,
+      "num_tokens": 25542455.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.0592689394950867,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4695993661880493,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.1024864912033081,
+      "mean_token_accuracy": 0.9591827690601349,
+      "num_tokens": 25723310.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.0781041979789734,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 1.3313041925430298,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.10552446842193604,
+      "mean_token_accuracy": 0.9576125502586365,
+      "num_tokens": 25904923.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0914432287216187,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.3578517436981201,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.11063914299011231,
+      "mean_token_accuracy": 0.9563858449459076,
+      "num_tokens": 26086154.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0916360139846801,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.3493428230285645,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.10228430032730103,
+      "mean_token_accuracy": 0.9597218215465546,
+      "num_tokens": 26268096.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0865699172019958,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.3002992868423462,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.10704560279846191,
+      "mean_token_accuracy": 0.9567803084850312,
+      "num_tokens": 26448598.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.0907732367515564,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.3105698823928833,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.10344587564468384,
+      "mean_token_accuracy": 0.9585539996623993,
+      "num_tokens": 26629572.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.097494399547577,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.5486772060394287,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.10484597682952881,
+      "mean_token_accuracy": 0.9578526735305786,
+      "num_tokens": 26810588.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 1.1259616188182953,
+      "eval_loss": 0.11920737475156784,
+      "eval_mean_token_accuracy": 0.9516198661676638,
+      "eval_num_tokens": 26919371.0,
+      "eval_runtime": 12.6599,
+      "eval_samples_per_second": 394.949,
+      "eval_steps_per_second": 12.401,
+      "step": 1486
+    },
+    {
+      "entropy": 1.1117471694946288,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.3559718132019043,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.09661188125610351,
+      "mean_token_accuracy": 0.9621427834033967,
+      "num_tokens": 26991831.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.0618590712547302,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.5231083631515503,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.07198524475097656,
+      "mean_token_accuracy": 0.9724888324737548,
+      "num_tokens": 27172578.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.0601553201675415,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0978407859802246,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.07232148051261902,
+      "mean_token_accuracy": 0.9721459746360779,
+      "num_tokens": 27353505.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.0545680522918701,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.375500202178955,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0737423300743103,
+      "mean_token_accuracy": 0.9723270058631897,
+      "num_tokens": 27534758.0,
+      "step": 1520
+    },
+    {
+      "entropy": 1.0429188370704652,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.5134543180465698,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.07129207253456116,
+      "mean_token_accuracy": 0.9730912566184997,
+      "num_tokens": 27715692.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0608030915260316,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.424338936805725,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.07388348579406738,
+      "mean_token_accuracy": 0.9727432250976562,
+      "num_tokens": 27896664.0,
+      "step": 1540
+    },
+    {
+      "entropy": 1.0577767491340637,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2193686962127686,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.07269638180732726,
+      "mean_token_accuracy": 0.9726069569587708,
+      "num_tokens": 28077786.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0627391576766967,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.4851799011230469,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.07443426847457886,
+      "mean_token_accuracy": 0.9716702818870544,
+      "num_tokens": 28259406.0,
+      "step": 1560
+    },
+    {
+      "entropy": 1.0606005311012268,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.136905312538147,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.07183161973953248,
+      "mean_token_accuracy": 0.9726755082607269,
+      "num_tokens": 28441336.0,
+      "step": 1570
+    },
+    {
+      "entropy": 1.0512076377868653,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.4116008281707764,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.07272077798843384,
+      "mean_token_accuracy": 0.972117805480957,
+      "num_tokens": 28621984.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0617831587791442,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.4194544553756714,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.07670866250991822,
+      "mean_token_accuracy": 0.9709574401378631,
+      "num_tokens": 28803335.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.054829752445221,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 1.0619711875915527,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.06954843401908875,
+      "mean_token_accuracy": 0.9732311069965363,
+      "num_tokens": 28984568.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.062503707408905,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1824331283569336,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.0742799162864685,
+      "mean_token_accuracy": 0.9718966007232666,
+      "num_tokens": 29165958.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0652349829673766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.2683000564575195,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.07545414566993713,
+      "mean_token_accuracy": 0.9704910278320312,
+      "num_tokens": 29347870.0,
+      "step": 1620
+    },
+    {
+      "entropy": 1.0519612908363343,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.257572889328003,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.075272136926651,
+      "mean_token_accuracy": 0.9712580978870392,
+      "num_tokens": 29529453.0,
+      "step": 1630
+    },
+    {
+      "entropy": 1.0668588399887085,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.3439997434616089,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.068633770942688,
+      "mean_token_accuracy": 0.9735071420669555,
+      "num_tokens": 29710695.0,
+      "step": 1640
+    },
+    {
+      "entropy": 1.0701637029647828,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.4352364540100098,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.07366064786911011,
+      "mean_token_accuracy": 0.9724323868751525,
+      "num_tokens": 29890993.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.065868353843689,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.3669353723526,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.0714600920677185,
+      "mean_token_accuracy": 0.9717992007732391,
+      "num_tokens": 30072284.0,
+      "step": 1660
+    },
+    {
+      "entropy": 1.077157938480377,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 1.3630682229995728,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.07237017154693604,
+      "mean_token_accuracy": 0.9729755759239197,
+      "num_tokens": 30253123.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0886191844940185,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.3787649869918823,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.07589051723480225,
+      "mean_token_accuracy": 0.9716001987457276,
+      "num_tokens": 30434100.0,
+      "step": 1680
+    },
+    {
+      "entropy": 1.080547845363617,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.4936959743499756,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.07393252849578857,
+      "mean_token_accuracy": 0.9709986329078675,
+      "num_tokens": 30615086.0,
+      "step": 1690
+    },
+    {
+      "entropy": 1.0851039648056031,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.4654748439788818,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.07487455606460572,
+      "mean_token_accuracy": 0.9710996568202972,
+      "num_tokens": 30795174.0,
+      "step": 1700
+    },
+    {
+      "entropy": 1.0743987679481506,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 1.354244589805603,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.07701593041419982,
+      "mean_token_accuracy": 0.9701604127883912,
+      "num_tokens": 30976627.0,
+      "step": 1710
+    },
+    {
+      "entropy": 1.0738040447235107,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.2830452919006348,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.07861257195472718,
+      "mean_token_accuracy": 0.9696074068546295,
+      "num_tokens": 31158542.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.0667856335639954,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.3329832553863525,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.07371129989624023,
+      "mean_token_accuracy": 0.9722550988197327,
+      "num_tokens": 31339091.0,
+      "step": 1730
+    },
+    {
+      "entropy": 1.0734196662902833,
+      "epoch": 2.34185733512786,
+      "grad_norm": 1.4483965635299683,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.07841320037841797,
+      "mean_token_accuracy": 0.9698444902896881,
+      "num_tokens": 31519961.0,
+      "step": 1740
+    },
+    {
+      "entropy": 1.0724906921386719,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.440190315246582,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.07452400922775268,
+      "mean_token_accuracy": 0.9715199947357178,
+      "num_tokens": 31701305.0,
+      "step": 1750
+    },
+    {
+      "entropy": 1.0770147442817688,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 1.17177414894104,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.07649941444396972,
+      "mean_token_accuracy": 0.9704680681228638,
+      "num_tokens": 31882106.0,
+      "step": 1760
+    },
+    {
+      "entropy": 1.0792171478271484,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.498350977897644,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.07850931882858277,
+      "mean_token_accuracy": 0.9699269771575928,
+      "num_tokens": 32062712.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.092840301990509,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.4153627157211304,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.07811332941055298,
+      "mean_token_accuracy": 0.9698202073574066,
+      "num_tokens": 32244156.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0767372131347657,
+      "epoch": 2.409152086137281,
+      "grad_norm": 1.3672188520431519,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.0765878677368164,
+      "mean_token_accuracy": 0.970400470495224,
+      "num_tokens": 32424975.0,
+      "step": 1790
+    },
+    {
+      "entropy": 1.0641321778297423,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.1141754388809204,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0744673728942871,
+      "mean_token_accuracy": 0.9713521003723145,
+      "num_tokens": 32606399.0,
+      "step": 1800
+    },
+    {
+      "entropy": 1.0847392082214355,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.372215986251831,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0788280725479126,
+      "mean_token_accuracy": 0.9688920557498932,
+      "num_tokens": 32787632.0,
+      "step": 1810
+    },
+    {
+      "entropy": 1.0952076196670533,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.3998148441314697,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.07322710752487183,
+      "mean_token_accuracy": 0.9713049352169036,
+      "num_tokens": 32969136.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.1047595977783202,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.538533091545105,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.07695393562316895,
+      "mean_token_accuracy": 0.9697132647037506,
+      "num_tokens": 33149585.0,
+      "step": 1830
+    },
+    {
+      "entropy": 1.1071401476860045,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.2797189950942993,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.07820409536361694,
+      "mean_token_accuracy": 0.9695189118385314,
+      "num_tokens": 33330604.0,
+      "step": 1840
+    },
+    {
+      "entropy": 1.0979747891426086,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 1.309956669807434,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.08005034327507018,
+      "mean_token_accuracy": 0.9683119177818298,
+      "num_tokens": 33510932.0,
+      "step": 1850
+    },
+    {
+      "entropy": 1.1024476408958435,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.2206803560256958,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.07752519249916076,
+      "mean_token_accuracy": 0.9697020888328552,
+      "num_tokens": 33692289.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0788621783256531,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.4230300188064575,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.08086194992065429,
+      "mean_token_accuracy": 0.9689899325370789,
+      "num_tokens": 33873643.0,
+      "step": 1870
+    },
+    {
+      "entropy": 1.0909191489219665,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 1.3553478717803955,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.07968194484710693,
+      "mean_token_accuracy": 0.9695583760738373,
+      "num_tokens": 34054342.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.0874631762504579,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 1.209466576576233,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.07668136358261109,
+      "mean_token_accuracy": 0.9706926643848419,
+      "num_tokens": 34235054.0,
+      "step": 1890
+    },
+    {
+      "entropy": 1.0850980758666993,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.2419084310531616,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.07621376514434815,
+      "mean_token_accuracy": 0.9708946466445922,
+      "num_tokens": 34415792.0,
+      "step": 1900
+    },
+    {
+      "entropy": 1.1024398922920227,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.5481151342391968,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.07621458768844605,
+      "mean_token_accuracy": 0.9696649491786957,
+      "num_tokens": 34596130.0,
+      "step": 1910
+    },
+    {
+      "entropy": 1.0917130827903747,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.4186631441116333,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.07716711163520813,
+      "mean_token_accuracy": 0.9699705123901368,
+      "num_tokens": 34776608.0,
+      "step": 1920
+    },
+    {
+      "entropy": 1.0845637321472168,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2549679279327393,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.07353252172470093,
+      "mean_token_accuracy": 0.9715336680412292,
+      "num_tokens": 34957854.0,
+      "step": 1930
+    },
+    {
+      "entropy": 1.070354163646698,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.390034794807434,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.07804152965545655,
+      "mean_token_accuracy": 0.9694301605224609,
+      "num_tokens": 35139425.0,
+      "step": 1940
+    },
+    {
+      "entropy": 1.0926494956016541,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.2890169620513916,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.07905603051185608,
+      "mean_token_accuracy": 0.9694173872470856,
+      "num_tokens": 35321009.0,
+      "step": 1950
+    },
+    {
+      "entropy": 1.090261971950531,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.201541543006897,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.07853106260299683,
+      "mean_token_accuracy": 0.9687644064426422,
+      "num_tokens": 35502338.0,
+      "step": 1960
+    },
+    {
+      "entropy": 1.110123836994171,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.3321393728256226,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.07721331715583801,
+      "mean_token_accuracy": 0.9703636646270752,
+      "num_tokens": 35684550.0,
+      "step": 1970
+    },
+    {
+      "entropy": 1.1154867172241212,
+      "epoch": 2.664872139973082,
+      "grad_norm": 1.5639152526855469,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.07769335508346557,
+      "mean_token_accuracy": 0.9694860637187958,
+      "num_tokens": 35866258.0,
+      "step": 1980
+    },
+    {
+      "entropy": 1.1126874089241028,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5160865783691406,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.07732025384902955,
+      "mean_token_accuracy": 0.9695809543132782,
+      "num_tokens": 36047771.0,
+      "step": 1990
+    },
+    {
+      "entropy": 1.1188251495361328,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.3689777851104736,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.07844393253326416,
+      "mean_token_accuracy": 0.9697122633457184,
+      "num_tokens": 36228702.0,
+      "step": 2000
+    },
+    {
+      "entropy": 1.1134975671768188,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 1.1956474781036377,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.07932614088058472,
+      "mean_token_accuracy": 0.9695560097694397,
+      "num_tokens": 36409577.0,
+      "step": 2010
+    },
+    {
+      "entropy": 1.1052889227867126,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.2844921350479126,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.07471654415130616,
+      "mean_token_accuracy": 0.9707616209983826,
+      "num_tokens": 36590635.0,
+      "step": 2020
+    },
+    {
+      "entropy": 1.1029305696487426,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3110395669937134,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.07654619216918945,
+      "mean_token_accuracy": 0.9699673116207123,
+      "num_tokens": 36772032.0,
+      "step": 2030
+    },
+    {
+      "entropy": 1.0967667698860168,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.2297992706298828,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.07549123167991638,
+      "mean_token_accuracy": 0.9705443382263184,
+      "num_tokens": 36953014.0,
+      "step": 2040
+    },
+    {
+      "entropy": 1.0954343318939208,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.5522037744522095,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.07590663433074951,
+      "mean_token_accuracy": 0.9709664165973664,
+      "num_tokens": 37134549.0,
+      "step": 2050
+    },
+    {
+      "entropy": 1.0984182119369508,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.3691717386245728,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.07795964479446411,
+      "mean_token_accuracy": 0.969748318195343,
+      "num_tokens": 37315651.0,
+      "step": 2060
+    },
+    {
+      "entropy": 1.0937008380889892,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 1.345433235168457,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.07556321024894715,
+      "mean_token_accuracy": 0.9707719802856445,
+      "num_tokens": 37497150.0,
+      "step": 2070
+    },
+    {
+      "entropy": 1.0978638410568238,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.3442333936691284,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.07603150010108947,
+      "mean_token_accuracy": 0.9707530319690705,
+      "num_tokens": 37678637.0,
+      "step": 2080
+    },
+    {
+      "entropy": 1.112320113182068,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.672094702720642,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.07702311277389526,
+      "mean_token_accuracy": 0.969988214969635,
+      "num_tokens": 37859530.0,
+      "step": 2090
+    },
+    {
+      "entropy": 1.1019986748695374,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.4613395929336548,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.07702183723449707,
+      "mean_token_accuracy": 0.9705046355724335,
+      "num_tokens": 38040160.0,
+      "step": 2100
+    },
+    {
+      "entropy": 1.1074181199073792,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.33228600025177,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.07809240221977234,
+      "mean_token_accuracy": 0.9702594041824341,
+      "num_tokens": 38221261.0,
+      "step": 2110
+    },
+    {
+      "entropy": 1.1087767481803894,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.3967821598052979,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.07641294002532958,
+      "mean_token_accuracy": 0.9694520056247711,
+      "num_tokens": 38402589.0,
+      "step": 2120
+    },
+    {
+      "entropy": 1.0902286529541017,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 1.3104654550552368,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.07864789366722107,
+      "mean_token_accuracy": 0.9693171322345734,
+      "num_tokens": 38583485.0,
+      "step": 2130
+    },
+    {
+      "entropy": 1.088974940776825,
+      "epoch": 2.88021534320323,
+      "grad_norm": 1.389603614807129,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.07767997980117798,
+      "mean_token_accuracy": 0.9691886186599732,
+      "num_tokens": 38764480.0,
+      "step": 2140
+    },
+    {
+      "entropy": 1.093208122253418,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.3133198022842407,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.08097923994064331,
+      "mean_token_accuracy": 0.968046635389328,
+      "num_tokens": 38946016.0,
+      "step": 2150
+    },
+    {
+      "entropy": 1.0920218586921693,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.3472810983657837,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.07504125237464905,
+      "mean_token_accuracy": 0.9705065429210663,
+      "num_tokens": 39127480.0,
+      "step": 2160
+    },
+    {
+      "entropy": 1.085507369041443,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.9070137739181519,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.07882450819015503,
+      "mean_token_accuracy": 0.9694548428058625,
+      "num_tokens": 39308552.0,
+      "step": 2170
+    },
+    {
+      "entropy": 1.1070378303527832,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.3059617280960083,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.08202729225158692,
+      "mean_token_accuracy": 0.9679245829582215,
+      "num_tokens": 39489844.0,
+      "step": 2180
+    },
+    {
+      "entropy": 1.0994928359985352,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.359965443611145,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.07510381340980529,
+      "mean_token_accuracy": 0.969987016916275,
+      "num_tokens": 39671941.0,
+      "step": 2190
+    },
+    {
+      "entropy": 1.0987842202186584,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.385233998298645,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.07863752245903015,
+      "mean_token_accuracy": 0.9685526669025422,
+      "num_tokens": 39852940.0,
+      "step": 2200
+    },
+    {
+      "entropy": 1.110793125629425,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.4456037282943726,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.07783852815628052,
+      "mean_token_accuracy": 0.9702686488628387,
+      "num_tokens": 40034313.0,
+      "step": 2210
+    },
+    {
+      "entropy": 1.1140072107315064,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.1889644861221313,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.07787686586380005,
+      "mean_token_accuracy": 0.9696961343288422,
+      "num_tokens": 40215866.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 1.11932077984901,
+      "eval_loss": 0.12462407350540161,
+      "eval_mean_token_accuracy": 0.951254228118119,
+      "eval_num_tokens": 40378902.0,
+      "eval_runtime": 13.5665,
+      "eval_samples_per_second": 368.556,
+      "eval_steps_per_second": 11.573,
+      "step": 2229
+    },
+    {
+      "entropy": 1.1124149322509767,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 1.1116644144058228,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.07439510822296143,
+      "mean_token_accuracy": 0.9713229060173034,
+      "num_tokens": 40396979.0,
+      "step": 2230
+    },
+    {
+      "entropy": 1.0716044545173644,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.1957958936691284,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.05146250128746033,
+      "mean_token_accuracy": 0.9813062250614166,
+      "num_tokens": 40578935.0,
+      "step": 2240
+    },
+    {
+      "entropy": 1.0567036390304565,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.337942361831665,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.052991294860839845,
+      "mean_token_accuracy": 0.9807850599288941,
+      "num_tokens": 40759500.0,
+      "step": 2250
+    },
+    {
+      "entropy": 1.056044340133667,
+      "epoch": 3.041722745625841,
+      "grad_norm": 1.4191840887069702,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.05163681507110596,
+      "mean_token_accuracy": 0.9813888549804688,
+      "num_tokens": 40940279.0,
+      "step": 2260
+    },
+    {
+      "entropy": 1.056977343559265,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.4481396675109863,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.05059016942977905,
+      "mean_token_accuracy": 0.9814287006855011,
+      "num_tokens": 41121782.0,
+      "step": 2270
+    },
+    {
+      "entropy": 1.0579043984413148,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.0136349201202393,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.05109646320343018,
+      "mean_token_accuracy": 0.9805608510971069,
+      "num_tokens": 41302614.0,
+      "step": 2280
+    },
+    {
+      "entropy": 1.0564674019813538,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.165916085243225,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.05656059980392456,
+      "mean_token_accuracy": 0.9801787078380585,
+      "num_tokens": 41484240.0,
+      "step": 2290
+    },
+    {
+      "entropy": 1.0617436170578003,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.500622272491455,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.05233668088912964,
+      "mean_token_accuracy": 0.9808355152606965,
+      "num_tokens": 41665160.0,
+      "step": 2300
+    },
+    {
+      "entropy": 1.0509375214576722,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.403001308441162,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.05074570775032043,
+      "mean_token_accuracy": 0.9807266294956207,
+      "num_tokens": 41846062.0,
+      "step": 2310
+    },
+    {
+      "entropy": 1.0536353230476379,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1046926975250244,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.04906890392303467,
+      "mean_token_accuracy": 0.9815255463123321,
+      "num_tokens": 42027115.0,
+      "step": 2320
+    },
+    {
+      "entropy": 1.0518895149230958,
+      "epoch": 3.135935397039031,
+      "grad_norm": 1.0369377136230469,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.04991455376148224,
+      "mean_token_accuracy": 0.9822782278060913,
+      "num_tokens": 42208594.0,
+      "step": 2330
+    },
+    {
+      "entropy": 1.0597471237182616,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.6156455278396606,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.049592965841293336,
+      "mean_token_accuracy": 0.9818629324436188,
+      "num_tokens": 42389065.0,
+      "step": 2340
+    },
+    {
+      "entropy": 1.0561720967292785,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.2941797971725464,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.05092741847038269,
+      "mean_token_accuracy": 0.9810208320617676,
+      "num_tokens": 42570319.0,
+      "step": 2350
+    },
+    {
+      "entropy": 1.0521231889724731,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.1866732835769653,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.051917088031768796,
+      "mean_token_accuracy": 0.981055635213852,
+      "num_tokens": 42751501.0,
+      "step": 2360
+    },
+    {
+      "entropy": 1.0555105328559875,
+      "epoch": 3.189771197846568,
+      "grad_norm": 1.2260069847106934,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.05160566568374634,
+      "mean_token_accuracy": 0.981193995475769,
+      "num_tokens": 42932643.0,
+      "step": 2370
+    },
+    {
+      "entropy": 1.0720075011253356,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.3786890506744385,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.053551071882247926,
+      "mean_token_accuracy": 0.9797191619873047,
+      "num_tokens": 43113204.0,
+      "step": 2380
+    },
+    {
+      "entropy": 1.0718719363212585,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.3200089931488037,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.054034185409545896,
+      "mean_token_accuracy": 0.9800151228904724,
+      "num_tokens": 43293433.0,
+      "step": 2390
+    },
+    {
+      "entropy": 1.0536052703857421,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.4560705423355103,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.05042939186096192,
+      "mean_token_accuracy": 0.9806897640228271,
+      "num_tokens": 43474189.0,
+      "step": 2400
+    },
+    {
+      "entropy": 1.0602891445159912,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.3123281002044678,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.05151479244232178,
+      "mean_token_accuracy": 0.980517429113388,
+      "num_tokens": 43655355.0,
+      "step": 2410
+    },
+    {
+      "entropy": 1.067916214466095,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.2652188539505005,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.05306308269500733,
+      "mean_token_accuracy": 0.9805097997188568,
+      "num_tokens": 43836752.0,
+      "step": 2420
+    },
+    {
+      "entropy": 1.067535150051117,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.6280162334442139,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.052641332149505615,
+      "mean_token_accuracy": 0.9803492069244385,
+      "num_tokens": 44018525.0,
+      "step": 2430
+    },
+    {
+      "entropy": 1.070455825328827,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 1.632573127746582,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.05452280044555664,
+      "mean_token_accuracy": 0.9793554663658142,
+      "num_tokens": 44199901.0,
+      "step": 2440
+    },
+    {
+      "entropy": 1.0741212248802186,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.3525495529174805,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.054098492860794066,
+      "mean_token_accuracy": 0.9804170608520508,
+      "num_tokens": 44381734.0,
+      "step": 2450
+    },
+    {
+      "entropy": 1.0836357831954957,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.3989282846450806,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.05158165693283081,
+      "mean_token_accuracy": 0.9810640931129455,
+      "num_tokens": 44563158.0,
+      "step": 2460
+    },
+    {
+      "entropy": 1.077314054965973,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1970937252044678,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.052333271503448485,
+      "mean_token_accuracy": 0.9808873236179352,
+      "num_tokens": 44744642.0,
+      "step": 2470
+    },
+    {
+      "entropy": 1.069891107082367,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.3197492361068726,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.05269680619239807,
+      "mean_token_accuracy": 0.980803519487381,
+      "num_tokens": 44926301.0,
+      "step": 2480
+    },
+    {
+      "entropy": 1.071691370010376,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.332481861114502,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.054685378074646,
+      "mean_token_accuracy": 0.9795110940933227,
+      "num_tokens": 45106935.0,
+      "step": 2490
+    },
+    {
+      "entropy": 1.0654058694839477,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 1.5478020906448364,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.05287986397743225,
+      "mean_token_accuracy": 0.9797925829887391,
+      "num_tokens": 45288325.0,
+      "step": 2500
+    },
+    {
+      "entropy": 1.0660210013389588,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.43852961063385,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.05411216616630554,
+      "mean_token_accuracy": 0.9790283799171448,
+      "num_tokens": 45469485.0,
+      "step": 2510
+    },
+    {
+      "entropy": 1.0649927496910094,
+      "epoch": 3.391655450874832,
+      "grad_norm": 1.3900808095932007,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.05290043354034424,
+      "mean_token_accuracy": 0.9808539688587189,
+      "num_tokens": 45650705.0,
+      "step": 2520
+    },
+    {
+      "entropy": 1.0768967866897583,
+      "epoch": 3.405114401076716,
+      "grad_norm": 1.3428847789764404,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.05374534130096435,
+      "mean_token_accuracy": 0.9802824795246124,
+      "num_tokens": 45832104.0,
+      "step": 2530
+    },
+    {
+      "entropy": 1.0660459756851197,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.3150343894958496,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.05366337299346924,
+      "mean_token_accuracy": 0.9796974420547485,
+      "num_tokens": 46013552.0,
+      "step": 2540
+    },
+    {
+      "entropy": 1.06330087184906,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.546858787536621,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.054223823547363284,
+      "mean_token_accuracy": 0.9796406507492066,
+      "num_tokens": 46194606.0,
+      "step": 2550
+    },
+    {
+      "entropy": 1.0564390897750855,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.3932640552520752,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.05237367153167725,
+      "mean_token_accuracy": 0.9805298745632172,
+      "num_tokens": 46375520.0,
+      "step": 2560
+    },
+    {
+      "entropy": 1.0519753217697143,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.8447308540344238,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.05499213933944702,
+      "mean_token_accuracy": 0.9787130057811737,
+      "num_tokens": 46556203.0,
+      "step": 2570
+    },
+    {
+      "entropy": 1.0693527460098267,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.3508682250976562,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.055162906646728516,
+      "mean_token_accuracy": 0.9794130921363831,
+      "num_tokens": 46737287.0,
+      "step": 2580
+    },
+    {
+      "entropy": 1.078024983406067,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 1.1582329273223877,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.05406954288482666,
+      "mean_token_accuracy": 0.9794795215129852,
+      "num_tokens": 46918363.0,
+      "step": 2590
+    },
+    {
+      "entropy": 1.0643377661705018,
+      "epoch": 3.499327052489906,
+      "grad_norm": 1.429380178451538,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.05309759378433228,
+      "mean_token_accuracy": 0.9806221067905426,
+      "num_tokens": 47099549.0,
+      "step": 2600
+    },
+    {
+      "entropy": 1.068834125995636,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.236444354057312,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.05395760536193848,
+      "mean_token_accuracy": 0.9798199534416199,
+      "num_tokens": 47280430.0,
+      "step": 2610
+    },
+    {
+      "entropy": 1.0729689002037048,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 1.32350754737854,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.05501353740692139,
+      "mean_token_accuracy": 0.9792946577072144,
+      "num_tokens": 47461390.0,
+      "step": 2620
+    },
+    {
+      "entropy": 1.0766199231147766,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2348153591156006,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.05568041205406189,
+      "mean_token_accuracy": 0.9796942174434662,
+      "num_tokens": 47642662.0,
+      "step": 2630
+    },
+    {
+      "entropy": 1.0804976940155029,
+      "epoch": 3.553162853297443,
+      "grad_norm": 1.1175553798675537,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.050679147243499756,
+      "mean_token_accuracy": 0.9814749479293823,
+      "num_tokens": 47824309.0,
+      "step": 2640
+    },
+    {
+      "entropy": 1.0689257264137269,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.6297036409378052,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.054615461826324464,
+      "mean_token_accuracy": 0.9796478092670441,
+      "num_tokens": 48005982.0,
+      "step": 2650
+    },
+    {
+      "entropy": 1.0696970105171204,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2477251291275024,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.05287024974822998,
+      "mean_token_accuracy": 0.979641854763031,
+      "num_tokens": 48187048.0,
+      "step": 2660
+    },
+    {
+      "entropy": 1.0800267577171325,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.3109506368637085,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.054475915431976316,
+      "mean_token_accuracy": 0.9794476389884949,
+      "num_tokens": 48368447.0,
+      "step": 2670
+    },
+    {
+      "entropy": 1.0848782062530518,
+      "epoch": 3.60699865410498,
+      "grad_norm": 1.2147419452667236,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.05212314128875732,
+      "mean_token_accuracy": 0.9798501491546631,
+      "num_tokens": 48549144.0,
+      "step": 2680
+    },
+    {
+      "entropy": 1.0681315302848815,
+      "epoch": 3.620457604306864,
+      "grad_norm": 1.3249231576919556,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.053503894805908205,
+      "mean_token_accuracy": 0.9800095915794372,
+      "num_tokens": 48731059.0,
+      "step": 2690
+    },
+    {
+      "entropy": 1.0629405975341797,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.2435307502746582,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.0541129469871521,
+      "mean_token_accuracy": 0.9795375943183899,
+      "num_tokens": 48912283.0,
+      "step": 2700
+    },
+    {
+      "entropy": 1.0608446598052979,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.3816258907318115,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.054210084676742556,
+      "mean_token_accuracy": 0.9793854176998138,
+      "num_tokens": 49094236.0,
+      "step": 2710
+    },
+    {
+      "entropy": 1.0596249461174012,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.132426381111145,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.05496481657028198,
+      "mean_token_accuracy": 0.9798576831817627,
+      "num_tokens": 49275540.0,
+      "step": 2720
+    },
+    {
+      "entropy": 1.060381281375885,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.4120980501174927,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.05694271326065063,
+      "mean_token_accuracy": 0.9788619399070739,
+      "num_tokens": 49456684.0,
+      "step": 2730
+    },
+    {
+      "entropy": 1.0594112157821656,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.428120493888855,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.05274912118911743,
+      "mean_token_accuracy": 0.9802964627742767,
+      "num_tokens": 49638099.0,
+      "step": 2740
+    },
+    {
+      "entropy": 1.0630932807922364,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.4599305391311646,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.057349550724029544,
+      "mean_token_accuracy": 0.9788366615772247,
+      "num_tokens": 49819751.0,
+      "step": 2750
+    },
+    {
+      "entropy": 1.0632095456123352,
+      "epoch": 3.714670255720054,
+      "grad_norm": 1.8489277362823486,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.053570222854614255,
+      "mean_token_accuracy": 0.980079036951065,
+      "num_tokens": 50000623.0,
+      "step": 2760
+    },
+    {
+      "entropy": 1.0583208560943604,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.4855526685714722,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.05508332252502442,
+      "mean_token_accuracy": 0.9789698243141174,
+      "num_tokens": 50181540.0,
+      "step": 2770
+    },
+    {
+      "entropy": 1.0690102458000184,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 1.553468942642212,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.05684031248092651,
+      "mean_token_accuracy": 0.9776920318603516,
+      "num_tokens": 50363406.0,
+      "step": 2780
+    },
+    {
+      "entropy": 1.0651323795318604,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 1.2951313257217407,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.055989408493041994,
+      "mean_token_accuracy": 0.978317403793335,
+      "num_tokens": 50544599.0,
+      "step": 2790
+    },
+    {
+      "entropy": 1.0571947336196899,
+      "epoch": 3.768506056527591,
+      "grad_norm": 1.1358306407928467,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.052672064304351805,
+      "mean_token_accuracy": 0.9797160506248475,
+      "num_tokens": 50726664.0,
+      "step": 2800
+    },
+    {
+      "entropy": 1.0475606441497802,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.4842852354049683,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.05601704120635986,
+      "mean_token_accuracy": 0.9794813454151153,
+      "num_tokens": 50907903.0,
+      "step": 2810
+    },
+    {
+      "entropy": 1.053141224384308,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.3320889472961426,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.05435909032821655,
+      "mean_token_accuracy": 0.9792923092842102,
+      "num_tokens": 51088734.0,
+      "step": 2820
+    },
+    {
+      "entropy": 1.0518054842948914,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.140783667564392,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.05576471090316772,
+      "mean_token_accuracy": 0.9789255380630493,
+      "num_tokens": 51269924.0,
+      "step": 2830
+    },
+    {
+      "entropy": 1.0490313410758971,
+      "epoch": 3.822341857335128,
+      "grad_norm": 1.5749891996383667,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.05655611753463745,
+      "mean_token_accuracy": 0.9785366714000702,
+      "num_tokens": 51450406.0,
+      "step": 2840
+    },
+    {
+      "entropy": 1.0508960008621215,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 1.2535520792007446,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.05710798501968384,
+      "mean_token_accuracy": 0.978301590681076,
+      "num_tokens": 51630630.0,
+      "step": 2850
+    },
+    {
+      "entropy": 1.0434152960777283,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.6045477390289307,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.0545048713684082,
+      "mean_token_accuracy": 0.9785818517208099,
+      "num_tokens": 51811252.0,
+      "step": 2860
+    },
+    {
+      "entropy": 1.046581244468689,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.2737947702407837,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.056829071044921874,
+      "mean_token_accuracy": 0.9779990613460541,
+      "num_tokens": 51992578.0,
+      "step": 2870
+    },
+    {
+      "entropy": 1.0562566757202148,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.2364535331726074,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.054702204465866086,
+      "mean_token_accuracy": 0.9796520650386811,
+      "num_tokens": 52173507.0,
+      "step": 2880
+    },
+    {
+      "entropy": 1.0444288849830627,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.2663625478744507,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.05222170352935791,
+      "mean_token_accuracy": 0.9805905818939209,
+      "num_tokens": 52354669.0,
+      "step": 2890
+    },
+    {
+      "entropy": 1.0429784655570984,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.2072714567184448,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.05404345989227295,
+      "mean_token_accuracy": 0.9802968561649322,
+      "num_tokens": 52535334.0,
+      "step": 2900
+    },
+    {
+      "entropy": 1.045020854473114,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.3894519805908203,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.05601906776428223,
+      "mean_token_accuracy": 0.9779200375080108,
+      "num_tokens": 52715952.0,
+      "step": 2910
+    },
+    {
+      "entropy": 1.0439318537712097,
+      "epoch": 3.930013458950202,
+      "grad_norm": 1.4112615585327148,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.05345791578292847,
+      "mean_token_accuracy": 0.9796200633049011,
+      "num_tokens": 52897155.0,
+      "step": 2920
+    },
+    {
+      "entropy": 1.0361477971076964,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 1.3158206939697266,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.05547314286231995,
+      "mean_token_accuracy": 0.9788706004619598,
+      "num_tokens": 53078916.0,
+      "step": 2930
+    },
+    {
+      "entropy": 1.0427968740463256,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2565301656723022,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.05601854920387268,
+      "mean_token_accuracy": 0.9791704952716828,
+      "num_tokens": 53258884.0,
+      "step": 2940
+    },
+    {
+      "entropy": 1.0374909162521362,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.4639637470245361,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.054512780904769895,
+      "mean_token_accuracy": 0.9793699860572815,
+      "num_tokens": 53440481.0,
+      "step": 2950
+    },
+    {
+      "entropy": 1.0511455059051513,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.242073655128479,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.055217862129211426,
+      "mean_token_accuracy": 0.9801203072071075,
+      "num_tokens": 53621142.0,
+      "step": 2960
+    },
+    {
+      "entropy": 1.0518356442451477,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.1518405675888062,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.05533761382102966,
+      "mean_token_accuracy": 0.9789008319377899,
+      "num_tokens": 53802588.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 1.0548353901334628,
+      "eval_loss": 0.13498826324939728,
+      "eval_mean_token_accuracy": 0.9507886888874564,
+      "eval_num_tokens": 53838633.0,
+      "eval_runtime": 12.6809,
+      "eval_samples_per_second": 394.295,
+      "eval_steps_per_second": 12.381,
+      "step": 2972
+    },
+    {
+      "entropy": 1.043621039390564,
+      "epoch": 4.010767160161508,
+      "grad_norm": 1.1016159057617188,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.03793205618858338,
+      "mean_token_accuracy": 0.9870165169239045,
+      "num_tokens": 53983430.0,
+      "step": 2980
+    },
+    {
+      "entropy": 1.0226780295372009,
+      "epoch": 4.024226110363392,
+      "grad_norm": 1.240837812423706,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.03402446508407593,
+      "mean_token_accuracy": 0.9883724093437195,
+      "num_tokens": 54164955.0,
+      "step": 2990
+    },
+    {
+      "entropy": 1.0194039583206176,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.1843734979629517,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.035861504077911374,
+      "mean_token_accuracy": 0.9872536063194275,
+      "num_tokens": 54346968.0,
+      "step": 3000
+    },
+    {
+      "entropy": 1.029161846637726,
+      "epoch": 4.05114401076716,
+      "grad_norm": 1.285415768623352,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.034106379747390746,
+      "mean_token_accuracy": 0.9882839441299438,
+      "num_tokens": 54528019.0,
+      "step": 3010
+    },
+    {
+      "entropy": 1.031964898109436,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4622068405151367,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.032776829600334165,
+      "mean_token_accuracy": 0.9882788896560669,
+      "num_tokens": 54708148.0,
+      "step": 3020
+    },
+    {
+      "entropy": 1.0282336235046388,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.1071696281433105,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.03698182702064514,
+      "mean_token_accuracy": 0.9865448236465454,
+      "num_tokens": 54888894.0,
+      "step": 3030
+    },
+    {
+      "entropy": 1.0233591556549073,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.3671438694000244,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.03386759161949158,
+      "mean_token_accuracy": 0.9879629850387573,
+      "num_tokens": 55069872.0,
+      "step": 3040
+    },
+    {
+      "entropy": 1.0250957012176514,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.1421459913253784,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.035586753487586976,
+      "mean_token_accuracy": 0.9869713068008423,
+      "num_tokens": 55250605.0,
+      "step": 3050
+    },
+    {
+      "entropy": 1.016488790512085,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.3857697248458862,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.039323312044143674,
+      "mean_token_accuracy": 0.986032885313034,
+      "num_tokens": 55431737.0,
+      "step": 3060
+    },
+    {
+      "entropy": 1.0130115926265717,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.1797211170196533,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.03508211970329285,
+      "mean_token_accuracy": 0.9872302651405335,
+      "num_tokens": 55612917.0,
+      "step": 3070
+    },
+    {
+      "entropy": 1.023509919643402,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.1484259366989136,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.0338476300239563,
+      "mean_token_accuracy": 0.9883188247680664,
+      "num_tokens": 55793687.0,
+      "step": 3080
+    },
+    {
+      "entropy": 1.0193628787994384,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.3184833526611328,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.03667713701725006,
+      "mean_token_accuracy": 0.9869367897510528,
+      "num_tokens": 55974723.0,
+      "step": 3090
+    },
+    {
+      "entropy": 1.013669216632843,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.2176531553268433,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.035006949305534364,
+      "mean_token_accuracy": 0.9868299901485443,
+      "num_tokens": 56156568.0,
+      "step": 3100
+    },
+    {
+      "entropy": 1.0072043538093567,
+      "epoch": 4.185733512786003,
+      "grad_norm": 1.4346243143081665,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.03560284972190857,
+      "mean_token_accuracy": 0.9869017779827118,
+      "num_tokens": 56337896.0,
+      "step": 3110
+    },
+    {
+      "entropy": 1.0102415680885315,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.2328612804412842,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.03608990907669067,
+      "mean_token_accuracy": 0.9874296009540557,
+      "num_tokens": 56519102.0,
+      "step": 3120
+    },
+    {
+      "entropy": 1.015284490585327,
+      "epoch": 4.212651413189771,
+      "grad_norm": 1.4385327100753784,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.03637963235378265,
+      "mean_token_accuracy": 0.986739182472229,
+      "num_tokens": 56699936.0,
+      "step": 3130
+    },
+    {
+      "entropy": 1.0054036319255828,
+      "epoch": 4.226110363391656,
+      "grad_norm": 1.1129945516586304,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.034222981333732604,
+      "mean_token_accuracy": 0.9883054196834564,
+      "num_tokens": 56882308.0,
+      "step": 3140
+    },
+    {
+      "entropy": 1.0061798334121703,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1598492860794067,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.03461446762084961,
+      "mean_token_accuracy": 0.9873203098773956,
+      "num_tokens": 57063770.0,
+      "step": 3150
+    },
+    {
+      "entropy": 1.0039719343185425,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.322458267211914,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.03701616525650024,
+      "mean_token_accuracy": 0.9862136900424957,
+      "num_tokens": 57244448.0,
+      "step": 3160
+    },
+    {
+      "entropy": 1.0048979759216308,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 1.570866584777832,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.036290496587753296,
+      "mean_token_accuracy": 0.9868158757686615,
+      "num_tokens": 57425448.0,
+      "step": 3170
+    },
+    {
+      "entropy": 1.0145976901054383,
+      "epoch": 4.279946164199193,
+      "grad_norm": 1.103385329246521,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.03472789227962494,
+      "mean_token_accuracy": 0.987658828496933,
+      "num_tokens": 57606369.0,
+      "step": 3180
+    },
+    {
+      "entropy": 1.0186771035194397,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.2541930675506592,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.03532317280769348,
+      "mean_token_accuracy": 0.9874464809894562,
+      "num_tokens": 57787685.0,
+      "step": 3190
+    },
+    {
+      "entropy": 1.0198698043823242,
+      "epoch": 4.306864064602961,
+      "grad_norm": 1.1590384244918823,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.036364412307739256,
+      "mean_token_accuracy": 0.9864933013916015,
+      "num_tokens": 57968903.0,
+      "step": 3200
+    },
+    {
+      "entropy": 1.0211624503135681,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.7207615375518799,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.03790584802627563,
+      "mean_token_accuracy": 0.9866538405418396,
+      "num_tokens": 58149469.0,
+      "step": 3210
+    },
+    {
+      "entropy": 1.0185290336608888,
+      "epoch": 4.33378196500673,
+      "grad_norm": 1.3042566776275635,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.03802136182785034,
+      "mean_token_accuracy": 0.9859855055809021,
+      "num_tokens": 58330867.0,
+      "step": 3220
+    },
+    {
+      "entropy": 1.0105828583240508,
+      "epoch": 4.347240915208614,
+      "grad_norm": 1.2536448240280151,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.034687772393226624,
+      "mean_token_accuracy": 0.9878224015235901,
+      "num_tokens": 58512529.0,
+      "step": 3230
+    },
+    {
+      "entropy": 1.0021630108356476,
+      "epoch": 4.360699865410498,
+      "grad_norm": 1.3816099166870117,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.035324835777282716,
+      "mean_token_accuracy": 0.9872316241264343,
+      "num_tokens": 58693573.0,
+      "step": 3240
+    },
+    {
+      "entropy": 1.007886278629303,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 1.3121757507324219,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.034709393978118896,
+      "mean_token_accuracy": 0.9877420663833618,
+      "num_tokens": 58874107.0,
+      "step": 3250
+    },
+    {
+      "entropy": 1.005664885044098,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.429775357246399,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.03714964389801025,
+      "mean_token_accuracy": 0.9861079812049866,
+      "num_tokens": 59054533.0,
+      "step": 3260
+    },
+    {
+      "entropy": 1.0047886371612549,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.2385882139205933,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.036937403678894046,
+      "mean_token_accuracy": 0.9863442242145538,
+      "num_tokens": 59236331.0,
+      "step": 3270
+    },
+    {
+      "entropy": 1.0209275126457213,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.476168155670166,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.03463808596134186,
+      "mean_token_accuracy": 0.9872730135917663,
+      "num_tokens": 59417193.0,
+      "step": 3280
+    },
+    {
+      "entropy": 1.021347463130951,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 1.4710102081298828,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.038167804479599,
+      "mean_token_accuracy": 0.986047238111496,
+      "num_tokens": 59598284.0,
+      "step": 3290
+    },
+    {
+      "entropy": 1.0282366394996643,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1043860912322998,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.036692625284194945,
+      "mean_token_accuracy": 0.9862713575363159,
+      "num_tokens": 59779329.0,
+      "step": 3300
+    },
+    {
+      "entropy": 1.021342945098877,
+      "epoch": 4.454912516823688,
+      "grad_norm": 1.1742897033691406,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.03571464121341705,
+      "mean_token_accuracy": 0.987505567073822,
+      "num_tokens": 59961340.0,
+      "step": 3310
+    },
+    {
+      "entropy": 1.0153375625610352,
+      "epoch": 4.468371467025572,
+      "grad_norm": 1.1321660280227661,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.036007928848266604,
+      "mean_token_accuracy": 0.9870990335941314,
+      "num_tokens": 60142647.0,
+      "step": 3320
+    },
+    {
+      "entropy": 1.0099608540534972,
+      "epoch": 4.481830417227457,
+      "grad_norm": 1.084017276763916,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.03562799394130707,
+      "mean_token_accuracy": 0.9865496158599854,
+      "num_tokens": 60324151.0,
+      "step": 3330
+    },
+    {
+      "entropy": 1.0173812866210938,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2779732942581177,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.03556201159954071,
+      "mean_token_accuracy": 0.9875519394874572,
+      "num_tokens": 60505023.0,
+      "step": 3340
+    },
+    {
+      "entropy": 1.0236618638038635,
+      "epoch": 4.508748317631225,
+      "grad_norm": 1.5285946130752563,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.03721941411495209,
+      "mean_token_accuracy": 0.9865630626678467,
+      "num_tokens": 60685921.0,
+      "step": 3350
+    },
+    {
+      "entropy": 1.0262888669967651,
+      "epoch": 4.522207267833109,
+      "grad_norm": 1.131675362586975,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.03602397441864014,
+      "mean_token_accuracy": 0.9864337563514709,
+      "num_tokens": 60867240.0,
+      "step": 3360
+    },
+    {
+      "entropy": 1.0292576909065247,
+      "epoch": 4.535666218034994,
+      "grad_norm": 1.5156011581420898,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.03973098993301392,
+      "mean_token_accuracy": 0.9855743169784545,
+      "num_tokens": 61047708.0,
+      "step": 3370
+    },
+    {
+      "entropy": 1.010222613811493,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0979945659637451,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.03533022105693817,
+      "mean_token_accuracy": 0.9869906663894653,
+      "num_tokens": 61229247.0,
+      "step": 3380
+    },
+    {
+      "entropy": 1.004716032743454,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.3082596063613892,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.033983749151229856,
+      "mean_token_accuracy": 0.9877199769020081,
+      "num_tokens": 61411059.0,
+      "step": 3390
+    },
+    {
+      "entropy": 1.0031994462013245,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.3033607006072998,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.03607205748558044,
+      "mean_token_accuracy": 0.9866301774978637,
+      "num_tokens": 61592538.0,
+      "step": 3400
+    },
+    {
+      "entropy": 1.0113391518592834,
+      "epoch": 4.589502018842531,
+      "grad_norm": 1.3490946292877197,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.03906603455543518,
+      "mean_token_accuracy": 0.9854673087596894,
+      "num_tokens": 61773613.0,
+      "step": 3410
+    },
+    {
+      "entropy": 1.0076410889625549,
+      "epoch": 4.602960969044415,
+      "grad_norm": 1.1719361543655396,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.037575924396514894,
+      "mean_token_accuracy": 0.9867486238479615,
+      "num_tokens": 61955690.0,
+      "step": 3420
+    },
+    {
+      "entropy": 1.0200914144515991,
+      "epoch": 4.616419919246299,
+      "grad_norm": 1.1205360889434814,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.03691520690917969,
+      "mean_token_accuracy": 0.9865606665611267,
+      "num_tokens": 62136047.0,
+      "step": 3430
+    },
+    {
+      "entropy": 1.0141666173934936,
+      "epoch": 4.629878869448183,
+      "grad_norm": 1.1805763244628906,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.035447508096694946,
+      "mean_token_accuracy": 0.9873166739940643,
+      "num_tokens": 62317192.0,
+      "step": 3440
+    },
+    {
+      "entropy": 1.0067960560321807,
+      "epoch": 4.643337819650068,
+      "grad_norm": 1.0286436080932617,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.03580200672149658,
+      "mean_token_accuracy": 0.9870070457458496,
+      "num_tokens": 62498860.0,
+      "step": 3450
+    },
+    {
+      "entropy": 1.0088169515132903,
+      "epoch": 4.656796769851952,
+      "grad_norm": 1.3638492822647095,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.03756250143051147,
+      "mean_token_accuracy": 0.9861987411975861,
+      "num_tokens": 62679272.0,
+      "step": 3460
+    },
+    {
+      "entropy": 1.0098722338676454,
+      "epoch": 4.670255720053836,
+      "grad_norm": 1.649560809135437,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.037615427374839784,
+      "mean_token_accuracy": 0.9864691913127899,
+      "num_tokens": 62860715.0,
+      "step": 3470
+    },
+    {
+      "entropy": 1.0174571394920349,
+      "epoch": 4.68371467025572,
+      "grad_norm": 1.2106678485870361,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.037718603014945985,
+      "mean_token_accuracy": 0.9855880260467529,
+      "num_tokens": 63041029.0,
+      "step": 3480
+    },
+    {
+      "entropy": 1.0129252314567565,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.1571707725524902,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.03724310994148254,
+      "mean_token_accuracy": 0.9865262687206269,
+      "num_tokens": 63222379.0,
+      "step": 3490
+    },
+    {
+      "entropy": 1.0124369859695435,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.0599361658096313,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.037783673405647276,
+      "mean_token_accuracy": 0.9859567046165466,
+      "num_tokens": 63403328.0,
+      "step": 3500
+    },
+    {
+      "entropy": 1.0041921436786652,
+      "epoch": 4.724091520861373,
+      "grad_norm": 1.2054452896118164,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.03585146069526672,
+      "mean_token_accuracy": 0.9868905484676361,
+      "num_tokens": 63584598.0,
+      "step": 3510
+    },
+    {
+      "entropy": 1.0015427589416503,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.5075141191482544,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.037428200244903564,
+      "mean_token_accuracy": 0.9867083311080933,
+      "num_tokens": 63766203.0,
+      "step": 3520
+    },
+    {
+      "entropy": 1.0039161920547486,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.310390830039978,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.037494581937789914,
+      "mean_token_accuracy": 0.9862947940826416,
+      "num_tokens": 63947714.0,
+      "step": 3530
+    },
+    {
+      "entropy": 1.0049733996391297,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.2912139892578125,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.03803396224975586,
+      "mean_token_accuracy": 0.9860390484333038,
+      "num_tokens": 64128798.0,
+      "step": 3540
+    },
+    {
+      "entropy": 1.0127450585365296,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.001133680343628,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0351622611284256,
+      "mean_token_accuracy": 0.9875370979309082,
+      "num_tokens": 64309577.0,
+      "step": 3550
+    },
+    {
+      "entropy": 1.0070195615291595,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 1.1870886087417603,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.036404851078987124,
+      "mean_token_accuracy": 0.9871752023696899,
+      "num_tokens": 64490707.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.9983750462532044,
+      "epoch": 4.804845222072679,
+      "grad_norm": 1.2374205589294434,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.03696496784687042,
+      "mean_token_accuracy": 0.9865773975849151,
+      "num_tokens": 64672104.0,
+      "step": 3570
+    },
+    {
+      "entropy": 1.0151559710502625,
+      "epoch": 4.818304172274562,
+      "grad_norm": 1.2001149654388428,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.03725983202457428,
+      "mean_token_accuracy": 0.9862341761589051,
+      "num_tokens": 64853353.0,
+      "step": 3580
+    },
+    {
+      "entropy": 1.0136774778366089,
+      "epoch": 4.831763122476447,
+      "grad_norm": 1.0128737688064575,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.03615986704826355,
+      "mean_token_accuracy": 0.9872736394405365,
+      "num_tokens": 65035009.0,
+      "step": 3590
+    },
+    {
+      "entropy": 1.0082051515579225,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.2552846670150757,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.03561249375343323,
+      "mean_token_accuracy": 0.9868618845939636,
+      "num_tokens": 65215981.0,
+      "step": 3600
+    },
+    {
+      "entropy": 1.0209908723831176,
+      "epoch": 4.858681022880216,
+      "grad_norm": 1.1914079189300537,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.03721153438091278,
+      "mean_token_accuracy": 0.9862311720848084,
+      "num_tokens": 65397088.0,
+      "step": 3610
+    },
+    {
+      "entropy": 1.014677894115448,
+      "epoch": 4.872139973082099,
+      "grad_norm": 1.3584535121917725,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.03897300064563751,
+      "mean_token_accuracy": 0.9851836740970612,
+      "num_tokens": 65578262.0,
+      "step": 3620
+    },
+    {
+      "entropy": 1.0207050919532776,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.2754287719726562,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.037926393747329715,
+      "mean_token_accuracy": 0.9859795570373535,
+      "num_tokens": 65758956.0,
+      "step": 3630
+    },
+    {
+      "entropy": 1.008733856678009,
+      "epoch": 4.899057873485868,
+      "grad_norm": 1.0815879106521606,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.036518561840057376,
+      "mean_token_accuracy": 0.9870649397373199,
+      "num_tokens": 65939970.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.9995620250701904,
+      "epoch": 4.912516823687753,
+      "grad_norm": 1.1773526668548584,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.035269084572792056,
+      "mean_token_accuracy": 0.9873197913169861,
+      "num_tokens": 66121045.0,
+      "step": 3650
+    },
+    {
+      "entropy": 1.0036249816417695,
+      "epoch": 4.925975773889636,
+      "grad_norm": 1.0941131114959717,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.0353422611951828,
+      "mean_token_accuracy": 0.9870408117771149,
+      "num_tokens": 66302324.0,
+      "step": 3660
+    },
+    {
+      "entropy": 1.004719352722168,
+      "epoch": 4.939434724091521,
+      "grad_norm": 1.5107269287109375,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.03743018507957459,
+      "mean_token_accuracy": 0.9857985079288483,
+      "num_tokens": 66483916.0,
+      "step": 3670
+    },
+    {
+      "entropy": 1.0076164484024048,
+      "epoch": 4.952893674293405,
+      "grad_norm": 1.3877427577972412,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.03857019543647766,
+      "mean_token_accuracy": 0.9856370270252228,
+      "num_tokens": 66664390.0,
+      "step": 3680
+    },
+    {
+      "entropy": 1.015517818927765,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1693767309188843,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.03634478151798248,
+      "mean_token_accuracy": 0.9868648707866668,
+      "num_tokens": 66845574.0,
+      "step": 3690
+    },
+    {
+      "entropy": 1.011656105518341,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.082209587097168,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.03513925671577454,
+      "mean_token_accuracy": 0.9872389912605286,
+      "num_tokens": 67027557.0,
+      "step": 3700
+    },
+    {
+      "entropy": 1.0161120653152467,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.266615629196167,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.0379630446434021,
+      "mean_token_accuracy": 0.9865313708782196,
+      "num_tokens": 67208088.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 1.0219634957374282,
+      "eval_loss": 0.14612138271331787,
+      "eval_mean_token_accuracy": 0.9504923186484416,
+      "eval_num_tokens": 67298266.0,
+      "eval_runtime": 12.6802,
+      "eval_samples_per_second": 394.317,
+      "eval_steps_per_second": 12.382,
+      "step": 3715
+    },
+    {
+      "entropy": 1.019774079322815,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8909980654716492,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.02882351279258728,
+      "mean_token_accuracy": 0.9900537073612213,
+      "num_tokens": 67388858.0,
+      "step": 3720
+    },
+    {
+      "entropy": 1.0057315409183503,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9996656775474548,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.02233442962169647,
+      "mean_token_accuracy": 0.9922977447509765,
+      "num_tokens": 67570084.0,
+      "step": 3730
+    },
+    {
+      "entropy": 1.0058627128601074,
+      "epoch": 5.03364737550471,
+      "grad_norm": 1.0265147686004639,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.021967282891273497,
+      "mean_token_accuracy": 0.9923401355743409,
+      "num_tokens": 67751587.0,
+      "step": 3740
+    },
+    {
+      "entropy": 1.0083157896995545,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 1.0139433145523071,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.022531555593013765,
+      "mean_token_accuracy": 0.9930226564407348,
+      "num_tokens": 67933269.0,
+      "step": 3750
+    },
+    {
+      "entropy": 1.0119245886802672,
+      "epoch": 5.060565275908479,
+      "grad_norm": 1.0655452013015747,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.021685637533664703,
+      "mean_token_accuracy": 0.9925794541835785,
+      "num_tokens": 68113971.0,
+      "step": 3760
+    },
+    {
+      "entropy": 1.0082412719726563,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.268751859664917,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.021513286232948303,
+      "mean_token_accuracy": 0.9927303433418274,
+      "num_tokens": 68295632.0,
+      "step": 3770
+    },
+    {
+      "entropy": 1.012417995929718,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.9899319410324097,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.02352543771266937,
+      "mean_token_accuracy": 0.992375648021698,
+      "num_tokens": 68476794.0,
+      "step": 3780
+    },
+    {
+      "entropy": 1.0069364666938783,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 1.1964375972747803,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.021382461488246917,
+      "mean_token_accuracy": 0.9928456962108612,
+      "num_tokens": 68658249.0,
+      "step": 3790
+    },
+    {
+      "entropy": 1.0040670156478881,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0445663928985596,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.02207571566104889,
+      "mean_token_accuracy": 0.9927122712135314,
+      "num_tokens": 68840525.0,
+      "step": 3800
+    },
+    {
+      "entropy": 1.0083145201206207,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.8981389403343201,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.022873447835445405,
+      "mean_token_accuracy": 0.9918712139129638,
+      "num_tokens": 69021896.0,
+      "step": 3810
+    },
+    {
+      "entropy": 1.0109572768211366,
+      "epoch": 5.141318977119784,
+      "grad_norm": 1.250670313835144,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.023577621579170226,
+      "mean_token_accuracy": 0.9919484734535218,
+      "num_tokens": 69203112.0,
+      "step": 3820
+    },
+    {
+      "entropy": 1.0164754271507264,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.3053207397460938,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.02436322420835495,
+      "mean_token_accuracy": 0.9921659469604492,
+      "num_tokens": 69384193.0,
+      "step": 3830
+    },
+    {
+      "entropy": 1.0034007966518401,
+      "epoch": 5.168236877523553,
+      "grad_norm": 1.4878787994384766,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.023177918791770936,
+      "mean_token_accuracy": 0.9919318974018096,
+      "num_tokens": 69565602.0,
+      "step": 3840
+    },
+    {
+      "entropy": 1.0074714839458465,
+      "epoch": 5.181695827725437,
+      "grad_norm": 1.0512391328811646,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.023409275710582732,
+      "mean_token_accuracy": 0.9918898940086365,
+      "num_tokens": 69746242.0,
+      "step": 3850
+    },
+    {
+      "entropy": 1.003154468536377,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.971832811832428,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.023223266005516052,
+      "mean_token_accuracy": 0.9920656085014343,
+      "num_tokens": 69927227.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.999264246225357,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.9266519546508789,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.02454002797603607,
+      "mean_token_accuracy": 0.9914694905281067,
+      "num_tokens": 70108580.0,
+      "step": 3870
+    },
+    {
+      "entropy": 1.002640813589096,
+      "epoch": 5.22207267833109,
+      "grad_norm": 1.5331412553787231,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.02458159625530243,
+      "mean_token_accuracy": 0.9919124245643616,
+      "num_tokens": 70288960.0,
+      "step": 3880
+    },
+    {
+      "entropy": 1.007998287677765,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.3164633512496948,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.02421087771654129,
+      "mean_token_accuracy": 0.9917516648769379,
+      "num_tokens": 70470266.0,
+      "step": 3890
+    },
+    {
+      "entropy": 1.0089037418365479,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1788280010223389,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.023224952816963195,
+      "mean_token_accuracy": 0.9921185493469238,
+      "num_tokens": 70650975.0,
+      "step": 3900
+    },
+    {
+      "entropy": 1.001932644844055,
+      "epoch": 5.262449528936743,
+      "grad_norm": 1.2484381198883057,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.02366664707660675,
+      "mean_token_accuracy": 0.9917596876621246,
+      "num_tokens": 70832549.0,
+      "step": 3910
+    },
+    {
+      "entropy": 1.004903507232666,
+      "epoch": 5.275908479138627,
+      "grad_norm": 1.1852507591247559,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.02296685576438904,
+      "mean_token_accuracy": 0.9918978750705719,
+      "num_tokens": 71014296.0,
+      "step": 3920
+    },
+    {
+      "entropy": 1.004357820749283,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.9933306574821472,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.02361467629671097,
+      "mean_token_accuracy": 0.991608303785324,
+      "num_tokens": 71195526.0,
+      "step": 3930
+    },
+    {
+      "entropy": 1.0020038604736328,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.4304245710372925,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.02268100529909134,
+      "mean_token_accuracy": 0.9917695164680481,
+      "num_tokens": 71377269.0,
+      "step": 3940
+    },
+    {
+      "entropy": 1.012037169933319,
+      "epoch": 5.31628532974428,
+      "grad_norm": 1.074124813079834,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.023454257845878602,
+      "mean_token_accuracy": 0.9922087669372559,
+      "num_tokens": 71558158.0,
+      "step": 3950
+    },
+    {
+      "entropy": 1.0094320178031921,
+      "epoch": 5.329744279946164,
+      "grad_norm": 1.0919133424758911,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.022751571238040925,
+      "mean_token_accuracy": 0.9921496629714965,
+      "num_tokens": 71738847.0,
+      "step": 3960
+    },
+    {
+      "entropy": 1.0063781797885896,
+      "epoch": 5.343203230148048,
+      "grad_norm": 1.1300997734069824,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.023900465667247774,
+      "mean_token_accuracy": 0.9914113521575928,
+      "num_tokens": 71918875.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.9972042560577392,
+      "epoch": 5.356662180349932,
+      "grad_norm": 1.165519118309021,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.021918240189552306,
+      "mean_token_accuracy": 0.9926528394222259,
+      "num_tokens": 72100551.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.9954554200172424,
+      "epoch": 5.370121130551817,
+      "grad_norm": 1.342820167541504,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.024170228838920595,
+      "mean_token_accuracy": 0.9913744986057281,
+      "num_tokens": 72281855.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.9931409180164337,
+      "epoch": 5.383580080753701,
+      "grad_norm": 1.2873902320861816,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.02334785759449005,
+      "mean_token_accuracy": 0.9921288907527923,
+      "num_tokens": 72462916.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.9915881812572479,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.9761195778846741,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.02302718609571457,
+      "mean_token_accuracy": 0.9924728453159333,
+      "num_tokens": 72644666.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.9931966602802277,
+      "epoch": 5.410497981157469,
+      "grad_norm": 1.4468477964401245,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.023682811856269838,
+      "mean_token_accuracy": 0.9917364478111267,
+      "num_tokens": 72825808.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.9975718319416046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.1559572219848633,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.02247478663921356,
+      "mean_token_accuracy": 0.992560338973999,
+      "num_tokens": 73007923.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.999228036403656,
+      "epoch": 5.437415881561238,
+      "grad_norm": 1.1230353116989136,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.022295859456062318,
+      "mean_token_accuracy": 0.9922321498394012,
+      "num_tokens": 73189074.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.9897105693817139,
+      "epoch": 5.450874831763122,
+      "grad_norm": 1.413734793663025,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.02201971709728241,
+      "mean_token_accuracy": 0.9927073001861573,
+      "num_tokens": 73369918.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.9986266851425171,
+      "epoch": 5.464333781965006,
+      "grad_norm": 1.0815346240997314,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.02358766496181488,
+      "mean_token_accuracy": 0.9911752700805664,
+      "num_tokens": 73550625.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.9953943610191345,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0506970882415771,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.022177645564079286,
+      "mean_token_accuracy": 0.9923022150993347,
+      "num_tokens": 73731436.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.9954936504364014,
+      "epoch": 5.491251682368775,
+      "grad_norm": 1.4665249586105347,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.024948063492774963,
+      "mean_token_accuracy": 0.9913177132606507,
+      "num_tokens": 73912097.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.9986696183681488,
+      "epoch": 5.504710632570659,
+      "grad_norm": 1.158179759979248,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.02555338144302368,
+      "mean_token_accuracy": 0.9908900320529938,
+      "num_tokens": 74093602.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.9955579698085785,
+      "epoch": 5.518169582772543,
+      "grad_norm": 1.0024714469909668,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.023432795703411103,
+      "mean_token_accuracy": 0.9919674038887024,
+      "num_tokens": 74274664.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.994187182188034,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.9290911555290222,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.023194606602191924,
+      "mean_token_accuracy": 0.9920254707336426,
+      "num_tokens": 74456436.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.9927510738372802,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.8986559510231018,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.02163631021976471,
+      "mean_token_accuracy": 0.9927416622638703,
+      "num_tokens": 74637700.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.9877451896667481,
+      "epoch": 5.558546433378196,
+      "grad_norm": 1.0314558744430542,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.0209101140499115,
+      "mean_token_accuracy": 0.9930387079715729,
+      "num_tokens": 74819310.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.9903139114379883,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 1.1819736957550049,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.02219167649745941,
+      "mean_token_accuracy": 0.992407250404358,
+      "num_tokens": 74999907.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.9891258835792541,
+      "epoch": 5.585464333781965,
+      "grad_norm": 1.0412501096725464,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.02201283872127533,
+      "mean_token_accuracy": 0.992091816663742,
+      "num_tokens": 75181189.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.9870602369308472,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0461134910583496,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.023040324449539185,
+      "mean_token_accuracy": 0.992098081111908,
+      "num_tokens": 75362769.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.9888905882835388,
+      "epoch": 5.612382234185733,
+      "grad_norm": 1.086546540260315,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.021401138603687288,
+      "mean_token_accuracy": 0.992308247089386,
+      "num_tokens": 75544519.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.9940549910068512,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 1.3897169828414917,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.023297235369682312,
+      "mean_token_accuracy": 0.9920303761959076,
+      "num_tokens": 75725651.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.9897378265857697,
+      "epoch": 5.639300134589502,
+      "grad_norm": 1.283887267112732,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.024343529343605043,
+      "mean_token_accuracy": 0.9916397035121918,
+      "num_tokens": 75907036.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.9989661812782288,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.9090050458908081,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.022062870860099792,
+      "mean_token_accuracy": 0.9921731591224671,
+      "num_tokens": 76088248.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.9966195583343506,
+      "epoch": 5.66621803499327,
+      "grad_norm": 1.2091970443725586,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.023516717553138732,
+      "mean_token_accuracy": 0.9918879508972168,
+      "num_tokens": 76269704.0,
+      "step": 4210
+    },
+    {
+      "entropy": 1.0081237196922301,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 1.0648062229156494,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.021994276344776152,
+      "mean_token_accuracy": 0.9925018906593323,
+      "num_tokens": 76451416.0,
+      "step": 4220
+    },
+    {
+      "entropy": 1.0061897933483124,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.188748836517334,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.022657927870750428,
+      "mean_token_accuracy": 0.9919365108013153,
+      "num_tokens": 76632827.0,
+      "step": 4230
+    },
+    {
+      "entropy": 1.0024715662002563,
+      "epoch": 5.706594885598923,
+      "grad_norm": 1.2140473127365112,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.02203955501317978,
+      "mean_token_accuracy": 0.9925529718399048,
+      "num_tokens": 76813534.0,
+      "step": 4240
+    },
+    {
+      "entropy": 1.0011352360248567,
+      "epoch": 5.720053835800807,
+      "grad_norm": 1.389479398727417,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.022718487679958342,
+      "mean_token_accuracy": 0.9922778069972992,
+      "num_tokens": 76994327.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.9971556246280671,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 1.1441208124160767,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.024178403615951537,
+      "mean_token_accuracy": 0.9913869142532349,
+      "num_tokens": 77174987.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.9961654126644135,
+      "epoch": 5.746971736204576,
+      "grad_norm": 1.1283735036849976,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.023389075696468354,
+      "mean_token_accuracy": 0.9921939969062805,
+      "num_tokens": 77356043.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.9927134871482849,
+      "epoch": 5.76043068640646,
+      "grad_norm": 1.201904535293579,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.022739484906196594,
+      "mean_token_accuracy": 0.9921142935752869,
+      "num_tokens": 77536396.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.9919731795787812,
+      "epoch": 5.773889636608344,
+      "grad_norm": 1.145776391029358,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.020524097979068755,
+      "mean_token_accuracy": 0.9926942765712738,
+      "num_tokens": 77717339.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.9935499131679535,
+      "epoch": 5.787348586810229,
+      "grad_norm": 1.099414348602295,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.02070924937725067,
+      "mean_token_accuracy": 0.9928180515766144,
+      "num_tokens": 77898360.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.9875575602054596,
+      "epoch": 5.800807537012113,
+      "grad_norm": 1.1510778665542603,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.02202107310295105,
+      "mean_token_accuracy": 0.9921730697154999,
+      "num_tokens": 78079699.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.9855418741703034,
+      "epoch": 5.814266487213997,
+      "grad_norm": 1.0748379230499268,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.022809866070747375,
+      "mean_token_accuracy": 0.9919199466705322,
+      "num_tokens": 78260900.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.977304744720459,
+      "epoch": 5.827725437415881,
+      "grad_norm": 1.091854453086853,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.02268455773591995,
+      "mean_token_accuracy": 0.9918094217777252,
+      "num_tokens": 78442168.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.9767287075519562,
+      "epoch": 5.841184387617766,
+      "grad_norm": 1.1712645292282104,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.022726841270923615,
+      "mean_token_accuracy": 0.9922167181968689,
+      "num_tokens": 78623589.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.974297159910202,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.9508016109466553,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.021290253102779388,
+      "mean_token_accuracy": 0.9926727712154388,
+      "num_tokens": 78804892.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.9795810282230377,
+      "epoch": 5.868102288021534,
+      "grad_norm": 1.3127822875976562,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.022329577803611757,
+      "mean_token_accuracy": 0.9927741348743438,
+      "num_tokens": 78986204.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.9879997313022614,
+      "epoch": 5.881561238223418,
+      "grad_norm": 1.0200371742248535,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.022206541895866395,
+      "mean_token_accuracy": 0.9921830892562866,
+      "num_tokens": 79166824.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.9801069736480713,
+      "epoch": 5.895020188425303,
+      "grad_norm": 1.4769432544708252,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.02101587802171707,
+      "mean_token_accuracy": 0.9930795848369598,
+      "num_tokens": 79347228.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.9727071464061737,
+      "epoch": 5.908479138627187,
+      "grad_norm": 1.7034555673599243,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.021690081059932708,
+      "mean_token_accuracy": 0.9926194071769714,
+      "num_tokens": 79529483.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.9782697260379791,
+      "epoch": 5.921938088829071,
+      "grad_norm": 1.5307973623275757,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.023112425208091737,
+      "mean_token_accuracy": 0.9919495701789856,
+      "num_tokens": 79710571.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.9831463754177093,
+      "epoch": 5.935397039030955,
+      "grad_norm": 1.1024526357650757,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.021427598595619202,
+      "mean_token_accuracy": 0.9924166023731231,
+      "num_tokens": 79890685.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.9811958551406861,
+      "epoch": 5.94885598923284,
+      "grad_norm": 1.7247848510742188,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.02339867949485779,
+      "mean_token_accuracy": 0.9916171073913574,
+      "num_tokens": 80071331.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.9830600321292877,
+      "epoch": 5.962314939434724,
+      "grad_norm": 1.0638445615768433,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.02204108238220215,
+      "mean_token_accuracy": 0.9925127327442169,
+      "num_tokens": 80251525.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.9840480864048005,
+      "epoch": 5.975773889636608,
+      "grad_norm": 1.1205570697784424,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.020443232357501985,
+      "mean_token_accuracy": 0.9932060956954956,
+      "num_tokens": 80432001.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.9841626048088074,
+      "epoch": 5.989232839838492,
+      "grad_norm": 1.2928433418273926,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.021673919260501863,
+      "mean_token_accuracy": 0.9931873917579651,
+      "num_tokens": 80613075.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.9904457463580332,
+      "eval_loss": 0.161154642701149,
+      "eval_mean_token_accuracy": 0.9516840383505366,
+      "eval_num_tokens": 80757887.0,
+      "eval_runtime": 12.7345,
+      "eval_samples_per_second": 392.633,
+      "eval_steps_per_second": 12.329,
+      "step": 4458
+    },
+    {
+      "entropy": 0.9850570559501648,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.8859100341796875,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.018883128464221955,
+      "mean_token_accuracy": 0.9933207035064697,
+      "num_tokens": 80794295.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.9796536922454834,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.8121339082717896,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.01288142502307892,
+      "mean_token_accuracy": 0.9960863590240479,
+      "num_tokens": 80975379.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.9699741184711457,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.8923460841178894,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.012284834682941437,
+      "mean_token_accuracy": 0.9960344016551972,
+      "num_tokens": 81155713.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.9629065454006195,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.6698539853096008,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.011575647443532944,
+      "mean_token_accuracy": 0.9962593257427216,
+      "num_tokens": 81336446.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.9579961061477661,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.7555826306343079,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.011635755002498627,
+      "mean_token_accuracy": 0.9965786218643189,
+      "num_tokens": 81517969.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.96330127120018,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.8983458280563354,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.012715654075145721,
+      "mean_token_accuracy": 0.9959319531917572,
+      "num_tokens": 81699419.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.9640884518623352,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.9172118902206421,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.012482476234436036,
+      "mean_token_accuracy": 0.9966461837291718,
+      "num_tokens": 81880956.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.9678717851638794,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.8765719532966614,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.012080545723438262,
+      "mean_token_accuracy": 0.9964690268039703,
+      "num_tokens": 82061897.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.9656756520271301,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.746626079082489,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.012495596706867219,
+      "mean_token_accuracy": 0.9959208548069001,
+      "num_tokens": 82242847.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.9631998062133789,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.9732807874679565,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.01309359222650528,
+      "mean_token_accuracy": 0.9960879981517792,
+      "num_tokens": 82424182.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.9686778664588929,
+      "epoch": 6.137281292059219,
+      "grad_norm": 1.040855884552002,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.013380664587020873,
+      "mean_token_accuracy": 0.9955641269683838,
+      "num_tokens": 82604728.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.9661992788314819,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 1.209067940711975,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.013264504075050355,
+      "mean_token_accuracy": 0.9958661019802093,
+      "num_tokens": 82786184.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.9628405570983887,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.9056928157806396,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.011586685478687287,
+      "mean_token_accuracy": 0.9962525606155396,
+      "num_tokens": 82968127.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.9664590656757355,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.8662444949150085,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.013227257132530212,
+      "mean_token_accuracy": 0.9958321928977967,
+      "num_tokens": 83149016.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.9584677755832672,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.8611947894096375,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.012484049052000045,
+      "mean_token_accuracy": 0.9961173057556152,
+      "num_tokens": 83330474.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.9613959610462188,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 1.0272881984710693,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.013143527507781982,
+      "mean_token_accuracy": 0.995938628911972,
+      "num_tokens": 83512502.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.9671844780445099,
+      "epoch": 6.218034993270525,
+      "grad_norm": 1.038604497909546,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.015797269344329835,
+      "mean_token_accuracy": 0.9949265837669372,
+      "num_tokens": 83693427.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.9725019454956054,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.7591229677200317,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.01309441477060318,
+      "mean_token_accuracy": 0.9960084438323975,
+      "num_tokens": 83874886.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.9710726916790009,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.9521571397781372,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.012572157382965087,
+      "mean_token_accuracy": 0.9960904121398926,
+      "num_tokens": 84055459.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.9712884068489075,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.812011182308197,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.014366535842418671,
+      "mean_token_accuracy": 0.9953844547271729,
+      "num_tokens": 84236017.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.9648497045040131,
+      "epoch": 6.271870794078062,
+      "grad_norm": 1.648348093032837,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.014780099689960479,
+      "mean_token_accuracy": 0.9955918669700623,
+      "num_tokens": 84417271.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.956806218624115,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.9592521786689758,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.014198371767997741,
+      "mean_token_accuracy": 0.9954844176769256,
+      "num_tokens": 84599326.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.9628090143203736,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.9466222524642944,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.013178789615631103,
+      "mean_token_accuracy": 0.995667016506195,
+      "num_tokens": 84779914.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.9617082297801971,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.9156956672668457,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.011176959425210953,
+      "mean_token_accuracy": 0.9965419709682465,
+      "num_tokens": 84960572.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.9526888430118561,
+      "epoch": 6.325706594885599,
+      "grad_norm": 1.107490062713623,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.01246263161301613,
+      "mean_token_accuracy": 0.9961843729019165,
+      "num_tokens": 85141667.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.952410751581192,
+      "epoch": 6.339165545087483,
+      "grad_norm": 1.06051766872406,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.01356920450925827,
+      "mean_token_accuracy": 0.9956991016864777,
+      "num_tokens": 85323361.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.9601339221000671,
+      "epoch": 6.352624495289367,
+      "grad_norm": 1.0533535480499268,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.012942484021186829,
+      "mean_token_accuracy": 0.9960961997509002,
+      "num_tokens": 85503828.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.9630769610404968,
+      "epoch": 6.366083445491252,
+      "grad_norm": 1.2243846654891968,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.012688860297203064,
+      "mean_token_accuracy": 0.9956350862979889,
+      "num_tokens": 85684804.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.9601665258407592,
+      "epoch": 6.379542395693136,
+      "grad_norm": 1.0219420194625854,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.012651938199996948,
+      "mean_token_accuracy": 0.9959766507148743,
+      "num_tokens": 85866033.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.9643365681171417,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.8759870529174805,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.013052919507026672,
+      "mean_token_accuracy": 0.9956570088863372,
+      "num_tokens": 86046793.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.9643962621688843,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.8771425485610962,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.013016942143440246,
+      "mean_token_accuracy": 0.9956202030181884,
+      "num_tokens": 86227594.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.9684974253177643,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.9857693314552307,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.01260487139225006,
+      "mean_token_accuracy": 0.9959571301937103,
+      "num_tokens": 86409012.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.9683425962924957,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.9220049381256104,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.012756258249282837,
+      "mean_token_accuracy": 0.9961577892303467,
+      "num_tokens": 86589770.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.9730850279331207,
+      "epoch": 6.446837146702557,
+      "grad_norm": 1.1092842817306519,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.011953817307949066,
+      "mean_token_accuracy": 0.9963648617267609,
+      "num_tokens": 86771224.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.9717083096504211,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.9012715220451355,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.0137409508228302,
+      "mean_token_accuracy": 0.9956796884536743,
+      "num_tokens": 86952165.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.9706744968891143,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.8237823247909546,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.013622400164604188,
+      "mean_token_accuracy": 0.9958401381969452,
+      "num_tokens": 87133948.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.968240886926651,
+      "epoch": 6.48721399730821,
+      "grad_norm": 0.9172955751419067,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.012218748033046723,
+      "mean_token_accuracy": 0.9961099803447724,
+      "num_tokens": 87315186.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.9701445996761322,
+      "epoch": 6.500672947510094,
+      "grad_norm": 1.1613796949386597,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.011357608437538146,
+      "mean_token_accuracy": 0.9961966574192047,
+      "num_tokens": 87496318.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.9753403663635254,
+      "epoch": 6.514131897711978,
+      "grad_norm": 1.0601547956466675,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.01224931925535202,
+      "mean_token_accuracy": 0.9964543879032135,
+      "num_tokens": 87676690.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.9595320463180542,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.8927657604217529,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.012476099282503128,
+      "mean_token_accuracy": 0.9961197137832641,
+      "num_tokens": 87858122.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.9639881610870361,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.9392650127410889,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.013359837234020233,
+      "mean_token_accuracy": 0.9956938862800598,
+      "num_tokens": 88039349.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.961546677350998,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.709064781665802,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.012232375890016555,
+      "mean_token_accuracy": 0.9963175058364868,
+      "num_tokens": 88221085.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.966980516910553,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 1.1382325887680054,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.013281235098838806,
+      "mean_token_accuracy": 0.9955899178981781,
+      "num_tokens": 88401750.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.967098617553711,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.7260605692863464,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.011995580792427064,
+      "mean_token_accuracy": 0.9963020443916321,
+      "num_tokens": 88582188.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.9611273169517517,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.9492855668067932,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.011460936814546584,
+      "mean_token_accuracy": 0.9965442717075348,
+      "num_tokens": 88764018.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.964341527223587,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.7225981950759888,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.011858727037906646,
+      "mean_token_accuracy": 0.996355015039444,
+      "num_tokens": 88944988.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.9631503582000732,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 0.996126651763916,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.013101565837860107,
+      "mean_token_accuracy": 0.9957616865634918,
+      "num_tokens": 89125768.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.9629738807678223,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.8971033096313477,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.012153981626033783,
+      "mean_token_accuracy": 0.9961553812026978,
+      "num_tokens": 89306694.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.9592839002609252,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.8334258198738098,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.012248124927282333,
+      "mean_token_accuracy": 0.9960185050964355,
+      "num_tokens": 89487628.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.9580614447593689,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.7489107251167297,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.012415885925292969,
+      "mean_token_accuracy": 0.9960191190242768,
+      "num_tokens": 89668907.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.9544502675533295,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.9663529992103577,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.012825769186019898,
+      "mean_token_accuracy": 0.9958091497421264,
+      "num_tokens": 89850200.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.9588588118553162,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.7662053108215332,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.012289515137672425,
+      "mean_token_accuracy": 0.9956038355827331,
+      "num_tokens": 90030738.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.9632011234760285,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.8352729082107544,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.01209896057844162,
+      "mean_token_accuracy": 0.9960383713245392,
+      "num_tokens": 90211066.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.962366932630539,
+      "epoch": 6.716016150740242,
+      "grad_norm": 1.271884799003601,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.013325554132461549,
+      "mean_token_accuracy": 0.9955421090126038,
+      "num_tokens": 90392202.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.9630074858665466,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.806925892829895,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.012963607907295227,
+      "mean_token_accuracy": 0.9955462396144867,
+      "num_tokens": 90573271.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.9631992042064667,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.9406610131263733,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.012753760814666748,
+      "mean_token_accuracy": 0.9954385042190552,
+      "num_tokens": 90753836.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.9599998474121094,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.7755982875823975,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.011351206153631211,
+      "mean_token_accuracy": 0.9965151965618133,
+      "num_tokens": 90935528.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.961510854959488,
+      "epoch": 6.769851951547779,
+      "grad_norm": 1.223897099494934,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.012862575054168702,
+      "mean_token_accuracy": 0.995742017030716,
+      "num_tokens": 91116520.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.9597096800804138,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.8847037553787231,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.01192454993724823,
+      "mean_token_accuracy": 0.9964925825595856,
+      "num_tokens": 91297847.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.961575037240982,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.7097873687744141,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.012100750207901,
+      "mean_token_accuracy": 0.9963988482952117,
+      "num_tokens": 91479371.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.9660714268684387,
+      "epoch": 6.810228802153432,
+      "grad_norm": 0.8912429213523865,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.011809615790843964,
+      "mean_token_accuracy": 0.9961784362792969,
+      "num_tokens": 91660175.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.9656784057617187,
+      "epoch": 6.823687752355316,
+      "grad_norm": 1.2362871170043945,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.013092437386512756,
+      "mean_token_accuracy": 0.9954990804195404,
+      "num_tokens": 91840728.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.9630216419696808,
+      "epoch": 6.837146702557201,
+      "grad_norm": 1.0081552267074585,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.013006564974784852,
+      "mean_token_accuracy": 0.995445728302002,
+      "num_tokens": 92021819.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.9650352597236633,
+      "epoch": 6.850605652759085,
+      "grad_norm": 1.0923664569854736,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.012209924310445786,
+      "mean_token_accuracy": 0.9961765468120575,
+      "num_tokens": 92203146.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.9654329180717468,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.8630998730659485,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.01286689043045044,
+      "mean_token_accuracy": 0.9955997586250305,
+      "num_tokens": 92384274.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.9633705973625183,
+      "epoch": 6.877523553162853,
+      "grad_norm": 0.9883823394775391,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.012318137288093566,
+      "mean_token_accuracy": 0.9964197039604187,
+      "num_tokens": 92564922.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.9619575142860413,
+      "epoch": 6.890982503364738,
+      "grad_norm": 1.0694894790649414,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.011325329542160034,
+      "mean_token_accuracy": 0.9967617630958557,
+      "num_tokens": 92745952.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.9608132243156433,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.850687563419342,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.012749406695365905,
+      "mean_token_accuracy": 0.9958359479904175,
+      "num_tokens": 92926966.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.9643485128879548,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.8775151968002319,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.012118304520845414,
+      "mean_token_accuracy": 0.996504932641983,
+      "num_tokens": 93108549.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.9634061992168427,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.925463855266571,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.011662732064723968,
+      "mean_token_accuracy": 0.9965654671192169,
+      "num_tokens": 93290512.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.9620663046836853,
+      "epoch": 6.944818304172275,
+      "grad_norm": 1.1143932342529297,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.01120457649230957,
+      "mean_token_accuracy": 0.9962508201599121,
+      "num_tokens": 93472398.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.9628985583782196,
+      "epoch": 6.958277254374159,
+      "grad_norm": 1.0271083116531372,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.012035199999809265,
+      "mean_token_accuracy": 0.996053409576416,
+      "num_tokens": 93654111.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.9658466935157776,
+      "epoch": 6.971736204576043,
+      "grad_norm": 1.0225892066955566,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.01229911893606186,
+      "mean_token_accuracy": 0.995949512720108,
+      "num_tokens": 93835692.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.9601108253002166,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.7876276969909668,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.011063569784164428,
+      "mean_token_accuracy": 0.9966089725494385,
+      "num_tokens": 94018119.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.9596541881561279,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.9803016185760498,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.011228029429912568,
+      "mean_token_accuracy": 0.9964218616485596,
+      "num_tokens": 94199483.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.9689761833020836,
+      "eval_loss": 0.1798103004693985,
+      "eval_mean_token_accuracy": 0.950934177371347,
+      "eval_num_tokens": 94217754.0,
+      "eval_runtime": 12.6557,
+      "eval_samples_per_second": 395.079,
+      "eval_steps_per_second": 12.405,
+      "step": 5201
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.455551417591005e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-5201/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-5201/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5201/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/checkpoint-5944/README.md b/spanish_L20_a100/seed_42/checkpoint-5944/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5944/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-5944/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-5944/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5944/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-5944/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-5944/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..41c6530f78f3bbce5adbfa01034015f31fcf73d0
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5944/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd9222ce1246a665fc58c694ddfba21c4c72b3479be0aa120529e2fd6670b949
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-5944/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-5944/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5944/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-5944/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-5944/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5944/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-5944/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-5944/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5944/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-5944/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-5944/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ee6d04799aaded032c9e5181f3267992f95bc05d
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5944/trainer_state.json
@@ -0,0 +1,6062 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.0,
+  "eval_steps": 500,
+  "global_step": 5944,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    },
+    {
+      "entropy": 1.049056887626648,
+      "epoch": 1.009421265141319,
+      "grad_norm": 1.4189648628234863,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.11506924629211426,
+      "mean_token_accuracy": 0.9560760498046875,
+      "num_tokens": 13586033.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0126458764076234,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.6394226551055908,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.10137178897857665,
+      "mean_token_accuracy": 0.9604915261268616,
+      "num_tokens": 13766883.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0174824237823485,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.3608797788619995,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.10521939992904664,
+      "mean_token_accuracy": 0.9592583775520325,
+      "num_tokens": 13949410.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0318764686584472,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.1440019607543945,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.1039464831352234,
+      "mean_token_accuracy": 0.9593731820583343,
+      "num_tokens": 14131283.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0228763103485108,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.609320878982544,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.10670688152313232,
+      "mean_token_accuracy": 0.9587047874927521,
+      "num_tokens": 14312694.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0123905301094056,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.4128942489624023,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.10546157360076905,
+      "mean_token_accuracy": 0.9586312890052795,
+      "num_tokens": 14493193.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.010130923986435,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.4825445413589478,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.09921822547912598,
+      "mean_token_accuracy": 0.9621170878410339,
+      "num_tokens": 14674762.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0216347932815553,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.2503618001937866,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.10619921684265136,
+      "mean_token_accuracy": 0.9584988236427308,
+      "num_tokens": 14856424.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.022511374950409,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.3209893703460693,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.10425064563751221,
+      "mean_token_accuracy": 0.9595829844474792,
+      "num_tokens": 15037452.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.007770985364914,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.4830716848373413,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.10189464092254638,
+      "mean_token_accuracy": 0.9609117627143859,
+      "num_tokens": 15218205.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0228153944015503,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.3191078901290894,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.10298972129821778,
+      "mean_token_accuracy": 0.9599695146083832,
+      "num_tokens": 15399322.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0262911200523377,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.2050191164016724,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.10532126426696778,
+      "mean_token_accuracy": 0.9590177297592163,
+      "num_tokens": 15581668.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0124290227890014,
+      "epoch": 1.17092866756393,
+      "grad_norm": 1.3545384407043457,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.10171200037002563,
+      "mean_token_accuracy": 0.9587810635566711,
+      "num_tokens": 15762970.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.028192389011383,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.55021071434021,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.10711442232131958,
+      "mean_token_accuracy": 0.956982696056366,
+      "num_tokens": 15944164.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0354714155197144,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2573878765106201,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.10722451210021973,
+      "mean_token_accuracy": 0.9568670630455017,
+      "num_tokens": 16125049.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.034978187084198,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.4448883533477783,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.10492000579833985,
+      "mean_token_accuracy": 0.9593439042568207,
+      "num_tokens": 16306183.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0321480870246886,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.4060254096984863,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.1017595648765564,
+      "mean_token_accuracy": 0.9604197800159454,
+      "num_tokens": 16487439.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.03044091463089,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.6413203477859497,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.1032058596611023,
+      "mean_token_accuracy": 0.9589122951030731,
+      "num_tokens": 16667826.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0394478559494018,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3867346048355103,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.10721726417541504,
+      "mean_token_accuracy": 0.9581708788871766,
+      "num_tokens": 16848488.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0355630755424499,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 1.3304182291030884,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.10916879177093505,
+      "mean_token_accuracy": 0.9566857099533081,
+      "num_tokens": 17028787.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0394182324409484,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.4853013753890991,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.10515115261077881,
+      "mean_token_accuracy": 0.9584204435348511,
+      "num_tokens": 17209699.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0405551195144653,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.1959068775177002,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.10281100273132324,
+      "mean_token_accuracy": 0.9599630236625671,
+      "num_tokens": 17391070.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0196930646896363,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.2734276056289673,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.10447280406951905,
+      "mean_token_accuracy": 0.9587895989418029,
+      "num_tokens": 17572309.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0284824252128602,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.1344295740127563,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.11111283302307129,
+      "mean_token_accuracy": 0.9560421764850616,
+      "num_tokens": 17753814.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0308443427085876,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.0935126543045044,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.10474474430084228,
+      "mean_token_accuracy": 0.9583515226840973,
+      "num_tokens": 17934774.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.0283282995224,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.5144932270050049,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.104243803024292,
+      "mean_token_accuracy": 0.9581555545330047,
+      "num_tokens": 18116015.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.0355387568473815,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.276055932044983,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.10444833040237426,
+      "mean_token_accuracy": 0.9587780058383941,
+      "num_tokens": 18296690.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.0340025186538697,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.4408668279647827,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.10817761421203613,
+      "mean_token_accuracy": 0.9573222637176514,
+      "num_tokens": 18477491.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.0480108737945557,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2114640474319458,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.10477652549743652,
+      "mean_token_accuracy": 0.9591076672077179,
+      "num_tokens": 18658366.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.0303086400032044,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.3319413661956787,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.10648446083068848,
+      "mean_token_accuracy": 0.9581128239631653,
+      "num_tokens": 18839989.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0405295491218567,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.4326776266098022,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.10707225799560546,
+      "mean_token_accuracy": 0.9574429094791412,
+      "num_tokens": 19020331.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.037307357788086,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.2203212976455688,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.10811934471130372,
+      "mean_token_accuracy": 0.9569429934024811,
+      "num_tokens": 19201621.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.048385226726532,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.2597253322601318,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.10502862930297852,
+      "mean_token_accuracy": 0.9592942714691162,
+      "num_tokens": 19382819.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.0431596279144286,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.288861632347107,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.1091151475906372,
+      "mean_token_accuracy": 0.9570925652980804,
+      "num_tokens": 19563291.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.0325645685195923,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.1845893859863281,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.10389597415924072,
+      "mean_token_accuracy": 0.9585334539413453,
+      "num_tokens": 19744388.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.025480568408966,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 1.3404650688171387,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.10605425834655761,
+      "mean_token_accuracy": 0.9583600521087646,
+      "num_tokens": 19925613.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.020683228969574,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5598406791687012,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.10631901025772095,
+      "mean_token_accuracy": 0.9573191046714783,
+      "num_tokens": 20106747.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.0185366213321685,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3814117908477783,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.1074562907218933,
+      "mean_token_accuracy": 0.9571435689926148,
+      "num_tokens": 20287872.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0160030841827392,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 1.5130647420883179,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.10584371089935303,
+      "mean_token_accuracy": 0.9584454715251922,
+      "num_tokens": 20469530.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.0295212030410767,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.137473464012146,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.10609703063964844,
+      "mean_token_accuracy": 0.9590815663337707,
+      "num_tokens": 20650652.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.0236494064331054,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.4402672052383423,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.10043272972106934,
+      "mean_token_accuracy": 0.9601598799228668,
+      "num_tokens": 20832844.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0228356838226318,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.6836261749267578,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.10231537818908691,
+      "mean_token_accuracy": 0.9592978358268738,
+      "num_tokens": 21014766.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.0204253315925598,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.7377065420150757,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.10434420108795166,
+      "mean_token_accuracy": 0.9578610718250274,
+      "num_tokens": 21195755.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.0251757979393006,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.337659239768982,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.10218650102615356,
+      "mean_token_accuracy": 0.9588291764259338,
+      "num_tokens": 21376476.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.021351706981659,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.278275728225708,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.10114043951034546,
+      "mean_token_accuracy": 0.9596256852149964,
+      "num_tokens": 21558292.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.0222898125648499,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.1806602478027344,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.10140385627746581,
+      "mean_token_accuracy": 0.9601265788078308,
+      "num_tokens": 21739640.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0402090787887572,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.2378828525543213,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.10622456073760986,
+      "mean_token_accuracy": 0.9573592960834503,
+      "num_tokens": 21921425.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.0412330985069276,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3350801467895508,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.10371602773666382,
+      "mean_token_accuracy": 0.9588219523429871,
+      "num_tokens": 22102605.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.0534109234809876,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 1.4873039722442627,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.10807039737701415,
+      "mean_token_accuracy": 0.9577955365180969,
+      "num_tokens": 22283687.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0713044166564942,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.1694425344467163,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.11005204916000366,
+      "mean_token_accuracy": 0.9568802535533905,
+      "num_tokens": 22464813.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.0745116233825684,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.5362104177474976,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.10913199186325073,
+      "mean_token_accuracy": 0.955611252784729,
+      "num_tokens": 22646088.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0925905466079713,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.3478912115097046,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.10608880519866944,
+      "mean_token_accuracy": 0.9566390633583068,
+      "num_tokens": 22826677.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0899853229522705,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.238072395324707,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.10292150974273681,
+      "mean_token_accuracy": 0.9586134016513824,
+      "num_tokens": 23008177.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0695149540901183,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.541548252105713,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.10405257940292359,
+      "mean_token_accuracy": 0.9580470204353333,
+      "num_tokens": 23190301.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.0614005208015442,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.4365121126174927,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.10439364910125733,
+      "mean_token_accuracy": 0.9584310293197632,
+      "num_tokens": 23371824.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.065568709373474,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.2848029136657715,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.104488205909729,
+      "mean_token_accuracy": 0.959163922071457,
+      "num_tokens": 23552849.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.0750214815139771,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 1.280882716178894,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.10620679855346679,
+      "mean_token_accuracy": 0.9575595021247864,
+      "num_tokens": 23733408.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.077055287361145,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.2904170751571655,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.11054887771606445,
+      "mean_token_accuracy": 0.9566682338714599,
+      "num_tokens": 23914203.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.081210732460022,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.129718542098999,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.10313529968261718,
+      "mean_token_accuracy": 0.9582066476345062,
+      "num_tokens": 24094078.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.0856151461601258,
+      "epoch": 1.80349932705249,
+      "grad_norm": 1.3682607412338257,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.10383679866790771,
+      "mean_token_accuracy": 0.9592101991176605,
+      "num_tokens": 24274071.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.0931835532188416,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.3282326459884644,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.10675904750823975,
+      "mean_token_accuracy": 0.9580830633640289,
+      "num_tokens": 24454766.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.0867947936058044,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.2584236860275269,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.10147552490234375,
+      "mean_token_accuracy": 0.9604024410247802,
+      "num_tokens": 24635903.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.0726998805999757,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 1.254036545753479,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.10524414777755738,
+      "mean_token_accuracy": 0.9582872509956359,
+      "num_tokens": 24816787.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.0692258477210999,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.41594660282135,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.10340676307678223,
+      "mean_token_accuracy": 0.9590053558349609,
+      "num_tokens": 24998543.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.0772343635559083,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.385327696800232,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.10295225381851196,
+      "mean_token_accuracy": 0.9591428697109222,
+      "num_tokens": 25180059.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.0874038934707642,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 1.0628604888916016,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.10015511512756348,
+      "mean_token_accuracy": 0.9605867803096771,
+      "num_tokens": 25361290.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0702470779418944,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.1819324493408203,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.104461407661438,
+      "mean_token_accuracy": 0.9581527650356293,
+      "num_tokens": 25542455.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.0592689394950867,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4695993661880493,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.1024864912033081,
+      "mean_token_accuracy": 0.9591827690601349,
+      "num_tokens": 25723310.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.0781041979789734,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 1.3313041925430298,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.10552446842193604,
+      "mean_token_accuracy": 0.9576125502586365,
+      "num_tokens": 25904923.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0914432287216187,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.3578517436981201,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.11063914299011231,
+      "mean_token_accuracy": 0.9563858449459076,
+      "num_tokens": 26086154.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0916360139846801,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.3493428230285645,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.10228430032730103,
+      "mean_token_accuracy": 0.9597218215465546,
+      "num_tokens": 26268096.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0865699172019958,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.3002992868423462,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.10704560279846191,
+      "mean_token_accuracy": 0.9567803084850312,
+      "num_tokens": 26448598.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.0907732367515564,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.3105698823928833,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.10344587564468384,
+      "mean_token_accuracy": 0.9585539996623993,
+      "num_tokens": 26629572.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.097494399547577,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.5486772060394287,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.10484597682952881,
+      "mean_token_accuracy": 0.9578526735305786,
+      "num_tokens": 26810588.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 1.1259616188182953,
+      "eval_loss": 0.11920737475156784,
+      "eval_mean_token_accuracy": 0.9516198661676638,
+      "eval_num_tokens": 26919371.0,
+      "eval_runtime": 12.6599,
+      "eval_samples_per_second": 394.949,
+      "eval_steps_per_second": 12.401,
+      "step": 1486
+    },
+    {
+      "entropy": 1.1117471694946288,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.3559718132019043,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.09661188125610351,
+      "mean_token_accuracy": 0.9621427834033967,
+      "num_tokens": 26991831.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.0618590712547302,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.5231083631515503,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.07198524475097656,
+      "mean_token_accuracy": 0.9724888324737548,
+      "num_tokens": 27172578.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.0601553201675415,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0978407859802246,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.07232148051261902,
+      "mean_token_accuracy": 0.9721459746360779,
+      "num_tokens": 27353505.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.0545680522918701,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.375500202178955,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0737423300743103,
+      "mean_token_accuracy": 0.9723270058631897,
+      "num_tokens": 27534758.0,
+      "step": 1520
+    },
+    {
+      "entropy": 1.0429188370704652,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.5134543180465698,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.07129207253456116,
+      "mean_token_accuracy": 0.9730912566184997,
+      "num_tokens": 27715692.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0608030915260316,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.424338936805725,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.07388348579406738,
+      "mean_token_accuracy": 0.9727432250976562,
+      "num_tokens": 27896664.0,
+      "step": 1540
+    },
+    {
+      "entropy": 1.0577767491340637,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2193686962127686,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.07269638180732726,
+      "mean_token_accuracy": 0.9726069569587708,
+      "num_tokens": 28077786.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0627391576766967,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.4851799011230469,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.07443426847457886,
+      "mean_token_accuracy": 0.9716702818870544,
+      "num_tokens": 28259406.0,
+      "step": 1560
+    },
+    {
+      "entropy": 1.0606005311012268,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.136905312538147,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.07183161973953248,
+      "mean_token_accuracy": 0.9726755082607269,
+      "num_tokens": 28441336.0,
+      "step": 1570
+    },
+    {
+      "entropy": 1.0512076377868653,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.4116008281707764,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.07272077798843384,
+      "mean_token_accuracy": 0.972117805480957,
+      "num_tokens": 28621984.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0617831587791442,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.4194544553756714,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.07670866250991822,
+      "mean_token_accuracy": 0.9709574401378631,
+      "num_tokens": 28803335.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.054829752445221,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 1.0619711875915527,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.06954843401908875,
+      "mean_token_accuracy": 0.9732311069965363,
+      "num_tokens": 28984568.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.062503707408905,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1824331283569336,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.0742799162864685,
+      "mean_token_accuracy": 0.9718966007232666,
+      "num_tokens": 29165958.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0652349829673766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.2683000564575195,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.07545414566993713,
+      "mean_token_accuracy": 0.9704910278320312,
+      "num_tokens": 29347870.0,
+      "step": 1620
+    },
+    {
+      "entropy": 1.0519612908363343,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.257572889328003,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.075272136926651,
+      "mean_token_accuracy": 0.9712580978870392,
+      "num_tokens": 29529453.0,
+      "step": 1630
+    },
+    {
+      "entropy": 1.0668588399887085,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.3439997434616089,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.068633770942688,
+      "mean_token_accuracy": 0.9735071420669555,
+      "num_tokens": 29710695.0,
+      "step": 1640
+    },
+    {
+      "entropy": 1.0701637029647828,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.4352364540100098,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.07366064786911011,
+      "mean_token_accuracy": 0.9724323868751525,
+      "num_tokens": 29890993.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.065868353843689,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.3669353723526,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.0714600920677185,
+      "mean_token_accuracy": 0.9717992007732391,
+      "num_tokens": 30072284.0,
+      "step": 1660
+    },
+    {
+      "entropy": 1.077157938480377,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 1.3630682229995728,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.07237017154693604,
+      "mean_token_accuracy": 0.9729755759239197,
+      "num_tokens": 30253123.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0886191844940185,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.3787649869918823,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.07589051723480225,
+      "mean_token_accuracy": 0.9716001987457276,
+      "num_tokens": 30434100.0,
+      "step": 1680
+    },
+    {
+      "entropy": 1.080547845363617,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.4936959743499756,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.07393252849578857,
+      "mean_token_accuracy": 0.9709986329078675,
+      "num_tokens": 30615086.0,
+      "step": 1690
+    },
+    {
+      "entropy": 1.0851039648056031,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.4654748439788818,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.07487455606460572,
+      "mean_token_accuracy": 0.9710996568202972,
+      "num_tokens": 30795174.0,
+      "step": 1700
+    },
+    {
+      "entropy": 1.0743987679481506,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 1.354244589805603,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.07701593041419982,
+      "mean_token_accuracy": 0.9701604127883912,
+      "num_tokens": 30976627.0,
+      "step": 1710
+    },
+    {
+      "entropy": 1.0738040447235107,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.2830452919006348,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.07861257195472718,
+      "mean_token_accuracy": 0.9696074068546295,
+      "num_tokens": 31158542.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.0667856335639954,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.3329832553863525,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.07371129989624023,
+      "mean_token_accuracy": 0.9722550988197327,
+      "num_tokens": 31339091.0,
+      "step": 1730
+    },
+    {
+      "entropy": 1.0734196662902833,
+      "epoch": 2.34185733512786,
+      "grad_norm": 1.4483965635299683,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.07841320037841797,
+      "mean_token_accuracy": 0.9698444902896881,
+      "num_tokens": 31519961.0,
+      "step": 1740
+    },
+    {
+      "entropy": 1.0724906921386719,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.440190315246582,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.07452400922775268,
+      "mean_token_accuracy": 0.9715199947357178,
+      "num_tokens": 31701305.0,
+      "step": 1750
+    },
+    {
+      "entropy": 1.0770147442817688,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 1.17177414894104,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.07649941444396972,
+      "mean_token_accuracy": 0.9704680681228638,
+      "num_tokens": 31882106.0,
+      "step": 1760
+    },
+    {
+      "entropy": 1.0792171478271484,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.498350977897644,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.07850931882858277,
+      "mean_token_accuracy": 0.9699269771575928,
+      "num_tokens": 32062712.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.092840301990509,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.4153627157211304,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.07811332941055298,
+      "mean_token_accuracy": 0.9698202073574066,
+      "num_tokens": 32244156.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0767372131347657,
+      "epoch": 2.409152086137281,
+      "grad_norm": 1.3672188520431519,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.0765878677368164,
+      "mean_token_accuracy": 0.970400470495224,
+      "num_tokens": 32424975.0,
+      "step": 1790
+    },
+    {
+      "entropy": 1.0641321778297423,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.1141754388809204,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0744673728942871,
+      "mean_token_accuracy": 0.9713521003723145,
+      "num_tokens": 32606399.0,
+      "step": 1800
+    },
+    {
+      "entropy": 1.0847392082214355,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.372215986251831,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0788280725479126,
+      "mean_token_accuracy": 0.9688920557498932,
+      "num_tokens": 32787632.0,
+      "step": 1810
+    },
+    {
+      "entropy": 1.0952076196670533,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.3998148441314697,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.07322710752487183,
+      "mean_token_accuracy": 0.9713049352169036,
+      "num_tokens": 32969136.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.1047595977783202,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.538533091545105,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.07695393562316895,
+      "mean_token_accuracy": 0.9697132647037506,
+      "num_tokens": 33149585.0,
+      "step": 1830
+    },
+    {
+      "entropy": 1.1071401476860045,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.2797189950942993,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.07820409536361694,
+      "mean_token_accuracy": 0.9695189118385314,
+      "num_tokens": 33330604.0,
+      "step": 1840
+    },
+    {
+      "entropy": 1.0979747891426086,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 1.309956669807434,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.08005034327507018,
+      "mean_token_accuracy": 0.9683119177818298,
+      "num_tokens": 33510932.0,
+      "step": 1850
+    },
+    {
+      "entropy": 1.1024476408958435,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.2206803560256958,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.07752519249916076,
+      "mean_token_accuracy": 0.9697020888328552,
+      "num_tokens": 33692289.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0788621783256531,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.4230300188064575,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.08086194992065429,
+      "mean_token_accuracy": 0.9689899325370789,
+      "num_tokens": 33873643.0,
+      "step": 1870
+    },
+    {
+      "entropy": 1.0909191489219665,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 1.3553478717803955,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.07968194484710693,
+      "mean_token_accuracy": 0.9695583760738373,
+      "num_tokens": 34054342.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.0874631762504579,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 1.209466576576233,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.07668136358261109,
+      "mean_token_accuracy": 0.9706926643848419,
+      "num_tokens": 34235054.0,
+      "step": 1890
+    },
+    {
+      "entropy": 1.0850980758666993,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.2419084310531616,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.07621376514434815,
+      "mean_token_accuracy": 0.9708946466445922,
+      "num_tokens": 34415792.0,
+      "step": 1900
+    },
+    {
+      "entropy": 1.1024398922920227,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.5481151342391968,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.07621458768844605,
+      "mean_token_accuracy": 0.9696649491786957,
+      "num_tokens": 34596130.0,
+      "step": 1910
+    },
+    {
+      "entropy": 1.0917130827903747,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.4186631441116333,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.07716711163520813,
+      "mean_token_accuracy": 0.9699705123901368,
+      "num_tokens": 34776608.0,
+      "step": 1920
+    },
+    {
+      "entropy": 1.0845637321472168,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2549679279327393,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.07353252172470093,
+      "mean_token_accuracy": 0.9715336680412292,
+      "num_tokens": 34957854.0,
+      "step": 1930
+    },
+    {
+      "entropy": 1.070354163646698,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.390034794807434,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.07804152965545655,
+      "mean_token_accuracy": 0.9694301605224609,
+      "num_tokens": 35139425.0,
+      "step": 1940
+    },
+    {
+      "entropy": 1.0926494956016541,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.2890169620513916,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.07905603051185608,
+      "mean_token_accuracy": 0.9694173872470856,
+      "num_tokens": 35321009.0,
+      "step": 1950
+    },
+    {
+      "entropy": 1.090261971950531,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.201541543006897,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.07853106260299683,
+      "mean_token_accuracy": 0.9687644064426422,
+      "num_tokens": 35502338.0,
+      "step": 1960
+    },
+    {
+      "entropy": 1.110123836994171,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.3321393728256226,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.07721331715583801,
+      "mean_token_accuracy": 0.9703636646270752,
+      "num_tokens": 35684550.0,
+      "step": 1970
+    },
+    {
+      "entropy": 1.1154867172241212,
+      "epoch": 2.664872139973082,
+      "grad_norm": 1.5639152526855469,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.07769335508346557,
+      "mean_token_accuracy": 0.9694860637187958,
+      "num_tokens": 35866258.0,
+      "step": 1980
+    },
+    {
+      "entropy": 1.1126874089241028,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5160865783691406,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.07732025384902955,
+      "mean_token_accuracy": 0.9695809543132782,
+      "num_tokens": 36047771.0,
+      "step": 1990
+    },
+    {
+      "entropy": 1.1188251495361328,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.3689777851104736,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.07844393253326416,
+      "mean_token_accuracy": 0.9697122633457184,
+      "num_tokens": 36228702.0,
+      "step": 2000
+    },
+    {
+      "entropy": 1.1134975671768188,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 1.1956474781036377,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.07932614088058472,
+      "mean_token_accuracy": 0.9695560097694397,
+      "num_tokens": 36409577.0,
+      "step": 2010
+    },
+    {
+      "entropy": 1.1052889227867126,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.2844921350479126,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.07471654415130616,
+      "mean_token_accuracy": 0.9707616209983826,
+      "num_tokens": 36590635.0,
+      "step": 2020
+    },
+    {
+      "entropy": 1.1029305696487426,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3110395669937134,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.07654619216918945,
+      "mean_token_accuracy": 0.9699673116207123,
+      "num_tokens": 36772032.0,
+      "step": 2030
+    },
+    {
+      "entropy": 1.0967667698860168,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.2297992706298828,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.07549123167991638,
+      "mean_token_accuracy": 0.9705443382263184,
+      "num_tokens": 36953014.0,
+      "step": 2040
+    },
+    {
+      "entropy": 1.0954343318939208,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.5522037744522095,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.07590663433074951,
+      "mean_token_accuracy": 0.9709664165973664,
+      "num_tokens": 37134549.0,
+      "step": 2050
+    },
+    {
+      "entropy": 1.0984182119369508,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.3691717386245728,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.07795964479446411,
+      "mean_token_accuracy": 0.969748318195343,
+      "num_tokens": 37315651.0,
+      "step": 2060
+    },
+    {
+      "entropy": 1.0937008380889892,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 1.345433235168457,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.07556321024894715,
+      "mean_token_accuracy": 0.9707719802856445,
+      "num_tokens": 37497150.0,
+      "step": 2070
+    },
+    {
+      "entropy": 1.0978638410568238,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.3442333936691284,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.07603150010108947,
+      "mean_token_accuracy": 0.9707530319690705,
+      "num_tokens": 37678637.0,
+      "step": 2080
+    },
+    {
+      "entropy": 1.112320113182068,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.672094702720642,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.07702311277389526,
+      "mean_token_accuracy": 0.969988214969635,
+      "num_tokens": 37859530.0,
+      "step": 2090
+    },
+    {
+      "entropy": 1.1019986748695374,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.4613395929336548,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.07702183723449707,
+      "mean_token_accuracy": 0.9705046355724335,
+      "num_tokens": 38040160.0,
+      "step": 2100
+    },
+    {
+      "entropy": 1.1074181199073792,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.33228600025177,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.07809240221977234,
+      "mean_token_accuracy": 0.9702594041824341,
+      "num_tokens": 38221261.0,
+      "step": 2110
+    },
+    {
+      "entropy": 1.1087767481803894,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.3967821598052979,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.07641294002532958,
+      "mean_token_accuracy": 0.9694520056247711,
+      "num_tokens": 38402589.0,
+      "step": 2120
+    },
+    {
+      "entropy": 1.0902286529541017,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 1.3104654550552368,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.07864789366722107,
+      "mean_token_accuracy": 0.9693171322345734,
+      "num_tokens": 38583485.0,
+      "step": 2130
+    },
+    {
+      "entropy": 1.088974940776825,
+      "epoch": 2.88021534320323,
+      "grad_norm": 1.389603614807129,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.07767997980117798,
+      "mean_token_accuracy": 0.9691886186599732,
+      "num_tokens": 38764480.0,
+      "step": 2140
+    },
+    {
+      "entropy": 1.093208122253418,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.3133198022842407,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.08097923994064331,
+      "mean_token_accuracy": 0.968046635389328,
+      "num_tokens": 38946016.0,
+      "step": 2150
+    },
+    {
+      "entropy": 1.0920218586921693,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.3472810983657837,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.07504125237464905,
+      "mean_token_accuracy": 0.9705065429210663,
+      "num_tokens": 39127480.0,
+      "step": 2160
+    },
+    {
+      "entropy": 1.085507369041443,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.9070137739181519,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.07882450819015503,
+      "mean_token_accuracy": 0.9694548428058625,
+      "num_tokens": 39308552.0,
+      "step": 2170
+    },
+    {
+      "entropy": 1.1070378303527832,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.3059617280960083,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.08202729225158692,
+      "mean_token_accuracy": 0.9679245829582215,
+      "num_tokens": 39489844.0,
+      "step": 2180
+    },
+    {
+      "entropy": 1.0994928359985352,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.359965443611145,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.07510381340980529,
+      "mean_token_accuracy": 0.969987016916275,
+      "num_tokens": 39671941.0,
+      "step": 2190
+    },
+    {
+      "entropy": 1.0987842202186584,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.385233998298645,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.07863752245903015,
+      "mean_token_accuracy": 0.9685526669025422,
+      "num_tokens": 39852940.0,
+      "step": 2200
+    },
+    {
+      "entropy": 1.110793125629425,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.4456037282943726,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.07783852815628052,
+      "mean_token_accuracy": 0.9702686488628387,
+      "num_tokens": 40034313.0,
+      "step": 2210
+    },
+    {
+      "entropy": 1.1140072107315064,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.1889644861221313,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.07787686586380005,
+      "mean_token_accuracy": 0.9696961343288422,
+      "num_tokens": 40215866.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 1.11932077984901,
+      "eval_loss": 0.12462407350540161,
+      "eval_mean_token_accuracy": 0.951254228118119,
+      "eval_num_tokens": 40378902.0,
+      "eval_runtime": 13.5665,
+      "eval_samples_per_second": 368.556,
+      "eval_steps_per_second": 11.573,
+      "step": 2229
+    },
+    {
+      "entropy": 1.1124149322509767,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 1.1116644144058228,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.07439510822296143,
+      "mean_token_accuracy": 0.9713229060173034,
+      "num_tokens": 40396979.0,
+      "step": 2230
+    },
+    {
+      "entropy": 1.0716044545173644,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.1957958936691284,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.05146250128746033,
+      "mean_token_accuracy": 0.9813062250614166,
+      "num_tokens": 40578935.0,
+      "step": 2240
+    },
+    {
+      "entropy": 1.0567036390304565,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.337942361831665,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.052991294860839845,
+      "mean_token_accuracy": 0.9807850599288941,
+      "num_tokens": 40759500.0,
+      "step": 2250
+    },
+    {
+      "entropy": 1.056044340133667,
+      "epoch": 3.041722745625841,
+      "grad_norm": 1.4191840887069702,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.05163681507110596,
+      "mean_token_accuracy": 0.9813888549804688,
+      "num_tokens": 40940279.0,
+      "step": 2260
+    },
+    {
+      "entropy": 1.056977343559265,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.4481396675109863,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.05059016942977905,
+      "mean_token_accuracy": 0.9814287006855011,
+      "num_tokens": 41121782.0,
+      "step": 2270
+    },
+    {
+      "entropy": 1.0579043984413148,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.0136349201202393,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.05109646320343018,
+      "mean_token_accuracy": 0.9805608510971069,
+      "num_tokens": 41302614.0,
+      "step": 2280
+    },
+    {
+      "entropy": 1.0564674019813538,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.165916085243225,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.05656059980392456,
+      "mean_token_accuracy": 0.9801787078380585,
+      "num_tokens": 41484240.0,
+      "step": 2290
+    },
+    {
+      "entropy": 1.0617436170578003,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.500622272491455,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.05233668088912964,
+      "mean_token_accuracy": 0.9808355152606965,
+      "num_tokens": 41665160.0,
+      "step": 2300
+    },
+    {
+      "entropy": 1.0509375214576722,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.403001308441162,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.05074570775032043,
+      "mean_token_accuracy": 0.9807266294956207,
+      "num_tokens": 41846062.0,
+      "step": 2310
+    },
+    {
+      "entropy": 1.0536353230476379,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1046926975250244,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.04906890392303467,
+      "mean_token_accuracy": 0.9815255463123321,
+      "num_tokens": 42027115.0,
+      "step": 2320
+    },
+    {
+      "entropy": 1.0518895149230958,
+      "epoch": 3.135935397039031,
+      "grad_norm": 1.0369377136230469,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.04991455376148224,
+      "mean_token_accuracy": 0.9822782278060913,
+      "num_tokens": 42208594.0,
+      "step": 2330
+    },
+    {
+      "entropy": 1.0597471237182616,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.6156455278396606,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.049592965841293336,
+      "mean_token_accuracy": 0.9818629324436188,
+      "num_tokens": 42389065.0,
+      "step": 2340
+    },
+    {
+      "entropy": 1.0561720967292785,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.2941797971725464,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.05092741847038269,
+      "mean_token_accuracy": 0.9810208320617676,
+      "num_tokens": 42570319.0,
+      "step": 2350
+    },
+    {
+      "entropy": 1.0521231889724731,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.1866732835769653,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.051917088031768796,
+      "mean_token_accuracy": 0.981055635213852,
+      "num_tokens": 42751501.0,
+      "step": 2360
+    },
+    {
+      "entropy": 1.0555105328559875,
+      "epoch": 3.189771197846568,
+      "grad_norm": 1.2260069847106934,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.05160566568374634,
+      "mean_token_accuracy": 0.981193995475769,
+      "num_tokens": 42932643.0,
+      "step": 2370
+    },
+    {
+      "entropy": 1.0720075011253356,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.3786890506744385,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.053551071882247926,
+      "mean_token_accuracy": 0.9797191619873047,
+      "num_tokens": 43113204.0,
+      "step": 2380
+    },
+    {
+      "entropy": 1.0718719363212585,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.3200089931488037,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.054034185409545896,
+      "mean_token_accuracy": 0.9800151228904724,
+      "num_tokens": 43293433.0,
+      "step": 2390
+    },
+    {
+      "entropy": 1.0536052703857421,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.4560705423355103,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.05042939186096192,
+      "mean_token_accuracy": 0.9806897640228271,
+      "num_tokens": 43474189.0,
+      "step": 2400
+    },
+    {
+      "entropy": 1.0602891445159912,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.3123281002044678,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.05151479244232178,
+      "mean_token_accuracy": 0.980517429113388,
+      "num_tokens": 43655355.0,
+      "step": 2410
+    },
+    {
+      "entropy": 1.067916214466095,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.2652188539505005,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.05306308269500733,
+      "mean_token_accuracy": 0.9805097997188568,
+      "num_tokens": 43836752.0,
+      "step": 2420
+    },
+    {
+      "entropy": 1.067535150051117,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.6280162334442139,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.052641332149505615,
+      "mean_token_accuracy": 0.9803492069244385,
+      "num_tokens": 44018525.0,
+      "step": 2430
+    },
+    {
+      "entropy": 1.070455825328827,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 1.632573127746582,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.05452280044555664,
+      "mean_token_accuracy": 0.9793554663658142,
+      "num_tokens": 44199901.0,
+      "step": 2440
+    },
+    {
+      "entropy": 1.0741212248802186,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.3525495529174805,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.054098492860794066,
+      "mean_token_accuracy": 0.9804170608520508,
+      "num_tokens": 44381734.0,
+      "step": 2450
+    },
+    {
+      "entropy": 1.0836357831954957,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.3989282846450806,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.05158165693283081,
+      "mean_token_accuracy": 0.9810640931129455,
+      "num_tokens": 44563158.0,
+      "step": 2460
+    },
+    {
+      "entropy": 1.077314054965973,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1970937252044678,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.052333271503448485,
+      "mean_token_accuracy": 0.9808873236179352,
+      "num_tokens": 44744642.0,
+      "step": 2470
+    },
+    {
+      "entropy": 1.069891107082367,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.3197492361068726,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.05269680619239807,
+      "mean_token_accuracy": 0.980803519487381,
+      "num_tokens": 44926301.0,
+      "step": 2480
+    },
+    {
+      "entropy": 1.071691370010376,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.332481861114502,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.054685378074646,
+      "mean_token_accuracy": 0.9795110940933227,
+      "num_tokens": 45106935.0,
+      "step": 2490
+    },
+    {
+      "entropy": 1.0654058694839477,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 1.5478020906448364,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.05287986397743225,
+      "mean_token_accuracy": 0.9797925829887391,
+      "num_tokens": 45288325.0,
+      "step": 2500
+    },
+    {
+      "entropy": 1.0660210013389588,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.43852961063385,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.05411216616630554,
+      "mean_token_accuracy": 0.9790283799171448,
+      "num_tokens": 45469485.0,
+      "step": 2510
+    },
+    {
+      "entropy": 1.0649927496910094,
+      "epoch": 3.391655450874832,
+      "grad_norm": 1.3900808095932007,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.05290043354034424,
+      "mean_token_accuracy": 0.9808539688587189,
+      "num_tokens": 45650705.0,
+      "step": 2520
+    },
+    {
+      "entropy": 1.0768967866897583,
+      "epoch": 3.405114401076716,
+      "grad_norm": 1.3428847789764404,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.05374534130096435,
+      "mean_token_accuracy": 0.9802824795246124,
+      "num_tokens": 45832104.0,
+      "step": 2530
+    },
+    {
+      "entropy": 1.0660459756851197,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.3150343894958496,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.05366337299346924,
+      "mean_token_accuracy": 0.9796974420547485,
+      "num_tokens": 46013552.0,
+      "step": 2540
+    },
+    {
+      "entropy": 1.06330087184906,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.546858787536621,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.054223823547363284,
+      "mean_token_accuracy": 0.9796406507492066,
+      "num_tokens": 46194606.0,
+      "step": 2550
+    },
+    {
+      "entropy": 1.0564390897750855,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.3932640552520752,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.05237367153167725,
+      "mean_token_accuracy": 0.9805298745632172,
+      "num_tokens": 46375520.0,
+      "step": 2560
+    },
+    {
+      "entropy": 1.0519753217697143,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.8447308540344238,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.05499213933944702,
+      "mean_token_accuracy": 0.9787130057811737,
+      "num_tokens": 46556203.0,
+      "step": 2570
+    },
+    {
+      "entropy": 1.0693527460098267,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.3508682250976562,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.055162906646728516,
+      "mean_token_accuracy": 0.9794130921363831,
+      "num_tokens": 46737287.0,
+      "step": 2580
+    },
+    {
+      "entropy": 1.078024983406067,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 1.1582329273223877,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.05406954288482666,
+      "mean_token_accuracy": 0.9794795215129852,
+      "num_tokens": 46918363.0,
+      "step": 2590
+    },
+    {
+      "entropy": 1.0643377661705018,
+      "epoch": 3.499327052489906,
+      "grad_norm": 1.429380178451538,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.05309759378433228,
+      "mean_token_accuracy": 0.9806221067905426,
+      "num_tokens": 47099549.0,
+      "step": 2600
+    },
+    {
+      "entropy": 1.068834125995636,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.236444354057312,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.05395760536193848,
+      "mean_token_accuracy": 0.9798199534416199,
+      "num_tokens": 47280430.0,
+      "step": 2610
+    },
+    {
+      "entropy": 1.0729689002037048,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 1.32350754737854,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.05501353740692139,
+      "mean_token_accuracy": 0.9792946577072144,
+      "num_tokens": 47461390.0,
+      "step": 2620
+    },
+    {
+      "entropy": 1.0766199231147766,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2348153591156006,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.05568041205406189,
+      "mean_token_accuracy": 0.9796942174434662,
+      "num_tokens": 47642662.0,
+      "step": 2630
+    },
+    {
+      "entropy": 1.0804976940155029,
+      "epoch": 3.553162853297443,
+      "grad_norm": 1.1175553798675537,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.050679147243499756,
+      "mean_token_accuracy": 0.9814749479293823,
+      "num_tokens": 47824309.0,
+      "step": 2640
+    },
+    {
+      "entropy": 1.0689257264137269,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.6297036409378052,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.054615461826324464,
+      "mean_token_accuracy": 0.9796478092670441,
+      "num_tokens": 48005982.0,
+      "step": 2650
+    },
+    {
+      "entropy": 1.0696970105171204,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2477251291275024,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.05287024974822998,
+      "mean_token_accuracy": 0.979641854763031,
+      "num_tokens": 48187048.0,
+      "step": 2660
+    },
+    {
+      "entropy": 1.0800267577171325,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.3109506368637085,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.054475915431976316,
+      "mean_token_accuracy": 0.9794476389884949,
+      "num_tokens": 48368447.0,
+      "step": 2670
+    },
+    {
+      "entropy": 1.0848782062530518,
+      "epoch": 3.60699865410498,
+      "grad_norm": 1.2147419452667236,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.05212314128875732,
+      "mean_token_accuracy": 0.9798501491546631,
+      "num_tokens": 48549144.0,
+      "step": 2680
+    },
+    {
+      "entropy": 1.0681315302848815,
+      "epoch": 3.620457604306864,
+      "grad_norm": 1.3249231576919556,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.053503894805908205,
+      "mean_token_accuracy": 0.9800095915794372,
+      "num_tokens": 48731059.0,
+      "step": 2690
+    },
+    {
+      "entropy": 1.0629405975341797,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.2435307502746582,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.0541129469871521,
+      "mean_token_accuracy": 0.9795375943183899,
+      "num_tokens": 48912283.0,
+      "step": 2700
+    },
+    {
+      "entropy": 1.0608446598052979,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.3816258907318115,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.054210084676742556,
+      "mean_token_accuracy": 0.9793854176998138,
+      "num_tokens": 49094236.0,
+      "step": 2710
+    },
+    {
+      "entropy": 1.0596249461174012,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.132426381111145,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.05496481657028198,
+      "mean_token_accuracy": 0.9798576831817627,
+      "num_tokens": 49275540.0,
+      "step": 2720
+    },
+    {
+      "entropy": 1.060381281375885,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.4120980501174927,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.05694271326065063,
+      "mean_token_accuracy": 0.9788619399070739,
+      "num_tokens": 49456684.0,
+      "step": 2730
+    },
+    {
+      "entropy": 1.0594112157821656,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.428120493888855,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.05274912118911743,
+      "mean_token_accuracy": 0.9802964627742767,
+      "num_tokens": 49638099.0,
+      "step": 2740
+    },
+    {
+      "entropy": 1.0630932807922364,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.4599305391311646,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.057349550724029544,
+      "mean_token_accuracy": 0.9788366615772247,
+      "num_tokens": 49819751.0,
+      "step": 2750
+    },
+    {
+      "entropy": 1.0632095456123352,
+      "epoch": 3.714670255720054,
+      "grad_norm": 1.8489277362823486,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.053570222854614255,
+      "mean_token_accuracy": 0.980079036951065,
+      "num_tokens": 50000623.0,
+      "step": 2760
+    },
+    {
+      "entropy": 1.0583208560943604,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.4855526685714722,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.05508332252502442,
+      "mean_token_accuracy": 0.9789698243141174,
+      "num_tokens": 50181540.0,
+      "step": 2770
+    },
+    {
+      "entropy": 1.0690102458000184,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 1.553468942642212,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.05684031248092651,
+      "mean_token_accuracy": 0.9776920318603516,
+      "num_tokens": 50363406.0,
+      "step": 2780
+    },
+    {
+      "entropy": 1.0651323795318604,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 1.2951313257217407,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.055989408493041994,
+      "mean_token_accuracy": 0.978317403793335,
+      "num_tokens": 50544599.0,
+      "step": 2790
+    },
+    {
+      "entropy": 1.0571947336196899,
+      "epoch": 3.768506056527591,
+      "grad_norm": 1.1358306407928467,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.052672064304351805,
+      "mean_token_accuracy": 0.9797160506248475,
+      "num_tokens": 50726664.0,
+      "step": 2800
+    },
+    {
+      "entropy": 1.0475606441497802,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.4842852354049683,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.05601704120635986,
+      "mean_token_accuracy": 0.9794813454151153,
+      "num_tokens": 50907903.0,
+      "step": 2810
+    },
+    {
+      "entropy": 1.053141224384308,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.3320889472961426,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.05435909032821655,
+      "mean_token_accuracy": 0.9792923092842102,
+      "num_tokens": 51088734.0,
+      "step": 2820
+    },
+    {
+      "entropy": 1.0518054842948914,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.140783667564392,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.05576471090316772,
+      "mean_token_accuracy": 0.9789255380630493,
+      "num_tokens": 51269924.0,
+      "step": 2830
+    },
+    {
+      "entropy": 1.0490313410758971,
+      "epoch": 3.822341857335128,
+      "grad_norm": 1.5749891996383667,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.05655611753463745,
+      "mean_token_accuracy": 0.9785366714000702,
+      "num_tokens": 51450406.0,
+      "step": 2840
+    },
+    {
+      "entropy": 1.0508960008621215,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 1.2535520792007446,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.05710798501968384,
+      "mean_token_accuracy": 0.978301590681076,
+      "num_tokens": 51630630.0,
+      "step": 2850
+    },
+    {
+      "entropy": 1.0434152960777283,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.6045477390289307,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.0545048713684082,
+      "mean_token_accuracy": 0.9785818517208099,
+      "num_tokens": 51811252.0,
+      "step": 2860
+    },
+    {
+      "entropy": 1.046581244468689,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.2737947702407837,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.056829071044921874,
+      "mean_token_accuracy": 0.9779990613460541,
+      "num_tokens": 51992578.0,
+      "step": 2870
+    },
+    {
+      "entropy": 1.0562566757202148,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.2364535331726074,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.054702204465866086,
+      "mean_token_accuracy": 0.9796520650386811,
+      "num_tokens": 52173507.0,
+      "step": 2880
+    },
+    {
+      "entropy": 1.0444288849830627,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.2663625478744507,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.05222170352935791,
+      "mean_token_accuracy": 0.9805905818939209,
+      "num_tokens": 52354669.0,
+      "step": 2890
+    },
+    {
+      "entropy": 1.0429784655570984,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.2072714567184448,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.05404345989227295,
+      "mean_token_accuracy": 0.9802968561649322,
+      "num_tokens": 52535334.0,
+      "step": 2900
+    },
+    {
+      "entropy": 1.045020854473114,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.3894519805908203,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.05601906776428223,
+      "mean_token_accuracy": 0.9779200375080108,
+      "num_tokens": 52715952.0,
+      "step": 2910
+    },
+    {
+      "entropy": 1.0439318537712097,
+      "epoch": 3.930013458950202,
+      "grad_norm": 1.4112615585327148,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.05345791578292847,
+      "mean_token_accuracy": 0.9796200633049011,
+      "num_tokens": 52897155.0,
+      "step": 2920
+    },
+    {
+      "entropy": 1.0361477971076964,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 1.3158206939697266,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.05547314286231995,
+      "mean_token_accuracy": 0.9788706004619598,
+      "num_tokens": 53078916.0,
+      "step": 2930
+    },
+    {
+      "entropy": 1.0427968740463256,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2565301656723022,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.05601854920387268,
+      "mean_token_accuracy": 0.9791704952716828,
+      "num_tokens": 53258884.0,
+      "step": 2940
+    },
+    {
+      "entropy": 1.0374909162521362,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.4639637470245361,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.054512780904769895,
+      "mean_token_accuracy": 0.9793699860572815,
+      "num_tokens": 53440481.0,
+      "step": 2950
+    },
+    {
+      "entropy": 1.0511455059051513,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.242073655128479,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.055217862129211426,
+      "mean_token_accuracy": 0.9801203072071075,
+      "num_tokens": 53621142.0,
+      "step": 2960
+    },
+    {
+      "entropy": 1.0518356442451477,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.1518405675888062,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.05533761382102966,
+      "mean_token_accuracy": 0.9789008319377899,
+      "num_tokens": 53802588.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 1.0548353901334628,
+      "eval_loss": 0.13498826324939728,
+      "eval_mean_token_accuracy": 0.9507886888874564,
+      "eval_num_tokens": 53838633.0,
+      "eval_runtime": 12.6809,
+      "eval_samples_per_second": 394.295,
+      "eval_steps_per_second": 12.381,
+      "step": 2972
+    },
+    {
+      "entropy": 1.043621039390564,
+      "epoch": 4.010767160161508,
+      "grad_norm": 1.1016159057617188,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.03793205618858338,
+      "mean_token_accuracy": 0.9870165169239045,
+      "num_tokens": 53983430.0,
+      "step": 2980
+    },
+    {
+      "entropy": 1.0226780295372009,
+      "epoch": 4.024226110363392,
+      "grad_norm": 1.240837812423706,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.03402446508407593,
+      "mean_token_accuracy": 0.9883724093437195,
+      "num_tokens": 54164955.0,
+      "step": 2990
+    },
+    {
+      "entropy": 1.0194039583206176,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.1843734979629517,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.035861504077911374,
+      "mean_token_accuracy": 0.9872536063194275,
+      "num_tokens": 54346968.0,
+      "step": 3000
+    },
+    {
+      "entropy": 1.029161846637726,
+      "epoch": 4.05114401076716,
+      "grad_norm": 1.285415768623352,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.034106379747390746,
+      "mean_token_accuracy": 0.9882839441299438,
+      "num_tokens": 54528019.0,
+      "step": 3010
+    },
+    {
+      "entropy": 1.031964898109436,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4622068405151367,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.032776829600334165,
+      "mean_token_accuracy": 0.9882788896560669,
+      "num_tokens": 54708148.0,
+      "step": 3020
+    },
+    {
+      "entropy": 1.0282336235046388,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.1071696281433105,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.03698182702064514,
+      "mean_token_accuracy": 0.9865448236465454,
+      "num_tokens": 54888894.0,
+      "step": 3030
+    },
+    {
+      "entropy": 1.0233591556549073,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.3671438694000244,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.03386759161949158,
+      "mean_token_accuracy": 0.9879629850387573,
+      "num_tokens": 55069872.0,
+      "step": 3040
+    },
+    {
+      "entropy": 1.0250957012176514,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.1421459913253784,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.035586753487586976,
+      "mean_token_accuracy": 0.9869713068008423,
+      "num_tokens": 55250605.0,
+      "step": 3050
+    },
+    {
+      "entropy": 1.016488790512085,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.3857697248458862,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.039323312044143674,
+      "mean_token_accuracy": 0.986032885313034,
+      "num_tokens": 55431737.0,
+      "step": 3060
+    },
+    {
+      "entropy": 1.0130115926265717,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.1797211170196533,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.03508211970329285,
+      "mean_token_accuracy": 0.9872302651405335,
+      "num_tokens": 55612917.0,
+      "step": 3070
+    },
+    {
+      "entropy": 1.023509919643402,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.1484259366989136,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.0338476300239563,
+      "mean_token_accuracy": 0.9883188247680664,
+      "num_tokens": 55793687.0,
+      "step": 3080
+    },
+    {
+      "entropy": 1.0193628787994384,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.3184833526611328,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.03667713701725006,
+      "mean_token_accuracy": 0.9869367897510528,
+      "num_tokens": 55974723.0,
+      "step": 3090
+    },
+    {
+      "entropy": 1.013669216632843,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.2176531553268433,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.035006949305534364,
+      "mean_token_accuracy": 0.9868299901485443,
+      "num_tokens": 56156568.0,
+      "step": 3100
+    },
+    {
+      "entropy": 1.0072043538093567,
+      "epoch": 4.185733512786003,
+      "grad_norm": 1.4346243143081665,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.03560284972190857,
+      "mean_token_accuracy": 0.9869017779827118,
+      "num_tokens": 56337896.0,
+      "step": 3110
+    },
+    {
+      "entropy": 1.0102415680885315,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.2328612804412842,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.03608990907669067,
+      "mean_token_accuracy": 0.9874296009540557,
+      "num_tokens": 56519102.0,
+      "step": 3120
+    },
+    {
+      "entropy": 1.015284490585327,
+      "epoch": 4.212651413189771,
+      "grad_norm": 1.4385327100753784,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.03637963235378265,
+      "mean_token_accuracy": 0.986739182472229,
+      "num_tokens": 56699936.0,
+      "step": 3130
+    },
+    {
+      "entropy": 1.0054036319255828,
+      "epoch": 4.226110363391656,
+      "grad_norm": 1.1129945516586304,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.034222981333732604,
+      "mean_token_accuracy": 0.9883054196834564,
+      "num_tokens": 56882308.0,
+      "step": 3140
+    },
+    {
+      "entropy": 1.0061798334121703,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1598492860794067,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.03461446762084961,
+      "mean_token_accuracy": 0.9873203098773956,
+      "num_tokens": 57063770.0,
+      "step": 3150
+    },
+    {
+      "entropy": 1.0039719343185425,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.322458267211914,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.03701616525650024,
+      "mean_token_accuracy": 0.9862136900424957,
+      "num_tokens": 57244448.0,
+      "step": 3160
+    },
+    {
+      "entropy": 1.0048979759216308,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 1.570866584777832,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.036290496587753296,
+      "mean_token_accuracy": 0.9868158757686615,
+      "num_tokens": 57425448.0,
+      "step": 3170
+    },
+    {
+      "entropy": 1.0145976901054383,
+      "epoch": 4.279946164199193,
+      "grad_norm": 1.103385329246521,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.03472789227962494,
+      "mean_token_accuracy": 0.987658828496933,
+      "num_tokens": 57606369.0,
+      "step": 3180
+    },
+    {
+      "entropy": 1.0186771035194397,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.2541930675506592,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.03532317280769348,
+      "mean_token_accuracy": 0.9874464809894562,
+      "num_tokens": 57787685.0,
+      "step": 3190
+    },
+    {
+      "entropy": 1.0198698043823242,
+      "epoch": 4.306864064602961,
+      "grad_norm": 1.1590384244918823,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.036364412307739256,
+      "mean_token_accuracy": 0.9864933013916015,
+      "num_tokens": 57968903.0,
+      "step": 3200
+    },
+    {
+      "entropy": 1.0211624503135681,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.7207615375518799,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.03790584802627563,
+      "mean_token_accuracy": 0.9866538405418396,
+      "num_tokens": 58149469.0,
+      "step": 3210
+    },
+    {
+      "entropy": 1.0185290336608888,
+      "epoch": 4.33378196500673,
+      "grad_norm": 1.3042566776275635,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.03802136182785034,
+      "mean_token_accuracy": 0.9859855055809021,
+      "num_tokens": 58330867.0,
+      "step": 3220
+    },
+    {
+      "entropy": 1.0105828583240508,
+      "epoch": 4.347240915208614,
+      "grad_norm": 1.2536448240280151,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.034687772393226624,
+      "mean_token_accuracy": 0.9878224015235901,
+      "num_tokens": 58512529.0,
+      "step": 3230
+    },
+    {
+      "entropy": 1.0021630108356476,
+      "epoch": 4.360699865410498,
+      "grad_norm": 1.3816099166870117,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.035324835777282716,
+      "mean_token_accuracy": 0.9872316241264343,
+      "num_tokens": 58693573.0,
+      "step": 3240
+    },
+    {
+      "entropy": 1.007886278629303,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 1.3121757507324219,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.034709393978118896,
+      "mean_token_accuracy": 0.9877420663833618,
+      "num_tokens": 58874107.0,
+      "step": 3250
+    },
+    {
+      "entropy": 1.005664885044098,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.429775357246399,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.03714964389801025,
+      "mean_token_accuracy": 0.9861079812049866,
+      "num_tokens": 59054533.0,
+      "step": 3260
+    },
+    {
+      "entropy": 1.0047886371612549,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.2385882139205933,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.036937403678894046,
+      "mean_token_accuracy": 0.9863442242145538,
+      "num_tokens": 59236331.0,
+      "step": 3270
+    },
+    {
+      "entropy": 1.0209275126457213,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.476168155670166,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.03463808596134186,
+      "mean_token_accuracy": 0.9872730135917663,
+      "num_tokens": 59417193.0,
+      "step": 3280
+    },
+    {
+      "entropy": 1.021347463130951,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 1.4710102081298828,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.038167804479599,
+      "mean_token_accuracy": 0.986047238111496,
+      "num_tokens": 59598284.0,
+      "step": 3290
+    },
+    {
+      "entropy": 1.0282366394996643,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1043860912322998,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.036692625284194945,
+      "mean_token_accuracy": 0.9862713575363159,
+      "num_tokens": 59779329.0,
+      "step": 3300
+    },
+    {
+      "entropy": 1.021342945098877,
+      "epoch": 4.454912516823688,
+      "grad_norm": 1.1742897033691406,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.03571464121341705,
+      "mean_token_accuracy": 0.987505567073822,
+      "num_tokens": 59961340.0,
+      "step": 3310
+    },
+    {
+      "entropy": 1.0153375625610352,
+      "epoch": 4.468371467025572,
+      "grad_norm": 1.1321660280227661,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.036007928848266604,
+      "mean_token_accuracy": 0.9870990335941314,
+      "num_tokens": 60142647.0,
+      "step": 3320
+    },
+    {
+      "entropy": 1.0099608540534972,
+      "epoch": 4.481830417227457,
+      "grad_norm": 1.084017276763916,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.03562799394130707,
+      "mean_token_accuracy": 0.9865496158599854,
+      "num_tokens": 60324151.0,
+      "step": 3330
+    },
+    {
+      "entropy": 1.0173812866210938,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2779732942581177,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.03556201159954071,
+      "mean_token_accuracy": 0.9875519394874572,
+      "num_tokens": 60505023.0,
+      "step": 3340
+    },
+    {
+      "entropy": 1.0236618638038635,
+      "epoch": 4.508748317631225,
+      "grad_norm": 1.5285946130752563,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.03721941411495209,
+      "mean_token_accuracy": 0.9865630626678467,
+      "num_tokens": 60685921.0,
+      "step": 3350
+    },
+    {
+      "entropy": 1.0262888669967651,
+      "epoch": 4.522207267833109,
+      "grad_norm": 1.131675362586975,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.03602397441864014,
+      "mean_token_accuracy": 0.9864337563514709,
+      "num_tokens": 60867240.0,
+      "step": 3360
+    },
+    {
+      "entropy": 1.0292576909065247,
+      "epoch": 4.535666218034994,
+      "grad_norm": 1.5156011581420898,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.03973098993301392,
+      "mean_token_accuracy": 0.9855743169784545,
+      "num_tokens": 61047708.0,
+      "step": 3370
+    },
+    {
+      "entropy": 1.010222613811493,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0979945659637451,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.03533022105693817,
+      "mean_token_accuracy": 0.9869906663894653,
+      "num_tokens": 61229247.0,
+      "step": 3380
+    },
+    {
+      "entropy": 1.004716032743454,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.3082596063613892,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.033983749151229856,
+      "mean_token_accuracy": 0.9877199769020081,
+      "num_tokens": 61411059.0,
+      "step": 3390
+    },
+    {
+      "entropy": 1.0031994462013245,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.3033607006072998,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.03607205748558044,
+      "mean_token_accuracy": 0.9866301774978637,
+      "num_tokens": 61592538.0,
+      "step": 3400
+    },
+    {
+      "entropy": 1.0113391518592834,
+      "epoch": 4.589502018842531,
+      "grad_norm": 1.3490946292877197,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.03906603455543518,
+      "mean_token_accuracy": 0.9854673087596894,
+      "num_tokens": 61773613.0,
+      "step": 3410
+    },
+    {
+      "entropy": 1.0076410889625549,
+      "epoch": 4.602960969044415,
+      "grad_norm": 1.1719361543655396,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.037575924396514894,
+      "mean_token_accuracy": 0.9867486238479615,
+      "num_tokens": 61955690.0,
+      "step": 3420
+    },
+    {
+      "entropy": 1.0200914144515991,
+      "epoch": 4.616419919246299,
+      "grad_norm": 1.1205360889434814,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.03691520690917969,
+      "mean_token_accuracy": 0.9865606665611267,
+      "num_tokens": 62136047.0,
+      "step": 3430
+    },
+    {
+      "entropy": 1.0141666173934936,
+      "epoch": 4.629878869448183,
+      "grad_norm": 1.1805763244628906,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.035447508096694946,
+      "mean_token_accuracy": 0.9873166739940643,
+      "num_tokens": 62317192.0,
+      "step": 3440
+    },
+    {
+      "entropy": 1.0067960560321807,
+      "epoch": 4.643337819650068,
+      "grad_norm": 1.0286436080932617,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.03580200672149658,
+      "mean_token_accuracy": 0.9870070457458496,
+      "num_tokens": 62498860.0,
+      "step": 3450
+    },
+    {
+      "entropy": 1.0088169515132903,
+      "epoch": 4.656796769851952,
+      "grad_norm": 1.3638492822647095,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.03756250143051147,
+      "mean_token_accuracy": 0.9861987411975861,
+      "num_tokens": 62679272.0,
+      "step": 3460
+    },
+    {
+      "entropy": 1.0098722338676454,
+      "epoch": 4.670255720053836,
+      "grad_norm": 1.649560809135437,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.037615427374839784,
+      "mean_token_accuracy": 0.9864691913127899,
+      "num_tokens": 62860715.0,
+      "step": 3470
+    },
+    {
+      "entropy": 1.0174571394920349,
+      "epoch": 4.68371467025572,
+      "grad_norm": 1.2106678485870361,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.037718603014945985,
+      "mean_token_accuracy": 0.9855880260467529,
+      "num_tokens": 63041029.0,
+      "step": 3480
+    },
+    {
+      "entropy": 1.0129252314567565,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.1571707725524902,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.03724310994148254,
+      "mean_token_accuracy": 0.9865262687206269,
+      "num_tokens": 63222379.0,
+      "step": 3490
+    },
+    {
+      "entropy": 1.0124369859695435,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.0599361658096313,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.037783673405647276,
+      "mean_token_accuracy": 0.9859567046165466,
+      "num_tokens": 63403328.0,
+      "step": 3500
+    },
+    {
+      "entropy": 1.0041921436786652,
+      "epoch": 4.724091520861373,
+      "grad_norm": 1.2054452896118164,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.03585146069526672,
+      "mean_token_accuracy": 0.9868905484676361,
+      "num_tokens": 63584598.0,
+      "step": 3510
+    },
+    {
+      "entropy": 1.0015427589416503,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.5075141191482544,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.037428200244903564,
+      "mean_token_accuracy": 0.9867083311080933,
+      "num_tokens": 63766203.0,
+      "step": 3520
+    },
+    {
+      "entropy": 1.0039161920547486,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.310390830039978,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.037494581937789914,
+      "mean_token_accuracy": 0.9862947940826416,
+      "num_tokens": 63947714.0,
+      "step": 3530
+    },
+    {
+      "entropy": 1.0049733996391297,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.2912139892578125,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.03803396224975586,
+      "mean_token_accuracy": 0.9860390484333038,
+      "num_tokens": 64128798.0,
+      "step": 3540
+    },
+    {
+      "entropy": 1.0127450585365296,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.001133680343628,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0351622611284256,
+      "mean_token_accuracy": 0.9875370979309082,
+      "num_tokens": 64309577.0,
+      "step": 3550
+    },
+    {
+      "entropy": 1.0070195615291595,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 1.1870886087417603,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.036404851078987124,
+      "mean_token_accuracy": 0.9871752023696899,
+      "num_tokens": 64490707.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.9983750462532044,
+      "epoch": 4.804845222072679,
+      "grad_norm": 1.2374205589294434,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.03696496784687042,
+      "mean_token_accuracy": 0.9865773975849151,
+      "num_tokens": 64672104.0,
+      "step": 3570
+    },
+    {
+      "entropy": 1.0151559710502625,
+      "epoch": 4.818304172274562,
+      "grad_norm": 1.2001149654388428,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.03725983202457428,
+      "mean_token_accuracy": 0.9862341761589051,
+      "num_tokens": 64853353.0,
+      "step": 3580
+    },
+    {
+      "entropy": 1.0136774778366089,
+      "epoch": 4.831763122476447,
+      "grad_norm": 1.0128737688064575,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.03615986704826355,
+      "mean_token_accuracy": 0.9872736394405365,
+      "num_tokens": 65035009.0,
+      "step": 3590
+    },
+    {
+      "entropy": 1.0082051515579225,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.2552846670150757,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.03561249375343323,
+      "mean_token_accuracy": 0.9868618845939636,
+      "num_tokens": 65215981.0,
+      "step": 3600
+    },
+    {
+      "entropy": 1.0209908723831176,
+      "epoch": 4.858681022880216,
+      "grad_norm": 1.1914079189300537,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.03721153438091278,
+      "mean_token_accuracy": 0.9862311720848084,
+      "num_tokens": 65397088.0,
+      "step": 3610
+    },
+    {
+      "entropy": 1.014677894115448,
+      "epoch": 4.872139973082099,
+      "grad_norm": 1.3584535121917725,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.03897300064563751,
+      "mean_token_accuracy": 0.9851836740970612,
+      "num_tokens": 65578262.0,
+      "step": 3620
+    },
+    {
+      "entropy": 1.0207050919532776,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.2754287719726562,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.037926393747329715,
+      "mean_token_accuracy": 0.9859795570373535,
+      "num_tokens": 65758956.0,
+      "step": 3630
+    },
+    {
+      "entropy": 1.008733856678009,
+      "epoch": 4.899057873485868,
+      "grad_norm": 1.0815879106521606,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.036518561840057376,
+      "mean_token_accuracy": 0.9870649397373199,
+      "num_tokens": 65939970.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.9995620250701904,
+      "epoch": 4.912516823687753,
+      "grad_norm": 1.1773526668548584,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.035269084572792056,
+      "mean_token_accuracy": 0.9873197913169861,
+      "num_tokens": 66121045.0,
+      "step": 3650
+    },
+    {
+      "entropy": 1.0036249816417695,
+      "epoch": 4.925975773889636,
+      "grad_norm": 1.0941131114959717,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.0353422611951828,
+      "mean_token_accuracy": 0.9870408117771149,
+      "num_tokens": 66302324.0,
+      "step": 3660
+    },
+    {
+      "entropy": 1.004719352722168,
+      "epoch": 4.939434724091521,
+      "grad_norm": 1.5107269287109375,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.03743018507957459,
+      "mean_token_accuracy": 0.9857985079288483,
+      "num_tokens": 66483916.0,
+      "step": 3670
+    },
+    {
+      "entropy": 1.0076164484024048,
+      "epoch": 4.952893674293405,
+      "grad_norm": 1.3877427577972412,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.03857019543647766,
+      "mean_token_accuracy": 0.9856370270252228,
+      "num_tokens": 66664390.0,
+      "step": 3680
+    },
+    {
+      "entropy": 1.015517818927765,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1693767309188843,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.03634478151798248,
+      "mean_token_accuracy": 0.9868648707866668,
+      "num_tokens": 66845574.0,
+      "step": 3690
+    },
+    {
+      "entropy": 1.011656105518341,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.082209587097168,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.03513925671577454,
+      "mean_token_accuracy": 0.9872389912605286,
+      "num_tokens": 67027557.0,
+      "step": 3700
+    },
+    {
+      "entropy": 1.0161120653152467,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.266615629196167,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.0379630446434021,
+      "mean_token_accuracy": 0.9865313708782196,
+      "num_tokens": 67208088.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 1.0219634957374282,
+      "eval_loss": 0.14612138271331787,
+      "eval_mean_token_accuracy": 0.9504923186484416,
+      "eval_num_tokens": 67298266.0,
+      "eval_runtime": 12.6802,
+      "eval_samples_per_second": 394.317,
+      "eval_steps_per_second": 12.382,
+      "step": 3715
+    },
+    {
+      "entropy": 1.019774079322815,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8909980654716492,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.02882351279258728,
+      "mean_token_accuracy": 0.9900537073612213,
+      "num_tokens": 67388858.0,
+      "step": 3720
+    },
+    {
+      "entropy": 1.0057315409183503,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9996656775474548,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.02233442962169647,
+      "mean_token_accuracy": 0.9922977447509765,
+      "num_tokens": 67570084.0,
+      "step": 3730
+    },
+    {
+      "entropy": 1.0058627128601074,
+      "epoch": 5.03364737550471,
+      "grad_norm": 1.0265147686004639,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.021967282891273497,
+      "mean_token_accuracy": 0.9923401355743409,
+      "num_tokens": 67751587.0,
+      "step": 3740
+    },
+    {
+      "entropy": 1.0083157896995545,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 1.0139433145523071,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.022531555593013765,
+      "mean_token_accuracy": 0.9930226564407348,
+      "num_tokens": 67933269.0,
+      "step": 3750
+    },
+    {
+      "entropy": 1.0119245886802672,
+      "epoch": 5.060565275908479,
+      "grad_norm": 1.0655452013015747,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.021685637533664703,
+      "mean_token_accuracy": 0.9925794541835785,
+      "num_tokens": 68113971.0,
+      "step": 3760
+    },
+    {
+      "entropy": 1.0082412719726563,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.268751859664917,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.021513286232948303,
+      "mean_token_accuracy": 0.9927303433418274,
+      "num_tokens": 68295632.0,
+      "step": 3770
+    },
+    {
+      "entropy": 1.012417995929718,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.9899319410324097,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.02352543771266937,
+      "mean_token_accuracy": 0.992375648021698,
+      "num_tokens": 68476794.0,
+      "step": 3780
+    },
+    {
+      "entropy": 1.0069364666938783,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 1.1964375972747803,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.021382461488246917,
+      "mean_token_accuracy": 0.9928456962108612,
+      "num_tokens": 68658249.0,
+      "step": 3790
+    },
+    {
+      "entropy": 1.0040670156478881,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0445663928985596,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.02207571566104889,
+      "mean_token_accuracy": 0.9927122712135314,
+      "num_tokens": 68840525.0,
+      "step": 3800
+    },
+    {
+      "entropy": 1.0083145201206207,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.8981389403343201,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.022873447835445405,
+      "mean_token_accuracy": 0.9918712139129638,
+      "num_tokens": 69021896.0,
+      "step": 3810
+    },
+    {
+      "entropy": 1.0109572768211366,
+      "epoch": 5.141318977119784,
+      "grad_norm": 1.250670313835144,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.023577621579170226,
+      "mean_token_accuracy": 0.9919484734535218,
+      "num_tokens": 69203112.0,
+      "step": 3820
+    },
+    {
+      "entropy": 1.0164754271507264,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.3053207397460938,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.02436322420835495,
+      "mean_token_accuracy": 0.9921659469604492,
+      "num_tokens": 69384193.0,
+      "step": 3830
+    },
+    {
+      "entropy": 1.0034007966518401,
+      "epoch": 5.168236877523553,
+      "grad_norm": 1.4878787994384766,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.023177918791770936,
+      "mean_token_accuracy": 0.9919318974018096,
+      "num_tokens": 69565602.0,
+      "step": 3840
+    },
+    {
+      "entropy": 1.0074714839458465,
+      "epoch": 5.181695827725437,
+      "grad_norm": 1.0512391328811646,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.023409275710582732,
+      "mean_token_accuracy": 0.9918898940086365,
+      "num_tokens": 69746242.0,
+      "step": 3850
+    },
+    {
+      "entropy": 1.003154468536377,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.971832811832428,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.023223266005516052,
+      "mean_token_accuracy": 0.9920656085014343,
+      "num_tokens": 69927227.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.999264246225357,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.9266519546508789,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.02454002797603607,
+      "mean_token_accuracy": 0.9914694905281067,
+      "num_tokens": 70108580.0,
+      "step": 3870
+    },
+    {
+      "entropy": 1.002640813589096,
+      "epoch": 5.22207267833109,
+      "grad_norm": 1.5331412553787231,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.02458159625530243,
+      "mean_token_accuracy": 0.9919124245643616,
+      "num_tokens": 70288960.0,
+      "step": 3880
+    },
+    {
+      "entropy": 1.007998287677765,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.3164633512496948,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.02421087771654129,
+      "mean_token_accuracy": 0.9917516648769379,
+      "num_tokens": 70470266.0,
+      "step": 3890
+    },
+    {
+      "entropy": 1.0089037418365479,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1788280010223389,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.023224952816963195,
+      "mean_token_accuracy": 0.9921185493469238,
+      "num_tokens": 70650975.0,
+      "step": 3900
+    },
+    {
+      "entropy": 1.001932644844055,
+      "epoch": 5.262449528936743,
+      "grad_norm": 1.2484381198883057,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.02366664707660675,
+      "mean_token_accuracy": 0.9917596876621246,
+      "num_tokens": 70832549.0,
+      "step": 3910
+    },
+    {
+      "entropy": 1.004903507232666,
+      "epoch": 5.275908479138627,
+      "grad_norm": 1.1852507591247559,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.02296685576438904,
+      "mean_token_accuracy": 0.9918978750705719,
+      "num_tokens": 71014296.0,
+      "step": 3920
+    },
+    {
+      "entropy": 1.004357820749283,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.9933306574821472,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.02361467629671097,
+      "mean_token_accuracy": 0.991608303785324,
+      "num_tokens": 71195526.0,
+      "step": 3930
+    },
+    {
+      "entropy": 1.0020038604736328,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.4304245710372925,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.02268100529909134,
+      "mean_token_accuracy": 0.9917695164680481,
+      "num_tokens": 71377269.0,
+      "step": 3940
+    },
+    {
+      "entropy": 1.012037169933319,
+      "epoch": 5.31628532974428,
+      "grad_norm": 1.074124813079834,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.023454257845878602,
+      "mean_token_accuracy": 0.9922087669372559,
+      "num_tokens": 71558158.0,
+      "step": 3950
+    },
+    {
+      "entropy": 1.0094320178031921,
+      "epoch": 5.329744279946164,
+      "grad_norm": 1.0919133424758911,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.022751571238040925,
+      "mean_token_accuracy": 0.9921496629714965,
+      "num_tokens": 71738847.0,
+      "step": 3960
+    },
+    {
+      "entropy": 1.0063781797885896,
+      "epoch": 5.343203230148048,
+      "grad_norm": 1.1300997734069824,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.023900465667247774,
+      "mean_token_accuracy": 0.9914113521575928,
+      "num_tokens": 71918875.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.9972042560577392,
+      "epoch": 5.356662180349932,
+      "grad_norm": 1.165519118309021,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.021918240189552306,
+      "mean_token_accuracy": 0.9926528394222259,
+      "num_tokens": 72100551.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.9954554200172424,
+      "epoch": 5.370121130551817,
+      "grad_norm": 1.342820167541504,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.024170228838920595,
+      "mean_token_accuracy": 0.9913744986057281,
+      "num_tokens": 72281855.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.9931409180164337,
+      "epoch": 5.383580080753701,
+      "grad_norm": 1.2873902320861816,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.02334785759449005,
+      "mean_token_accuracy": 0.9921288907527923,
+      "num_tokens": 72462916.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.9915881812572479,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.9761195778846741,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.02302718609571457,
+      "mean_token_accuracy": 0.9924728453159333,
+      "num_tokens": 72644666.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.9931966602802277,
+      "epoch": 5.410497981157469,
+      "grad_norm": 1.4468477964401245,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.023682811856269838,
+      "mean_token_accuracy": 0.9917364478111267,
+      "num_tokens": 72825808.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.9975718319416046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.1559572219848633,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.02247478663921356,
+      "mean_token_accuracy": 0.992560338973999,
+      "num_tokens": 73007923.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.999228036403656,
+      "epoch": 5.437415881561238,
+      "grad_norm": 1.1230353116989136,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.022295859456062318,
+      "mean_token_accuracy": 0.9922321498394012,
+      "num_tokens": 73189074.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.9897105693817139,
+      "epoch": 5.450874831763122,
+      "grad_norm": 1.413734793663025,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.02201971709728241,
+      "mean_token_accuracy": 0.9927073001861573,
+      "num_tokens": 73369918.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.9986266851425171,
+      "epoch": 5.464333781965006,
+      "grad_norm": 1.0815346240997314,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.02358766496181488,
+      "mean_token_accuracy": 0.9911752700805664,
+      "num_tokens": 73550625.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.9953943610191345,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0506970882415771,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.022177645564079286,
+      "mean_token_accuracy": 0.9923022150993347,
+      "num_tokens": 73731436.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.9954936504364014,
+      "epoch": 5.491251682368775,
+      "grad_norm": 1.4665249586105347,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.024948063492774963,
+      "mean_token_accuracy": 0.9913177132606507,
+      "num_tokens": 73912097.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.9986696183681488,
+      "epoch": 5.504710632570659,
+      "grad_norm": 1.158179759979248,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.02555338144302368,
+      "mean_token_accuracy": 0.9908900320529938,
+      "num_tokens": 74093602.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.9955579698085785,
+      "epoch": 5.518169582772543,
+      "grad_norm": 1.0024714469909668,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.023432795703411103,
+      "mean_token_accuracy": 0.9919674038887024,
+      "num_tokens": 74274664.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.994187182188034,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.9290911555290222,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.023194606602191924,
+      "mean_token_accuracy": 0.9920254707336426,
+      "num_tokens": 74456436.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.9927510738372802,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.8986559510231018,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.02163631021976471,
+      "mean_token_accuracy": 0.9927416622638703,
+      "num_tokens": 74637700.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.9877451896667481,
+      "epoch": 5.558546433378196,
+      "grad_norm": 1.0314558744430542,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.0209101140499115,
+      "mean_token_accuracy": 0.9930387079715729,
+      "num_tokens": 74819310.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.9903139114379883,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 1.1819736957550049,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.02219167649745941,
+      "mean_token_accuracy": 0.992407250404358,
+      "num_tokens": 74999907.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.9891258835792541,
+      "epoch": 5.585464333781965,
+      "grad_norm": 1.0412501096725464,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.02201283872127533,
+      "mean_token_accuracy": 0.992091816663742,
+      "num_tokens": 75181189.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.9870602369308472,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0461134910583496,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.023040324449539185,
+      "mean_token_accuracy": 0.992098081111908,
+      "num_tokens": 75362769.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.9888905882835388,
+      "epoch": 5.612382234185733,
+      "grad_norm": 1.086546540260315,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.021401138603687288,
+      "mean_token_accuracy": 0.992308247089386,
+      "num_tokens": 75544519.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.9940549910068512,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 1.3897169828414917,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.023297235369682312,
+      "mean_token_accuracy": 0.9920303761959076,
+      "num_tokens": 75725651.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.9897378265857697,
+      "epoch": 5.639300134589502,
+      "grad_norm": 1.283887267112732,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.024343529343605043,
+      "mean_token_accuracy": 0.9916397035121918,
+      "num_tokens": 75907036.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.9989661812782288,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.9090050458908081,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.022062870860099792,
+      "mean_token_accuracy": 0.9921731591224671,
+      "num_tokens": 76088248.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.9966195583343506,
+      "epoch": 5.66621803499327,
+      "grad_norm": 1.2091970443725586,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.023516717553138732,
+      "mean_token_accuracy": 0.9918879508972168,
+      "num_tokens": 76269704.0,
+      "step": 4210
+    },
+    {
+      "entropy": 1.0081237196922301,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 1.0648062229156494,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.021994276344776152,
+      "mean_token_accuracy": 0.9925018906593323,
+      "num_tokens": 76451416.0,
+      "step": 4220
+    },
+    {
+      "entropy": 1.0061897933483124,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.188748836517334,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.022657927870750428,
+      "mean_token_accuracy": 0.9919365108013153,
+      "num_tokens": 76632827.0,
+      "step": 4230
+    },
+    {
+      "entropy": 1.0024715662002563,
+      "epoch": 5.706594885598923,
+      "grad_norm": 1.2140473127365112,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.02203955501317978,
+      "mean_token_accuracy": 0.9925529718399048,
+      "num_tokens": 76813534.0,
+      "step": 4240
+    },
+    {
+      "entropy": 1.0011352360248567,
+      "epoch": 5.720053835800807,
+      "grad_norm": 1.389479398727417,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.022718487679958342,
+      "mean_token_accuracy": 0.9922778069972992,
+      "num_tokens": 76994327.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.9971556246280671,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 1.1441208124160767,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.024178403615951537,
+      "mean_token_accuracy": 0.9913869142532349,
+      "num_tokens": 77174987.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.9961654126644135,
+      "epoch": 5.746971736204576,
+      "grad_norm": 1.1283735036849976,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.023389075696468354,
+      "mean_token_accuracy": 0.9921939969062805,
+      "num_tokens": 77356043.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.9927134871482849,
+      "epoch": 5.76043068640646,
+      "grad_norm": 1.201904535293579,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.022739484906196594,
+      "mean_token_accuracy": 0.9921142935752869,
+      "num_tokens": 77536396.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.9919731795787812,
+      "epoch": 5.773889636608344,
+      "grad_norm": 1.145776391029358,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.020524097979068755,
+      "mean_token_accuracy": 0.9926942765712738,
+      "num_tokens": 77717339.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.9935499131679535,
+      "epoch": 5.787348586810229,
+      "grad_norm": 1.099414348602295,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.02070924937725067,
+      "mean_token_accuracy": 0.9928180515766144,
+      "num_tokens": 77898360.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.9875575602054596,
+      "epoch": 5.800807537012113,
+      "grad_norm": 1.1510778665542603,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.02202107310295105,
+      "mean_token_accuracy": 0.9921730697154999,
+      "num_tokens": 78079699.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.9855418741703034,
+      "epoch": 5.814266487213997,
+      "grad_norm": 1.0748379230499268,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.022809866070747375,
+      "mean_token_accuracy": 0.9919199466705322,
+      "num_tokens": 78260900.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.977304744720459,
+      "epoch": 5.827725437415881,
+      "grad_norm": 1.091854453086853,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.02268455773591995,
+      "mean_token_accuracy": 0.9918094217777252,
+      "num_tokens": 78442168.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.9767287075519562,
+      "epoch": 5.841184387617766,
+      "grad_norm": 1.1712645292282104,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.022726841270923615,
+      "mean_token_accuracy": 0.9922167181968689,
+      "num_tokens": 78623589.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.974297159910202,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.9508016109466553,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.021290253102779388,
+      "mean_token_accuracy": 0.9926727712154388,
+      "num_tokens": 78804892.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.9795810282230377,
+      "epoch": 5.868102288021534,
+      "grad_norm": 1.3127822875976562,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.022329577803611757,
+      "mean_token_accuracy": 0.9927741348743438,
+      "num_tokens": 78986204.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.9879997313022614,
+      "epoch": 5.881561238223418,
+      "grad_norm": 1.0200371742248535,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.022206541895866395,
+      "mean_token_accuracy": 0.9921830892562866,
+      "num_tokens": 79166824.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.9801069736480713,
+      "epoch": 5.895020188425303,
+      "grad_norm": 1.4769432544708252,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.02101587802171707,
+      "mean_token_accuracy": 0.9930795848369598,
+      "num_tokens": 79347228.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.9727071464061737,
+      "epoch": 5.908479138627187,
+      "grad_norm": 1.7034555673599243,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.021690081059932708,
+      "mean_token_accuracy": 0.9926194071769714,
+      "num_tokens": 79529483.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.9782697260379791,
+      "epoch": 5.921938088829071,
+      "grad_norm": 1.5307973623275757,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.023112425208091737,
+      "mean_token_accuracy": 0.9919495701789856,
+      "num_tokens": 79710571.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.9831463754177093,
+      "epoch": 5.935397039030955,
+      "grad_norm": 1.1024526357650757,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.021427598595619202,
+      "mean_token_accuracy": 0.9924166023731231,
+      "num_tokens": 79890685.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.9811958551406861,
+      "epoch": 5.94885598923284,
+      "grad_norm": 1.7247848510742188,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.02339867949485779,
+      "mean_token_accuracy": 0.9916171073913574,
+      "num_tokens": 80071331.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.9830600321292877,
+      "epoch": 5.962314939434724,
+      "grad_norm": 1.0638445615768433,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.02204108238220215,
+      "mean_token_accuracy": 0.9925127327442169,
+      "num_tokens": 80251525.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.9840480864048005,
+      "epoch": 5.975773889636608,
+      "grad_norm": 1.1205570697784424,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.020443232357501985,
+      "mean_token_accuracy": 0.9932060956954956,
+      "num_tokens": 80432001.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.9841626048088074,
+      "epoch": 5.989232839838492,
+      "grad_norm": 1.2928433418273926,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.021673919260501863,
+      "mean_token_accuracy": 0.9931873917579651,
+      "num_tokens": 80613075.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.9904457463580332,
+      "eval_loss": 0.161154642701149,
+      "eval_mean_token_accuracy": 0.9516840383505366,
+      "eval_num_tokens": 80757887.0,
+      "eval_runtime": 12.7345,
+      "eval_samples_per_second": 392.633,
+      "eval_steps_per_second": 12.329,
+      "step": 4458
+    },
+    {
+      "entropy": 0.9850570559501648,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.8859100341796875,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.018883128464221955,
+      "mean_token_accuracy": 0.9933207035064697,
+      "num_tokens": 80794295.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.9796536922454834,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.8121339082717896,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.01288142502307892,
+      "mean_token_accuracy": 0.9960863590240479,
+      "num_tokens": 80975379.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.9699741184711457,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.8923460841178894,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.012284834682941437,
+      "mean_token_accuracy": 0.9960344016551972,
+      "num_tokens": 81155713.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.9629065454006195,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.6698539853096008,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.011575647443532944,
+      "mean_token_accuracy": 0.9962593257427216,
+      "num_tokens": 81336446.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.9579961061477661,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.7555826306343079,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.011635755002498627,
+      "mean_token_accuracy": 0.9965786218643189,
+      "num_tokens": 81517969.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.96330127120018,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.8983458280563354,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.012715654075145721,
+      "mean_token_accuracy": 0.9959319531917572,
+      "num_tokens": 81699419.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.9640884518623352,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.9172118902206421,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.012482476234436036,
+      "mean_token_accuracy": 0.9966461837291718,
+      "num_tokens": 81880956.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.9678717851638794,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.8765719532966614,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.012080545723438262,
+      "mean_token_accuracy": 0.9964690268039703,
+      "num_tokens": 82061897.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.9656756520271301,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.746626079082489,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.012495596706867219,
+      "mean_token_accuracy": 0.9959208548069001,
+      "num_tokens": 82242847.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.9631998062133789,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.9732807874679565,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.01309359222650528,
+      "mean_token_accuracy": 0.9960879981517792,
+      "num_tokens": 82424182.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.9686778664588929,
+      "epoch": 6.137281292059219,
+      "grad_norm": 1.040855884552002,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.013380664587020873,
+      "mean_token_accuracy": 0.9955641269683838,
+      "num_tokens": 82604728.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.9661992788314819,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 1.209067940711975,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.013264504075050355,
+      "mean_token_accuracy": 0.9958661019802093,
+      "num_tokens": 82786184.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.9628405570983887,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.9056928157806396,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.011586685478687287,
+      "mean_token_accuracy": 0.9962525606155396,
+      "num_tokens": 82968127.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.9664590656757355,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.8662444949150085,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.013227257132530212,
+      "mean_token_accuracy": 0.9958321928977967,
+      "num_tokens": 83149016.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.9584677755832672,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.8611947894096375,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.012484049052000045,
+      "mean_token_accuracy": 0.9961173057556152,
+      "num_tokens": 83330474.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.9613959610462188,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 1.0272881984710693,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.013143527507781982,
+      "mean_token_accuracy": 0.995938628911972,
+      "num_tokens": 83512502.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.9671844780445099,
+      "epoch": 6.218034993270525,
+      "grad_norm": 1.038604497909546,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.015797269344329835,
+      "mean_token_accuracy": 0.9949265837669372,
+      "num_tokens": 83693427.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.9725019454956054,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.7591229677200317,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.01309441477060318,
+      "mean_token_accuracy": 0.9960084438323975,
+      "num_tokens": 83874886.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.9710726916790009,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.9521571397781372,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.012572157382965087,
+      "mean_token_accuracy": 0.9960904121398926,
+      "num_tokens": 84055459.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.9712884068489075,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.812011182308197,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.014366535842418671,
+      "mean_token_accuracy": 0.9953844547271729,
+      "num_tokens": 84236017.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.9648497045040131,
+      "epoch": 6.271870794078062,
+      "grad_norm": 1.648348093032837,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.014780099689960479,
+      "mean_token_accuracy": 0.9955918669700623,
+      "num_tokens": 84417271.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.956806218624115,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.9592521786689758,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.014198371767997741,
+      "mean_token_accuracy": 0.9954844176769256,
+      "num_tokens": 84599326.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.9628090143203736,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.9466222524642944,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.013178789615631103,
+      "mean_token_accuracy": 0.995667016506195,
+      "num_tokens": 84779914.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.9617082297801971,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.9156956672668457,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.011176959425210953,
+      "mean_token_accuracy": 0.9965419709682465,
+      "num_tokens": 84960572.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.9526888430118561,
+      "epoch": 6.325706594885599,
+      "grad_norm": 1.107490062713623,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.01246263161301613,
+      "mean_token_accuracy": 0.9961843729019165,
+      "num_tokens": 85141667.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.952410751581192,
+      "epoch": 6.339165545087483,
+      "grad_norm": 1.06051766872406,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.01356920450925827,
+      "mean_token_accuracy": 0.9956991016864777,
+      "num_tokens": 85323361.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.9601339221000671,
+      "epoch": 6.352624495289367,
+      "grad_norm": 1.0533535480499268,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.012942484021186829,
+      "mean_token_accuracy": 0.9960961997509002,
+      "num_tokens": 85503828.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.9630769610404968,
+      "epoch": 6.366083445491252,
+      "grad_norm": 1.2243846654891968,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.012688860297203064,
+      "mean_token_accuracy": 0.9956350862979889,
+      "num_tokens": 85684804.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.9601665258407592,
+      "epoch": 6.379542395693136,
+      "grad_norm": 1.0219420194625854,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.012651938199996948,
+      "mean_token_accuracy": 0.9959766507148743,
+      "num_tokens": 85866033.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.9643365681171417,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.8759870529174805,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.013052919507026672,
+      "mean_token_accuracy": 0.9956570088863372,
+      "num_tokens": 86046793.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.9643962621688843,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.8771425485610962,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.013016942143440246,
+      "mean_token_accuracy": 0.9956202030181884,
+      "num_tokens": 86227594.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.9684974253177643,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.9857693314552307,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.01260487139225006,
+      "mean_token_accuracy": 0.9959571301937103,
+      "num_tokens": 86409012.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.9683425962924957,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.9220049381256104,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.012756258249282837,
+      "mean_token_accuracy": 0.9961577892303467,
+      "num_tokens": 86589770.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.9730850279331207,
+      "epoch": 6.446837146702557,
+      "grad_norm": 1.1092842817306519,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.011953817307949066,
+      "mean_token_accuracy": 0.9963648617267609,
+      "num_tokens": 86771224.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.9717083096504211,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.9012715220451355,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.0137409508228302,
+      "mean_token_accuracy": 0.9956796884536743,
+      "num_tokens": 86952165.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.9706744968891143,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.8237823247909546,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.013622400164604188,
+      "mean_token_accuracy": 0.9958401381969452,
+      "num_tokens": 87133948.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.968240886926651,
+      "epoch": 6.48721399730821,
+      "grad_norm": 0.9172955751419067,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.012218748033046723,
+      "mean_token_accuracy": 0.9961099803447724,
+      "num_tokens": 87315186.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.9701445996761322,
+      "epoch": 6.500672947510094,
+      "grad_norm": 1.1613796949386597,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.011357608437538146,
+      "mean_token_accuracy": 0.9961966574192047,
+      "num_tokens": 87496318.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.9753403663635254,
+      "epoch": 6.514131897711978,
+      "grad_norm": 1.0601547956466675,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.01224931925535202,
+      "mean_token_accuracy": 0.9964543879032135,
+      "num_tokens": 87676690.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.9595320463180542,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.8927657604217529,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.012476099282503128,
+      "mean_token_accuracy": 0.9961197137832641,
+      "num_tokens": 87858122.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.9639881610870361,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.9392650127410889,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.013359837234020233,
+      "mean_token_accuracy": 0.9956938862800598,
+      "num_tokens": 88039349.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.961546677350998,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.709064781665802,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.012232375890016555,
+      "mean_token_accuracy": 0.9963175058364868,
+      "num_tokens": 88221085.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.966980516910553,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 1.1382325887680054,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.013281235098838806,
+      "mean_token_accuracy": 0.9955899178981781,
+      "num_tokens": 88401750.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.967098617553711,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.7260605692863464,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.011995580792427064,
+      "mean_token_accuracy": 0.9963020443916321,
+      "num_tokens": 88582188.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.9611273169517517,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.9492855668067932,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.011460936814546584,
+      "mean_token_accuracy": 0.9965442717075348,
+      "num_tokens": 88764018.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.964341527223587,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.7225981950759888,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.011858727037906646,
+      "mean_token_accuracy": 0.996355015039444,
+      "num_tokens": 88944988.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.9631503582000732,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 0.996126651763916,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.013101565837860107,
+      "mean_token_accuracy": 0.9957616865634918,
+      "num_tokens": 89125768.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.9629738807678223,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.8971033096313477,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.012153981626033783,
+      "mean_token_accuracy": 0.9961553812026978,
+      "num_tokens": 89306694.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.9592839002609252,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.8334258198738098,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.012248124927282333,
+      "mean_token_accuracy": 0.9960185050964355,
+      "num_tokens": 89487628.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.9580614447593689,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.7489107251167297,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.012415885925292969,
+      "mean_token_accuracy": 0.9960191190242768,
+      "num_tokens": 89668907.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.9544502675533295,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.9663529992103577,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.012825769186019898,
+      "mean_token_accuracy": 0.9958091497421264,
+      "num_tokens": 89850200.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.9588588118553162,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.7662053108215332,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.012289515137672425,
+      "mean_token_accuracy": 0.9956038355827331,
+      "num_tokens": 90030738.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.9632011234760285,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.8352729082107544,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.01209896057844162,
+      "mean_token_accuracy": 0.9960383713245392,
+      "num_tokens": 90211066.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.962366932630539,
+      "epoch": 6.716016150740242,
+      "grad_norm": 1.271884799003601,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.013325554132461549,
+      "mean_token_accuracy": 0.9955421090126038,
+      "num_tokens": 90392202.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.9630074858665466,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.806925892829895,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.012963607907295227,
+      "mean_token_accuracy": 0.9955462396144867,
+      "num_tokens": 90573271.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.9631992042064667,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.9406610131263733,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.012753760814666748,
+      "mean_token_accuracy": 0.9954385042190552,
+      "num_tokens": 90753836.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.9599998474121094,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.7755982875823975,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.011351206153631211,
+      "mean_token_accuracy": 0.9965151965618133,
+      "num_tokens": 90935528.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.961510854959488,
+      "epoch": 6.769851951547779,
+      "grad_norm": 1.223897099494934,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.012862575054168702,
+      "mean_token_accuracy": 0.995742017030716,
+      "num_tokens": 91116520.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.9597096800804138,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.8847037553787231,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.01192454993724823,
+      "mean_token_accuracy": 0.9964925825595856,
+      "num_tokens": 91297847.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.961575037240982,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.7097873687744141,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.012100750207901,
+      "mean_token_accuracy": 0.9963988482952117,
+      "num_tokens": 91479371.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.9660714268684387,
+      "epoch": 6.810228802153432,
+      "grad_norm": 0.8912429213523865,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.011809615790843964,
+      "mean_token_accuracy": 0.9961784362792969,
+      "num_tokens": 91660175.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.9656784057617187,
+      "epoch": 6.823687752355316,
+      "grad_norm": 1.2362871170043945,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.013092437386512756,
+      "mean_token_accuracy": 0.9954990804195404,
+      "num_tokens": 91840728.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.9630216419696808,
+      "epoch": 6.837146702557201,
+      "grad_norm": 1.0081552267074585,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.013006564974784852,
+      "mean_token_accuracy": 0.995445728302002,
+      "num_tokens": 92021819.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.9650352597236633,
+      "epoch": 6.850605652759085,
+      "grad_norm": 1.0923664569854736,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.012209924310445786,
+      "mean_token_accuracy": 0.9961765468120575,
+      "num_tokens": 92203146.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.9654329180717468,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.8630998730659485,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.01286689043045044,
+      "mean_token_accuracy": 0.9955997586250305,
+      "num_tokens": 92384274.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.9633705973625183,
+      "epoch": 6.877523553162853,
+      "grad_norm": 0.9883823394775391,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.012318137288093566,
+      "mean_token_accuracy": 0.9964197039604187,
+      "num_tokens": 92564922.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.9619575142860413,
+      "epoch": 6.890982503364738,
+      "grad_norm": 1.0694894790649414,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.011325329542160034,
+      "mean_token_accuracy": 0.9967617630958557,
+      "num_tokens": 92745952.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.9608132243156433,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.850687563419342,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.012749406695365905,
+      "mean_token_accuracy": 0.9958359479904175,
+      "num_tokens": 92926966.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.9643485128879548,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.8775151968002319,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.012118304520845414,
+      "mean_token_accuracy": 0.996504932641983,
+      "num_tokens": 93108549.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.9634061992168427,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.925463855266571,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.011662732064723968,
+      "mean_token_accuracy": 0.9965654671192169,
+      "num_tokens": 93290512.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.9620663046836853,
+      "epoch": 6.944818304172275,
+      "grad_norm": 1.1143932342529297,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.01120457649230957,
+      "mean_token_accuracy": 0.9962508201599121,
+      "num_tokens": 93472398.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.9628985583782196,
+      "epoch": 6.958277254374159,
+      "grad_norm": 1.0271083116531372,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.012035199999809265,
+      "mean_token_accuracy": 0.996053409576416,
+      "num_tokens": 93654111.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.9658466935157776,
+      "epoch": 6.971736204576043,
+      "grad_norm": 1.0225892066955566,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.01229911893606186,
+      "mean_token_accuracy": 0.995949512720108,
+      "num_tokens": 93835692.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.9601108253002166,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.7876276969909668,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.011063569784164428,
+      "mean_token_accuracy": 0.9966089725494385,
+      "num_tokens": 94018119.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.9596541881561279,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.9803016185760498,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.011228029429912568,
+      "mean_token_accuracy": 0.9964218616485596,
+      "num_tokens": 94199483.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.9689761833020836,
+      "eval_loss": 0.1798103004693985,
+      "eval_mean_token_accuracy": 0.950934177371347,
+      "eval_num_tokens": 94217754.0,
+      "eval_runtime": 12.6557,
+      "eval_samples_per_second": 395.079,
+      "eval_steps_per_second": 12.405,
+      "step": 5201
+    },
+    {
+      "entropy": 0.9555480360984803,
+      "epoch": 7.012113055181696,
+      "grad_norm": 0.7279857397079468,
+      "learning_rate": 2.2506648060191042e-05,
+      "loss": 0.007322490215301514,
+      "mean_token_accuracy": 0.9978758811950683,
+      "num_tokens": 94380656.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.9532195508480072,
+      "epoch": 7.02557200538358,
+      "grad_norm": 0.6351640820503235,
+      "learning_rate": 2.2321031102941066e-05,
+      "loss": 0.007188012450933456,
+      "mean_token_accuracy": 0.9980705499649047,
+      "num_tokens": 94561769.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.9503021717071534,
+      "epoch": 7.039030955585464,
+      "grad_norm": 0.768357515335083,
+      "learning_rate": 2.2135962530768055e-05,
+      "loss": 0.007337777316570282,
+      "mean_token_accuracy": 0.9979041218757629,
+      "num_tokens": 94742683.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.9469992637634277,
+      "epoch": 7.052489905787349,
+      "grad_norm": 0.6551904678344727,
+      "learning_rate": 2.1951446010313053e-05,
+      "loss": 0.00702056959271431,
+      "mean_token_accuracy": 0.9980945587158203,
+      "num_tokens": 94924159.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.9499239981174469,
+      "epoch": 7.065948855989233,
+      "grad_norm": 0.8986485600471497,
+      "learning_rate": 2.1767485197279662e-05,
+      "loss": 0.007891762256622314,
+      "mean_token_accuracy": 0.9978051722049713,
+      "num_tokens": 95105209.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.9502587139606475,
+      "epoch": 7.079407806191117,
+      "grad_norm": 0.4773225486278534,
+      "learning_rate": 2.1584083736361617e-05,
+      "loss": 0.0066356688737869264,
+      "mean_token_accuracy": 0.9983843564987183,
+      "num_tokens": 95286168.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.9485512435436249,
+      "epoch": 7.0928667563930015,
+      "grad_norm": 0.7873836159706116,
+      "learning_rate": 2.1401245261170582e-05,
+      "loss": 0.007745465636253357,
+      "mean_token_accuracy": 0.9978477001190186,
+      "num_tokens": 95467244.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.9444929420948028,
+      "epoch": 7.106325706594886,
+      "grad_norm": 0.9345479607582092,
+      "learning_rate": 2.121897339416416e-05,
+      "loss": 0.006544182449579239,
+      "mean_token_accuracy": 0.9983295381069184,
+      "num_tokens": 95648662.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.9411603569984436,
+      "epoch": 7.11978465679677,
+      "grad_norm": 0.8139040470123291,
+      "learning_rate": 2.103727174657412e-05,
+      "loss": 0.00677436888217926,
+      "mean_token_accuracy": 0.9982263207435608,
+      "num_tokens": 95829602.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.9461234629154205,
+      "epoch": 7.133243606998654,
+      "grad_norm": 0.7345687747001648,
+      "learning_rate": 2.085614391833486e-05,
+      "loss": 0.006355223059654236,
+      "mean_token_accuracy": 0.9985966920852661,
+      "num_tokens": 96010655.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.9512575507164002,
+      "epoch": 7.1467025572005385,
+      "grad_norm": 0.799483060836792,
+      "learning_rate": 2.0675593498012052e-05,
+      "loss": 0.006509234011173248,
+      "mean_token_accuracy": 0.9983614802360534,
+      "num_tokens": 96190810.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.9446900069713593,
+      "epoch": 7.160161507402423,
+      "grad_norm": 0.7756890654563904,
+      "learning_rate": 2.049562406273159e-05,
+      "loss": 0.0069117918610572815,
+      "mean_token_accuracy": 0.9983683228492737,
+      "num_tokens": 96371920.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.9443341970443726,
+      "epoch": 7.173620457604307,
+      "grad_norm": 0.7561301589012146,
+      "learning_rate": 2.0316239178108688e-05,
+      "loss": 0.006773699820041656,
+      "mean_token_accuracy": 0.9981043338775635,
+      "num_tokens": 96553406.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.9383498072624207,
+      "epoch": 7.187079407806191,
+      "grad_norm": 0.7623066306114197,
+      "learning_rate": 2.013744239817725e-05,
+      "loss": 0.006338060647249222,
+      "mean_token_accuracy": 0.9983479022979737,
+      "num_tokens": 96735706.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.9444310605525971,
+      "epoch": 7.2005383580080755,
+      "grad_norm": 0.4759703576564789,
+      "learning_rate": 1.9959237265319452e-05,
+      "loss": 0.006821495294570923,
+      "mean_token_accuracy": 0.9982879221439361,
+      "num_tokens": 96916401.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.9462945520877838,
+      "epoch": 7.21399730820996,
+      "grad_norm": 0.5842452049255371,
+      "learning_rate": 1.9781627310195528e-05,
+      "loss": 0.0067993059754371645,
+      "mean_token_accuracy": 0.9981901407241821,
+      "num_tokens": 97097081.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.9475398421287536,
+      "epoch": 7.227456258411844,
+      "grad_norm": 0.8197925686836243,
+      "learning_rate": 1.960461605167389e-05,
+      "loss": 0.006510883569717407,
+      "mean_token_accuracy": 0.9984028697013855,
+      "num_tokens": 97277934.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.941127610206604,
+      "epoch": 7.240915208613728,
+      "grad_norm": 0.37160298228263855,
+      "learning_rate": 1.9428206996761357e-05,
+      "loss": 0.006270279735326767,
+      "mean_token_accuracy": 0.9982465505599976,
+      "num_tokens": 97459337.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.9421460628509521,
+      "epoch": 7.2543741588156125,
+      "grad_norm": 0.545837938785553,
+      "learning_rate": 1.9252403640533678e-05,
+      "loss": 0.006034504622220993,
+      "mean_token_accuracy": 0.9986165881156921,
+      "num_tokens": 97640387.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.9390841007232666,
+      "epoch": 7.267833109017497,
+      "grad_norm": 0.7546548247337341,
+      "learning_rate": 1.9077209466066315e-05,
+      "loss": 0.007177229970693588,
+      "mean_token_accuracy": 0.9978139519691467,
+      "num_tokens": 97821851.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.9392310976982117,
+      "epoch": 7.281292059219381,
+      "grad_norm": 0.8076840043067932,
+      "learning_rate": 1.8902627944365396e-05,
+      "loss": 0.006884847581386566,
+      "mean_token_accuracy": 0.9983299553394318,
+      "num_tokens": 98003599.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.9376516520977021,
+      "epoch": 7.294751009421265,
+      "grad_norm": 0.6652392148971558,
+      "learning_rate": 1.872866253429897e-05,
+      "loss": 0.005731315910816192,
+      "mean_token_accuracy": 0.9984779059886932,
+      "num_tokens": 98184942.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.9412936687469482,
+      "epoch": 7.30820995962315,
+      "grad_norm": 0.7955637574195862,
+      "learning_rate": 1.8555316682528485e-05,
+      "loss": 0.0065481230616569516,
+      "mean_token_accuracy": 0.9982032060623169,
+      "num_tokens": 98366577.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.9405987918376922,
+      "epoch": 7.321668909825034,
+      "grad_norm": 0.89079350233078,
+      "learning_rate": 1.8382593823440474e-05,
+      "loss": 0.006681758165359497,
+      "mean_token_accuracy": 0.9983046293258667,
+      "num_tokens": 98547681.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.944351601600647,
+      "epoch": 7.335127860026918,
+      "grad_norm": 1.1325428485870361,
+      "learning_rate": 1.821049737907854e-05,
+      "loss": 0.006995528936386108,
+      "mean_token_accuracy": 0.9982137680053711,
+      "num_tokens": 98728624.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.9397885262966156,
+      "epoch": 7.348586810228802,
+      "grad_norm": 0.7245733737945557,
+      "learning_rate": 1.803903075907554e-05,
+      "loss": 0.007640085369348526,
+      "mean_token_accuracy": 0.9979041159152985,
+      "num_tokens": 98910132.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.9460044980049134,
+      "epoch": 7.362045760430687,
+      "grad_norm": 0.9225216507911682,
+      "learning_rate": 1.7868197360586054e-05,
+      "loss": 0.006461286544799804,
+      "mean_token_accuracy": 0.9983667492866516,
+      "num_tokens": 99091179.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.9402543127536773,
+      "epoch": 7.375504710632571,
+      "grad_norm": 0.7389097213745117,
+      "learning_rate": 1.769800056821904e-05,
+      "loss": 0.0065854907035827635,
+      "mean_token_accuracy": 0.9979489326477051,
+      "num_tokens": 99273068.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.9478578507900238,
+      "epoch": 7.388963660834455,
+      "grad_norm": 1.0250836610794067,
+      "learning_rate": 1.7528443753970814e-05,
+      "loss": 0.006443935632705689,
+      "mean_token_accuracy": 0.998309975862503,
+      "num_tokens": 99453517.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.9427528262138367,
+      "epoch": 7.402422611036339,
+      "grad_norm": 0.7677002549171448,
+      "learning_rate": 1.7359530277158227e-05,
+      "loss": 0.006590449810028076,
+      "mean_token_accuracy": 0.9982711434364319,
+      "num_tokens": 99634833.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.9481732070446014,
+      "epoch": 7.415881561238224,
+      "grad_norm": 0.9008362889289856,
+      "learning_rate": 1.7191263484352106e-05,
+      "loss": 0.006611865013837814,
+      "mean_token_accuracy": 0.9980513393878937,
+      "num_tokens": 99815381.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.9450771987438202,
+      "epoch": 7.429340511440108,
+      "grad_norm": 0.6731723546981812,
+      "learning_rate": 1.702364670931097e-05,
+      "loss": 0.006989569962024688,
+      "mean_token_accuracy": 0.9981782495975494,
+      "num_tokens": 99996758.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.9488214910030365,
+      "epoch": 7.442799461641992,
+      "grad_norm": 0.543502151966095,
+      "learning_rate": 1.685668327291495e-05,
+      "loss": 0.0056597858667373655,
+      "mean_token_accuracy": 0.9984924256801605,
+      "num_tokens": 100177864.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.9521346092224121,
+      "epoch": 7.456258411843876,
+      "grad_norm": 0.6044306755065918,
+      "learning_rate": 1.6690376483100034e-05,
+      "loss": 0.006509961187839508,
+      "mean_token_accuracy": 0.9981269180774689,
+      "num_tokens": 100358496.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.9487424850463867,
+      "epoch": 7.469717362045761,
+      "grad_norm": 0.8417298197746277,
+      "learning_rate": 1.6524729634792475e-05,
+      "loss": 0.006116047501564026,
+      "mean_token_accuracy": 0.9982541143894196,
+      "num_tokens": 100539331.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.9476044595241546,
+      "epoch": 7.483176312247645,
+      "grad_norm": 0.8056595325469971,
+      "learning_rate": 1.6359746009843567e-05,
+      "loss": 0.0057601254433393475,
+      "mean_token_accuracy": 0.9985706150531769,
+      "num_tokens": 100720450.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.9456092417240143,
+      "epoch": 7.496635262449529,
+      "grad_norm": 0.6036805510520935,
+      "learning_rate": 1.6195428876964592e-05,
+      "loss": 0.005734497308731079,
+      "mean_token_accuracy": 0.9984824895858765,
+      "num_tokens": 100902228.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.946131706237793,
+      "epoch": 7.510094212651413,
+      "grad_norm": 0.7125651240348816,
+      "learning_rate": 1.6031781491662056e-05,
+      "loss": 0.005733318626880646,
+      "mean_token_accuracy": 0.9984204411506653,
+      "num_tokens": 101083443.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.9503833770751953,
+      "epoch": 7.523553162853298,
+      "grad_norm": 0.6275620460510254,
+      "learning_rate": 1.58688070961732e-05,
+      "loss": 0.006019812077283859,
+      "mean_token_accuracy": 0.9983126580715179,
+      "num_tokens": 101264443.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.947618979215622,
+      "epoch": 7.537012113055182,
+      "grad_norm": 0.7690677046775818,
+      "learning_rate": 1.5706508919401804e-05,
+      "loss": 0.005866598710417748,
+      "mean_token_accuracy": 0.9985946953296662,
+      "num_tokens": 101445485.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.9441321969032288,
+      "epoch": 7.550471063257066,
+      "grad_norm": 0.6154682636260986,
+      "learning_rate": 1.554489017685412e-05,
+      "loss": 0.00593491718173027,
+      "mean_token_accuracy": 0.9986464619636536,
+      "num_tokens": 101626547.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.9452862203121185,
+      "epoch": 7.56393001345895,
+      "grad_norm": 0.8111712336540222,
+      "learning_rate": 1.538395407057525e-05,
+      "loss": 0.006593097746372223,
+      "mean_token_accuracy": 0.998135757446289,
+      "num_tokens": 101807733.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.9418992578983307,
+      "epoch": 7.577388963660835,
+      "grad_norm": 0.7640061974525452,
+      "learning_rate": 1.522370378908567e-05,
+      "loss": 0.006091068312525749,
+      "mean_token_accuracy": 0.9984888017177582,
+      "num_tokens": 101989357.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.9476424336433411,
+      "epoch": 7.590847913862719,
+      "grad_norm": 0.5616353750228882,
+      "learning_rate": 1.5064142507318051e-05,
+      "loss": 0.0060265038162469866,
+      "mean_token_accuracy": 0.9982646107673645,
+      "num_tokens": 102170318.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.9492906272411347,
+      "epoch": 7.604306864064603,
+      "grad_norm": 0.6769309043884277,
+      "learning_rate": 1.4905273386554386e-05,
+      "loss": 0.006565354764461517,
+      "mean_token_accuracy": 0.9979823470115662,
+      "num_tokens": 102351383.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.9434434294700622,
+      "epoch": 7.6177658142664875,
+      "grad_norm": 0.8144317865371704,
+      "learning_rate": 1.4747099574363332e-05,
+      "loss": 0.006383077055215835,
+      "mean_token_accuracy": 0.9983625173568725,
+      "num_tokens": 102532483.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.9410477578639984,
+      "epoch": 7.631224764468372,
+      "grad_norm": 0.7184897065162659,
+      "learning_rate": 1.4589624204537866e-05,
+      "loss": 0.00683993250131607,
+      "mean_token_accuracy": 0.9981952488422394,
+      "num_tokens": 102714429.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.9437882840633393,
+      "epoch": 7.644683714670256,
+      "grad_norm": 0.7210987210273743,
+      "learning_rate": 1.4432850397033176e-05,
+      "loss": 0.006337168067693711,
+      "mean_token_accuracy": 0.9981409192085267,
+      "num_tokens": 102895693.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.9423653900623321,
+      "epoch": 7.65814266487214,
+      "grad_norm": 0.7256081700325012,
+      "learning_rate": 1.427678125790487e-05,
+      "loss": 0.006422295421361923,
+      "mean_token_accuracy": 0.9981527864933014,
+      "num_tokens": 103076994.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.9402276873588562,
+      "epoch": 7.6716016150740245,
+      "grad_norm": 0.9724451899528503,
+      "learning_rate": 1.4121419879247422e-05,
+      "loss": 0.006309093534946441,
+      "mean_token_accuracy": 0.9982479989528656,
+      "num_tokens": 103258251.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.9454701244831085,
+      "epoch": 7.685060565275909,
+      "grad_norm": 0.40232348442077637,
+      "learning_rate": 1.3966769339132929e-05,
+      "loss": 0.005939274653792381,
+      "mean_token_accuracy": 0.998301374912262,
+      "num_tokens": 103438976.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.9492968201637269,
+      "epoch": 7.698519515477793,
+      "grad_norm": 0.598109781742096,
+      "learning_rate": 1.3812832701550099e-05,
+      "loss": 0.00723060742020607,
+      "mean_token_accuracy": 0.9979724586009979,
+      "num_tokens": 103619841.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.9433623433113099,
+      "epoch": 7.711978465679677,
+      "grad_norm": 0.7593542337417603,
+      "learning_rate": 1.365961301634357e-05,
+      "loss": 0.0071952566504478455,
+      "mean_token_accuracy": 0.9981424927711486,
+      "num_tokens": 103801692.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.9490162670612335,
+      "epoch": 7.7254374158815615,
+      "grad_norm": 0.6364531517028809,
+      "learning_rate": 1.350711331915348e-05,
+      "loss": 0.006192450225353241,
+      "mean_token_accuracy": 0.9984559118747711,
+      "num_tokens": 103982367.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.942770904302597,
+      "epoch": 7.738896366083446,
+      "grad_norm": 0.9279600977897644,
+      "learning_rate": 1.3355336631355314e-05,
+      "loss": 0.005739518254995346,
+      "mean_token_accuracy": 0.9985584795475007,
+      "num_tokens": 104164149.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.9427549839019775,
+      "epoch": 7.75235531628533,
+      "grad_norm": 0.9408928155899048,
+      "learning_rate": 1.3204285960000051e-05,
+      "loss": 0.005749309062957763,
+      "mean_token_accuracy": 0.9985000133514405,
+      "num_tokens": 104345175.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.9404960751533509,
+      "epoch": 7.765814266487214,
+      "grad_norm": 0.7260274291038513,
+      "learning_rate": 1.30539642977546e-05,
+      "loss": 0.006429329514503479,
+      "mean_token_accuracy": 0.9984842240810394,
+      "num_tokens": 104526928.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.9448217689990998,
+      "epoch": 7.7792732166890985,
+      "grad_norm": 0.6738228797912598,
+      "learning_rate": 1.2904374622842486e-05,
+      "loss": 0.005977935343980789,
+      "mean_token_accuracy": 0.9985457420349121,
+      "num_tokens": 104708230.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.9451113224029541,
+      "epoch": 7.792732166890983,
+      "grad_norm": 0.9410247802734375,
+      "learning_rate": 1.2755519898984846e-05,
+      "loss": 0.005663443356752396,
+      "mean_token_accuracy": 0.9984429180622101,
+      "num_tokens": 104888647.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.9483637988567353,
+      "epoch": 7.806191117092867,
+      "grad_norm": 0.5315118432044983,
+      "learning_rate": 1.2607403075341728e-05,
+      "loss": 0.005260968953371048,
+      "mean_token_accuracy": 0.9988214015960694,
+      "num_tokens": 105068604.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.9461100101470947,
+      "epoch": 7.819650067294751,
+      "grad_norm": 0.5080281496047974,
+      "learning_rate": 1.2460027086453658e-05,
+      "loss": 0.005320623517036438,
+      "mean_token_accuracy": 0.9986594915390015,
+      "num_tokens": 105249023.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.9451904594898224,
+      "epoch": 7.833109017496636,
+      "grad_norm": 0.6318531632423401,
+      "learning_rate": 1.2313394852183473e-05,
+      "loss": 0.005546718463301658,
+      "mean_token_accuracy": 0.9985934972763062,
+      "num_tokens": 105429833.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.9423372805118561,
+      "epoch": 7.84656796769852,
+      "grad_norm": 0.6764518022537231,
+      "learning_rate": 1.2167509277658518e-05,
+      "loss": 0.005748329311609268,
+      "mean_token_accuracy": 0.9986869513988494,
+      "num_tokens": 105611004.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.9406632006168365,
+      "epoch": 7.860026917900404,
+      "grad_norm": 0.6841229796409607,
+      "learning_rate": 1.2022373253213049e-05,
+      "loss": 0.005423003435134887,
+      "mean_token_accuracy": 0.9984905958175659,
+      "num_tokens": 105791366.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.9383720457553864,
+      "epoch": 7.873485868102288,
+      "grad_norm": 0.7743740677833557,
+      "learning_rate": 1.1877989654330973e-05,
+      "loss": 0.005470183119177819,
+      "mean_token_accuracy": 0.9987134814262391,
+      "num_tokens": 105973009.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.9405181050300598,
+      "epoch": 7.886944818304173,
+      "grad_norm": 0.6537275314331055,
+      "learning_rate": 1.1734361341588912e-05,
+      "loss": 0.005447608605027199,
+      "mean_token_accuracy": 0.9983970046043396,
+      "num_tokens": 106153797.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.935320508480072,
+      "epoch": 7.900403768506057,
+      "grad_norm": 0.8727037310600281,
+      "learning_rate": 1.1591491160599488e-05,
+      "loss": 0.004841895401477813,
+      "mean_token_accuracy": 0.9990314602851867,
+      "num_tokens": 106335272.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.9389428675174714,
+      "epoch": 7.913862718707941,
+      "grad_norm": 0.8051491379737854,
+      "learning_rate": 1.1449381941954956e-05,
+      "loss": 0.005671492218971253,
+      "mean_token_accuracy": 0.9985178470611572,
+      "num_tokens": 106516472.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.9336957454681396,
+      "epoch": 7.927321668909825,
+      "grad_norm": 0.6917922496795654,
+      "learning_rate": 1.130803650117116e-05,
+      "loss": 0.005551187694072724,
+      "mean_token_accuracy": 0.9983591079711914,
+      "num_tokens": 106697785.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.9350760340690613,
+      "epoch": 7.94078061911171,
+      "grad_norm": 0.7642367482185364,
+      "learning_rate": 1.1167457638631656e-05,
+      "loss": 0.005567619949579239,
+      "mean_token_accuracy": 0.9985032379627228,
+      "num_tokens": 106879069.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.9362707912921906,
+      "epoch": 7.954239569313594,
+      "grad_norm": 0.6716526746749878,
+      "learning_rate": 1.1027648139532355e-05,
+      "loss": 0.005269525200128555,
+      "mean_token_accuracy": 0.9987157702445983,
+      "num_tokens": 107060235.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.9299384176731109,
+      "epoch": 7.967698519515478,
+      "grad_norm": 0.6839614510536194,
+      "learning_rate": 1.0888610773826253e-05,
+      "loss": 0.00708087682723999,
+      "mean_token_accuracy": 0.9982430160045623,
+      "num_tokens": 107242113.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.9352255225181579,
+      "epoch": 7.981157469717362,
+      "grad_norm": 1.0992971658706665,
+      "learning_rate": 1.0750348296168583e-05,
+      "loss": 0.005663050711154938,
+      "mean_token_accuracy": 0.9984843015670777,
+      "num_tokens": 107423370.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.9345376133918762,
+      "epoch": 7.994616419919247,
+      "grad_norm": 0.771766722202301,
+      "learning_rate": 1.0612863445862236e-05,
+      "loss": 0.006203063577413559,
+      "mean_token_accuracy": 0.9982724368572236,
+      "num_tokens": 107604796.0,
+      "step": 5940
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.9459010544855884,
+      "eval_loss": 0.18965470790863037,
+      "eval_mean_token_accuracy": 0.9512642967473169,
+      "eval_num_tokens": 107677508.0,
+      "eval_runtime": 12.6472,
+      "eval_samples_per_second": 395.344,
+      "eval_steps_per_second": 12.414,
+      "step": 5944
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.09202200297839e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-5944/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-5944/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-5944/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/checkpoint-6687/README.md b/spanish_L20_a100/seed_42/checkpoint-6687/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-6687/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-6687/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-6687/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-6687/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-6687/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-6687/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..96de5adf5ce1d65f370574a8cdfc52ba91c2cc60
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-6687/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd986bbf494a9eec63df9063df4782ea6ebcf83a283d12d03c62bdf7ca273ab9
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-6687/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-6687/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-6687/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-6687/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-6687/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-6687/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-6687/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-6687/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-6687/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-6687/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-6687/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..b5f005dcfe4efb92f0c513b4b459df4e7745cb85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-6687/trainer_state.json
@@ -0,0 +1,6813 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.0,
+  "eval_steps": 500,
+  "global_step": 6687,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    },
+    {
+      "entropy": 1.049056887626648,
+      "epoch": 1.009421265141319,
+      "grad_norm": 1.4189648628234863,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.11506924629211426,
+      "mean_token_accuracy": 0.9560760498046875,
+      "num_tokens": 13586033.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0126458764076234,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.6394226551055908,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.10137178897857665,
+      "mean_token_accuracy": 0.9604915261268616,
+      "num_tokens": 13766883.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0174824237823485,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.3608797788619995,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.10521939992904664,
+      "mean_token_accuracy": 0.9592583775520325,
+      "num_tokens": 13949410.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0318764686584472,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.1440019607543945,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.1039464831352234,
+      "mean_token_accuracy": 0.9593731820583343,
+      "num_tokens": 14131283.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0228763103485108,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.609320878982544,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.10670688152313232,
+      "mean_token_accuracy": 0.9587047874927521,
+      "num_tokens": 14312694.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0123905301094056,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.4128942489624023,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.10546157360076905,
+      "mean_token_accuracy": 0.9586312890052795,
+      "num_tokens": 14493193.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.010130923986435,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.4825445413589478,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.09921822547912598,
+      "mean_token_accuracy": 0.9621170878410339,
+      "num_tokens": 14674762.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0216347932815553,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.2503618001937866,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.10619921684265136,
+      "mean_token_accuracy": 0.9584988236427308,
+      "num_tokens": 14856424.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.022511374950409,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.3209893703460693,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.10425064563751221,
+      "mean_token_accuracy": 0.9595829844474792,
+      "num_tokens": 15037452.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.007770985364914,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.4830716848373413,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.10189464092254638,
+      "mean_token_accuracy": 0.9609117627143859,
+      "num_tokens": 15218205.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0228153944015503,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.3191078901290894,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.10298972129821778,
+      "mean_token_accuracy": 0.9599695146083832,
+      "num_tokens": 15399322.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0262911200523377,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.2050191164016724,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.10532126426696778,
+      "mean_token_accuracy": 0.9590177297592163,
+      "num_tokens": 15581668.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0124290227890014,
+      "epoch": 1.17092866756393,
+      "grad_norm": 1.3545384407043457,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.10171200037002563,
+      "mean_token_accuracy": 0.9587810635566711,
+      "num_tokens": 15762970.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.028192389011383,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.55021071434021,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.10711442232131958,
+      "mean_token_accuracy": 0.956982696056366,
+      "num_tokens": 15944164.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0354714155197144,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2573878765106201,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.10722451210021973,
+      "mean_token_accuracy": 0.9568670630455017,
+      "num_tokens": 16125049.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.034978187084198,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.4448883533477783,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.10492000579833985,
+      "mean_token_accuracy": 0.9593439042568207,
+      "num_tokens": 16306183.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0321480870246886,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.4060254096984863,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.1017595648765564,
+      "mean_token_accuracy": 0.9604197800159454,
+      "num_tokens": 16487439.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.03044091463089,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.6413203477859497,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.1032058596611023,
+      "mean_token_accuracy": 0.9589122951030731,
+      "num_tokens": 16667826.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0394478559494018,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3867346048355103,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.10721726417541504,
+      "mean_token_accuracy": 0.9581708788871766,
+      "num_tokens": 16848488.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0355630755424499,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 1.3304182291030884,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.10916879177093505,
+      "mean_token_accuracy": 0.9566857099533081,
+      "num_tokens": 17028787.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0394182324409484,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.4853013753890991,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.10515115261077881,
+      "mean_token_accuracy": 0.9584204435348511,
+      "num_tokens": 17209699.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0405551195144653,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.1959068775177002,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.10281100273132324,
+      "mean_token_accuracy": 0.9599630236625671,
+      "num_tokens": 17391070.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0196930646896363,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.2734276056289673,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.10447280406951905,
+      "mean_token_accuracy": 0.9587895989418029,
+      "num_tokens": 17572309.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0284824252128602,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.1344295740127563,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.11111283302307129,
+      "mean_token_accuracy": 0.9560421764850616,
+      "num_tokens": 17753814.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0308443427085876,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.0935126543045044,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.10474474430084228,
+      "mean_token_accuracy": 0.9583515226840973,
+      "num_tokens": 17934774.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.0283282995224,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.5144932270050049,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.104243803024292,
+      "mean_token_accuracy": 0.9581555545330047,
+      "num_tokens": 18116015.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.0355387568473815,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.276055932044983,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.10444833040237426,
+      "mean_token_accuracy": 0.9587780058383941,
+      "num_tokens": 18296690.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.0340025186538697,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.4408668279647827,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.10817761421203613,
+      "mean_token_accuracy": 0.9573222637176514,
+      "num_tokens": 18477491.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.0480108737945557,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2114640474319458,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.10477652549743652,
+      "mean_token_accuracy": 0.9591076672077179,
+      "num_tokens": 18658366.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.0303086400032044,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.3319413661956787,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.10648446083068848,
+      "mean_token_accuracy": 0.9581128239631653,
+      "num_tokens": 18839989.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0405295491218567,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.4326776266098022,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.10707225799560546,
+      "mean_token_accuracy": 0.9574429094791412,
+      "num_tokens": 19020331.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.037307357788086,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.2203212976455688,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.10811934471130372,
+      "mean_token_accuracy": 0.9569429934024811,
+      "num_tokens": 19201621.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.048385226726532,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.2597253322601318,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.10502862930297852,
+      "mean_token_accuracy": 0.9592942714691162,
+      "num_tokens": 19382819.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.0431596279144286,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.288861632347107,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.1091151475906372,
+      "mean_token_accuracy": 0.9570925652980804,
+      "num_tokens": 19563291.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.0325645685195923,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.1845893859863281,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.10389597415924072,
+      "mean_token_accuracy": 0.9585334539413453,
+      "num_tokens": 19744388.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.025480568408966,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 1.3404650688171387,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.10605425834655761,
+      "mean_token_accuracy": 0.9583600521087646,
+      "num_tokens": 19925613.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.020683228969574,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5598406791687012,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.10631901025772095,
+      "mean_token_accuracy": 0.9573191046714783,
+      "num_tokens": 20106747.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.0185366213321685,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3814117908477783,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.1074562907218933,
+      "mean_token_accuracy": 0.9571435689926148,
+      "num_tokens": 20287872.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0160030841827392,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 1.5130647420883179,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.10584371089935303,
+      "mean_token_accuracy": 0.9584454715251922,
+      "num_tokens": 20469530.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.0295212030410767,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.137473464012146,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.10609703063964844,
+      "mean_token_accuracy": 0.9590815663337707,
+      "num_tokens": 20650652.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.0236494064331054,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.4402672052383423,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.10043272972106934,
+      "mean_token_accuracy": 0.9601598799228668,
+      "num_tokens": 20832844.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0228356838226318,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.6836261749267578,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.10231537818908691,
+      "mean_token_accuracy": 0.9592978358268738,
+      "num_tokens": 21014766.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.0204253315925598,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.7377065420150757,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.10434420108795166,
+      "mean_token_accuracy": 0.9578610718250274,
+      "num_tokens": 21195755.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.0251757979393006,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.337659239768982,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.10218650102615356,
+      "mean_token_accuracy": 0.9588291764259338,
+      "num_tokens": 21376476.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.021351706981659,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.278275728225708,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.10114043951034546,
+      "mean_token_accuracy": 0.9596256852149964,
+      "num_tokens": 21558292.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.0222898125648499,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.1806602478027344,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.10140385627746581,
+      "mean_token_accuracy": 0.9601265788078308,
+      "num_tokens": 21739640.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0402090787887572,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.2378828525543213,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.10622456073760986,
+      "mean_token_accuracy": 0.9573592960834503,
+      "num_tokens": 21921425.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.0412330985069276,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3350801467895508,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.10371602773666382,
+      "mean_token_accuracy": 0.9588219523429871,
+      "num_tokens": 22102605.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.0534109234809876,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 1.4873039722442627,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.10807039737701415,
+      "mean_token_accuracy": 0.9577955365180969,
+      "num_tokens": 22283687.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0713044166564942,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.1694425344467163,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.11005204916000366,
+      "mean_token_accuracy": 0.9568802535533905,
+      "num_tokens": 22464813.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.0745116233825684,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.5362104177474976,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.10913199186325073,
+      "mean_token_accuracy": 0.955611252784729,
+      "num_tokens": 22646088.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0925905466079713,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.3478912115097046,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.10608880519866944,
+      "mean_token_accuracy": 0.9566390633583068,
+      "num_tokens": 22826677.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0899853229522705,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.238072395324707,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.10292150974273681,
+      "mean_token_accuracy": 0.9586134016513824,
+      "num_tokens": 23008177.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0695149540901183,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.541548252105713,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.10405257940292359,
+      "mean_token_accuracy": 0.9580470204353333,
+      "num_tokens": 23190301.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.0614005208015442,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.4365121126174927,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.10439364910125733,
+      "mean_token_accuracy": 0.9584310293197632,
+      "num_tokens": 23371824.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.065568709373474,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.2848029136657715,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.104488205909729,
+      "mean_token_accuracy": 0.959163922071457,
+      "num_tokens": 23552849.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.0750214815139771,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 1.280882716178894,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.10620679855346679,
+      "mean_token_accuracy": 0.9575595021247864,
+      "num_tokens": 23733408.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.077055287361145,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.2904170751571655,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.11054887771606445,
+      "mean_token_accuracy": 0.9566682338714599,
+      "num_tokens": 23914203.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.081210732460022,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.129718542098999,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.10313529968261718,
+      "mean_token_accuracy": 0.9582066476345062,
+      "num_tokens": 24094078.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.0856151461601258,
+      "epoch": 1.80349932705249,
+      "grad_norm": 1.3682607412338257,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.10383679866790771,
+      "mean_token_accuracy": 0.9592101991176605,
+      "num_tokens": 24274071.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.0931835532188416,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.3282326459884644,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.10675904750823975,
+      "mean_token_accuracy": 0.9580830633640289,
+      "num_tokens": 24454766.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.0867947936058044,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.2584236860275269,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.10147552490234375,
+      "mean_token_accuracy": 0.9604024410247802,
+      "num_tokens": 24635903.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.0726998805999757,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 1.254036545753479,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.10524414777755738,
+      "mean_token_accuracy": 0.9582872509956359,
+      "num_tokens": 24816787.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.0692258477210999,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.41594660282135,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.10340676307678223,
+      "mean_token_accuracy": 0.9590053558349609,
+      "num_tokens": 24998543.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.0772343635559083,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.385327696800232,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.10295225381851196,
+      "mean_token_accuracy": 0.9591428697109222,
+      "num_tokens": 25180059.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.0874038934707642,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 1.0628604888916016,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.10015511512756348,
+      "mean_token_accuracy": 0.9605867803096771,
+      "num_tokens": 25361290.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0702470779418944,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.1819324493408203,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.104461407661438,
+      "mean_token_accuracy": 0.9581527650356293,
+      "num_tokens": 25542455.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.0592689394950867,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4695993661880493,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.1024864912033081,
+      "mean_token_accuracy": 0.9591827690601349,
+      "num_tokens": 25723310.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.0781041979789734,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 1.3313041925430298,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.10552446842193604,
+      "mean_token_accuracy": 0.9576125502586365,
+      "num_tokens": 25904923.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0914432287216187,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.3578517436981201,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.11063914299011231,
+      "mean_token_accuracy": 0.9563858449459076,
+      "num_tokens": 26086154.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0916360139846801,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.3493428230285645,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.10228430032730103,
+      "mean_token_accuracy": 0.9597218215465546,
+      "num_tokens": 26268096.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0865699172019958,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.3002992868423462,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.10704560279846191,
+      "mean_token_accuracy": 0.9567803084850312,
+      "num_tokens": 26448598.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.0907732367515564,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.3105698823928833,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.10344587564468384,
+      "mean_token_accuracy": 0.9585539996623993,
+      "num_tokens": 26629572.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.097494399547577,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.5486772060394287,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.10484597682952881,
+      "mean_token_accuracy": 0.9578526735305786,
+      "num_tokens": 26810588.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 1.1259616188182953,
+      "eval_loss": 0.11920737475156784,
+      "eval_mean_token_accuracy": 0.9516198661676638,
+      "eval_num_tokens": 26919371.0,
+      "eval_runtime": 12.6599,
+      "eval_samples_per_second": 394.949,
+      "eval_steps_per_second": 12.401,
+      "step": 1486
+    },
+    {
+      "entropy": 1.1117471694946288,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.3559718132019043,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.09661188125610351,
+      "mean_token_accuracy": 0.9621427834033967,
+      "num_tokens": 26991831.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.0618590712547302,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.5231083631515503,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.07198524475097656,
+      "mean_token_accuracy": 0.9724888324737548,
+      "num_tokens": 27172578.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.0601553201675415,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0978407859802246,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.07232148051261902,
+      "mean_token_accuracy": 0.9721459746360779,
+      "num_tokens": 27353505.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.0545680522918701,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.375500202178955,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0737423300743103,
+      "mean_token_accuracy": 0.9723270058631897,
+      "num_tokens": 27534758.0,
+      "step": 1520
+    },
+    {
+      "entropy": 1.0429188370704652,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.5134543180465698,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.07129207253456116,
+      "mean_token_accuracy": 0.9730912566184997,
+      "num_tokens": 27715692.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0608030915260316,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.424338936805725,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.07388348579406738,
+      "mean_token_accuracy": 0.9727432250976562,
+      "num_tokens": 27896664.0,
+      "step": 1540
+    },
+    {
+      "entropy": 1.0577767491340637,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2193686962127686,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.07269638180732726,
+      "mean_token_accuracy": 0.9726069569587708,
+      "num_tokens": 28077786.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0627391576766967,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.4851799011230469,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.07443426847457886,
+      "mean_token_accuracy": 0.9716702818870544,
+      "num_tokens": 28259406.0,
+      "step": 1560
+    },
+    {
+      "entropy": 1.0606005311012268,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.136905312538147,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.07183161973953248,
+      "mean_token_accuracy": 0.9726755082607269,
+      "num_tokens": 28441336.0,
+      "step": 1570
+    },
+    {
+      "entropy": 1.0512076377868653,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.4116008281707764,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.07272077798843384,
+      "mean_token_accuracy": 0.972117805480957,
+      "num_tokens": 28621984.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0617831587791442,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.4194544553756714,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.07670866250991822,
+      "mean_token_accuracy": 0.9709574401378631,
+      "num_tokens": 28803335.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.054829752445221,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 1.0619711875915527,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.06954843401908875,
+      "mean_token_accuracy": 0.9732311069965363,
+      "num_tokens": 28984568.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.062503707408905,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1824331283569336,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.0742799162864685,
+      "mean_token_accuracy": 0.9718966007232666,
+      "num_tokens": 29165958.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0652349829673766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.2683000564575195,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.07545414566993713,
+      "mean_token_accuracy": 0.9704910278320312,
+      "num_tokens": 29347870.0,
+      "step": 1620
+    },
+    {
+      "entropy": 1.0519612908363343,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.257572889328003,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.075272136926651,
+      "mean_token_accuracy": 0.9712580978870392,
+      "num_tokens": 29529453.0,
+      "step": 1630
+    },
+    {
+      "entropy": 1.0668588399887085,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.3439997434616089,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.068633770942688,
+      "mean_token_accuracy": 0.9735071420669555,
+      "num_tokens": 29710695.0,
+      "step": 1640
+    },
+    {
+      "entropy": 1.0701637029647828,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.4352364540100098,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.07366064786911011,
+      "mean_token_accuracy": 0.9724323868751525,
+      "num_tokens": 29890993.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.065868353843689,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.3669353723526,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.0714600920677185,
+      "mean_token_accuracy": 0.9717992007732391,
+      "num_tokens": 30072284.0,
+      "step": 1660
+    },
+    {
+      "entropy": 1.077157938480377,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 1.3630682229995728,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.07237017154693604,
+      "mean_token_accuracy": 0.9729755759239197,
+      "num_tokens": 30253123.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0886191844940185,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.3787649869918823,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.07589051723480225,
+      "mean_token_accuracy": 0.9716001987457276,
+      "num_tokens": 30434100.0,
+      "step": 1680
+    },
+    {
+      "entropy": 1.080547845363617,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.4936959743499756,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.07393252849578857,
+      "mean_token_accuracy": 0.9709986329078675,
+      "num_tokens": 30615086.0,
+      "step": 1690
+    },
+    {
+      "entropy": 1.0851039648056031,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.4654748439788818,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.07487455606460572,
+      "mean_token_accuracy": 0.9710996568202972,
+      "num_tokens": 30795174.0,
+      "step": 1700
+    },
+    {
+      "entropy": 1.0743987679481506,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 1.354244589805603,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.07701593041419982,
+      "mean_token_accuracy": 0.9701604127883912,
+      "num_tokens": 30976627.0,
+      "step": 1710
+    },
+    {
+      "entropy": 1.0738040447235107,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.2830452919006348,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.07861257195472718,
+      "mean_token_accuracy": 0.9696074068546295,
+      "num_tokens": 31158542.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.0667856335639954,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.3329832553863525,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.07371129989624023,
+      "mean_token_accuracy": 0.9722550988197327,
+      "num_tokens": 31339091.0,
+      "step": 1730
+    },
+    {
+      "entropy": 1.0734196662902833,
+      "epoch": 2.34185733512786,
+      "grad_norm": 1.4483965635299683,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.07841320037841797,
+      "mean_token_accuracy": 0.9698444902896881,
+      "num_tokens": 31519961.0,
+      "step": 1740
+    },
+    {
+      "entropy": 1.0724906921386719,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.440190315246582,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.07452400922775268,
+      "mean_token_accuracy": 0.9715199947357178,
+      "num_tokens": 31701305.0,
+      "step": 1750
+    },
+    {
+      "entropy": 1.0770147442817688,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 1.17177414894104,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.07649941444396972,
+      "mean_token_accuracy": 0.9704680681228638,
+      "num_tokens": 31882106.0,
+      "step": 1760
+    },
+    {
+      "entropy": 1.0792171478271484,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.498350977897644,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.07850931882858277,
+      "mean_token_accuracy": 0.9699269771575928,
+      "num_tokens": 32062712.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.092840301990509,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.4153627157211304,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.07811332941055298,
+      "mean_token_accuracy": 0.9698202073574066,
+      "num_tokens": 32244156.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0767372131347657,
+      "epoch": 2.409152086137281,
+      "grad_norm": 1.3672188520431519,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.0765878677368164,
+      "mean_token_accuracy": 0.970400470495224,
+      "num_tokens": 32424975.0,
+      "step": 1790
+    },
+    {
+      "entropy": 1.0641321778297423,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.1141754388809204,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0744673728942871,
+      "mean_token_accuracy": 0.9713521003723145,
+      "num_tokens": 32606399.0,
+      "step": 1800
+    },
+    {
+      "entropy": 1.0847392082214355,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.372215986251831,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0788280725479126,
+      "mean_token_accuracy": 0.9688920557498932,
+      "num_tokens": 32787632.0,
+      "step": 1810
+    },
+    {
+      "entropy": 1.0952076196670533,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.3998148441314697,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.07322710752487183,
+      "mean_token_accuracy": 0.9713049352169036,
+      "num_tokens": 32969136.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.1047595977783202,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.538533091545105,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.07695393562316895,
+      "mean_token_accuracy": 0.9697132647037506,
+      "num_tokens": 33149585.0,
+      "step": 1830
+    },
+    {
+      "entropy": 1.1071401476860045,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.2797189950942993,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.07820409536361694,
+      "mean_token_accuracy": 0.9695189118385314,
+      "num_tokens": 33330604.0,
+      "step": 1840
+    },
+    {
+      "entropy": 1.0979747891426086,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 1.309956669807434,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.08005034327507018,
+      "mean_token_accuracy": 0.9683119177818298,
+      "num_tokens": 33510932.0,
+      "step": 1850
+    },
+    {
+      "entropy": 1.1024476408958435,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.2206803560256958,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.07752519249916076,
+      "mean_token_accuracy": 0.9697020888328552,
+      "num_tokens": 33692289.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0788621783256531,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.4230300188064575,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.08086194992065429,
+      "mean_token_accuracy": 0.9689899325370789,
+      "num_tokens": 33873643.0,
+      "step": 1870
+    },
+    {
+      "entropy": 1.0909191489219665,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 1.3553478717803955,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.07968194484710693,
+      "mean_token_accuracy": 0.9695583760738373,
+      "num_tokens": 34054342.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.0874631762504579,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 1.209466576576233,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.07668136358261109,
+      "mean_token_accuracy": 0.9706926643848419,
+      "num_tokens": 34235054.0,
+      "step": 1890
+    },
+    {
+      "entropy": 1.0850980758666993,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.2419084310531616,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.07621376514434815,
+      "mean_token_accuracy": 0.9708946466445922,
+      "num_tokens": 34415792.0,
+      "step": 1900
+    },
+    {
+      "entropy": 1.1024398922920227,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.5481151342391968,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.07621458768844605,
+      "mean_token_accuracy": 0.9696649491786957,
+      "num_tokens": 34596130.0,
+      "step": 1910
+    },
+    {
+      "entropy": 1.0917130827903747,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.4186631441116333,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.07716711163520813,
+      "mean_token_accuracy": 0.9699705123901368,
+      "num_tokens": 34776608.0,
+      "step": 1920
+    },
+    {
+      "entropy": 1.0845637321472168,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2549679279327393,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.07353252172470093,
+      "mean_token_accuracy": 0.9715336680412292,
+      "num_tokens": 34957854.0,
+      "step": 1930
+    },
+    {
+      "entropy": 1.070354163646698,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.390034794807434,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.07804152965545655,
+      "mean_token_accuracy": 0.9694301605224609,
+      "num_tokens": 35139425.0,
+      "step": 1940
+    },
+    {
+      "entropy": 1.0926494956016541,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.2890169620513916,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.07905603051185608,
+      "mean_token_accuracy": 0.9694173872470856,
+      "num_tokens": 35321009.0,
+      "step": 1950
+    },
+    {
+      "entropy": 1.090261971950531,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.201541543006897,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.07853106260299683,
+      "mean_token_accuracy": 0.9687644064426422,
+      "num_tokens": 35502338.0,
+      "step": 1960
+    },
+    {
+      "entropy": 1.110123836994171,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.3321393728256226,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.07721331715583801,
+      "mean_token_accuracy": 0.9703636646270752,
+      "num_tokens": 35684550.0,
+      "step": 1970
+    },
+    {
+      "entropy": 1.1154867172241212,
+      "epoch": 2.664872139973082,
+      "grad_norm": 1.5639152526855469,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.07769335508346557,
+      "mean_token_accuracy": 0.9694860637187958,
+      "num_tokens": 35866258.0,
+      "step": 1980
+    },
+    {
+      "entropy": 1.1126874089241028,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5160865783691406,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.07732025384902955,
+      "mean_token_accuracy": 0.9695809543132782,
+      "num_tokens": 36047771.0,
+      "step": 1990
+    },
+    {
+      "entropy": 1.1188251495361328,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.3689777851104736,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.07844393253326416,
+      "mean_token_accuracy": 0.9697122633457184,
+      "num_tokens": 36228702.0,
+      "step": 2000
+    },
+    {
+      "entropy": 1.1134975671768188,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 1.1956474781036377,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.07932614088058472,
+      "mean_token_accuracy": 0.9695560097694397,
+      "num_tokens": 36409577.0,
+      "step": 2010
+    },
+    {
+      "entropy": 1.1052889227867126,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.2844921350479126,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.07471654415130616,
+      "mean_token_accuracy": 0.9707616209983826,
+      "num_tokens": 36590635.0,
+      "step": 2020
+    },
+    {
+      "entropy": 1.1029305696487426,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3110395669937134,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.07654619216918945,
+      "mean_token_accuracy": 0.9699673116207123,
+      "num_tokens": 36772032.0,
+      "step": 2030
+    },
+    {
+      "entropy": 1.0967667698860168,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.2297992706298828,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.07549123167991638,
+      "mean_token_accuracy": 0.9705443382263184,
+      "num_tokens": 36953014.0,
+      "step": 2040
+    },
+    {
+      "entropy": 1.0954343318939208,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.5522037744522095,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.07590663433074951,
+      "mean_token_accuracy": 0.9709664165973664,
+      "num_tokens": 37134549.0,
+      "step": 2050
+    },
+    {
+      "entropy": 1.0984182119369508,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.3691717386245728,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.07795964479446411,
+      "mean_token_accuracy": 0.969748318195343,
+      "num_tokens": 37315651.0,
+      "step": 2060
+    },
+    {
+      "entropy": 1.0937008380889892,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 1.345433235168457,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.07556321024894715,
+      "mean_token_accuracy": 0.9707719802856445,
+      "num_tokens": 37497150.0,
+      "step": 2070
+    },
+    {
+      "entropy": 1.0978638410568238,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.3442333936691284,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.07603150010108947,
+      "mean_token_accuracy": 0.9707530319690705,
+      "num_tokens": 37678637.0,
+      "step": 2080
+    },
+    {
+      "entropy": 1.112320113182068,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.672094702720642,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.07702311277389526,
+      "mean_token_accuracy": 0.969988214969635,
+      "num_tokens": 37859530.0,
+      "step": 2090
+    },
+    {
+      "entropy": 1.1019986748695374,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.4613395929336548,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.07702183723449707,
+      "mean_token_accuracy": 0.9705046355724335,
+      "num_tokens": 38040160.0,
+      "step": 2100
+    },
+    {
+      "entropy": 1.1074181199073792,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.33228600025177,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.07809240221977234,
+      "mean_token_accuracy": 0.9702594041824341,
+      "num_tokens": 38221261.0,
+      "step": 2110
+    },
+    {
+      "entropy": 1.1087767481803894,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.3967821598052979,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.07641294002532958,
+      "mean_token_accuracy": 0.9694520056247711,
+      "num_tokens": 38402589.0,
+      "step": 2120
+    },
+    {
+      "entropy": 1.0902286529541017,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 1.3104654550552368,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.07864789366722107,
+      "mean_token_accuracy": 0.9693171322345734,
+      "num_tokens": 38583485.0,
+      "step": 2130
+    },
+    {
+      "entropy": 1.088974940776825,
+      "epoch": 2.88021534320323,
+      "grad_norm": 1.389603614807129,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.07767997980117798,
+      "mean_token_accuracy": 0.9691886186599732,
+      "num_tokens": 38764480.0,
+      "step": 2140
+    },
+    {
+      "entropy": 1.093208122253418,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.3133198022842407,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.08097923994064331,
+      "mean_token_accuracy": 0.968046635389328,
+      "num_tokens": 38946016.0,
+      "step": 2150
+    },
+    {
+      "entropy": 1.0920218586921693,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.3472810983657837,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.07504125237464905,
+      "mean_token_accuracy": 0.9705065429210663,
+      "num_tokens": 39127480.0,
+      "step": 2160
+    },
+    {
+      "entropy": 1.085507369041443,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.9070137739181519,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.07882450819015503,
+      "mean_token_accuracy": 0.9694548428058625,
+      "num_tokens": 39308552.0,
+      "step": 2170
+    },
+    {
+      "entropy": 1.1070378303527832,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.3059617280960083,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.08202729225158692,
+      "mean_token_accuracy": 0.9679245829582215,
+      "num_tokens": 39489844.0,
+      "step": 2180
+    },
+    {
+      "entropy": 1.0994928359985352,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.359965443611145,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.07510381340980529,
+      "mean_token_accuracy": 0.969987016916275,
+      "num_tokens": 39671941.0,
+      "step": 2190
+    },
+    {
+      "entropy": 1.0987842202186584,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.385233998298645,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.07863752245903015,
+      "mean_token_accuracy": 0.9685526669025422,
+      "num_tokens": 39852940.0,
+      "step": 2200
+    },
+    {
+      "entropy": 1.110793125629425,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.4456037282943726,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.07783852815628052,
+      "mean_token_accuracy": 0.9702686488628387,
+      "num_tokens": 40034313.0,
+      "step": 2210
+    },
+    {
+      "entropy": 1.1140072107315064,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.1889644861221313,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.07787686586380005,
+      "mean_token_accuracy": 0.9696961343288422,
+      "num_tokens": 40215866.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 1.11932077984901,
+      "eval_loss": 0.12462407350540161,
+      "eval_mean_token_accuracy": 0.951254228118119,
+      "eval_num_tokens": 40378902.0,
+      "eval_runtime": 13.5665,
+      "eval_samples_per_second": 368.556,
+      "eval_steps_per_second": 11.573,
+      "step": 2229
+    },
+    {
+      "entropy": 1.1124149322509767,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 1.1116644144058228,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.07439510822296143,
+      "mean_token_accuracy": 0.9713229060173034,
+      "num_tokens": 40396979.0,
+      "step": 2230
+    },
+    {
+      "entropy": 1.0716044545173644,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.1957958936691284,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.05146250128746033,
+      "mean_token_accuracy": 0.9813062250614166,
+      "num_tokens": 40578935.0,
+      "step": 2240
+    },
+    {
+      "entropy": 1.0567036390304565,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.337942361831665,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.052991294860839845,
+      "mean_token_accuracy": 0.9807850599288941,
+      "num_tokens": 40759500.0,
+      "step": 2250
+    },
+    {
+      "entropy": 1.056044340133667,
+      "epoch": 3.041722745625841,
+      "grad_norm": 1.4191840887069702,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.05163681507110596,
+      "mean_token_accuracy": 0.9813888549804688,
+      "num_tokens": 40940279.0,
+      "step": 2260
+    },
+    {
+      "entropy": 1.056977343559265,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.4481396675109863,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.05059016942977905,
+      "mean_token_accuracy": 0.9814287006855011,
+      "num_tokens": 41121782.0,
+      "step": 2270
+    },
+    {
+      "entropy": 1.0579043984413148,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.0136349201202393,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.05109646320343018,
+      "mean_token_accuracy": 0.9805608510971069,
+      "num_tokens": 41302614.0,
+      "step": 2280
+    },
+    {
+      "entropy": 1.0564674019813538,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.165916085243225,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.05656059980392456,
+      "mean_token_accuracy": 0.9801787078380585,
+      "num_tokens": 41484240.0,
+      "step": 2290
+    },
+    {
+      "entropy": 1.0617436170578003,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.500622272491455,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.05233668088912964,
+      "mean_token_accuracy": 0.9808355152606965,
+      "num_tokens": 41665160.0,
+      "step": 2300
+    },
+    {
+      "entropy": 1.0509375214576722,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.403001308441162,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.05074570775032043,
+      "mean_token_accuracy": 0.9807266294956207,
+      "num_tokens": 41846062.0,
+      "step": 2310
+    },
+    {
+      "entropy": 1.0536353230476379,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1046926975250244,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.04906890392303467,
+      "mean_token_accuracy": 0.9815255463123321,
+      "num_tokens": 42027115.0,
+      "step": 2320
+    },
+    {
+      "entropy": 1.0518895149230958,
+      "epoch": 3.135935397039031,
+      "grad_norm": 1.0369377136230469,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.04991455376148224,
+      "mean_token_accuracy": 0.9822782278060913,
+      "num_tokens": 42208594.0,
+      "step": 2330
+    },
+    {
+      "entropy": 1.0597471237182616,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.6156455278396606,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.049592965841293336,
+      "mean_token_accuracy": 0.9818629324436188,
+      "num_tokens": 42389065.0,
+      "step": 2340
+    },
+    {
+      "entropy": 1.0561720967292785,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.2941797971725464,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.05092741847038269,
+      "mean_token_accuracy": 0.9810208320617676,
+      "num_tokens": 42570319.0,
+      "step": 2350
+    },
+    {
+      "entropy": 1.0521231889724731,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.1866732835769653,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.051917088031768796,
+      "mean_token_accuracy": 0.981055635213852,
+      "num_tokens": 42751501.0,
+      "step": 2360
+    },
+    {
+      "entropy": 1.0555105328559875,
+      "epoch": 3.189771197846568,
+      "grad_norm": 1.2260069847106934,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.05160566568374634,
+      "mean_token_accuracy": 0.981193995475769,
+      "num_tokens": 42932643.0,
+      "step": 2370
+    },
+    {
+      "entropy": 1.0720075011253356,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.3786890506744385,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.053551071882247926,
+      "mean_token_accuracy": 0.9797191619873047,
+      "num_tokens": 43113204.0,
+      "step": 2380
+    },
+    {
+      "entropy": 1.0718719363212585,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.3200089931488037,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.054034185409545896,
+      "mean_token_accuracy": 0.9800151228904724,
+      "num_tokens": 43293433.0,
+      "step": 2390
+    },
+    {
+      "entropy": 1.0536052703857421,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.4560705423355103,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.05042939186096192,
+      "mean_token_accuracy": 0.9806897640228271,
+      "num_tokens": 43474189.0,
+      "step": 2400
+    },
+    {
+      "entropy": 1.0602891445159912,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.3123281002044678,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.05151479244232178,
+      "mean_token_accuracy": 0.980517429113388,
+      "num_tokens": 43655355.0,
+      "step": 2410
+    },
+    {
+      "entropy": 1.067916214466095,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.2652188539505005,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.05306308269500733,
+      "mean_token_accuracy": 0.9805097997188568,
+      "num_tokens": 43836752.0,
+      "step": 2420
+    },
+    {
+      "entropy": 1.067535150051117,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.6280162334442139,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.052641332149505615,
+      "mean_token_accuracy": 0.9803492069244385,
+      "num_tokens": 44018525.0,
+      "step": 2430
+    },
+    {
+      "entropy": 1.070455825328827,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 1.632573127746582,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.05452280044555664,
+      "mean_token_accuracy": 0.9793554663658142,
+      "num_tokens": 44199901.0,
+      "step": 2440
+    },
+    {
+      "entropy": 1.0741212248802186,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.3525495529174805,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.054098492860794066,
+      "mean_token_accuracy": 0.9804170608520508,
+      "num_tokens": 44381734.0,
+      "step": 2450
+    },
+    {
+      "entropy": 1.0836357831954957,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.3989282846450806,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.05158165693283081,
+      "mean_token_accuracy": 0.9810640931129455,
+      "num_tokens": 44563158.0,
+      "step": 2460
+    },
+    {
+      "entropy": 1.077314054965973,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1970937252044678,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.052333271503448485,
+      "mean_token_accuracy": 0.9808873236179352,
+      "num_tokens": 44744642.0,
+      "step": 2470
+    },
+    {
+      "entropy": 1.069891107082367,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.3197492361068726,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.05269680619239807,
+      "mean_token_accuracy": 0.980803519487381,
+      "num_tokens": 44926301.0,
+      "step": 2480
+    },
+    {
+      "entropy": 1.071691370010376,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.332481861114502,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.054685378074646,
+      "mean_token_accuracy": 0.9795110940933227,
+      "num_tokens": 45106935.0,
+      "step": 2490
+    },
+    {
+      "entropy": 1.0654058694839477,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 1.5478020906448364,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.05287986397743225,
+      "mean_token_accuracy": 0.9797925829887391,
+      "num_tokens": 45288325.0,
+      "step": 2500
+    },
+    {
+      "entropy": 1.0660210013389588,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.43852961063385,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.05411216616630554,
+      "mean_token_accuracy": 0.9790283799171448,
+      "num_tokens": 45469485.0,
+      "step": 2510
+    },
+    {
+      "entropy": 1.0649927496910094,
+      "epoch": 3.391655450874832,
+      "grad_norm": 1.3900808095932007,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.05290043354034424,
+      "mean_token_accuracy": 0.9808539688587189,
+      "num_tokens": 45650705.0,
+      "step": 2520
+    },
+    {
+      "entropy": 1.0768967866897583,
+      "epoch": 3.405114401076716,
+      "grad_norm": 1.3428847789764404,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.05374534130096435,
+      "mean_token_accuracy": 0.9802824795246124,
+      "num_tokens": 45832104.0,
+      "step": 2530
+    },
+    {
+      "entropy": 1.0660459756851197,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.3150343894958496,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.05366337299346924,
+      "mean_token_accuracy": 0.9796974420547485,
+      "num_tokens": 46013552.0,
+      "step": 2540
+    },
+    {
+      "entropy": 1.06330087184906,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.546858787536621,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.054223823547363284,
+      "mean_token_accuracy": 0.9796406507492066,
+      "num_tokens": 46194606.0,
+      "step": 2550
+    },
+    {
+      "entropy": 1.0564390897750855,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.3932640552520752,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.05237367153167725,
+      "mean_token_accuracy": 0.9805298745632172,
+      "num_tokens": 46375520.0,
+      "step": 2560
+    },
+    {
+      "entropy": 1.0519753217697143,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.8447308540344238,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.05499213933944702,
+      "mean_token_accuracy": 0.9787130057811737,
+      "num_tokens": 46556203.0,
+      "step": 2570
+    },
+    {
+      "entropy": 1.0693527460098267,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.3508682250976562,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.055162906646728516,
+      "mean_token_accuracy": 0.9794130921363831,
+      "num_tokens": 46737287.0,
+      "step": 2580
+    },
+    {
+      "entropy": 1.078024983406067,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 1.1582329273223877,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.05406954288482666,
+      "mean_token_accuracy": 0.9794795215129852,
+      "num_tokens": 46918363.0,
+      "step": 2590
+    },
+    {
+      "entropy": 1.0643377661705018,
+      "epoch": 3.499327052489906,
+      "grad_norm": 1.429380178451538,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.05309759378433228,
+      "mean_token_accuracy": 0.9806221067905426,
+      "num_tokens": 47099549.0,
+      "step": 2600
+    },
+    {
+      "entropy": 1.068834125995636,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.236444354057312,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.05395760536193848,
+      "mean_token_accuracy": 0.9798199534416199,
+      "num_tokens": 47280430.0,
+      "step": 2610
+    },
+    {
+      "entropy": 1.0729689002037048,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 1.32350754737854,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.05501353740692139,
+      "mean_token_accuracy": 0.9792946577072144,
+      "num_tokens": 47461390.0,
+      "step": 2620
+    },
+    {
+      "entropy": 1.0766199231147766,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2348153591156006,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.05568041205406189,
+      "mean_token_accuracy": 0.9796942174434662,
+      "num_tokens": 47642662.0,
+      "step": 2630
+    },
+    {
+      "entropy": 1.0804976940155029,
+      "epoch": 3.553162853297443,
+      "grad_norm": 1.1175553798675537,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.050679147243499756,
+      "mean_token_accuracy": 0.9814749479293823,
+      "num_tokens": 47824309.0,
+      "step": 2640
+    },
+    {
+      "entropy": 1.0689257264137269,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.6297036409378052,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.054615461826324464,
+      "mean_token_accuracy": 0.9796478092670441,
+      "num_tokens": 48005982.0,
+      "step": 2650
+    },
+    {
+      "entropy": 1.0696970105171204,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2477251291275024,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.05287024974822998,
+      "mean_token_accuracy": 0.979641854763031,
+      "num_tokens": 48187048.0,
+      "step": 2660
+    },
+    {
+      "entropy": 1.0800267577171325,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.3109506368637085,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.054475915431976316,
+      "mean_token_accuracy": 0.9794476389884949,
+      "num_tokens": 48368447.0,
+      "step": 2670
+    },
+    {
+      "entropy": 1.0848782062530518,
+      "epoch": 3.60699865410498,
+      "grad_norm": 1.2147419452667236,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.05212314128875732,
+      "mean_token_accuracy": 0.9798501491546631,
+      "num_tokens": 48549144.0,
+      "step": 2680
+    },
+    {
+      "entropy": 1.0681315302848815,
+      "epoch": 3.620457604306864,
+      "grad_norm": 1.3249231576919556,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.053503894805908205,
+      "mean_token_accuracy": 0.9800095915794372,
+      "num_tokens": 48731059.0,
+      "step": 2690
+    },
+    {
+      "entropy": 1.0629405975341797,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.2435307502746582,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.0541129469871521,
+      "mean_token_accuracy": 0.9795375943183899,
+      "num_tokens": 48912283.0,
+      "step": 2700
+    },
+    {
+      "entropy": 1.0608446598052979,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.3816258907318115,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.054210084676742556,
+      "mean_token_accuracy": 0.9793854176998138,
+      "num_tokens": 49094236.0,
+      "step": 2710
+    },
+    {
+      "entropy": 1.0596249461174012,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.132426381111145,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.05496481657028198,
+      "mean_token_accuracy": 0.9798576831817627,
+      "num_tokens": 49275540.0,
+      "step": 2720
+    },
+    {
+      "entropy": 1.060381281375885,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.4120980501174927,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.05694271326065063,
+      "mean_token_accuracy": 0.9788619399070739,
+      "num_tokens": 49456684.0,
+      "step": 2730
+    },
+    {
+      "entropy": 1.0594112157821656,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.428120493888855,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.05274912118911743,
+      "mean_token_accuracy": 0.9802964627742767,
+      "num_tokens": 49638099.0,
+      "step": 2740
+    },
+    {
+      "entropy": 1.0630932807922364,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.4599305391311646,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.057349550724029544,
+      "mean_token_accuracy": 0.9788366615772247,
+      "num_tokens": 49819751.0,
+      "step": 2750
+    },
+    {
+      "entropy": 1.0632095456123352,
+      "epoch": 3.714670255720054,
+      "grad_norm": 1.8489277362823486,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.053570222854614255,
+      "mean_token_accuracy": 0.980079036951065,
+      "num_tokens": 50000623.0,
+      "step": 2760
+    },
+    {
+      "entropy": 1.0583208560943604,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.4855526685714722,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.05508332252502442,
+      "mean_token_accuracy": 0.9789698243141174,
+      "num_tokens": 50181540.0,
+      "step": 2770
+    },
+    {
+      "entropy": 1.0690102458000184,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 1.553468942642212,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.05684031248092651,
+      "mean_token_accuracy": 0.9776920318603516,
+      "num_tokens": 50363406.0,
+      "step": 2780
+    },
+    {
+      "entropy": 1.0651323795318604,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 1.2951313257217407,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.055989408493041994,
+      "mean_token_accuracy": 0.978317403793335,
+      "num_tokens": 50544599.0,
+      "step": 2790
+    },
+    {
+      "entropy": 1.0571947336196899,
+      "epoch": 3.768506056527591,
+      "grad_norm": 1.1358306407928467,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.052672064304351805,
+      "mean_token_accuracy": 0.9797160506248475,
+      "num_tokens": 50726664.0,
+      "step": 2800
+    },
+    {
+      "entropy": 1.0475606441497802,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.4842852354049683,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.05601704120635986,
+      "mean_token_accuracy": 0.9794813454151153,
+      "num_tokens": 50907903.0,
+      "step": 2810
+    },
+    {
+      "entropy": 1.053141224384308,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.3320889472961426,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.05435909032821655,
+      "mean_token_accuracy": 0.9792923092842102,
+      "num_tokens": 51088734.0,
+      "step": 2820
+    },
+    {
+      "entropy": 1.0518054842948914,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.140783667564392,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.05576471090316772,
+      "mean_token_accuracy": 0.9789255380630493,
+      "num_tokens": 51269924.0,
+      "step": 2830
+    },
+    {
+      "entropy": 1.0490313410758971,
+      "epoch": 3.822341857335128,
+      "grad_norm": 1.5749891996383667,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.05655611753463745,
+      "mean_token_accuracy": 0.9785366714000702,
+      "num_tokens": 51450406.0,
+      "step": 2840
+    },
+    {
+      "entropy": 1.0508960008621215,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 1.2535520792007446,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.05710798501968384,
+      "mean_token_accuracy": 0.978301590681076,
+      "num_tokens": 51630630.0,
+      "step": 2850
+    },
+    {
+      "entropy": 1.0434152960777283,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.6045477390289307,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.0545048713684082,
+      "mean_token_accuracy": 0.9785818517208099,
+      "num_tokens": 51811252.0,
+      "step": 2860
+    },
+    {
+      "entropy": 1.046581244468689,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.2737947702407837,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.056829071044921874,
+      "mean_token_accuracy": 0.9779990613460541,
+      "num_tokens": 51992578.0,
+      "step": 2870
+    },
+    {
+      "entropy": 1.0562566757202148,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.2364535331726074,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.054702204465866086,
+      "mean_token_accuracy": 0.9796520650386811,
+      "num_tokens": 52173507.0,
+      "step": 2880
+    },
+    {
+      "entropy": 1.0444288849830627,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.2663625478744507,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.05222170352935791,
+      "mean_token_accuracy": 0.9805905818939209,
+      "num_tokens": 52354669.0,
+      "step": 2890
+    },
+    {
+      "entropy": 1.0429784655570984,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.2072714567184448,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.05404345989227295,
+      "mean_token_accuracy": 0.9802968561649322,
+      "num_tokens": 52535334.0,
+      "step": 2900
+    },
+    {
+      "entropy": 1.045020854473114,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.3894519805908203,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.05601906776428223,
+      "mean_token_accuracy": 0.9779200375080108,
+      "num_tokens": 52715952.0,
+      "step": 2910
+    },
+    {
+      "entropy": 1.0439318537712097,
+      "epoch": 3.930013458950202,
+      "grad_norm": 1.4112615585327148,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.05345791578292847,
+      "mean_token_accuracy": 0.9796200633049011,
+      "num_tokens": 52897155.0,
+      "step": 2920
+    },
+    {
+      "entropy": 1.0361477971076964,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 1.3158206939697266,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.05547314286231995,
+      "mean_token_accuracy": 0.9788706004619598,
+      "num_tokens": 53078916.0,
+      "step": 2930
+    },
+    {
+      "entropy": 1.0427968740463256,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2565301656723022,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.05601854920387268,
+      "mean_token_accuracy": 0.9791704952716828,
+      "num_tokens": 53258884.0,
+      "step": 2940
+    },
+    {
+      "entropy": 1.0374909162521362,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.4639637470245361,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.054512780904769895,
+      "mean_token_accuracy": 0.9793699860572815,
+      "num_tokens": 53440481.0,
+      "step": 2950
+    },
+    {
+      "entropy": 1.0511455059051513,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.242073655128479,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.055217862129211426,
+      "mean_token_accuracy": 0.9801203072071075,
+      "num_tokens": 53621142.0,
+      "step": 2960
+    },
+    {
+      "entropy": 1.0518356442451477,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.1518405675888062,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.05533761382102966,
+      "mean_token_accuracy": 0.9789008319377899,
+      "num_tokens": 53802588.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 1.0548353901334628,
+      "eval_loss": 0.13498826324939728,
+      "eval_mean_token_accuracy": 0.9507886888874564,
+      "eval_num_tokens": 53838633.0,
+      "eval_runtime": 12.6809,
+      "eval_samples_per_second": 394.295,
+      "eval_steps_per_second": 12.381,
+      "step": 2972
+    },
+    {
+      "entropy": 1.043621039390564,
+      "epoch": 4.010767160161508,
+      "grad_norm": 1.1016159057617188,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.03793205618858338,
+      "mean_token_accuracy": 0.9870165169239045,
+      "num_tokens": 53983430.0,
+      "step": 2980
+    },
+    {
+      "entropy": 1.0226780295372009,
+      "epoch": 4.024226110363392,
+      "grad_norm": 1.240837812423706,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.03402446508407593,
+      "mean_token_accuracy": 0.9883724093437195,
+      "num_tokens": 54164955.0,
+      "step": 2990
+    },
+    {
+      "entropy": 1.0194039583206176,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.1843734979629517,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.035861504077911374,
+      "mean_token_accuracy": 0.9872536063194275,
+      "num_tokens": 54346968.0,
+      "step": 3000
+    },
+    {
+      "entropy": 1.029161846637726,
+      "epoch": 4.05114401076716,
+      "grad_norm": 1.285415768623352,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.034106379747390746,
+      "mean_token_accuracy": 0.9882839441299438,
+      "num_tokens": 54528019.0,
+      "step": 3010
+    },
+    {
+      "entropy": 1.031964898109436,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4622068405151367,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.032776829600334165,
+      "mean_token_accuracy": 0.9882788896560669,
+      "num_tokens": 54708148.0,
+      "step": 3020
+    },
+    {
+      "entropy": 1.0282336235046388,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.1071696281433105,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.03698182702064514,
+      "mean_token_accuracy": 0.9865448236465454,
+      "num_tokens": 54888894.0,
+      "step": 3030
+    },
+    {
+      "entropy": 1.0233591556549073,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.3671438694000244,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.03386759161949158,
+      "mean_token_accuracy": 0.9879629850387573,
+      "num_tokens": 55069872.0,
+      "step": 3040
+    },
+    {
+      "entropy": 1.0250957012176514,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.1421459913253784,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.035586753487586976,
+      "mean_token_accuracy": 0.9869713068008423,
+      "num_tokens": 55250605.0,
+      "step": 3050
+    },
+    {
+      "entropy": 1.016488790512085,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.3857697248458862,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.039323312044143674,
+      "mean_token_accuracy": 0.986032885313034,
+      "num_tokens": 55431737.0,
+      "step": 3060
+    },
+    {
+      "entropy": 1.0130115926265717,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.1797211170196533,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.03508211970329285,
+      "mean_token_accuracy": 0.9872302651405335,
+      "num_tokens": 55612917.0,
+      "step": 3070
+    },
+    {
+      "entropy": 1.023509919643402,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.1484259366989136,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.0338476300239563,
+      "mean_token_accuracy": 0.9883188247680664,
+      "num_tokens": 55793687.0,
+      "step": 3080
+    },
+    {
+      "entropy": 1.0193628787994384,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.3184833526611328,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.03667713701725006,
+      "mean_token_accuracy": 0.9869367897510528,
+      "num_tokens": 55974723.0,
+      "step": 3090
+    },
+    {
+      "entropy": 1.013669216632843,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.2176531553268433,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.035006949305534364,
+      "mean_token_accuracy": 0.9868299901485443,
+      "num_tokens": 56156568.0,
+      "step": 3100
+    },
+    {
+      "entropy": 1.0072043538093567,
+      "epoch": 4.185733512786003,
+      "grad_norm": 1.4346243143081665,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.03560284972190857,
+      "mean_token_accuracy": 0.9869017779827118,
+      "num_tokens": 56337896.0,
+      "step": 3110
+    },
+    {
+      "entropy": 1.0102415680885315,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.2328612804412842,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.03608990907669067,
+      "mean_token_accuracy": 0.9874296009540557,
+      "num_tokens": 56519102.0,
+      "step": 3120
+    },
+    {
+      "entropy": 1.015284490585327,
+      "epoch": 4.212651413189771,
+      "grad_norm": 1.4385327100753784,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.03637963235378265,
+      "mean_token_accuracy": 0.986739182472229,
+      "num_tokens": 56699936.0,
+      "step": 3130
+    },
+    {
+      "entropy": 1.0054036319255828,
+      "epoch": 4.226110363391656,
+      "grad_norm": 1.1129945516586304,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.034222981333732604,
+      "mean_token_accuracy": 0.9883054196834564,
+      "num_tokens": 56882308.0,
+      "step": 3140
+    },
+    {
+      "entropy": 1.0061798334121703,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1598492860794067,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.03461446762084961,
+      "mean_token_accuracy": 0.9873203098773956,
+      "num_tokens": 57063770.0,
+      "step": 3150
+    },
+    {
+      "entropy": 1.0039719343185425,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.322458267211914,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.03701616525650024,
+      "mean_token_accuracy": 0.9862136900424957,
+      "num_tokens": 57244448.0,
+      "step": 3160
+    },
+    {
+      "entropy": 1.0048979759216308,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 1.570866584777832,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.036290496587753296,
+      "mean_token_accuracy": 0.9868158757686615,
+      "num_tokens": 57425448.0,
+      "step": 3170
+    },
+    {
+      "entropy": 1.0145976901054383,
+      "epoch": 4.279946164199193,
+      "grad_norm": 1.103385329246521,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.03472789227962494,
+      "mean_token_accuracy": 0.987658828496933,
+      "num_tokens": 57606369.0,
+      "step": 3180
+    },
+    {
+      "entropy": 1.0186771035194397,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.2541930675506592,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.03532317280769348,
+      "mean_token_accuracy": 0.9874464809894562,
+      "num_tokens": 57787685.0,
+      "step": 3190
+    },
+    {
+      "entropy": 1.0198698043823242,
+      "epoch": 4.306864064602961,
+      "grad_norm": 1.1590384244918823,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.036364412307739256,
+      "mean_token_accuracy": 0.9864933013916015,
+      "num_tokens": 57968903.0,
+      "step": 3200
+    },
+    {
+      "entropy": 1.0211624503135681,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.7207615375518799,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.03790584802627563,
+      "mean_token_accuracy": 0.9866538405418396,
+      "num_tokens": 58149469.0,
+      "step": 3210
+    },
+    {
+      "entropy": 1.0185290336608888,
+      "epoch": 4.33378196500673,
+      "grad_norm": 1.3042566776275635,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.03802136182785034,
+      "mean_token_accuracy": 0.9859855055809021,
+      "num_tokens": 58330867.0,
+      "step": 3220
+    },
+    {
+      "entropy": 1.0105828583240508,
+      "epoch": 4.347240915208614,
+      "grad_norm": 1.2536448240280151,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.034687772393226624,
+      "mean_token_accuracy": 0.9878224015235901,
+      "num_tokens": 58512529.0,
+      "step": 3230
+    },
+    {
+      "entropy": 1.0021630108356476,
+      "epoch": 4.360699865410498,
+      "grad_norm": 1.3816099166870117,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.035324835777282716,
+      "mean_token_accuracy": 0.9872316241264343,
+      "num_tokens": 58693573.0,
+      "step": 3240
+    },
+    {
+      "entropy": 1.007886278629303,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 1.3121757507324219,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.034709393978118896,
+      "mean_token_accuracy": 0.9877420663833618,
+      "num_tokens": 58874107.0,
+      "step": 3250
+    },
+    {
+      "entropy": 1.005664885044098,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.429775357246399,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.03714964389801025,
+      "mean_token_accuracy": 0.9861079812049866,
+      "num_tokens": 59054533.0,
+      "step": 3260
+    },
+    {
+      "entropy": 1.0047886371612549,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.2385882139205933,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.036937403678894046,
+      "mean_token_accuracy": 0.9863442242145538,
+      "num_tokens": 59236331.0,
+      "step": 3270
+    },
+    {
+      "entropy": 1.0209275126457213,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.476168155670166,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.03463808596134186,
+      "mean_token_accuracy": 0.9872730135917663,
+      "num_tokens": 59417193.0,
+      "step": 3280
+    },
+    {
+      "entropy": 1.021347463130951,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 1.4710102081298828,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.038167804479599,
+      "mean_token_accuracy": 0.986047238111496,
+      "num_tokens": 59598284.0,
+      "step": 3290
+    },
+    {
+      "entropy": 1.0282366394996643,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1043860912322998,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.036692625284194945,
+      "mean_token_accuracy": 0.9862713575363159,
+      "num_tokens": 59779329.0,
+      "step": 3300
+    },
+    {
+      "entropy": 1.021342945098877,
+      "epoch": 4.454912516823688,
+      "grad_norm": 1.1742897033691406,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.03571464121341705,
+      "mean_token_accuracy": 0.987505567073822,
+      "num_tokens": 59961340.0,
+      "step": 3310
+    },
+    {
+      "entropy": 1.0153375625610352,
+      "epoch": 4.468371467025572,
+      "grad_norm": 1.1321660280227661,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.036007928848266604,
+      "mean_token_accuracy": 0.9870990335941314,
+      "num_tokens": 60142647.0,
+      "step": 3320
+    },
+    {
+      "entropy": 1.0099608540534972,
+      "epoch": 4.481830417227457,
+      "grad_norm": 1.084017276763916,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.03562799394130707,
+      "mean_token_accuracy": 0.9865496158599854,
+      "num_tokens": 60324151.0,
+      "step": 3330
+    },
+    {
+      "entropy": 1.0173812866210938,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2779732942581177,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.03556201159954071,
+      "mean_token_accuracy": 0.9875519394874572,
+      "num_tokens": 60505023.0,
+      "step": 3340
+    },
+    {
+      "entropy": 1.0236618638038635,
+      "epoch": 4.508748317631225,
+      "grad_norm": 1.5285946130752563,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.03721941411495209,
+      "mean_token_accuracy": 0.9865630626678467,
+      "num_tokens": 60685921.0,
+      "step": 3350
+    },
+    {
+      "entropy": 1.0262888669967651,
+      "epoch": 4.522207267833109,
+      "grad_norm": 1.131675362586975,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.03602397441864014,
+      "mean_token_accuracy": 0.9864337563514709,
+      "num_tokens": 60867240.0,
+      "step": 3360
+    },
+    {
+      "entropy": 1.0292576909065247,
+      "epoch": 4.535666218034994,
+      "grad_norm": 1.5156011581420898,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.03973098993301392,
+      "mean_token_accuracy": 0.9855743169784545,
+      "num_tokens": 61047708.0,
+      "step": 3370
+    },
+    {
+      "entropy": 1.010222613811493,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0979945659637451,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.03533022105693817,
+      "mean_token_accuracy": 0.9869906663894653,
+      "num_tokens": 61229247.0,
+      "step": 3380
+    },
+    {
+      "entropy": 1.004716032743454,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.3082596063613892,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.033983749151229856,
+      "mean_token_accuracy": 0.9877199769020081,
+      "num_tokens": 61411059.0,
+      "step": 3390
+    },
+    {
+      "entropy": 1.0031994462013245,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.3033607006072998,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.03607205748558044,
+      "mean_token_accuracy": 0.9866301774978637,
+      "num_tokens": 61592538.0,
+      "step": 3400
+    },
+    {
+      "entropy": 1.0113391518592834,
+      "epoch": 4.589502018842531,
+      "grad_norm": 1.3490946292877197,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.03906603455543518,
+      "mean_token_accuracy": 0.9854673087596894,
+      "num_tokens": 61773613.0,
+      "step": 3410
+    },
+    {
+      "entropy": 1.0076410889625549,
+      "epoch": 4.602960969044415,
+      "grad_norm": 1.1719361543655396,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.037575924396514894,
+      "mean_token_accuracy": 0.9867486238479615,
+      "num_tokens": 61955690.0,
+      "step": 3420
+    },
+    {
+      "entropy": 1.0200914144515991,
+      "epoch": 4.616419919246299,
+      "grad_norm": 1.1205360889434814,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.03691520690917969,
+      "mean_token_accuracy": 0.9865606665611267,
+      "num_tokens": 62136047.0,
+      "step": 3430
+    },
+    {
+      "entropy": 1.0141666173934936,
+      "epoch": 4.629878869448183,
+      "grad_norm": 1.1805763244628906,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.035447508096694946,
+      "mean_token_accuracy": 0.9873166739940643,
+      "num_tokens": 62317192.0,
+      "step": 3440
+    },
+    {
+      "entropy": 1.0067960560321807,
+      "epoch": 4.643337819650068,
+      "grad_norm": 1.0286436080932617,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.03580200672149658,
+      "mean_token_accuracy": 0.9870070457458496,
+      "num_tokens": 62498860.0,
+      "step": 3450
+    },
+    {
+      "entropy": 1.0088169515132903,
+      "epoch": 4.656796769851952,
+      "grad_norm": 1.3638492822647095,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.03756250143051147,
+      "mean_token_accuracy": 0.9861987411975861,
+      "num_tokens": 62679272.0,
+      "step": 3460
+    },
+    {
+      "entropy": 1.0098722338676454,
+      "epoch": 4.670255720053836,
+      "grad_norm": 1.649560809135437,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.037615427374839784,
+      "mean_token_accuracy": 0.9864691913127899,
+      "num_tokens": 62860715.0,
+      "step": 3470
+    },
+    {
+      "entropy": 1.0174571394920349,
+      "epoch": 4.68371467025572,
+      "grad_norm": 1.2106678485870361,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.037718603014945985,
+      "mean_token_accuracy": 0.9855880260467529,
+      "num_tokens": 63041029.0,
+      "step": 3480
+    },
+    {
+      "entropy": 1.0129252314567565,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.1571707725524902,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.03724310994148254,
+      "mean_token_accuracy": 0.9865262687206269,
+      "num_tokens": 63222379.0,
+      "step": 3490
+    },
+    {
+      "entropy": 1.0124369859695435,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.0599361658096313,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.037783673405647276,
+      "mean_token_accuracy": 0.9859567046165466,
+      "num_tokens": 63403328.0,
+      "step": 3500
+    },
+    {
+      "entropy": 1.0041921436786652,
+      "epoch": 4.724091520861373,
+      "grad_norm": 1.2054452896118164,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.03585146069526672,
+      "mean_token_accuracy": 0.9868905484676361,
+      "num_tokens": 63584598.0,
+      "step": 3510
+    },
+    {
+      "entropy": 1.0015427589416503,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.5075141191482544,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.037428200244903564,
+      "mean_token_accuracy": 0.9867083311080933,
+      "num_tokens": 63766203.0,
+      "step": 3520
+    },
+    {
+      "entropy": 1.0039161920547486,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.310390830039978,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.037494581937789914,
+      "mean_token_accuracy": 0.9862947940826416,
+      "num_tokens": 63947714.0,
+      "step": 3530
+    },
+    {
+      "entropy": 1.0049733996391297,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.2912139892578125,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.03803396224975586,
+      "mean_token_accuracy": 0.9860390484333038,
+      "num_tokens": 64128798.0,
+      "step": 3540
+    },
+    {
+      "entropy": 1.0127450585365296,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.001133680343628,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0351622611284256,
+      "mean_token_accuracy": 0.9875370979309082,
+      "num_tokens": 64309577.0,
+      "step": 3550
+    },
+    {
+      "entropy": 1.0070195615291595,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 1.1870886087417603,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.036404851078987124,
+      "mean_token_accuracy": 0.9871752023696899,
+      "num_tokens": 64490707.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.9983750462532044,
+      "epoch": 4.804845222072679,
+      "grad_norm": 1.2374205589294434,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.03696496784687042,
+      "mean_token_accuracy": 0.9865773975849151,
+      "num_tokens": 64672104.0,
+      "step": 3570
+    },
+    {
+      "entropy": 1.0151559710502625,
+      "epoch": 4.818304172274562,
+      "grad_norm": 1.2001149654388428,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.03725983202457428,
+      "mean_token_accuracy": 0.9862341761589051,
+      "num_tokens": 64853353.0,
+      "step": 3580
+    },
+    {
+      "entropy": 1.0136774778366089,
+      "epoch": 4.831763122476447,
+      "grad_norm": 1.0128737688064575,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.03615986704826355,
+      "mean_token_accuracy": 0.9872736394405365,
+      "num_tokens": 65035009.0,
+      "step": 3590
+    },
+    {
+      "entropy": 1.0082051515579225,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.2552846670150757,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.03561249375343323,
+      "mean_token_accuracy": 0.9868618845939636,
+      "num_tokens": 65215981.0,
+      "step": 3600
+    },
+    {
+      "entropy": 1.0209908723831176,
+      "epoch": 4.858681022880216,
+      "grad_norm": 1.1914079189300537,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.03721153438091278,
+      "mean_token_accuracy": 0.9862311720848084,
+      "num_tokens": 65397088.0,
+      "step": 3610
+    },
+    {
+      "entropy": 1.014677894115448,
+      "epoch": 4.872139973082099,
+      "grad_norm": 1.3584535121917725,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.03897300064563751,
+      "mean_token_accuracy": 0.9851836740970612,
+      "num_tokens": 65578262.0,
+      "step": 3620
+    },
+    {
+      "entropy": 1.0207050919532776,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.2754287719726562,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.037926393747329715,
+      "mean_token_accuracy": 0.9859795570373535,
+      "num_tokens": 65758956.0,
+      "step": 3630
+    },
+    {
+      "entropy": 1.008733856678009,
+      "epoch": 4.899057873485868,
+      "grad_norm": 1.0815879106521606,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.036518561840057376,
+      "mean_token_accuracy": 0.9870649397373199,
+      "num_tokens": 65939970.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.9995620250701904,
+      "epoch": 4.912516823687753,
+      "grad_norm": 1.1773526668548584,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.035269084572792056,
+      "mean_token_accuracy": 0.9873197913169861,
+      "num_tokens": 66121045.0,
+      "step": 3650
+    },
+    {
+      "entropy": 1.0036249816417695,
+      "epoch": 4.925975773889636,
+      "grad_norm": 1.0941131114959717,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.0353422611951828,
+      "mean_token_accuracy": 0.9870408117771149,
+      "num_tokens": 66302324.0,
+      "step": 3660
+    },
+    {
+      "entropy": 1.004719352722168,
+      "epoch": 4.939434724091521,
+      "grad_norm": 1.5107269287109375,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.03743018507957459,
+      "mean_token_accuracy": 0.9857985079288483,
+      "num_tokens": 66483916.0,
+      "step": 3670
+    },
+    {
+      "entropy": 1.0076164484024048,
+      "epoch": 4.952893674293405,
+      "grad_norm": 1.3877427577972412,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.03857019543647766,
+      "mean_token_accuracy": 0.9856370270252228,
+      "num_tokens": 66664390.0,
+      "step": 3680
+    },
+    {
+      "entropy": 1.015517818927765,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1693767309188843,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.03634478151798248,
+      "mean_token_accuracy": 0.9868648707866668,
+      "num_tokens": 66845574.0,
+      "step": 3690
+    },
+    {
+      "entropy": 1.011656105518341,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.082209587097168,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.03513925671577454,
+      "mean_token_accuracy": 0.9872389912605286,
+      "num_tokens": 67027557.0,
+      "step": 3700
+    },
+    {
+      "entropy": 1.0161120653152467,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.266615629196167,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.0379630446434021,
+      "mean_token_accuracy": 0.9865313708782196,
+      "num_tokens": 67208088.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 1.0219634957374282,
+      "eval_loss": 0.14612138271331787,
+      "eval_mean_token_accuracy": 0.9504923186484416,
+      "eval_num_tokens": 67298266.0,
+      "eval_runtime": 12.6802,
+      "eval_samples_per_second": 394.317,
+      "eval_steps_per_second": 12.382,
+      "step": 3715
+    },
+    {
+      "entropy": 1.019774079322815,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8909980654716492,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.02882351279258728,
+      "mean_token_accuracy": 0.9900537073612213,
+      "num_tokens": 67388858.0,
+      "step": 3720
+    },
+    {
+      "entropy": 1.0057315409183503,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9996656775474548,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.02233442962169647,
+      "mean_token_accuracy": 0.9922977447509765,
+      "num_tokens": 67570084.0,
+      "step": 3730
+    },
+    {
+      "entropy": 1.0058627128601074,
+      "epoch": 5.03364737550471,
+      "grad_norm": 1.0265147686004639,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.021967282891273497,
+      "mean_token_accuracy": 0.9923401355743409,
+      "num_tokens": 67751587.0,
+      "step": 3740
+    },
+    {
+      "entropy": 1.0083157896995545,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 1.0139433145523071,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.022531555593013765,
+      "mean_token_accuracy": 0.9930226564407348,
+      "num_tokens": 67933269.0,
+      "step": 3750
+    },
+    {
+      "entropy": 1.0119245886802672,
+      "epoch": 5.060565275908479,
+      "grad_norm": 1.0655452013015747,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.021685637533664703,
+      "mean_token_accuracy": 0.9925794541835785,
+      "num_tokens": 68113971.0,
+      "step": 3760
+    },
+    {
+      "entropy": 1.0082412719726563,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.268751859664917,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.021513286232948303,
+      "mean_token_accuracy": 0.9927303433418274,
+      "num_tokens": 68295632.0,
+      "step": 3770
+    },
+    {
+      "entropy": 1.012417995929718,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.9899319410324097,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.02352543771266937,
+      "mean_token_accuracy": 0.992375648021698,
+      "num_tokens": 68476794.0,
+      "step": 3780
+    },
+    {
+      "entropy": 1.0069364666938783,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 1.1964375972747803,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.021382461488246917,
+      "mean_token_accuracy": 0.9928456962108612,
+      "num_tokens": 68658249.0,
+      "step": 3790
+    },
+    {
+      "entropy": 1.0040670156478881,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0445663928985596,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.02207571566104889,
+      "mean_token_accuracy": 0.9927122712135314,
+      "num_tokens": 68840525.0,
+      "step": 3800
+    },
+    {
+      "entropy": 1.0083145201206207,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.8981389403343201,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.022873447835445405,
+      "mean_token_accuracy": 0.9918712139129638,
+      "num_tokens": 69021896.0,
+      "step": 3810
+    },
+    {
+      "entropy": 1.0109572768211366,
+      "epoch": 5.141318977119784,
+      "grad_norm": 1.250670313835144,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.023577621579170226,
+      "mean_token_accuracy": 0.9919484734535218,
+      "num_tokens": 69203112.0,
+      "step": 3820
+    },
+    {
+      "entropy": 1.0164754271507264,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.3053207397460938,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.02436322420835495,
+      "mean_token_accuracy": 0.9921659469604492,
+      "num_tokens": 69384193.0,
+      "step": 3830
+    },
+    {
+      "entropy": 1.0034007966518401,
+      "epoch": 5.168236877523553,
+      "grad_norm": 1.4878787994384766,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.023177918791770936,
+      "mean_token_accuracy": 0.9919318974018096,
+      "num_tokens": 69565602.0,
+      "step": 3840
+    },
+    {
+      "entropy": 1.0074714839458465,
+      "epoch": 5.181695827725437,
+      "grad_norm": 1.0512391328811646,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.023409275710582732,
+      "mean_token_accuracy": 0.9918898940086365,
+      "num_tokens": 69746242.0,
+      "step": 3850
+    },
+    {
+      "entropy": 1.003154468536377,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.971832811832428,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.023223266005516052,
+      "mean_token_accuracy": 0.9920656085014343,
+      "num_tokens": 69927227.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.999264246225357,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.9266519546508789,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.02454002797603607,
+      "mean_token_accuracy": 0.9914694905281067,
+      "num_tokens": 70108580.0,
+      "step": 3870
+    },
+    {
+      "entropy": 1.002640813589096,
+      "epoch": 5.22207267833109,
+      "grad_norm": 1.5331412553787231,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.02458159625530243,
+      "mean_token_accuracy": 0.9919124245643616,
+      "num_tokens": 70288960.0,
+      "step": 3880
+    },
+    {
+      "entropy": 1.007998287677765,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.3164633512496948,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.02421087771654129,
+      "mean_token_accuracy": 0.9917516648769379,
+      "num_tokens": 70470266.0,
+      "step": 3890
+    },
+    {
+      "entropy": 1.0089037418365479,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1788280010223389,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.023224952816963195,
+      "mean_token_accuracy": 0.9921185493469238,
+      "num_tokens": 70650975.0,
+      "step": 3900
+    },
+    {
+      "entropy": 1.001932644844055,
+      "epoch": 5.262449528936743,
+      "grad_norm": 1.2484381198883057,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.02366664707660675,
+      "mean_token_accuracy": 0.9917596876621246,
+      "num_tokens": 70832549.0,
+      "step": 3910
+    },
+    {
+      "entropy": 1.004903507232666,
+      "epoch": 5.275908479138627,
+      "grad_norm": 1.1852507591247559,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.02296685576438904,
+      "mean_token_accuracy": 0.9918978750705719,
+      "num_tokens": 71014296.0,
+      "step": 3920
+    },
+    {
+      "entropy": 1.004357820749283,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.9933306574821472,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.02361467629671097,
+      "mean_token_accuracy": 0.991608303785324,
+      "num_tokens": 71195526.0,
+      "step": 3930
+    },
+    {
+      "entropy": 1.0020038604736328,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.4304245710372925,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.02268100529909134,
+      "mean_token_accuracy": 0.9917695164680481,
+      "num_tokens": 71377269.0,
+      "step": 3940
+    },
+    {
+      "entropy": 1.012037169933319,
+      "epoch": 5.31628532974428,
+      "grad_norm": 1.074124813079834,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.023454257845878602,
+      "mean_token_accuracy": 0.9922087669372559,
+      "num_tokens": 71558158.0,
+      "step": 3950
+    },
+    {
+      "entropy": 1.0094320178031921,
+      "epoch": 5.329744279946164,
+      "grad_norm": 1.0919133424758911,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.022751571238040925,
+      "mean_token_accuracy": 0.9921496629714965,
+      "num_tokens": 71738847.0,
+      "step": 3960
+    },
+    {
+      "entropy": 1.0063781797885896,
+      "epoch": 5.343203230148048,
+      "grad_norm": 1.1300997734069824,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.023900465667247774,
+      "mean_token_accuracy": 0.9914113521575928,
+      "num_tokens": 71918875.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.9972042560577392,
+      "epoch": 5.356662180349932,
+      "grad_norm": 1.165519118309021,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.021918240189552306,
+      "mean_token_accuracy": 0.9926528394222259,
+      "num_tokens": 72100551.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.9954554200172424,
+      "epoch": 5.370121130551817,
+      "grad_norm": 1.342820167541504,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.024170228838920595,
+      "mean_token_accuracy": 0.9913744986057281,
+      "num_tokens": 72281855.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.9931409180164337,
+      "epoch": 5.383580080753701,
+      "grad_norm": 1.2873902320861816,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.02334785759449005,
+      "mean_token_accuracy": 0.9921288907527923,
+      "num_tokens": 72462916.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.9915881812572479,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.9761195778846741,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.02302718609571457,
+      "mean_token_accuracy": 0.9924728453159333,
+      "num_tokens": 72644666.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.9931966602802277,
+      "epoch": 5.410497981157469,
+      "grad_norm": 1.4468477964401245,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.023682811856269838,
+      "mean_token_accuracy": 0.9917364478111267,
+      "num_tokens": 72825808.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.9975718319416046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.1559572219848633,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.02247478663921356,
+      "mean_token_accuracy": 0.992560338973999,
+      "num_tokens": 73007923.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.999228036403656,
+      "epoch": 5.437415881561238,
+      "grad_norm": 1.1230353116989136,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.022295859456062318,
+      "mean_token_accuracy": 0.9922321498394012,
+      "num_tokens": 73189074.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.9897105693817139,
+      "epoch": 5.450874831763122,
+      "grad_norm": 1.413734793663025,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.02201971709728241,
+      "mean_token_accuracy": 0.9927073001861573,
+      "num_tokens": 73369918.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.9986266851425171,
+      "epoch": 5.464333781965006,
+      "grad_norm": 1.0815346240997314,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.02358766496181488,
+      "mean_token_accuracy": 0.9911752700805664,
+      "num_tokens": 73550625.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.9953943610191345,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0506970882415771,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.022177645564079286,
+      "mean_token_accuracy": 0.9923022150993347,
+      "num_tokens": 73731436.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.9954936504364014,
+      "epoch": 5.491251682368775,
+      "grad_norm": 1.4665249586105347,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.024948063492774963,
+      "mean_token_accuracy": 0.9913177132606507,
+      "num_tokens": 73912097.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.9986696183681488,
+      "epoch": 5.504710632570659,
+      "grad_norm": 1.158179759979248,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.02555338144302368,
+      "mean_token_accuracy": 0.9908900320529938,
+      "num_tokens": 74093602.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.9955579698085785,
+      "epoch": 5.518169582772543,
+      "grad_norm": 1.0024714469909668,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.023432795703411103,
+      "mean_token_accuracy": 0.9919674038887024,
+      "num_tokens": 74274664.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.994187182188034,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.9290911555290222,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.023194606602191924,
+      "mean_token_accuracy": 0.9920254707336426,
+      "num_tokens": 74456436.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.9927510738372802,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.8986559510231018,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.02163631021976471,
+      "mean_token_accuracy": 0.9927416622638703,
+      "num_tokens": 74637700.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.9877451896667481,
+      "epoch": 5.558546433378196,
+      "grad_norm": 1.0314558744430542,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.0209101140499115,
+      "mean_token_accuracy": 0.9930387079715729,
+      "num_tokens": 74819310.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.9903139114379883,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 1.1819736957550049,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.02219167649745941,
+      "mean_token_accuracy": 0.992407250404358,
+      "num_tokens": 74999907.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.9891258835792541,
+      "epoch": 5.585464333781965,
+      "grad_norm": 1.0412501096725464,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.02201283872127533,
+      "mean_token_accuracy": 0.992091816663742,
+      "num_tokens": 75181189.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.9870602369308472,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0461134910583496,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.023040324449539185,
+      "mean_token_accuracy": 0.992098081111908,
+      "num_tokens": 75362769.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.9888905882835388,
+      "epoch": 5.612382234185733,
+      "grad_norm": 1.086546540260315,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.021401138603687288,
+      "mean_token_accuracy": 0.992308247089386,
+      "num_tokens": 75544519.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.9940549910068512,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 1.3897169828414917,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.023297235369682312,
+      "mean_token_accuracy": 0.9920303761959076,
+      "num_tokens": 75725651.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.9897378265857697,
+      "epoch": 5.639300134589502,
+      "grad_norm": 1.283887267112732,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.024343529343605043,
+      "mean_token_accuracy": 0.9916397035121918,
+      "num_tokens": 75907036.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.9989661812782288,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.9090050458908081,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.022062870860099792,
+      "mean_token_accuracy": 0.9921731591224671,
+      "num_tokens": 76088248.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.9966195583343506,
+      "epoch": 5.66621803499327,
+      "grad_norm": 1.2091970443725586,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.023516717553138732,
+      "mean_token_accuracy": 0.9918879508972168,
+      "num_tokens": 76269704.0,
+      "step": 4210
+    },
+    {
+      "entropy": 1.0081237196922301,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 1.0648062229156494,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.021994276344776152,
+      "mean_token_accuracy": 0.9925018906593323,
+      "num_tokens": 76451416.0,
+      "step": 4220
+    },
+    {
+      "entropy": 1.0061897933483124,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.188748836517334,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.022657927870750428,
+      "mean_token_accuracy": 0.9919365108013153,
+      "num_tokens": 76632827.0,
+      "step": 4230
+    },
+    {
+      "entropy": 1.0024715662002563,
+      "epoch": 5.706594885598923,
+      "grad_norm": 1.2140473127365112,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.02203955501317978,
+      "mean_token_accuracy": 0.9925529718399048,
+      "num_tokens": 76813534.0,
+      "step": 4240
+    },
+    {
+      "entropy": 1.0011352360248567,
+      "epoch": 5.720053835800807,
+      "grad_norm": 1.389479398727417,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.022718487679958342,
+      "mean_token_accuracy": 0.9922778069972992,
+      "num_tokens": 76994327.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.9971556246280671,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 1.1441208124160767,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.024178403615951537,
+      "mean_token_accuracy": 0.9913869142532349,
+      "num_tokens": 77174987.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.9961654126644135,
+      "epoch": 5.746971736204576,
+      "grad_norm": 1.1283735036849976,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.023389075696468354,
+      "mean_token_accuracy": 0.9921939969062805,
+      "num_tokens": 77356043.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.9927134871482849,
+      "epoch": 5.76043068640646,
+      "grad_norm": 1.201904535293579,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.022739484906196594,
+      "mean_token_accuracy": 0.9921142935752869,
+      "num_tokens": 77536396.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.9919731795787812,
+      "epoch": 5.773889636608344,
+      "grad_norm": 1.145776391029358,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.020524097979068755,
+      "mean_token_accuracy": 0.9926942765712738,
+      "num_tokens": 77717339.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.9935499131679535,
+      "epoch": 5.787348586810229,
+      "grad_norm": 1.099414348602295,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.02070924937725067,
+      "mean_token_accuracy": 0.9928180515766144,
+      "num_tokens": 77898360.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.9875575602054596,
+      "epoch": 5.800807537012113,
+      "grad_norm": 1.1510778665542603,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.02202107310295105,
+      "mean_token_accuracy": 0.9921730697154999,
+      "num_tokens": 78079699.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.9855418741703034,
+      "epoch": 5.814266487213997,
+      "grad_norm": 1.0748379230499268,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.022809866070747375,
+      "mean_token_accuracy": 0.9919199466705322,
+      "num_tokens": 78260900.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.977304744720459,
+      "epoch": 5.827725437415881,
+      "grad_norm": 1.091854453086853,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.02268455773591995,
+      "mean_token_accuracy": 0.9918094217777252,
+      "num_tokens": 78442168.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.9767287075519562,
+      "epoch": 5.841184387617766,
+      "grad_norm": 1.1712645292282104,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.022726841270923615,
+      "mean_token_accuracy": 0.9922167181968689,
+      "num_tokens": 78623589.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.974297159910202,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.9508016109466553,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.021290253102779388,
+      "mean_token_accuracy": 0.9926727712154388,
+      "num_tokens": 78804892.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.9795810282230377,
+      "epoch": 5.868102288021534,
+      "grad_norm": 1.3127822875976562,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.022329577803611757,
+      "mean_token_accuracy": 0.9927741348743438,
+      "num_tokens": 78986204.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.9879997313022614,
+      "epoch": 5.881561238223418,
+      "grad_norm": 1.0200371742248535,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.022206541895866395,
+      "mean_token_accuracy": 0.9921830892562866,
+      "num_tokens": 79166824.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.9801069736480713,
+      "epoch": 5.895020188425303,
+      "grad_norm": 1.4769432544708252,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.02101587802171707,
+      "mean_token_accuracy": 0.9930795848369598,
+      "num_tokens": 79347228.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.9727071464061737,
+      "epoch": 5.908479138627187,
+      "grad_norm": 1.7034555673599243,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.021690081059932708,
+      "mean_token_accuracy": 0.9926194071769714,
+      "num_tokens": 79529483.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.9782697260379791,
+      "epoch": 5.921938088829071,
+      "grad_norm": 1.5307973623275757,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.023112425208091737,
+      "mean_token_accuracy": 0.9919495701789856,
+      "num_tokens": 79710571.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.9831463754177093,
+      "epoch": 5.935397039030955,
+      "grad_norm": 1.1024526357650757,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.021427598595619202,
+      "mean_token_accuracy": 0.9924166023731231,
+      "num_tokens": 79890685.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.9811958551406861,
+      "epoch": 5.94885598923284,
+      "grad_norm": 1.7247848510742188,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.02339867949485779,
+      "mean_token_accuracy": 0.9916171073913574,
+      "num_tokens": 80071331.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.9830600321292877,
+      "epoch": 5.962314939434724,
+      "grad_norm": 1.0638445615768433,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.02204108238220215,
+      "mean_token_accuracy": 0.9925127327442169,
+      "num_tokens": 80251525.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.9840480864048005,
+      "epoch": 5.975773889636608,
+      "grad_norm": 1.1205570697784424,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.020443232357501985,
+      "mean_token_accuracy": 0.9932060956954956,
+      "num_tokens": 80432001.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.9841626048088074,
+      "epoch": 5.989232839838492,
+      "grad_norm": 1.2928433418273926,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.021673919260501863,
+      "mean_token_accuracy": 0.9931873917579651,
+      "num_tokens": 80613075.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.9904457463580332,
+      "eval_loss": 0.161154642701149,
+      "eval_mean_token_accuracy": 0.9516840383505366,
+      "eval_num_tokens": 80757887.0,
+      "eval_runtime": 12.7345,
+      "eval_samples_per_second": 392.633,
+      "eval_steps_per_second": 12.329,
+      "step": 4458
+    },
+    {
+      "entropy": 0.9850570559501648,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.8859100341796875,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.018883128464221955,
+      "mean_token_accuracy": 0.9933207035064697,
+      "num_tokens": 80794295.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.9796536922454834,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.8121339082717896,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.01288142502307892,
+      "mean_token_accuracy": 0.9960863590240479,
+      "num_tokens": 80975379.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.9699741184711457,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.8923460841178894,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.012284834682941437,
+      "mean_token_accuracy": 0.9960344016551972,
+      "num_tokens": 81155713.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.9629065454006195,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.6698539853096008,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.011575647443532944,
+      "mean_token_accuracy": 0.9962593257427216,
+      "num_tokens": 81336446.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.9579961061477661,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.7555826306343079,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.011635755002498627,
+      "mean_token_accuracy": 0.9965786218643189,
+      "num_tokens": 81517969.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.96330127120018,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.8983458280563354,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.012715654075145721,
+      "mean_token_accuracy": 0.9959319531917572,
+      "num_tokens": 81699419.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.9640884518623352,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.9172118902206421,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.012482476234436036,
+      "mean_token_accuracy": 0.9966461837291718,
+      "num_tokens": 81880956.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.9678717851638794,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.8765719532966614,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.012080545723438262,
+      "mean_token_accuracy": 0.9964690268039703,
+      "num_tokens": 82061897.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.9656756520271301,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.746626079082489,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.012495596706867219,
+      "mean_token_accuracy": 0.9959208548069001,
+      "num_tokens": 82242847.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.9631998062133789,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.9732807874679565,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.01309359222650528,
+      "mean_token_accuracy": 0.9960879981517792,
+      "num_tokens": 82424182.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.9686778664588929,
+      "epoch": 6.137281292059219,
+      "grad_norm": 1.040855884552002,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.013380664587020873,
+      "mean_token_accuracy": 0.9955641269683838,
+      "num_tokens": 82604728.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.9661992788314819,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 1.209067940711975,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.013264504075050355,
+      "mean_token_accuracy": 0.9958661019802093,
+      "num_tokens": 82786184.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.9628405570983887,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.9056928157806396,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.011586685478687287,
+      "mean_token_accuracy": 0.9962525606155396,
+      "num_tokens": 82968127.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.9664590656757355,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.8662444949150085,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.013227257132530212,
+      "mean_token_accuracy": 0.9958321928977967,
+      "num_tokens": 83149016.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.9584677755832672,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.8611947894096375,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.012484049052000045,
+      "mean_token_accuracy": 0.9961173057556152,
+      "num_tokens": 83330474.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.9613959610462188,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 1.0272881984710693,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.013143527507781982,
+      "mean_token_accuracy": 0.995938628911972,
+      "num_tokens": 83512502.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.9671844780445099,
+      "epoch": 6.218034993270525,
+      "grad_norm": 1.038604497909546,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.015797269344329835,
+      "mean_token_accuracy": 0.9949265837669372,
+      "num_tokens": 83693427.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.9725019454956054,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.7591229677200317,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.01309441477060318,
+      "mean_token_accuracy": 0.9960084438323975,
+      "num_tokens": 83874886.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.9710726916790009,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.9521571397781372,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.012572157382965087,
+      "mean_token_accuracy": 0.9960904121398926,
+      "num_tokens": 84055459.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.9712884068489075,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.812011182308197,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.014366535842418671,
+      "mean_token_accuracy": 0.9953844547271729,
+      "num_tokens": 84236017.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.9648497045040131,
+      "epoch": 6.271870794078062,
+      "grad_norm": 1.648348093032837,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.014780099689960479,
+      "mean_token_accuracy": 0.9955918669700623,
+      "num_tokens": 84417271.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.956806218624115,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.9592521786689758,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.014198371767997741,
+      "mean_token_accuracy": 0.9954844176769256,
+      "num_tokens": 84599326.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.9628090143203736,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.9466222524642944,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.013178789615631103,
+      "mean_token_accuracy": 0.995667016506195,
+      "num_tokens": 84779914.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.9617082297801971,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.9156956672668457,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.011176959425210953,
+      "mean_token_accuracy": 0.9965419709682465,
+      "num_tokens": 84960572.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.9526888430118561,
+      "epoch": 6.325706594885599,
+      "grad_norm": 1.107490062713623,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.01246263161301613,
+      "mean_token_accuracy": 0.9961843729019165,
+      "num_tokens": 85141667.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.952410751581192,
+      "epoch": 6.339165545087483,
+      "grad_norm": 1.06051766872406,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.01356920450925827,
+      "mean_token_accuracy": 0.9956991016864777,
+      "num_tokens": 85323361.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.9601339221000671,
+      "epoch": 6.352624495289367,
+      "grad_norm": 1.0533535480499268,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.012942484021186829,
+      "mean_token_accuracy": 0.9960961997509002,
+      "num_tokens": 85503828.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.9630769610404968,
+      "epoch": 6.366083445491252,
+      "grad_norm": 1.2243846654891968,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.012688860297203064,
+      "mean_token_accuracy": 0.9956350862979889,
+      "num_tokens": 85684804.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.9601665258407592,
+      "epoch": 6.379542395693136,
+      "grad_norm": 1.0219420194625854,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.012651938199996948,
+      "mean_token_accuracy": 0.9959766507148743,
+      "num_tokens": 85866033.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.9643365681171417,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.8759870529174805,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.013052919507026672,
+      "mean_token_accuracy": 0.9956570088863372,
+      "num_tokens": 86046793.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.9643962621688843,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.8771425485610962,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.013016942143440246,
+      "mean_token_accuracy": 0.9956202030181884,
+      "num_tokens": 86227594.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.9684974253177643,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.9857693314552307,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.01260487139225006,
+      "mean_token_accuracy": 0.9959571301937103,
+      "num_tokens": 86409012.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.9683425962924957,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.9220049381256104,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.012756258249282837,
+      "mean_token_accuracy": 0.9961577892303467,
+      "num_tokens": 86589770.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.9730850279331207,
+      "epoch": 6.446837146702557,
+      "grad_norm": 1.1092842817306519,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.011953817307949066,
+      "mean_token_accuracy": 0.9963648617267609,
+      "num_tokens": 86771224.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.9717083096504211,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.9012715220451355,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.0137409508228302,
+      "mean_token_accuracy": 0.9956796884536743,
+      "num_tokens": 86952165.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.9706744968891143,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.8237823247909546,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.013622400164604188,
+      "mean_token_accuracy": 0.9958401381969452,
+      "num_tokens": 87133948.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.968240886926651,
+      "epoch": 6.48721399730821,
+      "grad_norm": 0.9172955751419067,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.012218748033046723,
+      "mean_token_accuracy": 0.9961099803447724,
+      "num_tokens": 87315186.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.9701445996761322,
+      "epoch": 6.500672947510094,
+      "grad_norm": 1.1613796949386597,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.011357608437538146,
+      "mean_token_accuracy": 0.9961966574192047,
+      "num_tokens": 87496318.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.9753403663635254,
+      "epoch": 6.514131897711978,
+      "grad_norm": 1.0601547956466675,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.01224931925535202,
+      "mean_token_accuracy": 0.9964543879032135,
+      "num_tokens": 87676690.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.9595320463180542,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.8927657604217529,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.012476099282503128,
+      "mean_token_accuracy": 0.9961197137832641,
+      "num_tokens": 87858122.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.9639881610870361,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.9392650127410889,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.013359837234020233,
+      "mean_token_accuracy": 0.9956938862800598,
+      "num_tokens": 88039349.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.961546677350998,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.709064781665802,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.012232375890016555,
+      "mean_token_accuracy": 0.9963175058364868,
+      "num_tokens": 88221085.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.966980516910553,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 1.1382325887680054,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.013281235098838806,
+      "mean_token_accuracy": 0.9955899178981781,
+      "num_tokens": 88401750.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.967098617553711,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.7260605692863464,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.011995580792427064,
+      "mean_token_accuracy": 0.9963020443916321,
+      "num_tokens": 88582188.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.9611273169517517,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.9492855668067932,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.011460936814546584,
+      "mean_token_accuracy": 0.9965442717075348,
+      "num_tokens": 88764018.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.964341527223587,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.7225981950759888,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.011858727037906646,
+      "mean_token_accuracy": 0.996355015039444,
+      "num_tokens": 88944988.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.9631503582000732,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 0.996126651763916,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.013101565837860107,
+      "mean_token_accuracy": 0.9957616865634918,
+      "num_tokens": 89125768.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.9629738807678223,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.8971033096313477,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.012153981626033783,
+      "mean_token_accuracy": 0.9961553812026978,
+      "num_tokens": 89306694.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.9592839002609252,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.8334258198738098,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.012248124927282333,
+      "mean_token_accuracy": 0.9960185050964355,
+      "num_tokens": 89487628.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.9580614447593689,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.7489107251167297,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.012415885925292969,
+      "mean_token_accuracy": 0.9960191190242768,
+      "num_tokens": 89668907.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.9544502675533295,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.9663529992103577,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.012825769186019898,
+      "mean_token_accuracy": 0.9958091497421264,
+      "num_tokens": 89850200.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.9588588118553162,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.7662053108215332,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.012289515137672425,
+      "mean_token_accuracy": 0.9956038355827331,
+      "num_tokens": 90030738.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.9632011234760285,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.8352729082107544,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.01209896057844162,
+      "mean_token_accuracy": 0.9960383713245392,
+      "num_tokens": 90211066.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.962366932630539,
+      "epoch": 6.716016150740242,
+      "grad_norm": 1.271884799003601,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.013325554132461549,
+      "mean_token_accuracy": 0.9955421090126038,
+      "num_tokens": 90392202.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.9630074858665466,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.806925892829895,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.012963607907295227,
+      "mean_token_accuracy": 0.9955462396144867,
+      "num_tokens": 90573271.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.9631992042064667,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.9406610131263733,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.012753760814666748,
+      "mean_token_accuracy": 0.9954385042190552,
+      "num_tokens": 90753836.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.9599998474121094,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.7755982875823975,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.011351206153631211,
+      "mean_token_accuracy": 0.9965151965618133,
+      "num_tokens": 90935528.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.961510854959488,
+      "epoch": 6.769851951547779,
+      "grad_norm": 1.223897099494934,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.012862575054168702,
+      "mean_token_accuracy": 0.995742017030716,
+      "num_tokens": 91116520.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.9597096800804138,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.8847037553787231,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.01192454993724823,
+      "mean_token_accuracy": 0.9964925825595856,
+      "num_tokens": 91297847.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.961575037240982,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.7097873687744141,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.012100750207901,
+      "mean_token_accuracy": 0.9963988482952117,
+      "num_tokens": 91479371.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.9660714268684387,
+      "epoch": 6.810228802153432,
+      "grad_norm": 0.8912429213523865,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.011809615790843964,
+      "mean_token_accuracy": 0.9961784362792969,
+      "num_tokens": 91660175.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.9656784057617187,
+      "epoch": 6.823687752355316,
+      "grad_norm": 1.2362871170043945,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.013092437386512756,
+      "mean_token_accuracy": 0.9954990804195404,
+      "num_tokens": 91840728.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.9630216419696808,
+      "epoch": 6.837146702557201,
+      "grad_norm": 1.0081552267074585,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.013006564974784852,
+      "mean_token_accuracy": 0.995445728302002,
+      "num_tokens": 92021819.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.9650352597236633,
+      "epoch": 6.850605652759085,
+      "grad_norm": 1.0923664569854736,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.012209924310445786,
+      "mean_token_accuracy": 0.9961765468120575,
+      "num_tokens": 92203146.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.9654329180717468,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.8630998730659485,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.01286689043045044,
+      "mean_token_accuracy": 0.9955997586250305,
+      "num_tokens": 92384274.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.9633705973625183,
+      "epoch": 6.877523553162853,
+      "grad_norm": 0.9883823394775391,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.012318137288093566,
+      "mean_token_accuracy": 0.9964197039604187,
+      "num_tokens": 92564922.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.9619575142860413,
+      "epoch": 6.890982503364738,
+      "grad_norm": 1.0694894790649414,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.011325329542160034,
+      "mean_token_accuracy": 0.9967617630958557,
+      "num_tokens": 92745952.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.9608132243156433,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.850687563419342,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.012749406695365905,
+      "mean_token_accuracy": 0.9958359479904175,
+      "num_tokens": 92926966.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.9643485128879548,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.8775151968002319,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.012118304520845414,
+      "mean_token_accuracy": 0.996504932641983,
+      "num_tokens": 93108549.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.9634061992168427,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.925463855266571,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.011662732064723968,
+      "mean_token_accuracy": 0.9965654671192169,
+      "num_tokens": 93290512.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.9620663046836853,
+      "epoch": 6.944818304172275,
+      "grad_norm": 1.1143932342529297,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.01120457649230957,
+      "mean_token_accuracy": 0.9962508201599121,
+      "num_tokens": 93472398.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.9628985583782196,
+      "epoch": 6.958277254374159,
+      "grad_norm": 1.0271083116531372,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.012035199999809265,
+      "mean_token_accuracy": 0.996053409576416,
+      "num_tokens": 93654111.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.9658466935157776,
+      "epoch": 6.971736204576043,
+      "grad_norm": 1.0225892066955566,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.01229911893606186,
+      "mean_token_accuracy": 0.995949512720108,
+      "num_tokens": 93835692.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.9601108253002166,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.7876276969909668,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.011063569784164428,
+      "mean_token_accuracy": 0.9966089725494385,
+      "num_tokens": 94018119.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.9596541881561279,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.9803016185760498,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.011228029429912568,
+      "mean_token_accuracy": 0.9964218616485596,
+      "num_tokens": 94199483.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.9689761833020836,
+      "eval_loss": 0.1798103004693985,
+      "eval_mean_token_accuracy": 0.950934177371347,
+      "eval_num_tokens": 94217754.0,
+      "eval_runtime": 12.6557,
+      "eval_samples_per_second": 395.079,
+      "eval_steps_per_second": 12.405,
+      "step": 5201
+    },
+    {
+      "entropy": 0.9555480360984803,
+      "epoch": 7.012113055181696,
+      "grad_norm": 0.7279857397079468,
+      "learning_rate": 2.2506648060191042e-05,
+      "loss": 0.007322490215301514,
+      "mean_token_accuracy": 0.9978758811950683,
+      "num_tokens": 94380656.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.9532195508480072,
+      "epoch": 7.02557200538358,
+      "grad_norm": 0.6351640820503235,
+      "learning_rate": 2.2321031102941066e-05,
+      "loss": 0.007188012450933456,
+      "mean_token_accuracy": 0.9980705499649047,
+      "num_tokens": 94561769.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.9503021717071534,
+      "epoch": 7.039030955585464,
+      "grad_norm": 0.768357515335083,
+      "learning_rate": 2.2135962530768055e-05,
+      "loss": 0.007337777316570282,
+      "mean_token_accuracy": 0.9979041218757629,
+      "num_tokens": 94742683.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.9469992637634277,
+      "epoch": 7.052489905787349,
+      "grad_norm": 0.6551904678344727,
+      "learning_rate": 2.1951446010313053e-05,
+      "loss": 0.00702056959271431,
+      "mean_token_accuracy": 0.9980945587158203,
+      "num_tokens": 94924159.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.9499239981174469,
+      "epoch": 7.065948855989233,
+      "grad_norm": 0.8986485600471497,
+      "learning_rate": 2.1767485197279662e-05,
+      "loss": 0.007891762256622314,
+      "mean_token_accuracy": 0.9978051722049713,
+      "num_tokens": 95105209.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.9502587139606475,
+      "epoch": 7.079407806191117,
+      "grad_norm": 0.4773225486278534,
+      "learning_rate": 2.1584083736361617e-05,
+      "loss": 0.0066356688737869264,
+      "mean_token_accuracy": 0.9983843564987183,
+      "num_tokens": 95286168.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.9485512435436249,
+      "epoch": 7.0928667563930015,
+      "grad_norm": 0.7873836159706116,
+      "learning_rate": 2.1401245261170582e-05,
+      "loss": 0.007745465636253357,
+      "mean_token_accuracy": 0.9978477001190186,
+      "num_tokens": 95467244.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.9444929420948028,
+      "epoch": 7.106325706594886,
+      "grad_norm": 0.9345479607582092,
+      "learning_rate": 2.121897339416416e-05,
+      "loss": 0.006544182449579239,
+      "mean_token_accuracy": 0.9983295381069184,
+      "num_tokens": 95648662.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.9411603569984436,
+      "epoch": 7.11978465679677,
+      "grad_norm": 0.8139040470123291,
+      "learning_rate": 2.103727174657412e-05,
+      "loss": 0.00677436888217926,
+      "mean_token_accuracy": 0.9982263207435608,
+      "num_tokens": 95829602.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.9461234629154205,
+      "epoch": 7.133243606998654,
+      "grad_norm": 0.7345687747001648,
+      "learning_rate": 2.085614391833486e-05,
+      "loss": 0.006355223059654236,
+      "mean_token_accuracy": 0.9985966920852661,
+      "num_tokens": 96010655.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.9512575507164002,
+      "epoch": 7.1467025572005385,
+      "grad_norm": 0.799483060836792,
+      "learning_rate": 2.0675593498012052e-05,
+      "loss": 0.006509234011173248,
+      "mean_token_accuracy": 0.9983614802360534,
+      "num_tokens": 96190810.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.9446900069713593,
+      "epoch": 7.160161507402423,
+      "grad_norm": 0.7756890654563904,
+      "learning_rate": 2.049562406273159e-05,
+      "loss": 0.0069117918610572815,
+      "mean_token_accuracy": 0.9983683228492737,
+      "num_tokens": 96371920.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.9443341970443726,
+      "epoch": 7.173620457604307,
+      "grad_norm": 0.7561301589012146,
+      "learning_rate": 2.0316239178108688e-05,
+      "loss": 0.006773699820041656,
+      "mean_token_accuracy": 0.9981043338775635,
+      "num_tokens": 96553406.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.9383498072624207,
+      "epoch": 7.187079407806191,
+      "grad_norm": 0.7623066306114197,
+      "learning_rate": 2.013744239817725e-05,
+      "loss": 0.006338060647249222,
+      "mean_token_accuracy": 0.9983479022979737,
+      "num_tokens": 96735706.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.9444310605525971,
+      "epoch": 7.2005383580080755,
+      "grad_norm": 0.4759703576564789,
+      "learning_rate": 1.9959237265319452e-05,
+      "loss": 0.006821495294570923,
+      "mean_token_accuracy": 0.9982879221439361,
+      "num_tokens": 96916401.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.9462945520877838,
+      "epoch": 7.21399730820996,
+      "grad_norm": 0.5842452049255371,
+      "learning_rate": 1.9781627310195528e-05,
+      "loss": 0.0067993059754371645,
+      "mean_token_accuracy": 0.9981901407241821,
+      "num_tokens": 97097081.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.9475398421287536,
+      "epoch": 7.227456258411844,
+      "grad_norm": 0.8197925686836243,
+      "learning_rate": 1.960461605167389e-05,
+      "loss": 0.006510883569717407,
+      "mean_token_accuracy": 0.9984028697013855,
+      "num_tokens": 97277934.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.941127610206604,
+      "epoch": 7.240915208613728,
+      "grad_norm": 0.37160298228263855,
+      "learning_rate": 1.9428206996761357e-05,
+      "loss": 0.006270279735326767,
+      "mean_token_accuracy": 0.9982465505599976,
+      "num_tokens": 97459337.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.9421460628509521,
+      "epoch": 7.2543741588156125,
+      "grad_norm": 0.545837938785553,
+      "learning_rate": 1.9252403640533678e-05,
+      "loss": 0.006034504622220993,
+      "mean_token_accuracy": 0.9986165881156921,
+      "num_tokens": 97640387.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.9390841007232666,
+      "epoch": 7.267833109017497,
+      "grad_norm": 0.7546548247337341,
+      "learning_rate": 1.9077209466066315e-05,
+      "loss": 0.007177229970693588,
+      "mean_token_accuracy": 0.9978139519691467,
+      "num_tokens": 97821851.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.9392310976982117,
+      "epoch": 7.281292059219381,
+      "grad_norm": 0.8076840043067932,
+      "learning_rate": 1.8902627944365396e-05,
+      "loss": 0.006884847581386566,
+      "mean_token_accuracy": 0.9983299553394318,
+      "num_tokens": 98003599.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.9376516520977021,
+      "epoch": 7.294751009421265,
+      "grad_norm": 0.6652392148971558,
+      "learning_rate": 1.872866253429897e-05,
+      "loss": 0.005731315910816192,
+      "mean_token_accuracy": 0.9984779059886932,
+      "num_tokens": 98184942.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.9412936687469482,
+      "epoch": 7.30820995962315,
+      "grad_norm": 0.7955637574195862,
+      "learning_rate": 1.8555316682528485e-05,
+      "loss": 0.0065481230616569516,
+      "mean_token_accuracy": 0.9982032060623169,
+      "num_tokens": 98366577.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.9405987918376922,
+      "epoch": 7.321668909825034,
+      "grad_norm": 0.89079350233078,
+      "learning_rate": 1.8382593823440474e-05,
+      "loss": 0.006681758165359497,
+      "mean_token_accuracy": 0.9983046293258667,
+      "num_tokens": 98547681.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.944351601600647,
+      "epoch": 7.335127860026918,
+      "grad_norm": 1.1325428485870361,
+      "learning_rate": 1.821049737907854e-05,
+      "loss": 0.006995528936386108,
+      "mean_token_accuracy": 0.9982137680053711,
+      "num_tokens": 98728624.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.9397885262966156,
+      "epoch": 7.348586810228802,
+      "grad_norm": 0.7245733737945557,
+      "learning_rate": 1.803903075907554e-05,
+      "loss": 0.007640085369348526,
+      "mean_token_accuracy": 0.9979041159152985,
+      "num_tokens": 98910132.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.9460044980049134,
+      "epoch": 7.362045760430687,
+      "grad_norm": 0.9225216507911682,
+      "learning_rate": 1.7868197360586054e-05,
+      "loss": 0.006461286544799804,
+      "mean_token_accuracy": 0.9983667492866516,
+      "num_tokens": 99091179.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.9402543127536773,
+      "epoch": 7.375504710632571,
+      "grad_norm": 0.7389097213745117,
+      "learning_rate": 1.769800056821904e-05,
+      "loss": 0.0065854907035827635,
+      "mean_token_accuracy": 0.9979489326477051,
+      "num_tokens": 99273068.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.9478578507900238,
+      "epoch": 7.388963660834455,
+      "grad_norm": 1.0250836610794067,
+      "learning_rate": 1.7528443753970814e-05,
+      "loss": 0.006443935632705689,
+      "mean_token_accuracy": 0.998309975862503,
+      "num_tokens": 99453517.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.9427528262138367,
+      "epoch": 7.402422611036339,
+      "grad_norm": 0.7677002549171448,
+      "learning_rate": 1.7359530277158227e-05,
+      "loss": 0.006590449810028076,
+      "mean_token_accuracy": 0.9982711434364319,
+      "num_tokens": 99634833.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.9481732070446014,
+      "epoch": 7.415881561238224,
+      "grad_norm": 0.9008362889289856,
+      "learning_rate": 1.7191263484352106e-05,
+      "loss": 0.006611865013837814,
+      "mean_token_accuracy": 0.9980513393878937,
+      "num_tokens": 99815381.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.9450771987438202,
+      "epoch": 7.429340511440108,
+      "grad_norm": 0.6731723546981812,
+      "learning_rate": 1.702364670931097e-05,
+      "loss": 0.006989569962024688,
+      "mean_token_accuracy": 0.9981782495975494,
+      "num_tokens": 99996758.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.9488214910030365,
+      "epoch": 7.442799461641992,
+      "grad_norm": 0.543502151966095,
+      "learning_rate": 1.685668327291495e-05,
+      "loss": 0.0056597858667373655,
+      "mean_token_accuracy": 0.9984924256801605,
+      "num_tokens": 100177864.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.9521346092224121,
+      "epoch": 7.456258411843876,
+      "grad_norm": 0.6044306755065918,
+      "learning_rate": 1.6690376483100034e-05,
+      "loss": 0.006509961187839508,
+      "mean_token_accuracy": 0.9981269180774689,
+      "num_tokens": 100358496.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.9487424850463867,
+      "epoch": 7.469717362045761,
+      "grad_norm": 0.8417298197746277,
+      "learning_rate": 1.6524729634792475e-05,
+      "loss": 0.006116047501564026,
+      "mean_token_accuracy": 0.9982541143894196,
+      "num_tokens": 100539331.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.9476044595241546,
+      "epoch": 7.483176312247645,
+      "grad_norm": 0.8056595325469971,
+      "learning_rate": 1.6359746009843567e-05,
+      "loss": 0.0057601254433393475,
+      "mean_token_accuracy": 0.9985706150531769,
+      "num_tokens": 100720450.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.9456092417240143,
+      "epoch": 7.496635262449529,
+      "grad_norm": 0.6036805510520935,
+      "learning_rate": 1.6195428876964592e-05,
+      "loss": 0.005734497308731079,
+      "mean_token_accuracy": 0.9984824895858765,
+      "num_tokens": 100902228.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.946131706237793,
+      "epoch": 7.510094212651413,
+      "grad_norm": 0.7125651240348816,
+      "learning_rate": 1.6031781491662056e-05,
+      "loss": 0.005733318626880646,
+      "mean_token_accuracy": 0.9984204411506653,
+      "num_tokens": 101083443.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.9503833770751953,
+      "epoch": 7.523553162853298,
+      "grad_norm": 0.6275620460510254,
+      "learning_rate": 1.58688070961732e-05,
+      "loss": 0.006019812077283859,
+      "mean_token_accuracy": 0.9983126580715179,
+      "num_tokens": 101264443.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.947618979215622,
+      "epoch": 7.537012113055182,
+      "grad_norm": 0.7690677046775818,
+      "learning_rate": 1.5706508919401804e-05,
+      "loss": 0.005866598710417748,
+      "mean_token_accuracy": 0.9985946953296662,
+      "num_tokens": 101445485.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.9441321969032288,
+      "epoch": 7.550471063257066,
+      "grad_norm": 0.6154682636260986,
+      "learning_rate": 1.554489017685412e-05,
+      "loss": 0.00593491718173027,
+      "mean_token_accuracy": 0.9986464619636536,
+      "num_tokens": 101626547.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.9452862203121185,
+      "epoch": 7.56393001345895,
+      "grad_norm": 0.8111712336540222,
+      "learning_rate": 1.538395407057525e-05,
+      "loss": 0.006593097746372223,
+      "mean_token_accuracy": 0.998135757446289,
+      "num_tokens": 101807733.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.9418992578983307,
+      "epoch": 7.577388963660835,
+      "grad_norm": 0.7640061974525452,
+      "learning_rate": 1.522370378908567e-05,
+      "loss": 0.006091068312525749,
+      "mean_token_accuracy": 0.9984888017177582,
+      "num_tokens": 101989357.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.9476424336433411,
+      "epoch": 7.590847913862719,
+      "grad_norm": 0.5616353750228882,
+      "learning_rate": 1.5064142507318051e-05,
+      "loss": 0.0060265038162469866,
+      "mean_token_accuracy": 0.9982646107673645,
+      "num_tokens": 102170318.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.9492906272411347,
+      "epoch": 7.604306864064603,
+      "grad_norm": 0.6769309043884277,
+      "learning_rate": 1.4905273386554386e-05,
+      "loss": 0.006565354764461517,
+      "mean_token_accuracy": 0.9979823470115662,
+      "num_tokens": 102351383.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.9434434294700622,
+      "epoch": 7.6177658142664875,
+      "grad_norm": 0.8144317865371704,
+      "learning_rate": 1.4747099574363332e-05,
+      "loss": 0.006383077055215835,
+      "mean_token_accuracy": 0.9983625173568725,
+      "num_tokens": 102532483.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.9410477578639984,
+      "epoch": 7.631224764468372,
+      "grad_norm": 0.7184897065162659,
+      "learning_rate": 1.4589624204537866e-05,
+      "loss": 0.00683993250131607,
+      "mean_token_accuracy": 0.9981952488422394,
+      "num_tokens": 102714429.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.9437882840633393,
+      "epoch": 7.644683714670256,
+      "grad_norm": 0.7210987210273743,
+      "learning_rate": 1.4432850397033176e-05,
+      "loss": 0.006337168067693711,
+      "mean_token_accuracy": 0.9981409192085267,
+      "num_tokens": 102895693.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.9423653900623321,
+      "epoch": 7.65814266487214,
+      "grad_norm": 0.7256081700325012,
+      "learning_rate": 1.427678125790487e-05,
+      "loss": 0.006422295421361923,
+      "mean_token_accuracy": 0.9981527864933014,
+      "num_tokens": 103076994.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.9402276873588562,
+      "epoch": 7.6716016150740245,
+      "grad_norm": 0.9724451899528503,
+      "learning_rate": 1.4121419879247422e-05,
+      "loss": 0.006309093534946441,
+      "mean_token_accuracy": 0.9982479989528656,
+      "num_tokens": 103258251.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.9454701244831085,
+      "epoch": 7.685060565275909,
+      "grad_norm": 0.40232348442077637,
+      "learning_rate": 1.3966769339132929e-05,
+      "loss": 0.005939274653792381,
+      "mean_token_accuracy": 0.998301374912262,
+      "num_tokens": 103438976.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.9492968201637269,
+      "epoch": 7.698519515477793,
+      "grad_norm": 0.598109781742096,
+      "learning_rate": 1.3812832701550099e-05,
+      "loss": 0.00723060742020607,
+      "mean_token_accuracy": 0.9979724586009979,
+      "num_tokens": 103619841.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.9433623433113099,
+      "epoch": 7.711978465679677,
+      "grad_norm": 0.7593542337417603,
+      "learning_rate": 1.365961301634357e-05,
+      "loss": 0.0071952566504478455,
+      "mean_token_accuracy": 0.9981424927711486,
+      "num_tokens": 103801692.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.9490162670612335,
+      "epoch": 7.7254374158815615,
+      "grad_norm": 0.6364531517028809,
+      "learning_rate": 1.350711331915348e-05,
+      "loss": 0.006192450225353241,
+      "mean_token_accuracy": 0.9984559118747711,
+      "num_tokens": 103982367.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.942770904302597,
+      "epoch": 7.738896366083446,
+      "grad_norm": 0.9279600977897644,
+      "learning_rate": 1.3355336631355314e-05,
+      "loss": 0.005739518254995346,
+      "mean_token_accuracy": 0.9985584795475007,
+      "num_tokens": 104164149.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.9427549839019775,
+      "epoch": 7.75235531628533,
+      "grad_norm": 0.9408928155899048,
+      "learning_rate": 1.3204285960000051e-05,
+      "loss": 0.005749309062957763,
+      "mean_token_accuracy": 0.9985000133514405,
+      "num_tokens": 104345175.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.9404960751533509,
+      "epoch": 7.765814266487214,
+      "grad_norm": 0.7260274291038513,
+      "learning_rate": 1.30539642977546e-05,
+      "loss": 0.006429329514503479,
+      "mean_token_accuracy": 0.9984842240810394,
+      "num_tokens": 104526928.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.9448217689990998,
+      "epoch": 7.7792732166890985,
+      "grad_norm": 0.6738228797912598,
+      "learning_rate": 1.2904374622842486e-05,
+      "loss": 0.005977935343980789,
+      "mean_token_accuracy": 0.9985457420349121,
+      "num_tokens": 104708230.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.9451113224029541,
+      "epoch": 7.792732166890983,
+      "grad_norm": 0.9410247802734375,
+      "learning_rate": 1.2755519898984846e-05,
+      "loss": 0.005663443356752396,
+      "mean_token_accuracy": 0.9984429180622101,
+      "num_tokens": 104888647.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.9483637988567353,
+      "epoch": 7.806191117092867,
+      "grad_norm": 0.5315118432044983,
+      "learning_rate": 1.2607403075341728e-05,
+      "loss": 0.005260968953371048,
+      "mean_token_accuracy": 0.9988214015960694,
+      "num_tokens": 105068604.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.9461100101470947,
+      "epoch": 7.819650067294751,
+      "grad_norm": 0.5080281496047974,
+      "learning_rate": 1.2460027086453658e-05,
+      "loss": 0.005320623517036438,
+      "mean_token_accuracy": 0.9986594915390015,
+      "num_tokens": 105249023.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.9451904594898224,
+      "epoch": 7.833109017496636,
+      "grad_norm": 0.6318531632423401,
+      "learning_rate": 1.2313394852183473e-05,
+      "loss": 0.005546718463301658,
+      "mean_token_accuracy": 0.9985934972763062,
+      "num_tokens": 105429833.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.9423372805118561,
+      "epoch": 7.84656796769852,
+      "grad_norm": 0.6764518022537231,
+      "learning_rate": 1.2167509277658518e-05,
+      "loss": 0.005748329311609268,
+      "mean_token_accuracy": 0.9986869513988494,
+      "num_tokens": 105611004.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.9406632006168365,
+      "epoch": 7.860026917900404,
+      "grad_norm": 0.6841229796409607,
+      "learning_rate": 1.2022373253213049e-05,
+      "loss": 0.005423003435134887,
+      "mean_token_accuracy": 0.9984905958175659,
+      "num_tokens": 105791366.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.9383720457553864,
+      "epoch": 7.873485868102288,
+      "grad_norm": 0.7743740677833557,
+      "learning_rate": 1.1877989654330973e-05,
+      "loss": 0.005470183119177819,
+      "mean_token_accuracy": 0.9987134814262391,
+      "num_tokens": 105973009.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.9405181050300598,
+      "epoch": 7.886944818304173,
+      "grad_norm": 0.6537275314331055,
+      "learning_rate": 1.1734361341588912e-05,
+      "loss": 0.005447608605027199,
+      "mean_token_accuracy": 0.9983970046043396,
+      "num_tokens": 106153797.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.935320508480072,
+      "epoch": 7.900403768506057,
+      "grad_norm": 0.8727037310600281,
+      "learning_rate": 1.1591491160599488e-05,
+      "loss": 0.004841895401477813,
+      "mean_token_accuracy": 0.9990314602851867,
+      "num_tokens": 106335272.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.9389428675174714,
+      "epoch": 7.913862718707941,
+      "grad_norm": 0.8051491379737854,
+      "learning_rate": 1.1449381941954956e-05,
+      "loss": 0.005671492218971253,
+      "mean_token_accuracy": 0.9985178470611572,
+      "num_tokens": 106516472.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.9336957454681396,
+      "epoch": 7.927321668909825,
+      "grad_norm": 0.6917922496795654,
+      "learning_rate": 1.130803650117116e-05,
+      "loss": 0.005551187694072724,
+      "mean_token_accuracy": 0.9983591079711914,
+      "num_tokens": 106697785.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.9350760340690613,
+      "epoch": 7.94078061911171,
+      "grad_norm": 0.7642367482185364,
+      "learning_rate": 1.1167457638631656e-05,
+      "loss": 0.005567619949579239,
+      "mean_token_accuracy": 0.9985032379627228,
+      "num_tokens": 106879069.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.9362707912921906,
+      "epoch": 7.954239569313594,
+      "grad_norm": 0.6716526746749878,
+      "learning_rate": 1.1027648139532355e-05,
+      "loss": 0.005269525200128555,
+      "mean_token_accuracy": 0.9987157702445983,
+      "num_tokens": 107060235.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.9299384176731109,
+      "epoch": 7.967698519515478,
+      "grad_norm": 0.6839614510536194,
+      "learning_rate": 1.0888610773826253e-05,
+      "loss": 0.00708087682723999,
+      "mean_token_accuracy": 0.9982430160045623,
+      "num_tokens": 107242113.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.9352255225181579,
+      "epoch": 7.981157469717362,
+      "grad_norm": 1.0992971658706665,
+      "learning_rate": 1.0750348296168583e-05,
+      "loss": 0.005663050711154938,
+      "mean_token_accuracy": 0.9984843015670777,
+      "num_tokens": 107423370.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.9345376133918762,
+      "epoch": 7.994616419919247,
+      "grad_norm": 0.771766722202301,
+      "learning_rate": 1.0612863445862236e-05,
+      "loss": 0.006203063577413559,
+      "mean_token_accuracy": 0.9982724368572236,
+      "num_tokens": 107604796.0,
+      "step": 5940
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.9459010544855884,
+      "eval_loss": 0.18965470790863037,
+      "eval_mean_token_accuracy": 0.9512642967473169,
+      "eval_num_tokens": 107677508.0,
+      "eval_runtime": 12.6472,
+      "eval_samples_per_second": 395.344,
+      "eval_steps_per_second": 12.414,
+      "step": 5944
+    },
+    {
+      "entropy": 0.9325907230377197,
+      "epoch": 8.00807537012113,
+      "grad_norm": 0.40456992387771606,
+      "learning_rate": 1.0476158946803482e-05,
+      "loss": 0.0041702061891555784,
+      "mean_token_accuracy": 0.9988861680030823,
+      "num_tokens": 107786362.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.9357569575309753,
+      "epoch": 8.021534320323015,
+      "grad_norm": 0.5090038180351257,
+      "learning_rate": 1.0340237507428009e-05,
+      "loss": 0.0035715073347091673,
+      "mean_token_accuracy": 0.9994814991950989,
+      "num_tokens": 107967199.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.9353849172592164,
+      "epoch": 8.034993270524899,
+      "grad_norm": 0.47324568033218384,
+      "learning_rate": 1.020510182065727e-05,
+      "loss": 0.0035488776862621307,
+      "mean_token_accuracy": 0.9993434071540832,
+      "num_tokens": 108147567.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.9339651763439178,
+      "epoch": 8.048452220726784,
+      "grad_norm": 0.4190789461135864,
+      "learning_rate": 1.0070754563845114e-05,
+      "loss": 0.003430064395070076,
+      "mean_token_accuracy": 0.9994158089160919,
+      "num_tokens": 108328451.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.9365652620792388,
+      "epoch": 8.061911170928667,
+      "grad_norm": 0.37395232915878296,
+      "learning_rate": 9.937198398724751e-06,
+      "loss": 0.0032767456024885176,
+      "mean_token_accuracy": 0.9995010256767273,
+      "num_tokens": 108509228.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.9290562212467194,
+      "epoch": 8.075370121130552,
+      "grad_norm": 0.2843457758426666,
+      "learning_rate": 9.804435971356025e-06,
+      "loss": 0.0030744092538952826,
+      "mean_token_accuracy": 0.9995416045188904,
+      "num_tokens": 108691291.0,
+      "step": 6000
+    },
+    {
+      "entropy": 0.9333357572555542,
+      "epoch": 8.088829071332436,
+      "grad_norm": 0.36153435707092285,
+      "learning_rate": 9.672469912072956e-06,
+      "loss": 0.0031955156475305556,
+      "mean_token_accuracy": 0.9994354546070099,
+      "num_tokens": 108872687.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.9334155559539795,
+      "epoch": 8.10228802153432,
+      "grad_norm": 0.35333436727523804,
+      "learning_rate": 9.541302835431681e-06,
+      "loss": 0.0032100573182106016,
+      "mean_token_accuracy": 0.9995277643203735,
+      "num_tokens": 109053638.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.9349715650081635,
+      "epoch": 8.115746971736204,
+      "grad_norm": 0.493518590927124,
+      "learning_rate": 9.410937340158599e-06,
+      "loss": 0.003197438269853592,
+      "mean_token_accuracy": 0.9993844270706177,
+      "num_tokens": 109234361.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.9317407608032227,
+      "epoch": 8.12920592193809,
+      "grad_norm": 0.32104817032814026,
+      "learning_rate": 9.28137600909892e-06,
+      "loss": 0.003348514437675476,
+      "mean_token_accuracy": 0.9993906795978547,
+      "num_tokens": 109415438.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.9339701652526855,
+      "epoch": 8.142664872139973,
+      "grad_norm": 0.5996560454368591,
+      "learning_rate": 9.15262140916548e-06,
+      "loss": 0.0037448287010192873,
+      "mean_token_accuracy": 0.9991378724575043,
+      "num_tokens": 109596756.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.9327323913574219,
+      "epoch": 8.156123822341858,
+      "grad_norm": 0.7599992752075195,
+      "learning_rate": 9.024676091287881e-06,
+      "loss": 0.003381679952144623,
+      "mean_token_accuracy": 0.999460369348526,
+      "num_tokens": 109778064.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.931190824508667,
+      "epoch": 8.169582772543741,
+      "grad_norm": 0.4155709445476532,
+      "learning_rate": 8.897542590361963e-06,
+      "loss": 0.003494596481323242,
+      "mean_token_accuracy": 0.9995332777500152,
+      "num_tokens": 109959186.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.9299486935138702,
+      "epoch": 8.183041722745626,
+      "grad_norm": 0.45130202174186707,
+      "learning_rate": 8.77122342519956e-06,
+      "loss": 0.003702623024582863,
+      "mean_token_accuracy": 0.9991639256477356,
+      "num_tokens": 110140682.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.9285082340240478,
+      "epoch": 8.19650067294751,
+      "grad_norm": 0.5240904688835144,
+      "learning_rate": 8.64572109847862e-06,
+      "loss": 0.0037362042814493178,
+      "mean_token_accuracy": 0.9991588532924652,
+      "num_tokens": 110322435.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.9306709587574005,
+      "epoch": 8.209959623149395,
+      "grad_norm": 0.4319130480289459,
+      "learning_rate": 8.521038096693635e-06,
+      "loss": 0.0034312527626752854,
+      "mean_token_accuracy": 0.9994248628616333,
+      "num_tokens": 110503926.0,
+      "step": 6100
+    },
+    {
+      "entropy": 0.9292944014072418,
+      "epoch": 8.223418573351278,
+      "grad_norm": 0.4186641275882721,
+      "learning_rate": 8.397176890106317e-06,
+      "loss": 0.003543431684374809,
+      "mean_token_accuracy": 0.9993721604347229,
+      "num_tokens": 110685574.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.9340905666351318,
+      "epoch": 8.236877523553163,
+      "grad_norm": 0.44297507405281067,
+      "learning_rate": 8.274139932696717e-06,
+      "loss": 0.00324522964656353,
+      "mean_token_accuracy": 0.9993358612060547,
+      "num_tokens": 110866354.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.9321817755699158,
+      "epoch": 8.250336473755047,
+      "grad_norm": 0.6701546907424927,
+      "learning_rate": 8.151929662114566e-06,
+      "loss": 0.0037943169474601744,
+      "mean_token_accuracy": 0.9993444323539734,
+      "num_tokens": 111047803.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.9300240755081177,
+      "epoch": 8.263795423956932,
+      "grad_norm": 0.3400649428367615,
+      "learning_rate": 8.030548499631014e-06,
+      "loss": 0.0031486738473176954,
+      "mean_token_accuracy": 0.9994214653968811,
+      "num_tokens": 111229226.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.9348362684249878,
+      "epoch": 8.277254374158815,
+      "grad_norm": 0.3338402509689331,
+      "learning_rate": 7.909998850090628e-06,
+      "loss": 0.003268548846244812,
+      "mean_token_accuracy": 0.9993820369243622,
+      "num_tokens": 111410053.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.9298858225345612,
+      "epoch": 8.2907133243607,
+      "grad_norm": 0.3765404224395752,
+      "learning_rate": 7.790283101863771e-06,
+      "loss": 0.0032513745129108427,
+      "mean_token_accuracy": 0.999443256855011,
+      "num_tokens": 111591318.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.9351325869560242,
+      "epoch": 8.304172274562584,
+      "grad_norm": 0.45491793751716614,
+      "learning_rate": 7.671403626799262e-06,
+      "loss": 0.0032870754599571227,
+      "mean_token_accuracy": 0.9993524074554443,
+      "num_tokens": 111772032.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.9296027898788453,
+      "epoch": 8.317631224764469,
+      "grad_norm": 0.40146195888519287,
+      "learning_rate": 7.553362780177398e-06,
+      "loss": 0.003901958093047142,
+      "mean_token_accuracy": 0.9991503238677979,
+      "num_tokens": 111953168.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.9276497423648834,
+      "epoch": 8.331090174966352,
+      "grad_norm": 0.3894268274307251,
+      "learning_rate": 7.43616290066329e-06,
+      "loss": 0.0032478343695402146,
+      "mean_token_accuracy": 0.9992106795310974,
+      "num_tokens": 112135017.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.9245680391788482,
+      "epoch": 8.344549125168237,
+      "grad_norm": 0.5110934376716614,
+      "learning_rate": 7.3198063102605185e-06,
+      "loss": 0.00324617363512516,
+      "mean_token_accuracy": 0.9995225965976715,
+      "num_tokens": 112317445.0,
+      "step": 6200
+    },
+    {
+      "entropy": 0.9306623697280884,
+      "epoch": 8.35800807537012,
+      "grad_norm": 0.39472904801368713,
+      "learning_rate": 7.204295314265136e-06,
+      "loss": 0.0035538677126169204,
+      "mean_token_accuracy": 0.9991783916950225,
+      "num_tokens": 112498753.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.9369937419891358,
+      "epoch": 8.371467025572006,
+      "grad_norm": 0.6480345129966736,
+      "learning_rate": 7.089632201219997e-06,
+      "loss": 0.0035104259848594667,
+      "mean_token_accuracy": 0.9993222296237946,
+      "num_tokens": 112678638.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.9279407441616059,
+      "epoch": 8.38492597577389,
+      "grad_norm": 0.37613216042518616,
+      "learning_rate": 6.975819242869414e-06,
+      "loss": 0.003352391347289085,
+      "mean_token_accuracy": 0.9994183361530304,
+      "num_tokens": 112860513.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.9334605932235718,
+      "epoch": 8.398384925975774,
+      "grad_norm": 0.5418120622634888,
+      "learning_rate": 6.862858694114138e-06,
+      "loss": 0.003238685056567192,
+      "mean_token_accuracy": 0.9992875456809998,
+      "num_tokens": 113041014.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.9314471304416656,
+      "epoch": 8.411843876177658,
+      "grad_norm": 0.4664161205291748,
+      "learning_rate": 6.750752792966702e-06,
+      "loss": 0.003342866525053978,
+      "mean_token_accuracy": 0.9993779182434082,
+      "num_tokens": 113221507.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.9293332099914551,
+      "epoch": 8.425302826379543,
+      "grad_norm": 0.5400107502937317,
+      "learning_rate": 6.639503760507065e-06,
+      "loss": 0.003294713795185089,
+      "mean_token_accuracy": 0.99941286444664,
+      "num_tokens": 113402195.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.9276987969875335,
+      "epoch": 8.438761776581426,
+      "grad_norm": 0.4761334955692291,
+      "learning_rate": 6.529113800838632e-06,
+      "loss": 0.0031884826719760897,
+      "mean_token_accuracy": 0.9994406461715698,
+      "num_tokens": 113583396.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.9320788323879242,
+      "epoch": 8.452220726783311,
+      "grad_norm": 0.7529016733169556,
+      "learning_rate": 6.41958510104454e-06,
+      "loss": 0.0033160604536533354,
+      "mean_token_accuracy": 0.999411803483963,
+      "num_tokens": 113763905.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.930284959077835,
+      "epoch": 8.465679676985195,
+      "grad_norm": 0.3949340879917145,
+      "learning_rate": 6.31091983114438e-06,
+      "loss": 0.003930982202291489,
+      "mean_token_accuracy": 0.999339497089386,
+      "num_tokens": 113944757.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.92959144115448,
+      "epoch": 8.47913862718708,
+      "grad_norm": 0.5477228164672852,
+      "learning_rate": 6.203120144051161e-06,
+      "loss": 0.0033377643674612047,
+      "mean_token_accuracy": 0.9992568790912628,
+      "num_tokens": 114126403.0,
+      "step": 6300
+    },
+    {
+      "entropy": 0.931204092502594,
+      "epoch": 8.492597577388963,
+      "grad_norm": 0.5682660937309265,
+      "learning_rate": 6.0961881755286775e-06,
+      "loss": 0.0033181462436914443,
+      "mean_token_accuracy": 0.9993676483631134,
+      "num_tokens": 114307860.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.9299099445343018,
+      "epoch": 8.506056527590848,
+      "grad_norm": 0.48706290125846863,
+      "learning_rate": 5.990126044149203e-06,
+      "loss": 0.0034614600241184234,
+      "mean_token_accuracy": 0.9993451237678528,
+      "num_tokens": 114489346.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.9328338205814362,
+      "epoch": 8.519515477792732,
+      "grad_norm": 0.502772331237793,
+      "learning_rate": 5.884935851251483e-06,
+      "loss": 0.0033218566328287124,
+      "mean_token_accuracy": 0.9993848502635956,
+      "num_tokens": 114670149.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.9331273257732391,
+      "epoch": 8.532974427994617,
+      "grad_norm": 0.3274834156036377,
+      "learning_rate": 5.780619680899146e-06,
+      "loss": 0.0031801596283912658,
+      "mean_token_accuracy": 0.9995054185390473,
+      "num_tokens": 114851185.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.9312391161918641,
+      "epoch": 8.5464333781965,
+      "grad_norm": 0.3442527651786804,
+      "learning_rate": 5.677179599839366e-06,
+      "loss": 0.0031024085357785223,
+      "mean_token_accuracy": 0.999411153793335,
+      "num_tokens": 115032612.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.9326258182525635,
+      "epoch": 8.559892328398385,
+      "grad_norm": 0.510369598865509,
+      "learning_rate": 5.574617657461967e-06,
+      "loss": 0.0038056544959545135,
+      "mean_token_accuracy": 0.9992052555084229,
+      "num_tokens": 115214161.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.9323939800262451,
+      "epoch": 8.573351278600269,
+      "grad_norm": 0.4948248267173767,
+      "learning_rate": 5.472935885758767e-06,
+      "loss": 0.0032266426831483843,
+      "mean_token_accuracy": 0.9994330883026123,
+      "num_tokens": 115395414.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.9314477443695068,
+      "epoch": 8.586810228802154,
+      "grad_norm": 0.5037770867347717,
+      "learning_rate": 5.372136299283365e-06,
+      "loss": 0.0034978736191987993,
+      "mean_token_accuracy": 0.9991765201091767,
+      "num_tokens": 115576607.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.9309112429618835,
+      "epoch": 8.600269179004037,
+      "grad_norm": 0.3841687738895416,
+      "learning_rate": 5.272220895111196e-06,
+      "loss": 0.0030230386182665825,
+      "mean_token_accuracy": 0.9993924856185913,
+      "num_tokens": 115757822.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.9289616286754608,
+      "epoch": 8.613728129205922,
+      "grad_norm": 0.5192815661430359,
+      "learning_rate": 5.173191652799986e-06,
+      "loss": 0.003417670726776123,
+      "mean_token_accuracy": 0.9993391752243042,
+      "num_tokens": 115939024.0,
+      "step": 6400
+    },
+    {
+      "entropy": 0.9319016218185425,
+      "epoch": 8.627187079407806,
+      "grad_norm": 0.4927634596824646,
+      "learning_rate": 5.075050534350512e-06,
+      "loss": 0.003317685052752495,
+      "mean_token_accuracy": 0.9993408203125,
+      "num_tokens": 116120013.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.932750004529953,
+      "epoch": 8.640646029609691,
+      "grad_norm": 0.5494413375854492,
+      "learning_rate": 4.977799484167767e-06,
+      "loss": 0.003227224573493004,
+      "mean_token_accuracy": 0.9993854761123657,
+      "num_tokens": 116300658.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.9317887723445892,
+      "epoch": 8.654104979811574,
+      "grad_norm": 0.6813295483589172,
+      "learning_rate": 4.8814404290223655e-06,
+      "loss": 0.002960843779146671,
+      "mean_token_accuracy": 0.9993847787380219,
+      "num_tokens": 116481203.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.9327813625335694,
+      "epoch": 8.66756393001346,
+      "grad_norm": 0.3762431740760803,
+      "learning_rate": 4.7859752780124555e-06,
+      "loss": 0.0030213048681616782,
+      "mean_token_accuracy": 0.9996239364147186,
+      "num_tokens": 116662086.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.9316872417926788,
+      "epoch": 8.681022880215343,
+      "grad_norm": 0.5861575603485107,
+      "learning_rate": 4.691405922525838e-06,
+      "loss": 0.003052959218621254,
+      "mean_token_accuracy": 0.9994400024414063,
+      "num_tokens": 116843008.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.9310117483139038,
+      "epoch": 8.694481830417228,
+      "grad_norm": 0.5606390833854675,
+      "learning_rate": 4.597734236202511e-06,
+      "loss": 0.0031916122883558274,
+      "mean_token_accuracy": 0.9994355738162994,
+      "num_tokens": 117024241.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.9291791677474975,
+      "epoch": 8.707940780619111,
+      "grad_norm": 0.5419787764549255,
+      "learning_rate": 4.504962074897556e-06,
+      "loss": 0.00293008740991354,
+      "mean_token_accuracy": 0.9994637906551361,
+      "num_tokens": 117205874.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.929755812883377,
+      "epoch": 8.721399730820997,
+      "grad_norm": 0.5368452668190002,
+      "learning_rate": 4.413091276644349e-06,
+      "loss": 0.0030313016846776008,
+      "mean_token_accuracy": 0.9993925631046295,
+      "num_tokens": 117387081.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.9327210128307343,
+      "epoch": 8.73485868102288,
+      "grad_norm": 0.4368033707141876,
+      "learning_rate": 4.322123661618166e-06,
+      "loss": 0.003084339387714863,
+      "mean_token_accuracy": 0.9995987117290497,
+      "num_tokens": 117567583.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.9297920942306519,
+      "epoch": 8.748317631224765,
+      "grad_norm": 0.5393041372299194,
+      "learning_rate": 4.23206103210011e-06,
+      "loss": 0.0031465519219636916,
+      "mean_token_accuracy": 0.9994846880435944,
+      "num_tokens": 117749088.0,
+      "step": 6500
+    },
+    {
+      "entropy": 0.9288382709026337,
+      "epoch": 8.761776581426648,
+      "grad_norm": 0.3374616205692291,
+      "learning_rate": 4.142905172441408e-06,
+      "loss": 0.0027178578078746797,
+      "mean_token_accuracy": 0.9996258854866028,
+      "num_tokens": 117930285.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.9260033547878266,
+      "epoch": 8.775235531628534,
+      "grad_norm": 0.20111319422721863,
+      "learning_rate": 4.054657849028043e-06,
+      "loss": 0.003253967314958572,
+      "mean_token_accuracy": 0.9993975520133972,
+      "num_tokens": 118112036.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.9301682531833648,
+      "epoch": 8.788694481830417,
+      "grad_norm": 0.4301469326019287,
+      "learning_rate": 3.9673208102457895e-06,
+      "loss": 0.0030012888833880424,
+      "mean_token_accuracy": 0.9995055079460144,
+      "num_tokens": 118293038.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.9303179025650025,
+      "epoch": 8.802153432032302,
+      "grad_norm": 0.3773118853569031,
+      "learning_rate": 3.880895786445543e-06,
+      "loss": 0.003305017575621605,
+      "mean_token_accuracy": 0.9994101107120514,
+      "num_tokens": 118474004.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.9334917306900025,
+      "epoch": 8.815612382234185,
+      "grad_norm": 0.39689475297927856,
+      "learning_rate": 3.795384489909054e-06,
+      "loss": 0.0031069353222846984,
+      "mean_token_accuracy": 0.9994810342788696,
+      "num_tokens": 118654235.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.9243707060813904,
+      "epoch": 8.82907133243607,
+      "grad_norm": 0.3700069785118103,
+      "learning_rate": 3.7107886148150084e-06,
+      "loss": 0.003253418952226639,
+      "mean_token_accuracy": 0.9993774831295014,
+      "num_tokens": 118835998.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.9256916821002961,
+      "epoch": 8.842530282637954,
+      "grad_norm": 0.5005241632461548,
+      "learning_rate": 3.6271098372054334e-06,
+      "loss": 0.003014829009771347,
+      "mean_token_accuracy": 0.9993231058120727,
+      "num_tokens": 119017476.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.929146409034729,
+      "epoch": 8.855989232839839,
+      "grad_norm": 0.4028603434562683,
+      "learning_rate": 3.5443498149525357e-06,
+      "loss": 0.0029433896765112878,
+      "mean_token_accuracy": 0.9995526134967804,
+      "num_tokens": 119198443.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.9259550392627716,
+      "epoch": 8.869448183041722,
+      "grad_norm": 0.46131935715675354,
+      "learning_rate": 3.462510187725809e-06,
+      "loss": 0.0030248038470745086,
+      "mean_token_accuracy": 0.999508935213089,
+      "num_tokens": 119380252.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.9258705079555511,
+      "epoch": 8.882907133243608,
+      "grad_norm": 0.48013147711753845,
+      "learning_rate": 3.3815925769595745e-06,
+      "loss": 0.0031910501420497894,
+      "mean_token_accuracy": 0.9992617785930633,
+      "num_tokens": 119561793.0,
+      "step": 6600
+    },
+    {
+      "entropy": 0.9294628858566284,
+      "epoch": 8.896366083445491,
+      "grad_norm": 0.5725185871124268,
+      "learning_rate": 3.30159858582087e-06,
+      "loss": 0.0035464100539684294,
+      "mean_token_accuracy": 0.999199116230011,
+      "num_tokens": 119742430.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.9263964056968689,
+      "epoch": 8.909825033647376,
+      "grad_norm": 0.4084455966949463,
+      "learning_rate": 3.2225297991776536e-06,
+      "loss": 0.002907513827085495,
+      "mean_token_accuracy": 0.9995316982269287,
+      "num_tokens": 119923690.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.9269788503646851,
+      "epoch": 8.92328398384926,
+      "grad_norm": 0.3295418322086334,
+      "learning_rate": 3.1443877835674208e-06,
+      "loss": 0.002725762873888016,
+      "mean_token_accuracy": 0.9996701359748841,
+      "num_tokens": 120104753.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.929550188779831,
+      "epoch": 8.936742934051145,
+      "grad_norm": 0.5200356841087341,
+      "learning_rate": 3.0671740871661635e-06,
+      "loss": 0.0031205663457512856,
+      "mean_token_accuracy": 0.9993889212608338,
+      "num_tokens": 120285587.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.9283001899719239,
+      "epoch": 8.950201884253028,
+      "grad_norm": 0.4056023061275482,
+      "learning_rate": 2.9908902397577043e-06,
+      "loss": 0.00296444371342659,
+      "mean_token_accuracy": 0.9994133710861206,
+      "num_tokens": 120467044.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.9271095216274261,
+      "epoch": 8.963660834454913,
+      "grad_norm": 0.3392156660556793,
+      "learning_rate": 2.91553775270339e-06,
+      "loss": 0.003089761920273304,
+      "mean_token_accuracy": 0.9994589149951935,
+      "num_tokens": 120648306.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.9254171431064606,
+      "epoch": 8.977119784656796,
+      "grad_norm": 0.4930778741836548,
+      "learning_rate": 2.8411181189121284e-06,
+      "loss": 0.0031807802617549895,
+      "mean_token_accuracy": 0.9994666993618011,
+      "num_tokens": 120829851.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.9281193077564239,
+      "epoch": 8.990578734858682,
+      "grad_norm": 0.476330041885376,
+      "learning_rate": 2.767632812810839e-06,
+      "loss": 0.002921390533447266,
+      "mean_token_accuracy": 0.9995766639709472,
+      "num_tokens": 121011008.0,
+      "step": 6680
+    },
+    {
+      "epoch": 9.0,
+      "eval_entropy": 0.9369368272222531,
+      "eval_loss": 0.20012438297271729,
+      "eval_mean_token_accuracy": 0.9520987811361908,
+      "eval_num_tokens": 121137269.0,
+      "eval_runtime": 12.7296,
+      "eval_samples_per_second": 392.786,
+      "eval_steps_per_second": 12.333,
+      "step": 6687
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.728510286583824e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-6687/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-6687/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-6687/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/checkpoint-743/README.md b/spanish_L20_a100/seed_42/checkpoint-743/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-743/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-743/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-743/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-743/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-743/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-743/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d9def39dff8dfc4044d0260bbcac4e5b295bc4ba
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-743/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0ca13043ebf27367ef7696b297bf67cbbb1b0b470c14359c0c4b51bf94b71f6
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-743/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-743/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-743/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-743/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-743/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-743/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-743/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-743/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-743/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-743/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-743/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..beddb43862f4ab77856188d12aeb6a49fe82f022
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-743/trainer_state.json
@@ -0,0 +1,785 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 743,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.365536320697139e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-743/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-743/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-743/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/checkpoint-7430/README.md b/spanish_L20_a100/seed_42/checkpoint-7430/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-7430/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-7430/adapter_config.json b/spanish_L20_a100/seed_42/checkpoint-7430/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..3bb243291b5c7f9ec1cfe4b01e4aa7c7084d58f5
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-7430/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "v_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100/seed_42/checkpoint-7430/adapter_model.safetensors b/spanish_L20_a100/seed_42/checkpoint-7430/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..702c5fbfebe1da3a268a134ebbf5e1173d4e5ae3
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-7430/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af7e5dfe3f75354a8b3ef971d391984b2997392a3ad823e9de15325f5be34cf6
+size 80792096
diff --git a/spanish_L20_a100/seed_42/checkpoint-7430/chat_template.jinja b/spanish_L20_a100/seed_42/checkpoint-7430/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-7430/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100/seed_42/checkpoint-7430/tokenizer.json b/spanish_L20_a100/seed_42/checkpoint-7430/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-7430/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/checkpoint-7430/tokenizer_config.json b/spanish_L20_a100/seed_42/checkpoint-7430/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-7430/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-7430/trainer_state.json b/spanish_L20_a100/seed_42/checkpoint-7430/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..a27c0ebda42172f0bfd6c72d44993029055e439b
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-7430/trainer_state.json
@@ -0,0 +1,7574 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 7430,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.3158730387687683,
+      "epoch": 0.013458950201884253,
+      "grad_norm": 1.118133783340454,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.39357497692108157,
+      "mean_token_accuracy": 0.8673052072525025,
+      "num_tokens": 180584.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.3121831893920899,
+      "epoch": 0.026917900403768506,
+      "grad_norm": 1.1106795072555542,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.3878319501876831,
+      "mean_token_accuracy": 0.8720601975917817,
+      "num_tokens": 361697.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3157818913459778,
+      "epoch": 0.040376850605652756,
+      "grad_norm": 0.9109562635421753,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.3660596370697021,
+      "mean_token_accuracy": 0.8742272794246674,
+      "num_tokens": 543160.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.3119020223617555,
+      "epoch": 0.05383580080753701,
+      "grad_norm": 1.0371071100234985,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.3286003112792969,
+      "mean_token_accuracy": 0.8857202768325806,
+      "num_tokens": 725185.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.3058258891105652,
+      "epoch": 0.06729475100942127,
+      "grad_norm": 0.9179689288139343,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.2818267822265625,
+      "mean_token_accuracy": 0.8993496894836426,
+      "num_tokens": 906523.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.300613534450531,
+      "epoch": 0.08075370121130551,
+      "grad_norm": 0.9701176881790161,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.2788205146789551,
+      "mean_token_accuracy": 0.8970906794071197,
+      "num_tokens": 1087805.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2840707421302795,
+      "epoch": 0.09421265141318977,
+      "grad_norm": 0.9395180344581604,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.24806475639343262,
+      "mean_token_accuracy": 0.9066538572311401,
+      "num_tokens": 1269198.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2761542201042175,
+      "epoch": 0.10767160161507403,
+      "grad_norm": 1.1721688508987427,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.22089595794677735,
+      "mean_token_accuracy": 0.9154297649860382,
+      "num_tokens": 1450767.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2688742518424987,
+      "epoch": 0.12113055181695828,
+      "grad_norm": 1.1889774799346924,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.21888947486877441,
+      "mean_token_accuracy": 0.916539442539215,
+      "num_tokens": 1632132.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2648243069648744,
+      "epoch": 0.13458950201884254,
+      "grad_norm": 1.1281704902648926,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.21190347671508789,
+      "mean_token_accuracy": 0.9193436801433563,
+      "num_tokens": 1812719.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.2527987241744996,
+      "epoch": 0.1480484522207268,
+      "grad_norm": 1.0904334783554077,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.19499809741973878,
+      "mean_token_accuracy": 0.9249548614025116,
+      "num_tokens": 1994458.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.26150963306427,
+      "epoch": 0.16150740242261102,
+      "grad_norm": 1.3606208562850952,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.2026348352432251,
+      "mean_token_accuracy": 0.9223432123661042,
+      "num_tokens": 2175775.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2535589694976808,
+      "epoch": 0.17496635262449528,
+      "grad_norm": 1.410483956336975,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.1906351327896118,
+      "mean_token_accuracy": 0.9269951701164245,
+      "num_tokens": 2356948.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2413362622261048,
+      "epoch": 0.18842530282637954,
+      "grad_norm": 1.1648170948028564,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.18144721984863282,
+      "mean_token_accuracy": 0.9301130294799804,
+      "num_tokens": 2538027.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.2315245509147643,
+      "epoch": 0.2018842530282638,
+      "grad_norm": 1.4551143646240234,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.17774341106414795,
+      "mean_token_accuracy": 0.9320161044597626,
+      "num_tokens": 2719132.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2315849184989929,
+      "epoch": 0.21534320323014805,
+      "grad_norm": 1.330846905708313,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.17371795177459717,
+      "mean_token_accuracy": 0.9324863314628601,
+      "num_tokens": 2900232.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2322124123573304,
+      "epoch": 0.2288021534320323,
+      "grad_norm": 1.3832001686096191,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.1807727575302124,
+      "mean_token_accuracy": 0.9306116938591004,
+      "num_tokens": 3081545.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.217245590686798,
+      "epoch": 0.24226110363391656,
+      "grad_norm": 1.3935562372207642,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.16498433351516723,
+      "mean_token_accuracy": 0.9357629179954529,
+      "num_tokens": 3263501.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.222991979122162,
+      "epoch": 0.2557200538358008,
+      "grad_norm": 1.4280253648757935,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.16378771066665648,
+      "mean_token_accuracy": 0.9364173293113709,
+      "num_tokens": 3443966.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.224669301509857,
+      "epoch": 0.2691790040376851,
+      "grad_norm": 1.4656109809875488,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.16791722774505616,
+      "mean_token_accuracy": 0.934725958108902,
+      "num_tokens": 3625337.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.2104426741600036,
+      "epoch": 0.28263795423956933,
+      "grad_norm": 1.6048086881637573,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.16221411228179933,
+      "mean_token_accuracy": 0.93584925532341,
+      "num_tokens": 3806470.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.2051853775978087,
+      "epoch": 0.2960969044414536,
+      "grad_norm": 1.2150211334228516,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.15338623523712158,
+      "mean_token_accuracy": 0.9399264812469482,
+      "num_tokens": 3987424.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1891771316528321,
+      "epoch": 0.30955585464333785,
+      "grad_norm": 1.4724675416946411,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.1591094970703125,
+      "mean_token_accuracy": 0.9368935227394104,
+      "num_tokens": 4168428.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.1861667037010193,
+      "epoch": 0.32301480484522205,
+      "grad_norm": 1.2883172035217285,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.15525293350219727,
+      "mean_token_accuracy": 0.9381617724895477,
+      "num_tokens": 4349308.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.1858779191970825,
+      "epoch": 0.3364737550471063,
+      "grad_norm": 1.2819322347640991,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.1584002733230591,
+      "mean_token_accuracy": 0.937253201007843,
+      "num_tokens": 4531293.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1851443648338318,
+      "epoch": 0.34993270524899056,
+      "grad_norm": 1.1528538465499878,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 0.15353044271469116,
+      "mean_token_accuracy": 0.9391701340675354,
+      "num_tokens": 4712762.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.182717990875244,
+      "epoch": 0.3633916554508748,
+      "grad_norm": 1.4254769086837769,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 0.14897005558013915,
+      "mean_token_accuracy": 0.9409211337566376,
+      "num_tokens": 4893612.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1652904391288756,
+      "epoch": 0.3768506056527591,
+      "grad_norm": 1.0718536376953125,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.1478255271911621,
+      "mean_token_accuracy": 0.9414895117282868,
+      "num_tokens": 5075569.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.164769423007965,
+      "epoch": 0.39030955585464333,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 0.1419359803199768,
+      "mean_token_accuracy": 0.9446478188037872,
+      "num_tokens": 5257025.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.1616728186607361,
+      "epoch": 0.4037685060565276,
+      "grad_norm": 1.4268428087234497,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.1495826005935669,
+      "mean_token_accuracy": 0.9422928750514984,
+      "num_tokens": 5438486.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.159751296043396,
+      "epoch": 0.41722745625841184,
+      "grad_norm": 1.5416603088378906,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.14568655490875243,
+      "mean_token_accuracy": 0.9432988226413727,
+      "num_tokens": 5619167.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1518551111221313,
+      "epoch": 0.4306864064602961,
+      "grad_norm": 1.2034111022949219,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 0.14470478296279907,
+      "mean_token_accuracy": 0.9438421249389648,
+      "num_tokens": 5800167.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1474129199981689,
+      "epoch": 0.44414535666218036,
+      "grad_norm": 1.2916200160980225,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 0.14664748907089234,
+      "mean_token_accuracy": 0.9410681962966919,
+      "num_tokens": 5980988.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1333133339881898,
+      "epoch": 0.4576043068640646,
+      "grad_norm": 1.4899367094039917,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 0.14436495304107666,
+      "mean_token_accuracy": 0.9434790909290314,
+      "num_tokens": 6161672.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1283114314079286,
+      "epoch": 0.47106325706594887,
+      "grad_norm": 1.2886645793914795,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 0.14442335367202758,
+      "mean_token_accuracy": 0.9413226485252381,
+      "num_tokens": 6343223.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.1332532525062562,
+      "epoch": 0.4845222072678331,
+      "grad_norm": 1.8172247409820557,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 0.14870083332061768,
+      "mean_token_accuracy": 0.9412250339984893,
+      "num_tokens": 6523859.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1274277448654175,
+      "epoch": 0.4979811574697174,
+      "grad_norm": 1.7064348459243774,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 0.15062304735183715,
+      "mean_token_accuracy": 0.9394510447978973,
+      "num_tokens": 6705578.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.1206182718276978,
+      "epoch": 0.5114401076716016,
+      "grad_norm": 1.2750648260116577,
+      "learning_rate": 9.999975729865971e-05,
+      "loss": 0.14098584651947021,
+      "mean_token_accuracy": 0.9451459109783172,
+      "num_tokens": 6887129.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1138589024543761,
+      "epoch": 0.5248990578734859,
+      "grad_norm": 1.4869309663772583,
+      "learning_rate": 9.999856856307314e-05,
+      "loss": 0.1419318675994873,
+      "mean_token_accuracy": 0.9448270738124848,
+      "num_tokens": 7068169.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.0940229773521424,
+      "epoch": 0.5383580080753702,
+      "grad_norm": 1.3280725479125977,
+      "learning_rate": 9.999638923896533e-05,
+      "loss": 0.1438223123550415,
+      "mean_token_accuracy": 0.9430389642715454,
+      "num_tokens": 7249540.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.0885985612869262,
+      "epoch": 0.5518169582772544,
+      "grad_norm": 1.4446754455566406,
+      "learning_rate": 9.999321936951374e-05,
+      "loss": 0.1442363977432251,
+      "mean_token_accuracy": 0.940515261888504,
+      "num_tokens": 7429589.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.0717393517494203,
+      "epoch": 0.5652759084791387,
+      "grad_norm": 1.3230535984039307,
+      "learning_rate": 9.998905901752091e-05,
+      "loss": 0.14062190055847168,
+      "mean_token_accuracy": 0.944873696565628,
+      "num_tokens": 7611183.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.0707338809967042,
+      "epoch": 0.5787348586810229,
+      "grad_norm": 1.4193118810653687,
+      "learning_rate": 9.998390826541315e-05,
+      "loss": 0.13488584756851196,
+      "mean_token_accuracy": 0.9468543648719787,
+      "num_tokens": 7792490.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.076532781124115,
+      "epoch": 0.5921938088829072,
+      "grad_norm": 1.4312297105789185,
+      "learning_rate": 9.997776721523888e-05,
+      "loss": 0.13298068046569825,
+      "mean_token_accuracy": 0.9474191963672638,
+      "num_tokens": 7973319.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.0722519278526306,
+      "epoch": 0.6056527590847914,
+      "grad_norm": 1.3930675983428955,
+      "learning_rate": 9.99706359886667e-05,
+      "loss": 0.1368749260902405,
+      "mean_token_accuracy": 0.9454025149345398,
+      "num_tokens": 8154292.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.089442777633667,
+      "epoch": 0.6191117092866757,
+      "grad_norm": 1.4437086582183838,
+      "learning_rate": 9.996251472698281e-05,
+      "loss": 0.13591750860214233,
+      "mean_token_accuracy": 0.9460393130779267,
+      "num_tokens": 8334902.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.0922252893447877,
+      "epoch": 0.6325706594885598,
+      "grad_norm": 1.3557511568069458,
+      "learning_rate": 9.995340359108844e-05,
+      "loss": 0.13737888336181642,
+      "mean_token_accuracy": 0.9453330874443054,
+      "num_tokens": 8515896.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.0872651576995849,
+      "epoch": 0.6460296096904441,
+      "grad_norm": 1.2690744400024414,
+      "learning_rate": 9.994330276149649e-05,
+      "loss": 0.13499363660812377,
+      "mean_token_accuracy": 0.9465942263603211,
+      "num_tokens": 8696576.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.0870140314102172,
+      "epoch": 0.6594885598923284,
+      "grad_norm": 1.204150915145874,
+      "learning_rate": 9.993221243832797e-05,
+      "loss": 0.13191089630126954,
+      "mean_token_accuracy": 0.9466091632843018,
+      "num_tokens": 8878911.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.0668653488159179,
+      "epoch": 0.6729475100942126,
+      "grad_norm": 1.0860817432403564,
+      "learning_rate": 9.992013284130816e-05,
+      "loss": 0.1315389394760132,
+      "mean_token_accuracy": 0.9467469096183777,
+      "num_tokens": 9060738.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.076583206653595,
+      "epoch": 0.6864064602960969,
+      "grad_norm": 1.3365256786346436,
+      "learning_rate": 9.990706420976206e-05,
+      "loss": 0.13684618473052979,
+      "mean_token_accuracy": 0.9469292283058166,
+      "num_tokens": 9241241.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.0677453875541687,
+      "epoch": 0.6998654104979811,
+      "grad_norm": 1.3905078172683716,
+      "learning_rate": 9.989300680260985e-05,
+      "loss": 0.13291587829589843,
+      "mean_token_accuracy": 0.9476063907146454,
+      "num_tokens": 9422765.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0732616305351257,
+      "epoch": 0.7133243606998654,
+      "grad_norm": 1.5653958320617676,
+      "learning_rate": 9.98779608983616e-05,
+      "loss": 0.13485829830169677,
+      "mean_token_accuracy": 0.9460451543331146,
+      "num_tokens": 9604372.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.0796746015548706,
+      "epoch": 0.7267833109017496,
+      "grad_norm": 1.067662000656128,
+      "learning_rate": 9.986192679511189e-05,
+      "loss": 0.1378292918205261,
+      "mean_token_accuracy": 0.9459867894649505,
+      "num_tokens": 9784879.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0722593426704408,
+      "epoch": 0.7402422611036339,
+      "grad_norm": 1.4049694538116455,
+      "learning_rate": 9.984490481053372e-05,
+      "loss": 0.1350596070289612,
+      "mean_token_accuracy": 0.9467743813991547,
+      "num_tokens": 9965547.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0819878458976746,
+      "epoch": 0.7537012113055181,
+      "grad_norm": 1.2730485200881958,
+      "learning_rate": 9.982689528187244e-05,
+      "loss": 0.13021925687789918,
+      "mean_token_accuracy": 0.9474351227283477,
+      "num_tokens": 10146381.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.052109134197235,
+      "epoch": 0.7671601615074024,
+      "grad_norm": 1.491249918937683,
+      "learning_rate": 9.98078985659389e-05,
+      "loss": 0.1340065121650696,
+      "mean_token_accuracy": 0.9467333614826202,
+      "num_tokens": 10328316.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.064994204044342,
+      "epoch": 0.7806191117092867,
+      "grad_norm": 1.5963112115859985,
+      "learning_rate": 9.978791503910246e-05,
+      "loss": 0.13630943298339843,
+      "mean_token_accuracy": 0.9467737913131714,
+      "num_tokens": 10508389.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.059787142276764,
+      "epoch": 0.7940780619111709,
+      "grad_norm": 1.3717354536056519,
+      "learning_rate": 9.97669450972835e-05,
+      "loss": 0.12716681957244874,
+      "mean_token_accuracy": 0.9494995296001434,
+      "num_tokens": 10689408.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.0495775938034058,
+      "epoch": 0.8075370121130552,
+      "grad_norm": 1.289700984954834,
+      "learning_rate": 9.974498915594557e-05,
+      "loss": 0.1314472198486328,
+      "mean_token_accuracy": 0.9477905452251434,
+      "num_tokens": 10870096.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0443841576576234,
+      "epoch": 0.8209959623149394,
+      "grad_norm": 1.3570683002471924,
+      "learning_rate": 9.97220476500872e-05,
+      "loss": 0.1300058126449585,
+      "mean_token_accuracy": 0.9484966337680817,
+      "num_tokens": 11051491.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0572444915771484,
+      "epoch": 0.8344549125168237,
+      "grad_norm": 1.2557934522628784,
+      "learning_rate": 9.969812103423325e-05,
+      "loss": 0.12966666221618653,
+      "mean_token_accuracy": 0.9475180447101593,
+      "num_tokens": 11232341.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0466806173324585,
+      "epoch": 0.847913862718708,
+      "grad_norm": 1.159976840019226,
+      "learning_rate": 9.967320978242592e-05,
+      "loss": 0.13168529272079468,
+      "mean_token_accuracy": 0.9476813077926636,
+      "num_tokens": 11413810.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.0480555057525636,
+      "epoch": 0.8613728129205922,
+      "grad_norm": 1.2526041269302368,
+      "learning_rate": 9.964731438821533e-05,
+      "loss": 0.13281782865524291,
+      "mean_token_accuracy": 0.9469167709350585,
+      "num_tokens": 11595017.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0574912667274474,
+      "epoch": 0.8748317631224765,
+      "grad_norm": 1.2427246570587158,
+      "learning_rate": 9.962043536464978e-05,
+      "loss": 0.1324108362197876,
+      "mean_token_accuracy": 0.9479593813419342,
+      "num_tokens": 11775825.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.0491220474243164,
+      "epoch": 0.8882907133243607,
+      "grad_norm": 1.2983356714248657,
+      "learning_rate": 9.959257324426556e-05,
+      "loss": 0.12988500595092772,
+      "mean_token_accuracy": 0.9481373846530914,
+      "num_tokens": 11956910.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0449891567230225,
+      "epoch": 0.901749663526245,
+      "grad_norm": 1.5811021327972412,
+      "learning_rate": 9.95637285790764e-05,
+      "loss": 0.1291733741760254,
+      "mean_token_accuracy": 0.9491346716880799,
+      "num_tokens": 12138027.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0314183115959168,
+      "epoch": 0.9152086137281292,
+      "grad_norm": 1.3054865598678589,
+      "learning_rate": 9.953390194056258e-05,
+      "loss": 0.1284323215484619,
+      "mean_token_accuracy": 0.948948085308075,
+      "num_tokens": 12319160.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.032126522064209,
+      "epoch": 0.9286675639300135,
+      "grad_norm": 1.8568792343139648,
+      "learning_rate": 9.950309391965947e-05,
+      "loss": 0.1263338088989258,
+      "mean_token_accuracy": 0.9493363380432129,
+      "num_tokens": 12500653.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0391790747642518,
+      "epoch": 0.9421265141318977,
+      "grad_norm": 1.2497203350067139,
+      "learning_rate": 9.947130512674602e-05,
+      "loss": 0.13258013725280762,
+      "mean_token_accuracy": 0.9460227429866791,
+      "num_tokens": 12681172.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0239480495452882,
+      "epoch": 0.955585464333782,
+      "grad_norm": 1.4836376905441284,
+      "learning_rate": 9.943853619163255e-05,
+      "loss": 0.13169267177581787,
+      "mean_token_accuracy": 0.9465819537639618,
+      "num_tokens": 12861964.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.032191848754883,
+      "epoch": 0.9690444145356663,
+      "grad_norm": 1.2642196416854858,
+      "learning_rate": 9.94047877635482e-05,
+      "loss": 0.12583615779876708,
+      "mean_token_accuracy": 0.9503573417663574,
+      "num_tokens": 13043409.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.0391651749610902,
+      "epoch": 0.9825033647375505,
+      "grad_norm": 1.4302080869674683,
+      "learning_rate": 9.93700605111283e-05,
+      "loss": 0.12409310340881348,
+      "mean_token_accuracy": 0.9505755603313446,
+      "num_tokens": 13225072.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.0451523542404175,
+      "epoch": 0.9959623149394348,
+      "grad_norm": 1.4081453084945679,
+      "learning_rate": 9.933435512240084e-05,
+      "loss": 0.12939380407333373,
+      "mean_token_accuracy": 0.9479136288166046,
+      "num_tokens": 13405436.0,
+      "step": 740
+    },
+    {
+      "epoch": 1.0,
+      "eval_entropy": 1.0626230862489932,
+      "eval_loss": 0.12674932181835175,
+      "eval_mean_token_accuracy": 0.9496257267180522,
+      "eval_num_tokens": 13459679.0,
+      "eval_runtime": 13.2455,
+      "eval_samples_per_second": 377.486,
+      "eval_steps_per_second": 11.853,
+      "step": 743
+    },
+    {
+      "entropy": 1.049056887626648,
+      "epoch": 1.009421265141319,
+      "grad_norm": 1.4189648628234863,
+      "learning_rate": 9.929767230477305e-05,
+      "loss": 0.11506924629211426,
+      "mean_token_accuracy": 0.9560760498046875,
+      "num_tokens": 13586033.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0126458764076234,
+      "epoch": 1.0228802153432033,
+      "grad_norm": 1.6394226551055908,
+      "learning_rate": 9.92600127850173e-05,
+      "loss": 0.10137178897857665,
+      "mean_token_accuracy": 0.9604915261268616,
+      "num_tokens": 13766883.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0174824237823485,
+      "epoch": 1.0363391655450875,
+      "grad_norm": 1.3608797788619995,
+      "learning_rate": 9.922137730925673e-05,
+      "loss": 0.10521939992904664,
+      "mean_token_accuracy": 0.9592583775520325,
+      "num_tokens": 13949410.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0318764686584472,
+      "epoch": 1.0497981157469718,
+      "grad_norm": 1.1440019607543945,
+      "learning_rate": 9.918176664295041e-05,
+      "loss": 0.1039464831352234,
+      "mean_token_accuracy": 0.9593731820583343,
+      "num_tokens": 14131283.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0228763103485108,
+      "epoch": 1.063257065948856,
+      "grad_norm": 1.609320878982544,
+      "learning_rate": 9.914118157087824e-05,
+      "loss": 0.10670688152313232,
+      "mean_token_accuracy": 0.9587047874927521,
+      "num_tokens": 14312694.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0123905301094056,
+      "epoch": 1.0767160161507403,
+      "grad_norm": 1.4128942489624023,
+      "learning_rate": 9.909962289712538e-05,
+      "loss": 0.10546157360076905,
+      "mean_token_accuracy": 0.9586312890052795,
+      "num_tokens": 14493193.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.010130923986435,
+      "epoch": 1.0901749663526246,
+      "grad_norm": 1.4825445413589478,
+      "learning_rate": 9.905709144506629e-05,
+      "loss": 0.09921822547912598,
+      "mean_token_accuracy": 0.9621170878410339,
+      "num_tokens": 14674762.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0216347932815553,
+      "epoch": 1.1036339165545088,
+      "grad_norm": 1.2503618001937866,
+      "learning_rate": 9.901358805734846e-05,
+      "loss": 0.10619921684265136,
+      "mean_token_accuracy": 0.9584988236427308,
+      "num_tokens": 14856424.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.022511374950409,
+      "epoch": 1.117092866756393,
+      "grad_norm": 1.3209893703460693,
+      "learning_rate": 9.89691135958757e-05,
+      "loss": 0.10425064563751221,
+      "mean_token_accuracy": 0.9595829844474792,
+      "num_tokens": 15037452.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.007770985364914,
+      "epoch": 1.1305518169582773,
+      "grad_norm": 1.4830716848373413,
+      "learning_rate": 9.892366894179105e-05,
+      "loss": 0.10189464092254638,
+      "mean_token_accuracy": 0.9609117627143859,
+      "num_tokens": 15218205.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0228153944015503,
+      "epoch": 1.1440107671601616,
+      "grad_norm": 1.3191078901290894,
+      "learning_rate": 9.887725499545937e-05,
+      "loss": 0.10298972129821778,
+      "mean_token_accuracy": 0.9599695146083832,
+      "num_tokens": 15399322.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0262911200523377,
+      "epoch": 1.1574697173620458,
+      "grad_norm": 1.2050191164016724,
+      "learning_rate": 9.882987267644939e-05,
+      "loss": 0.10532126426696778,
+      "mean_token_accuracy": 0.9590177297592163,
+      "num_tokens": 15581668.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0124290227890014,
+      "epoch": 1.17092866756393,
+      "grad_norm": 1.3545384407043457,
+      "learning_rate": 9.878152292351563e-05,
+      "loss": 0.10171200037002563,
+      "mean_token_accuracy": 0.9587810635566711,
+      "num_tokens": 15762970.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.028192389011383,
+      "epoch": 1.1843876177658144,
+      "grad_norm": 1.55021071434021,
+      "learning_rate": 9.873220669457975e-05,
+      "loss": 0.10711442232131958,
+      "mean_token_accuracy": 0.956982696056366,
+      "num_tokens": 15944164.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0354714155197144,
+      "epoch": 1.1978465679676986,
+      "grad_norm": 1.2573878765106201,
+      "learning_rate": 9.868192496671147e-05,
+      "loss": 0.10722451210021973,
+      "mean_token_accuracy": 0.9568670630455017,
+      "num_tokens": 16125049.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.034978187084198,
+      "epoch": 1.2113055181695827,
+      "grad_norm": 1.4448883533477783,
+      "learning_rate": 9.86306787361094e-05,
+      "loss": 0.10492000579833985,
+      "mean_token_accuracy": 0.9593439042568207,
+      "num_tokens": 16306183.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0321480870246886,
+      "epoch": 1.224764468371467,
+      "grad_norm": 1.4060254096984863,
+      "learning_rate": 9.857846901808117e-05,
+      "loss": 0.1017595648765564,
+      "mean_token_accuracy": 0.9604197800159454,
+      "num_tokens": 16487439.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.03044091463089,
+      "epoch": 1.2382234185733512,
+      "grad_norm": 1.6413203477859497,
+      "learning_rate": 9.852529684702329e-05,
+      "loss": 0.1032058596611023,
+      "mean_token_accuracy": 0.9589122951030731,
+      "num_tokens": 16667826.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0394478559494018,
+      "epoch": 1.2516823687752354,
+      "grad_norm": 1.3867346048355103,
+      "learning_rate": 9.847116327640082e-05,
+      "loss": 0.10721726417541504,
+      "mean_token_accuracy": 0.9581708788871766,
+      "num_tokens": 16848488.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0355630755424499,
+      "epoch": 1.2651413189771197,
+      "grad_norm": 1.3304182291030884,
+      "learning_rate": 9.841606937872632e-05,
+      "loss": 0.10916879177093505,
+      "mean_token_accuracy": 0.9566857099533081,
+      "num_tokens": 17028787.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0394182324409484,
+      "epoch": 1.278600269179004,
+      "grad_norm": 1.4853013753890991,
+      "learning_rate": 9.836001624553869e-05,
+      "loss": 0.10515115261077881,
+      "mean_token_accuracy": 0.9584204435348511,
+      "num_tokens": 17209699.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0405551195144653,
+      "epoch": 1.2920592193808882,
+      "grad_norm": 1.1959068775177002,
+      "learning_rate": 9.830300498738152e-05,
+      "loss": 0.10281100273132324,
+      "mean_token_accuracy": 0.9599630236625671,
+      "num_tokens": 17391070.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0196930646896363,
+      "epoch": 1.3055181695827724,
+      "grad_norm": 1.2734276056289673,
+      "learning_rate": 9.824503673378112e-05,
+      "loss": 0.10447280406951905,
+      "mean_token_accuracy": 0.9587895989418029,
+      "num_tokens": 17572309.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0284824252128602,
+      "epoch": 1.3189771197846567,
+      "grad_norm": 1.1344295740127563,
+      "learning_rate": 9.81861126332241e-05,
+      "loss": 0.11111283302307129,
+      "mean_token_accuracy": 0.9560421764850616,
+      "num_tokens": 17753814.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0308443427085876,
+      "epoch": 1.332436069986541,
+      "grad_norm": 1.0935126543045044,
+      "learning_rate": 9.812623385313461e-05,
+      "loss": 0.10474474430084228,
+      "mean_token_accuracy": 0.9583515226840973,
+      "num_tokens": 17934774.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.0283282995224,
+      "epoch": 1.3458950201884252,
+      "grad_norm": 1.5144932270050049,
+      "learning_rate": 9.806540157985131e-05,
+      "loss": 0.104243803024292,
+      "mean_token_accuracy": 0.9581555545330047,
+      "num_tokens": 18116015.0,
+      "step": 1000
+    },
+    {
+      "entropy": 1.0355387568473815,
+      "epoch": 1.3593539703903095,
+      "grad_norm": 1.276055932044983,
+      "learning_rate": 9.800361701860368e-05,
+      "loss": 0.10444833040237426,
+      "mean_token_accuracy": 0.9587780058383941,
+      "num_tokens": 18296690.0,
+      "step": 1010
+    },
+    {
+      "entropy": 1.0340025186538697,
+      "epoch": 1.3728129205921937,
+      "grad_norm": 1.4408668279647827,
+      "learning_rate": 9.794088139348835e-05,
+      "loss": 0.10817761421203613,
+      "mean_token_accuracy": 0.9573222637176514,
+      "num_tokens": 18477491.0,
+      "step": 1020
+    },
+    {
+      "entropy": 1.0480108737945557,
+      "epoch": 1.386271870794078,
+      "grad_norm": 1.2114640474319458,
+      "learning_rate": 9.787719594744468e-05,
+      "loss": 0.10477652549743652,
+      "mean_token_accuracy": 0.9591076672077179,
+      "num_tokens": 18658366.0,
+      "step": 1030
+    },
+    {
+      "entropy": 1.0303086400032044,
+      "epoch": 1.3997308209959622,
+      "grad_norm": 1.3319413661956787,
+      "learning_rate": 9.781256194223023e-05,
+      "loss": 0.10648446083068848,
+      "mean_token_accuracy": 0.9581128239631653,
+      "num_tokens": 18839989.0,
+      "step": 1040
+    },
+    {
+      "entropy": 1.0405295491218567,
+      "epoch": 1.4131897711978465,
+      "grad_norm": 1.4326776266098022,
+      "learning_rate": 9.774698065839577e-05,
+      "loss": 0.10707225799560546,
+      "mean_token_accuracy": 0.9574429094791412,
+      "num_tokens": 19020331.0,
+      "step": 1050
+    },
+    {
+      "entropy": 1.037307357788086,
+      "epoch": 1.4266487213997308,
+      "grad_norm": 1.2203212976455688,
+      "learning_rate": 9.768045339525979e-05,
+      "loss": 0.10811934471130372,
+      "mean_token_accuracy": 0.9569429934024811,
+      "num_tokens": 19201621.0,
+      "step": 1060
+    },
+    {
+      "entropy": 1.048385226726532,
+      "epoch": 1.440107671601615,
+      "grad_norm": 1.2597253322601318,
+      "learning_rate": 9.76129814708829e-05,
+      "loss": 0.10502862930297852,
+      "mean_token_accuracy": 0.9592942714691162,
+      "num_tokens": 19382819.0,
+      "step": 1070
+    },
+    {
+      "entropy": 1.0431596279144286,
+      "epoch": 1.4535666218034993,
+      "grad_norm": 1.288861632347107,
+      "learning_rate": 9.754456622204167e-05,
+      "loss": 0.1091151475906372,
+      "mean_token_accuracy": 0.9570925652980804,
+      "num_tokens": 19563291.0,
+      "step": 1080
+    },
+    {
+      "entropy": 1.0325645685195923,
+      "epoch": 1.4670255720053835,
+      "grad_norm": 1.1845893859863281,
+      "learning_rate": 9.747520900420209e-05,
+      "loss": 0.10389597415924072,
+      "mean_token_accuracy": 0.9585334539413453,
+      "num_tokens": 19744388.0,
+      "step": 1090
+    },
+    {
+      "entropy": 1.025480568408966,
+      "epoch": 1.4804845222072678,
+      "grad_norm": 1.3404650688171387,
+      "learning_rate": 9.740491119149277e-05,
+      "loss": 0.10605425834655761,
+      "mean_token_accuracy": 0.9583600521087646,
+      "num_tokens": 19925613.0,
+      "step": 1100
+    },
+    {
+      "entropy": 1.020683228969574,
+      "epoch": 1.493943472409152,
+      "grad_norm": 1.5598406791687012,
+      "learning_rate": 9.733367417667773e-05,
+      "loss": 0.10631901025772095,
+      "mean_token_accuracy": 0.9573191046714783,
+      "num_tokens": 20106747.0,
+      "step": 1110
+    },
+    {
+      "entropy": 1.0185366213321685,
+      "epoch": 1.5074024226110363,
+      "grad_norm": 1.3814117908477783,
+      "learning_rate": 9.726149937112873e-05,
+      "loss": 0.1074562907218933,
+      "mean_token_accuracy": 0.9571435689926148,
+      "num_tokens": 20287872.0,
+      "step": 1120
+    },
+    {
+      "entropy": 1.0160030841827392,
+      "epoch": 1.5208613728129206,
+      "grad_norm": 1.5130647420883179,
+      "learning_rate": 9.718838820479743e-05,
+      "loss": 0.10584371089935303,
+      "mean_token_accuracy": 0.9584454715251922,
+      "num_tokens": 20469530.0,
+      "step": 1130
+    },
+    {
+      "entropy": 1.0295212030410767,
+      "epoch": 1.5343203230148048,
+      "grad_norm": 1.137473464012146,
+      "learning_rate": 9.711434212618691e-05,
+      "loss": 0.10609703063964844,
+      "mean_token_accuracy": 0.9590815663337707,
+      "num_tokens": 20650652.0,
+      "step": 1140
+    },
+    {
+      "entropy": 1.0236494064331054,
+      "epoch": 1.547779273216689,
+      "grad_norm": 1.4402672052383423,
+      "learning_rate": 9.703936260232308e-05,
+      "loss": 0.10043272972106934,
+      "mean_token_accuracy": 0.9601598799228668,
+      "num_tokens": 20832844.0,
+      "step": 1150
+    },
+    {
+      "entropy": 1.0228356838226318,
+      "epoch": 1.5612382234185733,
+      "grad_norm": 1.6836261749267578,
+      "learning_rate": 9.696345111872557e-05,
+      "loss": 0.10231537818908691,
+      "mean_token_accuracy": 0.9592978358268738,
+      "num_tokens": 21014766.0,
+      "step": 1160
+    },
+    {
+      "entropy": 1.0204253315925598,
+      "epoch": 1.5746971736204576,
+      "grad_norm": 1.7377065420150757,
+      "learning_rate": 9.688660917937838e-05,
+      "loss": 0.10434420108795166,
+      "mean_token_accuracy": 0.9578610718250274,
+      "num_tokens": 21195755.0,
+      "step": 1170
+    },
+    {
+      "entropy": 1.0251757979393006,
+      "epoch": 1.5881561238223418,
+      "grad_norm": 1.337659239768982,
+      "learning_rate": 9.68088383066999e-05,
+      "loss": 0.10218650102615356,
+      "mean_token_accuracy": 0.9588291764259338,
+      "num_tokens": 21376476.0,
+      "step": 1180
+    },
+    {
+      "entropy": 1.021351706981659,
+      "epoch": 1.601615074024226,
+      "grad_norm": 1.278275728225708,
+      "learning_rate": 9.673014004151292e-05,
+      "loss": 0.10114043951034546,
+      "mean_token_accuracy": 0.9596256852149964,
+      "num_tokens": 21558292.0,
+      "step": 1190
+    },
+    {
+      "entropy": 1.0222898125648499,
+      "epoch": 1.6150740242261103,
+      "grad_norm": 1.1806602478027344,
+      "learning_rate": 9.665051594301407e-05,
+      "loss": 0.10140385627746581,
+      "mean_token_accuracy": 0.9601265788078308,
+      "num_tokens": 21739640.0,
+      "step": 1200
+    },
+    {
+      "entropy": 1.0402090787887572,
+      "epoch": 1.6285329744279946,
+      "grad_norm": 1.2378828525543213,
+      "learning_rate": 9.656996758874284e-05,
+      "loss": 0.10622456073760986,
+      "mean_token_accuracy": 0.9573592960834503,
+      "num_tokens": 21921425.0,
+      "step": 1210
+    },
+    {
+      "entropy": 1.0412330985069276,
+      "epoch": 1.6419919246298789,
+      "grad_norm": 1.3350801467895508,
+      "learning_rate": 9.648849657455044e-05,
+      "loss": 0.10371602773666382,
+      "mean_token_accuracy": 0.9588219523429871,
+      "num_tokens": 22102605.0,
+      "step": 1220
+    },
+    {
+      "entropy": 1.0534109234809876,
+      "epoch": 1.6554508748317631,
+      "grad_norm": 1.4873039722442627,
+      "learning_rate": 9.640610451456811e-05,
+      "loss": 0.10807039737701415,
+      "mean_token_accuracy": 0.9577955365180969,
+      "num_tokens": 22283687.0,
+      "step": 1230
+    },
+    {
+      "entropy": 1.0713044166564942,
+      "epoch": 1.6689098250336474,
+      "grad_norm": 1.1694425344467163,
+      "learning_rate": 9.632279304117517e-05,
+      "loss": 0.11005204916000366,
+      "mean_token_accuracy": 0.9568802535533905,
+      "num_tokens": 22464813.0,
+      "step": 1240
+    },
+    {
+      "entropy": 1.0745116233825684,
+      "epoch": 1.6823687752355316,
+      "grad_norm": 1.5362104177474976,
+      "learning_rate": 9.623856380496664e-05,
+      "loss": 0.10913199186325073,
+      "mean_token_accuracy": 0.955611252784729,
+      "num_tokens": 22646088.0,
+      "step": 1250
+    },
+    {
+      "entropy": 1.0925905466079713,
+      "epoch": 1.695827725437416,
+      "grad_norm": 1.3478912115097046,
+      "learning_rate": 9.615341847472059e-05,
+      "loss": 0.10608880519866944,
+      "mean_token_accuracy": 0.9566390633583068,
+      "num_tokens": 22826677.0,
+      "step": 1260
+    },
+    {
+      "entropy": 1.0899853229522705,
+      "epoch": 1.7092866756393001,
+      "grad_norm": 1.238072395324707,
+      "learning_rate": 9.606735873736505e-05,
+      "loss": 0.10292150974273681,
+      "mean_token_accuracy": 0.9586134016513824,
+      "num_tokens": 23008177.0,
+      "step": 1270
+    },
+    {
+      "entropy": 1.0695149540901183,
+      "epoch": 1.7227456258411844,
+      "grad_norm": 1.541548252105713,
+      "learning_rate": 9.598038629794461e-05,
+      "loss": 0.10405257940292359,
+      "mean_token_accuracy": 0.9580470204353333,
+      "num_tokens": 23190301.0,
+      "step": 1280
+    },
+    {
+      "entropy": 1.0614005208015442,
+      "epoch": 1.7362045760430687,
+      "grad_norm": 1.4365121126174927,
+      "learning_rate": 9.589250287958657e-05,
+      "loss": 0.10439364910125733,
+      "mean_token_accuracy": 0.9584310293197632,
+      "num_tokens": 23371824.0,
+      "step": 1290
+    },
+    {
+      "entropy": 1.065568709373474,
+      "epoch": 1.749663526244953,
+      "grad_norm": 1.2848029136657715,
+      "learning_rate": 9.580371022346693e-05,
+      "loss": 0.104488205909729,
+      "mean_token_accuracy": 0.959163922071457,
+      "num_tokens": 23552849.0,
+      "step": 1300
+    },
+    {
+      "entropy": 1.0750214815139771,
+      "epoch": 1.7631224764468372,
+      "grad_norm": 1.280882716178894,
+      "learning_rate": 9.571401008877572e-05,
+      "loss": 0.10620679855346679,
+      "mean_token_accuracy": 0.9575595021247864,
+      "num_tokens": 23733408.0,
+      "step": 1310
+    },
+    {
+      "entropy": 1.077055287361145,
+      "epoch": 1.7765814266487214,
+      "grad_norm": 1.2904170751571655,
+      "learning_rate": 9.562340425268233e-05,
+      "loss": 0.11054887771606445,
+      "mean_token_accuracy": 0.9566682338714599,
+      "num_tokens": 23914203.0,
+      "step": 1320
+    },
+    {
+      "entropy": 1.081210732460022,
+      "epoch": 1.7900403768506057,
+      "grad_norm": 1.129718542098999,
+      "learning_rate": 9.553189451030019e-05,
+      "loss": 0.10313529968261718,
+      "mean_token_accuracy": 0.9582066476345062,
+      "num_tokens": 24094078.0,
+      "step": 1330
+    },
+    {
+      "entropy": 1.0856151461601258,
+      "epoch": 1.80349932705249,
+      "grad_norm": 1.3682607412338257,
+      "learning_rate": 9.543948267465115e-05,
+      "loss": 0.10383679866790771,
+      "mean_token_accuracy": 0.9592101991176605,
+      "num_tokens": 24274071.0,
+      "step": 1340
+    },
+    {
+      "entropy": 1.0931835532188416,
+      "epoch": 1.8169582772543742,
+      "grad_norm": 1.3282326459884644,
+      "learning_rate": 9.534617057662977e-05,
+      "loss": 0.10675904750823975,
+      "mean_token_accuracy": 0.9580830633640289,
+      "num_tokens": 24454766.0,
+      "step": 1350
+    },
+    {
+      "entropy": 1.0867947936058044,
+      "epoch": 1.8304172274562585,
+      "grad_norm": 1.2584236860275269,
+      "learning_rate": 9.525196006496679e-05,
+      "loss": 0.10147552490234375,
+      "mean_token_accuracy": 0.9604024410247802,
+      "num_tokens": 24635903.0,
+      "step": 1360
+    },
+    {
+      "entropy": 1.0726998805999757,
+      "epoch": 1.8438761776581427,
+      "grad_norm": 1.254036545753479,
+      "learning_rate": 9.515685300619271e-05,
+      "loss": 0.10524414777755738,
+      "mean_token_accuracy": 0.9582872509956359,
+      "num_tokens": 24816787.0,
+      "step": 1370
+    },
+    {
+      "entropy": 1.0692258477210999,
+      "epoch": 1.857335127860027,
+      "grad_norm": 1.41594660282135,
+      "learning_rate": 9.506085128460065e-05,
+      "loss": 0.10340676307678223,
+      "mean_token_accuracy": 0.9590053558349609,
+      "num_tokens": 24998543.0,
+      "step": 1380
+    },
+    {
+      "entropy": 1.0772343635559083,
+      "epoch": 1.8707940780619112,
+      "grad_norm": 1.385327696800232,
+      "learning_rate": 9.496395680220918e-05,
+      "loss": 0.10295225381851196,
+      "mean_token_accuracy": 0.9591428697109222,
+      "num_tokens": 25180059.0,
+      "step": 1390
+    },
+    {
+      "entropy": 1.0874038934707642,
+      "epoch": 1.8842530282637955,
+      "grad_norm": 1.0628604888916016,
+      "learning_rate": 9.486617147872446e-05,
+      "loss": 0.10015511512756348,
+      "mean_token_accuracy": 0.9605867803096771,
+      "num_tokens": 25361290.0,
+      "step": 1400
+    },
+    {
+      "entropy": 1.0702470779418944,
+      "epoch": 1.8977119784656797,
+      "grad_norm": 1.1819324493408203,
+      "learning_rate": 9.476749725150235e-05,
+      "loss": 0.104461407661438,
+      "mean_token_accuracy": 0.9581527650356293,
+      "num_tokens": 25542455.0,
+      "step": 1410
+    },
+    {
+      "entropy": 1.0592689394950867,
+      "epoch": 1.911170928667564,
+      "grad_norm": 1.4695993661880493,
+      "learning_rate": 9.466793607550995e-05,
+      "loss": 0.1024864912033081,
+      "mean_token_accuracy": 0.9591827690601349,
+      "num_tokens": 25723310.0,
+      "step": 1420
+    },
+    {
+      "entropy": 1.0781041979789734,
+      "epoch": 1.9246298788694483,
+      "grad_norm": 1.3313041925430298,
+      "learning_rate": 9.45674899232869e-05,
+      "loss": 0.10552446842193604,
+      "mean_token_accuracy": 0.9576125502586365,
+      "num_tokens": 25904923.0,
+      "step": 1430
+    },
+    {
+      "entropy": 1.0914432287216187,
+      "epoch": 1.9380888290713325,
+      "grad_norm": 1.3578517436981201,
+      "learning_rate": 9.446616078490626e-05,
+      "loss": 0.11063914299011231,
+      "mean_token_accuracy": 0.9563858449459076,
+      "num_tokens": 26086154.0,
+      "step": 1440
+    },
+    {
+      "entropy": 1.0916360139846801,
+      "epoch": 1.9515477792732168,
+      "grad_norm": 1.3493428230285645,
+      "learning_rate": 9.436395066793518e-05,
+      "loss": 0.10228430032730103,
+      "mean_token_accuracy": 0.9597218215465546,
+      "num_tokens": 26268096.0,
+      "step": 1450
+    },
+    {
+      "entropy": 1.0865699172019958,
+      "epoch": 1.965006729475101,
+      "grad_norm": 1.3002992868423462,
+      "learning_rate": 9.426086159739496e-05,
+      "loss": 0.10704560279846191,
+      "mean_token_accuracy": 0.9567803084850312,
+      "num_tokens": 26448598.0,
+      "step": 1460
+    },
+    {
+      "entropy": 1.0907732367515564,
+      "epoch": 1.9784656796769853,
+      "grad_norm": 1.3105698823928833,
+      "learning_rate": 9.415689561572107e-05,
+      "loss": 0.10344587564468384,
+      "mean_token_accuracy": 0.9585539996623993,
+      "num_tokens": 26629572.0,
+      "step": 1470
+    },
+    {
+      "entropy": 1.097494399547577,
+      "epoch": 1.9919246298788695,
+      "grad_norm": 1.5486772060394287,
+      "learning_rate": 9.405205478272267e-05,
+      "loss": 0.10484597682952881,
+      "mean_token_accuracy": 0.9578526735305786,
+      "num_tokens": 26810588.0,
+      "step": 1480
+    },
+    {
+      "epoch": 2.0,
+      "eval_entropy": 1.1259616188182953,
+      "eval_loss": 0.11920737475156784,
+      "eval_mean_token_accuracy": 0.9516198661676638,
+      "eval_num_tokens": 26919371.0,
+      "eval_runtime": 12.6599,
+      "eval_samples_per_second": 394.949,
+      "eval_steps_per_second": 12.401,
+      "step": 1486
+    },
+    {
+      "entropy": 1.1117471694946288,
+      "epoch": 2.005383580080754,
+      "grad_norm": 1.3559718132019043,
+      "learning_rate": 9.394634117554173e-05,
+      "loss": 0.09661188125610351,
+      "mean_token_accuracy": 0.9621427834033967,
+      "num_tokens": 26991831.0,
+      "step": 1490
+    },
+    {
+      "entropy": 1.0618590712547302,
+      "epoch": 2.018842530282638,
+      "grad_norm": 1.5231083631515503,
+      "learning_rate": 9.38397568886119e-05,
+      "loss": 0.07198524475097656,
+      "mean_token_accuracy": 0.9724888324737548,
+      "num_tokens": 27172578.0,
+      "step": 1500
+    },
+    {
+      "entropy": 1.0601553201675415,
+      "epoch": 2.0323014804845223,
+      "grad_norm": 1.0978407859802246,
+      "learning_rate": 9.373230403361712e-05,
+      "loss": 0.07232148051261902,
+      "mean_token_accuracy": 0.9721459746360779,
+      "num_tokens": 27353505.0,
+      "step": 1510
+    },
+    {
+      "entropy": 1.0545680522918701,
+      "epoch": 2.0457604306864066,
+      "grad_norm": 1.375500202178955,
+      "learning_rate": 9.362398473944958e-05,
+      "loss": 0.0737423300743103,
+      "mean_token_accuracy": 0.9723270058631897,
+      "num_tokens": 27534758.0,
+      "step": 1520
+    },
+    {
+      "entropy": 1.0429188370704652,
+      "epoch": 2.059219380888291,
+      "grad_norm": 1.5134543180465698,
+      "learning_rate": 9.35148011521677e-05,
+      "loss": 0.07129207253456116,
+      "mean_token_accuracy": 0.9730912566184997,
+      "num_tokens": 27715692.0,
+      "step": 1530
+    },
+    {
+      "entropy": 1.0608030915260316,
+      "epoch": 2.072678331090175,
+      "grad_norm": 1.424338936805725,
+      "learning_rate": 9.340475543495364e-05,
+      "loss": 0.07388348579406738,
+      "mean_token_accuracy": 0.9727432250976562,
+      "num_tokens": 27896664.0,
+      "step": 1540
+    },
+    {
+      "entropy": 1.0577767491340637,
+      "epoch": 2.0861372812920593,
+      "grad_norm": 1.2193686962127686,
+      "learning_rate": 9.329384976807023e-05,
+      "loss": 0.07269638180732726,
+      "mean_token_accuracy": 0.9726069569587708,
+      "num_tokens": 28077786.0,
+      "step": 1550
+    },
+    {
+      "entropy": 1.0627391576766967,
+      "epoch": 2.0995962314939436,
+      "grad_norm": 1.4851799011230469,
+      "learning_rate": 9.318208634881802e-05,
+      "loss": 0.07443426847457886,
+      "mean_token_accuracy": 0.9716702818870544,
+      "num_tokens": 28259406.0,
+      "step": 1560
+    },
+    {
+      "entropy": 1.0606005311012268,
+      "epoch": 2.113055181695828,
+      "grad_norm": 1.136905312538147,
+      "learning_rate": 9.306946739149161e-05,
+      "loss": 0.07183161973953248,
+      "mean_token_accuracy": 0.9726755082607269,
+      "num_tokens": 28441336.0,
+      "step": 1570
+    },
+    {
+      "entropy": 1.0512076377868653,
+      "epoch": 2.126514131897712,
+      "grad_norm": 1.4116008281707764,
+      "learning_rate": 9.29559951273358e-05,
+      "loss": 0.07272077798843384,
+      "mean_token_accuracy": 0.972117805480957,
+      "num_tokens": 28621984.0,
+      "step": 1580
+    },
+    {
+      "entropy": 1.0617831587791442,
+      "epoch": 2.1399730820995964,
+      "grad_norm": 1.4194544553756714,
+      "learning_rate": 9.284167180450141e-05,
+      "loss": 0.07670866250991822,
+      "mean_token_accuracy": 0.9709574401378631,
+      "num_tokens": 28803335.0,
+      "step": 1590
+    },
+    {
+      "entropy": 1.054829752445221,
+      "epoch": 2.1534320323014806,
+      "grad_norm": 1.0619711875915527,
+      "learning_rate": 9.272649968800069e-05,
+      "loss": 0.06954843401908875,
+      "mean_token_accuracy": 0.9732311069965363,
+      "num_tokens": 28984568.0,
+      "step": 1600
+    },
+    {
+      "entropy": 1.062503707408905,
+      "epoch": 2.166890982503365,
+      "grad_norm": 1.1824331283569336,
+      "learning_rate": 9.26104810596625e-05,
+      "loss": 0.0742799162864685,
+      "mean_token_accuracy": 0.9718966007232666,
+      "num_tokens": 29165958.0,
+      "step": 1610
+    },
+    {
+      "entropy": 1.0652349829673766,
+      "epoch": 2.180349932705249,
+      "grad_norm": 1.2683000564575195,
+      "learning_rate": 9.249361821808708e-05,
+      "loss": 0.07545414566993713,
+      "mean_token_accuracy": 0.9704910278320312,
+      "num_tokens": 29347870.0,
+      "step": 1620
+    },
+    {
+      "entropy": 1.0519612908363343,
+      "epoch": 2.1938088829071334,
+      "grad_norm": 1.257572889328003,
+      "learning_rate": 9.237591347860052e-05,
+      "loss": 0.075272136926651,
+      "mean_token_accuracy": 0.9712580978870392,
+      "num_tokens": 29529453.0,
+      "step": 1630
+    },
+    {
+      "entropy": 1.0668588399887085,
+      "epoch": 2.2072678331090176,
+      "grad_norm": 1.3439997434616089,
+      "learning_rate": 9.225736917320886e-05,
+      "loss": 0.068633770942688,
+      "mean_token_accuracy": 0.9735071420669555,
+      "num_tokens": 29710695.0,
+      "step": 1640
+    },
+    {
+      "entropy": 1.0701637029647828,
+      "epoch": 2.220726783310902,
+      "grad_norm": 1.4352364540100098,
+      "learning_rate": 9.213798765055187e-05,
+      "loss": 0.07366064786911011,
+      "mean_token_accuracy": 0.9724323868751525,
+      "num_tokens": 29890993.0,
+      "step": 1650
+    },
+    {
+      "entropy": 1.065868353843689,
+      "epoch": 2.234185733512786,
+      "grad_norm": 1.3669353723526,
+      "learning_rate": 9.20177712758566e-05,
+      "loss": 0.0714600920677185,
+      "mean_token_accuracy": 0.9717992007732391,
+      "num_tokens": 30072284.0,
+      "step": 1660
+    },
+    {
+      "entropy": 1.077157938480377,
+      "epoch": 2.2476446837146704,
+      "grad_norm": 1.3630682229995728,
+      "learning_rate": 9.189672243089046e-05,
+      "loss": 0.07237017154693604,
+      "mean_token_accuracy": 0.9729755759239197,
+      "num_tokens": 30253123.0,
+      "step": 1670
+    },
+    {
+      "entropy": 1.0886191844940185,
+      "epoch": 2.2611036339165547,
+      "grad_norm": 1.3787649869918823,
+      "learning_rate": 9.177484351391402e-05,
+      "loss": 0.07589051723480225,
+      "mean_token_accuracy": 0.9716001987457276,
+      "num_tokens": 30434100.0,
+      "step": 1680
+    },
+    {
+      "entropy": 1.080547845363617,
+      "epoch": 2.274562584118439,
+      "grad_norm": 1.4936959743499756,
+      "learning_rate": 9.165213693963355e-05,
+      "loss": 0.07393252849578857,
+      "mean_token_accuracy": 0.9709986329078675,
+      "num_tokens": 30615086.0,
+      "step": 1690
+    },
+    {
+      "entropy": 1.0851039648056031,
+      "epoch": 2.288021534320323,
+      "grad_norm": 1.4654748439788818,
+      "learning_rate": 9.152860513915314e-05,
+      "loss": 0.07487455606460572,
+      "mean_token_accuracy": 0.9710996568202972,
+      "num_tokens": 30795174.0,
+      "step": 1700
+    },
+    {
+      "entropy": 1.0743987679481506,
+      "epoch": 2.3014804845222074,
+      "grad_norm": 1.354244589805603,
+      "learning_rate": 9.140425055992648e-05,
+      "loss": 0.07701593041419982,
+      "mean_token_accuracy": 0.9701604127883912,
+      "num_tokens": 30976627.0,
+      "step": 1710
+    },
+    {
+      "entropy": 1.0738040447235107,
+      "epoch": 2.3149394347240917,
+      "grad_norm": 1.2830452919006348,
+      "learning_rate": 9.127907566570853e-05,
+      "loss": 0.07861257195472718,
+      "mean_token_accuracy": 0.9696074068546295,
+      "num_tokens": 31158542.0,
+      "step": 1720
+    },
+    {
+      "entropy": 1.0667856335639954,
+      "epoch": 2.328398384925976,
+      "grad_norm": 1.3329832553863525,
+      "learning_rate": 9.115308293650653e-05,
+      "loss": 0.07371129989624023,
+      "mean_token_accuracy": 0.9722550988197327,
+      "num_tokens": 31339091.0,
+      "step": 1730
+    },
+    {
+      "entropy": 1.0734196662902833,
+      "epoch": 2.34185733512786,
+      "grad_norm": 1.4483965635299683,
+      "learning_rate": 9.102627486853099e-05,
+      "loss": 0.07841320037841797,
+      "mean_token_accuracy": 0.9698444902896881,
+      "num_tokens": 31519961.0,
+      "step": 1740
+    },
+    {
+      "entropy": 1.0724906921386719,
+      "epoch": 2.3553162853297445,
+      "grad_norm": 1.440190315246582,
+      "learning_rate": 9.089865397414614e-05,
+      "loss": 0.07452400922775268,
+      "mean_token_accuracy": 0.9715199947357178,
+      "num_tokens": 31701305.0,
+      "step": 1750
+    },
+    {
+      "entropy": 1.0770147442817688,
+      "epoch": 2.3687752355316287,
+      "grad_norm": 1.17177414894104,
+      "learning_rate": 9.077022278182024e-05,
+      "loss": 0.07649941444396972,
+      "mean_token_accuracy": 0.9704680681228638,
+      "num_tokens": 31882106.0,
+      "step": 1760
+    },
+    {
+      "entropy": 1.0792171478271484,
+      "epoch": 2.382234185733513,
+      "grad_norm": 1.498350977897644,
+      "learning_rate": 9.064098383607545e-05,
+      "loss": 0.07850931882858277,
+      "mean_token_accuracy": 0.9699269771575928,
+      "num_tokens": 32062712.0,
+      "step": 1770
+    },
+    {
+      "entropy": 1.092840301990509,
+      "epoch": 2.3956931359353972,
+      "grad_norm": 1.4153627157211304,
+      "learning_rate": 9.051093969743738e-05,
+      "loss": 0.07811332941055298,
+      "mean_token_accuracy": 0.9698202073574066,
+      "num_tokens": 32244156.0,
+      "step": 1780
+    },
+    {
+      "entropy": 1.0767372131347657,
+      "epoch": 2.409152086137281,
+      "grad_norm": 1.3672188520431519,
+      "learning_rate": 9.03800929423844e-05,
+      "loss": 0.0765878677368164,
+      "mean_token_accuracy": 0.970400470495224,
+      "num_tokens": 32424975.0,
+      "step": 1790
+    },
+    {
+      "entropy": 1.0641321778297423,
+      "epoch": 2.4226110363391653,
+      "grad_norm": 1.1141754388809204,
+      "learning_rate": 9.024844616329662e-05,
+      "loss": 0.0744673728942871,
+      "mean_token_accuracy": 0.9713521003723145,
+      "num_tokens": 32606399.0,
+      "step": 1800
+    },
+    {
+      "entropy": 1.0847392082214355,
+      "epoch": 2.4360699865410496,
+      "grad_norm": 1.372215986251831,
+      "learning_rate": 9.011600196840447e-05,
+      "loss": 0.0788280725479126,
+      "mean_token_accuracy": 0.9688920557498932,
+      "num_tokens": 32787632.0,
+      "step": 1810
+    },
+    {
+      "entropy": 1.0952076196670533,
+      "epoch": 2.449528936742934,
+      "grad_norm": 1.3998148441314697,
+      "learning_rate": 8.998276298173707e-05,
+      "loss": 0.07322710752487183,
+      "mean_token_accuracy": 0.9713049352169036,
+      "num_tokens": 32969136.0,
+      "step": 1820
+    },
+    {
+      "entropy": 1.1047595977783202,
+      "epoch": 2.462987886944818,
+      "grad_norm": 1.538533091545105,
+      "learning_rate": 8.984873184307017e-05,
+      "loss": 0.07695393562316895,
+      "mean_token_accuracy": 0.9697132647037506,
+      "num_tokens": 33149585.0,
+      "step": 1830
+    },
+    {
+      "entropy": 1.1071401476860045,
+      "epoch": 2.4764468371467023,
+      "grad_norm": 1.2797189950942993,
+      "learning_rate": 8.971391120787397e-05,
+      "loss": 0.07820409536361694,
+      "mean_token_accuracy": 0.9695189118385314,
+      "num_tokens": 33330604.0,
+      "step": 1840
+    },
+    {
+      "entropy": 1.0979747891426086,
+      "epoch": 2.4899057873485866,
+      "grad_norm": 1.309956669807434,
+      "learning_rate": 8.957830374726042e-05,
+      "loss": 0.08005034327507018,
+      "mean_token_accuracy": 0.9683119177818298,
+      "num_tokens": 33510932.0,
+      "step": 1850
+    },
+    {
+      "entropy": 1.1024476408958435,
+      "epoch": 2.503364737550471,
+      "grad_norm": 1.2206803560256958,
+      "learning_rate": 8.944191214793028e-05,
+      "loss": 0.07752519249916076,
+      "mean_token_accuracy": 0.9697020888328552,
+      "num_tokens": 33692289.0,
+      "step": 1860
+    },
+    {
+      "entropy": 1.0788621783256531,
+      "epoch": 2.516823687752355,
+      "grad_norm": 1.4230300188064575,
+      "learning_rate": 8.930473911212e-05,
+      "loss": 0.08086194992065429,
+      "mean_token_accuracy": 0.9689899325370789,
+      "num_tokens": 33873643.0,
+      "step": 1870
+    },
+    {
+      "entropy": 1.0909191489219665,
+      "epoch": 2.5302826379542394,
+      "grad_norm": 1.3553478717803955,
+      "learning_rate": 8.916678735754809e-05,
+      "loss": 0.07968194484710693,
+      "mean_token_accuracy": 0.9695583760738373,
+      "num_tokens": 34054342.0,
+      "step": 1880
+    },
+    {
+      "entropy": 1.0874631762504579,
+      "epoch": 2.5437415881561236,
+      "grad_norm": 1.209466576576233,
+      "learning_rate": 8.902805961736123e-05,
+      "loss": 0.07668136358261109,
+      "mean_token_accuracy": 0.9706926643848419,
+      "num_tokens": 34235054.0,
+      "step": 1890
+    },
+    {
+      "entropy": 1.0850980758666993,
+      "epoch": 2.557200538358008,
+      "grad_norm": 1.2419084310531616,
+      "learning_rate": 8.88885586400803e-05,
+      "loss": 0.07621376514434815,
+      "mean_token_accuracy": 0.9708946466445922,
+      "num_tokens": 34415792.0,
+      "step": 1900
+    },
+    {
+      "entropy": 1.1024398922920227,
+      "epoch": 2.570659488559892,
+      "grad_norm": 1.5481151342391968,
+      "learning_rate": 8.874828718954576e-05,
+      "loss": 0.07621458768844605,
+      "mean_token_accuracy": 0.9696649491786957,
+      "num_tokens": 34596130.0,
+      "step": 1910
+    },
+    {
+      "entropy": 1.0917130827903747,
+      "epoch": 2.5841184387617764,
+      "grad_norm": 1.4186631441116333,
+      "learning_rate": 8.86072480448629e-05,
+      "loss": 0.07716711163520813,
+      "mean_token_accuracy": 0.9699705123901368,
+      "num_tokens": 34776608.0,
+      "step": 1920
+    },
+    {
+      "entropy": 1.0845637321472168,
+      "epoch": 2.5975773889636606,
+      "grad_norm": 1.2549679279327393,
+      "learning_rate": 8.84654440003469e-05,
+      "loss": 0.07353252172470093,
+      "mean_token_accuracy": 0.9715336680412292,
+      "num_tokens": 34957854.0,
+      "step": 1930
+    },
+    {
+      "entropy": 1.070354163646698,
+      "epoch": 2.611036339165545,
+      "grad_norm": 1.390034794807434,
+      "learning_rate": 8.83228778654674e-05,
+      "loss": 0.07804152965545655,
+      "mean_token_accuracy": 0.9694301605224609,
+      "num_tokens": 35139425.0,
+      "step": 1940
+    },
+    {
+      "entropy": 1.0926494956016541,
+      "epoch": 2.624495289367429,
+      "grad_norm": 1.2890169620513916,
+      "learning_rate": 8.817955246479276e-05,
+      "loss": 0.07905603051185608,
+      "mean_token_accuracy": 0.9694173872470856,
+      "num_tokens": 35321009.0,
+      "step": 1950
+    },
+    {
+      "entropy": 1.090261971950531,
+      "epoch": 2.6379542395693134,
+      "grad_norm": 1.201541543006897,
+      "learning_rate": 8.803547063793422e-05,
+      "loss": 0.07853106260299683,
+      "mean_token_accuracy": 0.9687644064426422,
+      "num_tokens": 35502338.0,
+      "step": 1960
+    },
+    {
+      "entropy": 1.110123836994171,
+      "epoch": 2.6514131897711977,
+      "grad_norm": 1.3321393728256226,
+      "learning_rate": 8.789063523948958e-05,
+      "loss": 0.07721331715583801,
+      "mean_token_accuracy": 0.9703636646270752,
+      "num_tokens": 35684550.0,
+      "step": 1970
+    },
+    {
+      "entropy": 1.1154867172241212,
+      "epoch": 2.664872139973082,
+      "grad_norm": 1.5639152526855469,
+      "learning_rate": 8.774504913898663e-05,
+      "loss": 0.07769335508346557,
+      "mean_token_accuracy": 0.9694860637187958,
+      "num_tokens": 35866258.0,
+      "step": 1980
+    },
+    {
+      "entropy": 1.1126874089241028,
+      "epoch": 2.678331090174966,
+      "grad_norm": 1.5160865783691406,
+      "learning_rate": 8.75987152208264e-05,
+      "loss": 0.07732025384902955,
+      "mean_token_accuracy": 0.9695809543132782,
+      "num_tokens": 36047771.0,
+      "step": 1990
+    },
+    {
+      "entropy": 1.1188251495361328,
+      "epoch": 2.6917900403768504,
+      "grad_norm": 1.3689777851104736,
+      "learning_rate": 8.745163638422583e-05,
+      "loss": 0.07844393253326416,
+      "mean_token_accuracy": 0.9697122633457184,
+      "num_tokens": 36228702.0,
+      "step": 2000
+    },
+    {
+      "entropy": 1.1134975671768188,
+      "epoch": 2.7052489905787347,
+      "grad_norm": 1.1956474781036377,
+      "learning_rate": 8.730381554316051e-05,
+      "loss": 0.07932614088058472,
+      "mean_token_accuracy": 0.9695560097694397,
+      "num_tokens": 36409577.0,
+      "step": 2010
+    },
+    {
+      "entropy": 1.1052889227867126,
+      "epoch": 2.718707940780619,
+      "grad_norm": 1.2844921350479126,
+      "learning_rate": 8.715525562630687e-05,
+      "loss": 0.07471654415130616,
+      "mean_token_accuracy": 0.9707616209983826,
+      "num_tokens": 36590635.0,
+      "step": 2020
+    },
+    {
+      "entropy": 1.1029305696487426,
+      "epoch": 2.732166890982503,
+      "grad_norm": 1.3110395669937134,
+      "learning_rate": 8.700595957698411e-05,
+      "loss": 0.07654619216918945,
+      "mean_token_accuracy": 0.9699673116207123,
+      "num_tokens": 36772032.0,
+      "step": 2030
+    },
+    {
+      "entropy": 1.0967667698860168,
+      "epoch": 2.7456258411843875,
+      "grad_norm": 1.2297992706298828,
+      "learning_rate": 8.685593035309598e-05,
+      "loss": 0.07549123167991638,
+      "mean_token_accuracy": 0.9705443382263184,
+      "num_tokens": 36953014.0,
+      "step": 2040
+    },
+    {
+      "entropy": 1.0954343318939208,
+      "epoch": 2.7590847913862717,
+      "grad_norm": 1.5522037744522095,
+      "learning_rate": 8.670517092707213e-05,
+      "loss": 0.07590663433074951,
+      "mean_token_accuracy": 0.9709664165973664,
+      "num_tokens": 37134549.0,
+      "step": 2050
+    },
+    {
+      "entropy": 1.0984182119369508,
+      "epoch": 2.772543741588156,
+      "grad_norm": 1.3691717386245728,
+      "learning_rate": 8.655368428580919e-05,
+      "loss": 0.07795964479446411,
+      "mean_token_accuracy": 0.969748318195343,
+      "num_tokens": 37315651.0,
+      "step": 2060
+    },
+    {
+      "entropy": 1.0937008380889892,
+      "epoch": 2.7860026917900402,
+      "grad_norm": 1.345433235168457,
+      "learning_rate": 8.640147343061165e-05,
+      "loss": 0.07556321024894715,
+      "mean_token_accuracy": 0.9707719802856445,
+      "num_tokens": 37497150.0,
+      "step": 2070
+    },
+    {
+      "entropy": 1.0978638410568238,
+      "epoch": 2.7994616419919245,
+      "grad_norm": 1.3442333936691284,
+      "learning_rate": 8.624854137713234e-05,
+      "loss": 0.07603150010108947,
+      "mean_token_accuracy": 0.9707530319690705,
+      "num_tokens": 37678637.0,
+      "step": 2080
+    },
+    {
+      "entropy": 1.112320113182068,
+      "epoch": 2.8129205921938087,
+      "grad_norm": 1.672094702720642,
+      "learning_rate": 8.609489115531278e-05,
+      "loss": 0.07702311277389526,
+      "mean_token_accuracy": 0.969988214969635,
+      "num_tokens": 37859530.0,
+      "step": 2090
+    },
+    {
+      "entropy": 1.1019986748695374,
+      "epoch": 2.826379542395693,
+      "grad_norm": 1.4613395929336548,
+      "learning_rate": 8.594052580932301e-05,
+      "loss": 0.07702183723449707,
+      "mean_token_accuracy": 0.9705046355724335,
+      "num_tokens": 38040160.0,
+      "step": 2100
+    },
+    {
+      "entropy": 1.1074181199073792,
+      "epoch": 2.8398384925975773,
+      "grad_norm": 1.33228600025177,
+      "learning_rate": 8.578544839750141e-05,
+      "loss": 0.07809240221977234,
+      "mean_token_accuracy": 0.9702594041824341,
+      "num_tokens": 38221261.0,
+      "step": 2110
+    },
+    {
+      "entropy": 1.1087767481803894,
+      "epoch": 2.8532974427994615,
+      "grad_norm": 1.3967821598052979,
+      "learning_rate": 8.562966199229399e-05,
+      "loss": 0.07641294002532958,
+      "mean_token_accuracy": 0.9694520056247711,
+      "num_tokens": 38402589.0,
+      "step": 2120
+    },
+    {
+      "entropy": 1.0902286529541017,
+      "epoch": 2.8667563930013458,
+      "grad_norm": 1.3104654550552368,
+      "learning_rate": 8.547316968019363e-05,
+      "loss": 0.07864789366722107,
+      "mean_token_accuracy": 0.9693171322345734,
+      "num_tokens": 38583485.0,
+      "step": 2130
+    },
+    {
+      "entropy": 1.088974940776825,
+      "epoch": 2.88021534320323,
+      "grad_norm": 1.389603614807129,
+      "learning_rate": 8.531597456167885e-05,
+      "loss": 0.07767997980117798,
+      "mean_token_accuracy": 0.9691886186599732,
+      "num_tokens": 38764480.0,
+      "step": 2140
+    },
+    {
+      "entropy": 1.093208122253418,
+      "epoch": 2.8936742934051143,
+      "grad_norm": 1.3133198022842407,
+      "learning_rate": 8.515807975115239e-05,
+      "loss": 0.08097923994064331,
+      "mean_token_accuracy": 0.968046635389328,
+      "num_tokens": 38946016.0,
+      "step": 2150
+    },
+    {
+      "entropy": 1.0920218586921693,
+      "epoch": 2.9071332436069985,
+      "grad_norm": 1.3472810983657837,
+      "learning_rate": 8.499948837687959e-05,
+      "loss": 0.07504125237464905,
+      "mean_token_accuracy": 0.9705065429210663,
+      "num_tokens": 39127480.0,
+      "step": 2160
+    },
+    {
+      "entropy": 1.085507369041443,
+      "epoch": 2.920592193808883,
+      "grad_norm": 1.9070137739181519,
+      "learning_rate": 8.484020358092625e-05,
+      "loss": 0.07882450819015503,
+      "mean_token_accuracy": 0.9694548428058625,
+      "num_tokens": 39308552.0,
+      "step": 2170
+    },
+    {
+      "entropy": 1.1070378303527832,
+      "epoch": 2.934051144010767,
+      "grad_norm": 1.3059617280960083,
+      "learning_rate": 8.468022851909657e-05,
+      "loss": 0.08202729225158692,
+      "mean_token_accuracy": 0.9679245829582215,
+      "num_tokens": 39489844.0,
+      "step": 2180
+    },
+    {
+      "entropy": 1.0994928359985352,
+      "epoch": 2.9475100942126513,
+      "grad_norm": 1.359965443611145,
+      "learning_rate": 8.451956636087046e-05,
+      "loss": 0.07510381340980529,
+      "mean_token_accuracy": 0.969987016916275,
+      "num_tokens": 39671941.0,
+      "step": 2190
+    },
+    {
+      "entropy": 1.0987842202186584,
+      "epoch": 2.9609690444145356,
+      "grad_norm": 1.385233998298645,
+      "learning_rate": 8.435822028934087e-05,
+      "loss": 0.07863752245903015,
+      "mean_token_accuracy": 0.9685526669025422,
+      "num_tokens": 39852940.0,
+      "step": 2200
+    },
+    {
+      "entropy": 1.110793125629425,
+      "epoch": 2.97442799461642,
+      "grad_norm": 1.4456037282943726,
+      "learning_rate": 8.41961935011506e-05,
+      "loss": 0.07783852815628052,
+      "mean_token_accuracy": 0.9702686488628387,
+      "num_tokens": 40034313.0,
+      "step": 2210
+    },
+    {
+      "entropy": 1.1140072107315064,
+      "epoch": 2.987886944818304,
+      "grad_norm": 1.1889644861221313,
+      "learning_rate": 8.403348920642911e-05,
+      "loss": 0.07787686586380005,
+      "mean_token_accuracy": 0.9696961343288422,
+      "num_tokens": 40215866.0,
+      "step": 2220
+    },
+    {
+      "epoch": 3.0,
+      "eval_entropy": 1.11932077984901,
+      "eval_loss": 0.12462407350540161,
+      "eval_mean_token_accuracy": 0.951254228118119,
+      "eval_num_tokens": 40378902.0,
+      "eval_runtime": 13.5665,
+      "eval_samples_per_second": 368.556,
+      "eval_steps_per_second": 11.573,
+      "step": 2229
+    },
+    {
+      "entropy": 1.1124149322509767,
+      "epoch": 3.0013458950201883,
+      "grad_norm": 1.1116644144058228,
+      "learning_rate": 8.387011062872883e-05,
+      "loss": 0.07439510822296143,
+      "mean_token_accuracy": 0.9713229060173034,
+      "num_tokens": 40396979.0,
+      "step": 2230
+    },
+    {
+      "entropy": 1.0716044545173644,
+      "epoch": 3.0148048452220726,
+      "grad_norm": 1.1957958936691284,
+      "learning_rate": 8.370606100496128e-05,
+      "loss": 0.05146250128746033,
+      "mean_token_accuracy": 0.9813062250614166,
+      "num_tokens": 40578935.0,
+      "step": 2240
+    },
+    {
+      "entropy": 1.0567036390304565,
+      "epoch": 3.028263795423957,
+      "grad_norm": 1.337942361831665,
+      "learning_rate": 8.354134358533301e-05,
+      "loss": 0.052991294860839845,
+      "mean_token_accuracy": 0.9807850599288941,
+      "num_tokens": 40759500.0,
+      "step": 2250
+    },
+    {
+      "entropy": 1.056044340133667,
+      "epoch": 3.041722745625841,
+      "grad_norm": 1.4191840887069702,
+      "learning_rate": 8.337596163328114e-05,
+      "loss": 0.05163681507110596,
+      "mean_token_accuracy": 0.9813888549804688,
+      "num_tokens": 40940279.0,
+      "step": 2260
+    },
+    {
+      "entropy": 1.056977343559265,
+      "epoch": 3.0551816958277254,
+      "grad_norm": 1.4481396675109863,
+      "learning_rate": 8.320991842540875e-05,
+      "loss": 0.05059016942977905,
+      "mean_token_accuracy": 0.9814287006855011,
+      "num_tokens": 41121782.0,
+      "step": 2270
+    },
+    {
+      "entropy": 1.0579043984413148,
+      "epoch": 3.0686406460296096,
+      "grad_norm": 1.0136349201202393,
+      "learning_rate": 8.304321725141995e-05,
+      "loss": 0.05109646320343018,
+      "mean_token_accuracy": 0.9805608510971069,
+      "num_tokens": 41302614.0,
+      "step": 2280
+    },
+    {
+      "entropy": 1.0564674019813538,
+      "epoch": 3.082099596231494,
+      "grad_norm": 1.165916085243225,
+      "learning_rate": 8.287586141405464e-05,
+      "loss": 0.05656059980392456,
+      "mean_token_accuracy": 0.9801787078380585,
+      "num_tokens": 41484240.0,
+      "step": 2290
+    },
+    {
+      "entropy": 1.0617436170578003,
+      "epoch": 3.095558546433378,
+      "grad_norm": 1.500622272491455,
+      "learning_rate": 8.27078542290232e-05,
+      "loss": 0.05233668088912964,
+      "mean_token_accuracy": 0.9808355152606965,
+      "num_tokens": 41665160.0,
+      "step": 2300
+    },
+    {
+      "entropy": 1.0509375214576722,
+      "epoch": 3.1090174966352624,
+      "grad_norm": 1.403001308441162,
+      "learning_rate": 8.253919902494071e-05,
+      "loss": 0.05074570775032043,
+      "mean_token_accuracy": 0.9807266294956207,
+      "num_tokens": 41846062.0,
+      "step": 2310
+    },
+    {
+      "entropy": 1.0536353230476379,
+      "epoch": 3.1224764468371466,
+      "grad_norm": 1.1046926975250244,
+      "learning_rate": 8.236989914326101e-05,
+      "loss": 0.04906890392303467,
+      "mean_token_accuracy": 0.9815255463123321,
+      "num_tokens": 42027115.0,
+      "step": 2320
+    },
+    {
+      "entropy": 1.0518895149230958,
+      "epoch": 3.135935397039031,
+      "grad_norm": 1.0369377136230469,
+      "learning_rate": 8.21999579382105e-05,
+      "loss": 0.04991455376148224,
+      "mean_token_accuracy": 0.9822782278060913,
+      "num_tokens": 42208594.0,
+      "step": 2330
+    },
+    {
+      "entropy": 1.0597471237182616,
+      "epoch": 3.149394347240915,
+      "grad_norm": 1.6156455278396606,
+      "learning_rate": 8.202937877672175e-05,
+      "loss": 0.049592965841293336,
+      "mean_token_accuracy": 0.9818629324436188,
+      "num_tokens": 42389065.0,
+      "step": 2340
+    },
+    {
+      "entropy": 1.0561720967292785,
+      "epoch": 3.1628532974427994,
+      "grad_norm": 1.2941797971725464,
+      "learning_rate": 8.185816503836665e-05,
+      "loss": 0.05092741847038269,
+      "mean_token_accuracy": 0.9810208320617676,
+      "num_tokens": 42570319.0,
+      "step": 2350
+    },
+    {
+      "entropy": 1.0521231889724731,
+      "epoch": 3.1763122476446837,
+      "grad_norm": 1.1866732835769653,
+      "learning_rate": 8.168632011528961e-05,
+      "loss": 0.051917088031768796,
+      "mean_token_accuracy": 0.981055635213852,
+      "num_tokens": 42751501.0,
+      "step": 2360
+    },
+    {
+      "entropy": 1.0555105328559875,
+      "epoch": 3.189771197846568,
+      "grad_norm": 1.2260069847106934,
+      "learning_rate": 8.15138474121403e-05,
+      "loss": 0.05160566568374634,
+      "mean_token_accuracy": 0.981193995475769,
+      "num_tokens": 42932643.0,
+      "step": 2370
+    },
+    {
+      "entropy": 1.0720075011253356,
+      "epoch": 3.203230148048452,
+      "grad_norm": 1.3786890506744385,
+      "learning_rate": 8.134075034600609e-05,
+      "loss": 0.053551071882247926,
+      "mean_token_accuracy": 0.9797191619873047,
+      "num_tokens": 43113204.0,
+      "step": 2380
+    },
+    {
+      "entropy": 1.0718719363212585,
+      "epoch": 3.2166890982503364,
+      "grad_norm": 1.3200089931488037,
+      "learning_rate": 8.116703234634453e-05,
+      "loss": 0.054034185409545896,
+      "mean_token_accuracy": 0.9800151228904724,
+      "num_tokens": 43293433.0,
+      "step": 2390
+    },
+    {
+      "entropy": 1.0536052703857421,
+      "epoch": 3.2301480484522207,
+      "grad_norm": 1.4560705423355103,
+      "learning_rate": 8.099269685491528e-05,
+      "loss": 0.05042939186096192,
+      "mean_token_accuracy": 0.9806897640228271,
+      "num_tokens": 43474189.0,
+      "step": 2400
+    },
+    {
+      "entropy": 1.0602891445159912,
+      "epoch": 3.243606998654105,
+      "grad_norm": 1.3123281002044678,
+      "learning_rate": 8.081774732571196e-05,
+      "loss": 0.05151479244232178,
+      "mean_token_accuracy": 0.980517429113388,
+      "num_tokens": 43655355.0,
+      "step": 2410
+    },
+    {
+      "entropy": 1.067916214466095,
+      "epoch": 3.257065948855989,
+      "grad_norm": 1.2652188539505005,
+      "learning_rate": 8.06421872248937e-05,
+      "loss": 0.05306308269500733,
+      "mean_token_accuracy": 0.9805097997188568,
+      "num_tokens": 43836752.0,
+      "step": 2420
+    },
+    {
+      "entropy": 1.067535150051117,
+      "epoch": 3.2705248990578735,
+      "grad_norm": 1.6280162334442139,
+      "learning_rate": 8.046602003071648e-05,
+      "loss": 0.052641332149505615,
+      "mean_token_accuracy": 0.9803492069244385,
+      "num_tokens": 44018525.0,
+      "step": 2430
+    },
+    {
+      "entropy": 1.070455825328827,
+      "epoch": 3.2839838492597577,
+      "grad_norm": 1.632573127746582,
+      "learning_rate": 8.028924923346426e-05,
+      "loss": 0.05452280044555664,
+      "mean_token_accuracy": 0.9793554663658142,
+      "num_tokens": 44199901.0,
+      "step": 2440
+    },
+    {
+      "entropy": 1.0741212248802186,
+      "epoch": 3.297442799461642,
+      "grad_norm": 1.3525495529174805,
+      "learning_rate": 8.011187833537972e-05,
+      "loss": 0.054098492860794066,
+      "mean_token_accuracy": 0.9804170608520508,
+      "num_tokens": 44381734.0,
+      "step": 2450
+    },
+    {
+      "entropy": 1.0836357831954957,
+      "epoch": 3.3109017496635262,
+      "grad_norm": 1.3989282846450806,
+      "learning_rate": 7.993391085059502e-05,
+      "loss": 0.05158165693283081,
+      "mean_token_accuracy": 0.9810640931129455,
+      "num_tokens": 44563158.0,
+      "step": 2460
+    },
+    {
+      "entropy": 1.077314054965973,
+      "epoch": 3.3243606998654105,
+      "grad_norm": 1.1970937252044678,
+      "learning_rate": 7.975535030506203e-05,
+      "loss": 0.052333271503448485,
+      "mean_token_accuracy": 0.9808873236179352,
+      "num_tokens": 44744642.0,
+      "step": 2470
+    },
+    {
+      "entropy": 1.069891107082367,
+      "epoch": 3.3378196500672948,
+      "grad_norm": 1.3197492361068726,
+      "learning_rate": 7.957620023648256e-05,
+      "loss": 0.05269680619239807,
+      "mean_token_accuracy": 0.980803519487381,
+      "num_tokens": 44926301.0,
+      "step": 2480
+    },
+    {
+      "entropy": 1.071691370010376,
+      "epoch": 3.351278600269179,
+      "grad_norm": 1.332481861114502,
+      "learning_rate": 7.939646419423826e-05,
+      "loss": 0.054685378074646,
+      "mean_token_accuracy": 0.9795110940933227,
+      "num_tokens": 45106935.0,
+      "step": 2490
+    },
+    {
+      "entropy": 1.0654058694839477,
+      "epoch": 3.3647375504710633,
+      "grad_norm": 1.5478020906448364,
+      "learning_rate": 7.92161457393203e-05,
+      "loss": 0.05287986397743225,
+      "mean_token_accuracy": 0.9797925829887391,
+      "num_tokens": 45288325.0,
+      "step": 2500
+    },
+    {
+      "entropy": 1.0660210013389588,
+      "epoch": 3.3781965006729475,
+      "grad_norm": 1.43852961063385,
+      "learning_rate": 7.903524844425878e-05,
+      "loss": 0.05411216616630554,
+      "mean_token_accuracy": 0.9790283799171448,
+      "num_tokens": 45469485.0,
+      "step": 2510
+    },
+    {
+      "entropy": 1.0649927496910094,
+      "epoch": 3.391655450874832,
+      "grad_norm": 1.3900808095932007,
+      "learning_rate": 7.885377589305197e-05,
+      "loss": 0.05290043354034424,
+      "mean_token_accuracy": 0.9808539688587189,
+      "num_tokens": 45650705.0,
+      "step": 2520
+    },
+    {
+      "entropy": 1.0768967866897583,
+      "epoch": 3.405114401076716,
+      "grad_norm": 1.3428847789764404,
+      "learning_rate": 7.867173168109534e-05,
+      "loss": 0.05374534130096435,
+      "mean_token_accuracy": 0.9802824795246124,
+      "num_tokens": 45832104.0,
+      "step": 2530
+    },
+    {
+      "entropy": 1.0660459756851197,
+      "epoch": 3.4185733512786003,
+      "grad_norm": 1.3150343894958496,
+      "learning_rate": 7.84891194151103e-05,
+      "loss": 0.05366337299346924,
+      "mean_token_accuracy": 0.9796974420547485,
+      "num_tokens": 46013552.0,
+      "step": 2540
+    },
+    {
+      "entropy": 1.06330087184906,
+      "epoch": 3.4320323014804845,
+      "grad_norm": 1.546858787536621,
+      "learning_rate": 7.830594271307267e-05,
+      "loss": 0.054223823547363284,
+      "mean_token_accuracy": 0.9796406507492066,
+      "num_tokens": 46194606.0,
+      "step": 2550
+    },
+    {
+      "entropy": 1.0564390897750855,
+      "epoch": 3.445491251682369,
+      "grad_norm": 1.3932640552520752,
+      "learning_rate": 7.812220520414115e-05,
+      "loss": 0.05237367153167725,
+      "mean_token_accuracy": 0.9805298745632172,
+      "num_tokens": 46375520.0,
+      "step": 2560
+    },
+    {
+      "entropy": 1.0519753217697143,
+      "epoch": 3.458950201884253,
+      "grad_norm": 1.8447308540344238,
+      "learning_rate": 7.793791052858528e-05,
+      "loss": 0.05499213933944702,
+      "mean_token_accuracy": 0.9787130057811737,
+      "num_tokens": 46556203.0,
+      "step": 2570
+    },
+    {
+      "entropy": 1.0693527460098267,
+      "epoch": 3.4724091520861373,
+      "grad_norm": 1.3508682250976562,
+      "learning_rate": 7.775306233771343e-05,
+      "loss": 0.055162906646728516,
+      "mean_token_accuracy": 0.9794130921363831,
+      "num_tokens": 46737287.0,
+      "step": 2580
+    },
+    {
+      "entropy": 1.078024983406067,
+      "epoch": 3.4858681022880216,
+      "grad_norm": 1.1582329273223877,
+      "learning_rate": 7.756766429380033e-05,
+      "loss": 0.05406954288482666,
+      "mean_token_accuracy": 0.9794795215129852,
+      "num_tokens": 46918363.0,
+      "step": 2590
+    },
+    {
+      "entropy": 1.0643377661705018,
+      "epoch": 3.499327052489906,
+      "grad_norm": 1.429380178451538,
+      "learning_rate": 7.738172007001465e-05,
+      "loss": 0.05309759378433228,
+      "mean_token_accuracy": 0.9806221067905426,
+      "num_tokens": 47099549.0,
+      "step": 2600
+    },
+    {
+      "entropy": 1.068834125995636,
+      "epoch": 3.51278600269179,
+      "grad_norm": 1.236444354057312,
+      "learning_rate": 7.719523335034612e-05,
+      "loss": 0.05395760536193848,
+      "mean_token_accuracy": 0.9798199534416199,
+      "num_tokens": 47280430.0,
+      "step": 2610
+    },
+    {
+      "entropy": 1.0729689002037048,
+      "epoch": 3.5262449528936743,
+      "grad_norm": 1.32350754737854,
+      "learning_rate": 7.70082078295326e-05,
+      "loss": 0.05501353740692139,
+      "mean_token_accuracy": 0.9792946577072144,
+      "num_tokens": 47461390.0,
+      "step": 2620
+    },
+    {
+      "entropy": 1.0766199231147766,
+      "epoch": 3.5397039030955586,
+      "grad_norm": 1.2348153591156006,
+      "learning_rate": 7.682064721298683e-05,
+      "loss": 0.05568041205406189,
+      "mean_token_accuracy": 0.9796942174434662,
+      "num_tokens": 47642662.0,
+      "step": 2630
+    },
+    {
+      "entropy": 1.0804976940155029,
+      "epoch": 3.553162853297443,
+      "grad_norm": 1.1175553798675537,
+      "learning_rate": 7.663255521672308e-05,
+      "loss": 0.050679147243499756,
+      "mean_token_accuracy": 0.9814749479293823,
+      "num_tokens": 47824309.0,
+      "step": 2640
+    },
+    {
+      "entropy": 1.0689257264137269,
+      "epoch": 3.566621803499327,
+      "grad_norm": 1.6297036409378052,
+      "learning_rate": 7.64439355672835e-05,
+      "loss": 0.054615461826324464,
+      "mean_token_accuracy": 0.9796478092670441,
+      "num_tokens": 48005982.0,
+      "step": 2650
+    },
+    {
+      "entropy": 1.0696970105171204,
+      "epoch": 3.5800807537012114,
+      "grad_norm": 1.2477251291275024,
+      "learning_rate": 7.625479200166425e-05,
+      "loss": 0.05287024974822998,
+      "mean_token_accuracy": 0.979641854763031,
+      "num_tokens": 48187048.0,
+      "step": 2660
+    },
+    {
+      "entropy": 1.0800267577171325,
+      "epoch": 3.5935397039030956,
+      "grad_norm": 1.3109506368637085,
+      "learning_rate": 7.606512826724155e-05,
+      "loss": 0.054475915431976316,
+      "mean_token_accuracy": 0.9794476389884949,
+      "num_tokens": 48368447.0,
+      "step": 2670
+    },
+    {
+      "entropy": 1.0848782062530518,
+      "epoch": 3.60699865410498,
+      "grad_norm": 1.2147419452667236,
+      "learning_rate": 7.587494812169728e-05,
+      "loss": 0.05212314128875732,
+      "mean_token_accuracy": 0.9798501491546631,
+      "num_tokens": 48549144.0,
+      "step": 2680
+    },
+    {
+      "entropy": 1.0681315302848815,
+      "epoch": 3.620457604306864,
+      "grad_norm": 1.3249231576919556,
+      "learning_rate": 7.568425533294476e-05,
+      "loss": 0.053503894805908205,
+      "mean_token_accuracy": 0.9800095915794372,
+      "num_tokens": 48731059.0,
+      "step": 2690
+    },
+    {
+      "entropy": 1.0629405975341797,
+      "epoch": 3.6339165545087484,
+      "grad_norm": 1.2435307502746582,
+      "learning_rate": 7.549305367905385e-05,
+      "loss": 0.0541129469871521,
+      "mean_token_accuracy": 0.9795375943183899,
+      "num_tokens": 48912283.0,
+      "step": 2700
+    },
+    {
+      "entropy": 1.0608446598052979,
+      "epoch": 3.6473755047106327,
+      "grad_norm": 1.3816258907318115,
+      "learning_rate": 7.53013469481763e-05,
+      "loss": 0.054210084676742556,
+      "mean_token_accuracy": 0.9793854176998138,
+      "num_tokens": 49094236.0,
+      "step": 2710
+    },
+    {
+      "entropy": 1.0596249461174012,
+      "epoch": 3.660834454912517,
+      "grad_norm": 1.132426381111145,
+      "learning_rate": 7.510913893847058e-05,
+      "loss": 0.05496481657028198,
+      "mean_token_accuracy": 0.9798576831817627,
+      "num_tokens": 49275540.0,
+      "step": 2720
+    },
+    {
+      "entropy": 1.060381281375885,
+      "epoch": 3.674293405114401,
+      "grad_norm": 1.4120980501174927,
+      "learning_rate": 7.491643345802667e-05,
+      "loss": 0.05694271326065063,
+      "mean_token_accuracy": 0.9788619399070739,
+      "num_tokens": 49456684.0,
+      "step": 2730
+    },
+    {
+      "entropy": 1.0594112157821656,
+      "epoch": 3.6877523553162854,
+      "grad_norm": 1.428120493888855,
+      "learning_rate": 7.472323432479062e-05,
+      "loss": 0.05274912118911743,
+      "mean_token_accuracy": 0.9802964627742767,
+      "num_tokens": 49638099.0,
+      "step": 2740
+    },
+    {
+      "entropy": 1.0630932807922364,
+      "epoch": 3.7012113055181697,
+      "grad_norm": 1.4599305391311646,
+      "learning_rate": 7.452954536648888e-05,
+      "loss": 0.057349550724029544,
+      "mean_token_accuracy": 0.9788366615772247,
+      "num_tokens": 49819751.0,
+      "step": 2750
+    },
+    {
+      "entropy": 1.0632095456123352,
+      "epoch": 3.714670255720054,
+      "grad_norm": 1.8489277362823486,
+      "learning_rate": 7.433537042055248e-05,
+      "loss": 0.053570222854614255,
+      "mean_token_accuracy": 0.980079036951065,
+      "num_tokens": 50000623.0,
+      "step": 2760
+    },
+    {
+      "entropy": 1.0583208560943604,
+      "epoch": 3.728129205921938,
+      "grad_norm": 1.4855526685714722,
+      "learning_rate": 7.414071333404104e-05,
+      "loss": 0.05508332252502442,
+      "mean_token_accuracy": 0.9789698243141174,
+      "num_tokens": 50181540.0,
+      "step": 2770
+    },
+    {
+      "entropy": 1.0690102458000184,
+      "epoch": 3.7415881561238225,
+      "grad_norm": 1.553468942642212,
+      "learning_rate": 7.394557796356644e-05,
+      "loss": 0.05684031248092651,
+      "mean_token_accuracy": 0.9776920318603516,
+      "num_tokens": 50363406.0,
+      "step": 2780
+    },
+    {
+      "entropy": 1.0651323795318604,
+      "epoch": 3.7550471063257067,
+      "grad_norm": 1.2951313257217407,
+      "learning_rate": 7.374996817521653e-05,
+      "loss": 0.055989408493041994,
+      "mean_token_accuracy": 0.978317403793335,
+      "num_tokens": 50544599.0,
+      "step": 2790
+    },
+    {
+      "entropy": 1.0571947336196899,
+      "epoch": 3.768506056527591,
+      "grad_norm": 1.1358306407928467,
+      "learning_rate": 7.35538878444785e-05,
+      "loss": 0.052672064304351805,
+      "mean_token_accuracy": 0.9797160506248475,
+      "num_tokens": 50726664.0,
+      "step": 2800
+    },
+    {
+      "entropy": 1.0475606441497802,
+      "epoch": 3.781965006729475,
+      "grad_norm": 1.4842852354049683,
+      "learning_rate": 7.335734085616206e-05,
+      "loss": 0.05601704120635986,
+      "mean_token_accuracy": 0.9794813454151153,
+      "num_tokens": 50907903.0,
+      "step": 2810
+    },
+    {
+      "entropy": 1.053141224384308,
+      "epoch": 3.7954239569313595,
+      "grad_norm": 1.3320889472961426,
+      "learning_rate": 7.316033110432249e-05,
+      "loss": 0.05435909032821655,
+      "mean_token_accuracy": 0.9792923092842102,
+      "num_tokens": 51088734.0,
+      "step": 2820
+    },
+    {
+      "entropy": 1.0518054842948914,
+      "epoch": 3.8088829071332437,
+      "grad_norm": 1.140783667564392,
+      "learning_rate": 7.296286249218352e-05,
+      "loss": 0.05576471090316772,
+      "mean_token_accuracy": 0.9789255380630493,
+      "num_tokens": 51269924.0,
+      "step": 2830
+    },
+    {
+      "entropy": 1.0490313410758971,
+      "epoch": 3.822341857335128,
+      "grad_norm": 1.5749891996383667,
+      "learning_rate": 7.276493893205995e-05,
+      "loss": 0.05655611753463745,
+      "mean_token_accuracy": 0.9785366714000702,
+      "num_tokens": 51450406.0,
+      "step": 2840
+    },
+    {
+      "entropy": 1.0508960008621215,
+      "epoch": 3.8358008075370122,
+      "grad_norm": 1.2535520792007446,
+      "learning_rate": 7.256656434528018e-05,
+      "loss": 0.05710798501968384,
+      "mean_token_accuracy": 0.978301590681076,
+      "num_tokens": 51630630.0,
+      "step": 2850
+    },
+    {
+      "entropy": 1.0434152960777283,
+      "epoch": 3.8492597577388965,
+      "grad_norm": 1.6045477390289307,
+      "learning_rate": 7.236774266210852e-05,
+      "loss": 0.0545048713684082,
+      "mean_token_accuracy": 0.9785818517208099,
+      "num_tokens": 51811252.0,
+      "step": 2860
+    },
+    {
+      "entropy": 1.046581244468689,
+      "epoch": 3.8627187079407808,
+      "grad_norm": 1.2737947702407837,
+      "learning_rate": 7.216847782166727e-05,
+      "loss": 0.056829071044921874,
+      "mean_token_accuracy": 0.9779990613460541,
+      "num_tokens": 51992578.0,
+      "step": 2870
+    },
+    {
+      "entropy": 1.0562566757202148,
+      "epoch": 3.876177658142665,
+      "grad_norm": 1.2364535331726074,
+      "learning_rate": 7.196877377185872e-05,
+      "loss": 0.054702204465866086,
+      "mean_token_accuracy": 0.9796520650386811,
+      "num_tokens": 52173507.0,
+      "step": 2880
+    },
+    {
+      "entropy": 1.0444288849830627,
+      "epoch": 3.8896366083445493,
+      "grad_norm": 1.2663625478744507,
+      "learning_rate": 7.176863446928694e-05,
+      "loss": 0.05222170352935791,
+      "mean_token_accuracy": 0.9805905818939209,
+      "num_tokens": 52354669.0,
+      "step": 2890
+    },
+    {
+      "entropy": 1.0429784655570984,
+      "epoch": 3.9030955585464335,
+      "grad_norm": 1.2072714567184448,
+      "learning_rate": 7.156806387917937e-05,
+      "loss": 0.05404345989227295,
+      "mean_token_accuracy": 0.9802968561649322,
+      "num_tokens": 52535334.0,
+      "step": 2900
+    },
+    {
+      "entropy": 1.045020854473114,
+      "epoch": 3.916554508748318,
+      "grad_norm": 1.3894519805908203,
+      "learning_rate": 7.136706597530825e-05,
+      "loss": 0.05601906776428223,
+      "mean_token_accuracy": 0.9779200375080108,
+      "num_tokens": 52715952.0,
+      "step": 2910
+    },
+    {
+      "entropy": 1.0439318537712097,
+      "epoch": 3.930013458950202,
+      "grad_norm": 1.4112615585327148,
+      "learning_rate": 7.116564473991192e-05,
+      "loss": 0.05345791578292847,
+      "mean_token_accuracy": 0.9796200633049011,
+      "num_tokens": 52897155.0,
+      "step": 2920
+    },
+    {
+      "entropy": 1.0361477971076964,
+      "epoch": 3.9434724091520863,
+      "grad_norm": 1.3158206939697266,
+      "learning_rate": 7.096380416361588e-05,
+      "loss": 0.05547314286231995,
+      "mean_token_accuracy": 0.9788706004619598,
+      "num_tokens": 53078916.0,
+      "step": 2930
+    },
+    {
+      "entropy": 1.0427968740463256,
+      "epoch": 3.9569313593539706,
+      "grad_norm": 1.2565301656723022,
+      "learning_rate": 7.076154824535381e-05,
+      "loss": 0.05601854920387268,
+      "mean_token_accuracy": 0.9791704952716828,
+      "num_tokens": 53258884.0,
+      "step": 2940
+    },
+    {
+      "entropy": 1.0374909162521362,
+      "epoch": 3.970390309555855,
+      "grad_norm": 1.4639637470245361,
+      "learning_rate": 7.055888099228825e-05,
+      "loss": 0.054512780904769895,
+      "mean_token_accuracy": 0.9793699860572815,
+      "num_tokens": 53440481.0,
+      "step": 2950
+    },
+    {
+      "entropy": 1.0511455059051513,
+      "epoch": 3.983849259757739,
+      "grad_norm": 1.242073655128479,
+      "learning_rate": 7.035580641973119e-05,
+      "loss": 0.055217862129211426,
+      "mean_token_accuracy": 0.9801203072071075,
+      "num_tokens": 53621142.0,
+      "step": 2960
+    },
+    {
+      "entropy": 1.0518356442451477,
+      "epoch": 3.9973082099596233,
+      "grad_norm": 1.1518405675888062,
+      "learning_rate": 7.015232855106468e-05,
+      "loss": 0.05533761382102966,
+      "mean_token_accuracy": 0.9789008319377899,
+      "num_tokens": 53802588.0,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0,
+      "eval_entropy": 1.0548353901334628,
+      "eval_loss": 0.13498826324939728,
+      "eval_mean_token_accuracy": 0.9507886888874564,
+      "eval_num_tokens": 53838633.0,
+      "eval_runtime": 12.6809,
+      "eval_samples_per_second": 394.295,
+      "eval_steps_per_second": 12.381,
+      "step": 2972
+    },
+    {
+      "entropy": 1.043621039390564,
+      "epoch": 4.010767160161508,
+      "grad_norm": 1.1016159057617188,
+      "learning_rate": 6.994845141766093e-05,
+      "loss": 0.03793205618858338,
+      "mean_token_accuracy": 0.9870165169239045,
+      "num_tokens": 53983430.0,
+      "step": 2980
+    },
+    {
+      "entropy": 1.0226780295372009,
+      "epoch": 4.024226110363392,
+      "grad_norm": 1.240837812423706,
+      "learning_rate": 6.974417905880255e-05,
+      "loss": 0.03402446508407593,
+      "mean_token_accuracy": 0.9883724093437195,
+      "num_tokens": 54164955.0,
+      "step": 2990
+    },
+    {
+      "entropy": 1.0194039583206176,
+      "epoch": 4.037685060565276,
+      "grad_norm": 1.1843734979629517,
+      "learning_rate": 6.953951552160248e-05,
+      "loss": 0.035861504077911374,
+      "mean_token_accuracy": 0.9872536063194275,
+      "num_tokens": 54346968.0,
+      "step": 3000
+    },
+    {
+      "entropy": 1.029161846637726,
+      "epoch": 4.05114401076716,
+      "grad_norm": 1.285415768623352,
+      "learning_rate": 6.933446486092381e-05,
+      "loss": 0.034106379747390746,
+      "mean_token_accuracy": 0.9882839441299438,
+      "num_tokens": 54528019.0,
+      "step": 3010
+    },
+    {
+      "entropy": 1.031964898109436,
+      "epoch": 4.064602960969045,
+      "grad_norm": 1.4622068405151367,
+      "learning_rate": 6.912903113929947e-05,
+      "loss": 0.032776829600334165,
+      "mean_token_accuracy": 0.9882788896560669,
+      "num_tokens": 54708148.0,
+      "step": 3020
+    },
+    {
+      "entropy": 1.0282336235046388,
+      "epoch": 4.078061911170929,
+      "grad_norm": 1.1071696281433105,
+      "learning_rate": 6.892321842685171e-05,
+      "loss": 0.03698182702064514,
+      "mean_token_accuracy": 0.9865448236465454,
+      "num_tokens": 54888894.0,
+      "step": 3030
+    },
+    {
+      "entropy": 1.0233591556549073,
+      "epoch": 4.091520861372813,
+      "grad_norm": 1.3671438694000244,
+      "learning_rate": 6.871703080121148e-05,
+      "loss": 0.03386759161949158,
+      "mean_token_accuracy": 0.9879629850387573,
+      "num_tokens": 55069872.0,
+      "step": 3040
+    },
+    {
+      "entropy": 1.0250957012176514,
+      "epoch": 4.104979811574697,
+      "grad_norm": 1.1421459913253784,
+      "learning_rate": 6.851047234743763e-05,
+      "loss": 0.035586753487586976,
+      "mean_token_accuracy": 0.9869713068008423,
+      "num_tokens": 55250605.0,
+      "step": 3050
+    },
+    {
+      "entropy": 1.016488790512085,
+      "epoch": 4.118438761776582,
+      "grad_norm": 1.3857697248458862,
+      "learning_rate": 6.830354715793598e-05,
+      "loss": 0.039323312044143674,
+      "mean_token_accuracy": 0.986032885313034,
+      "num_tokens": 55431737.0,
+      "step": 3060
+    },
+    {
+      "entropy": 1.0130115926265717,
+      "epoch": 4.131897711978466,
+      "grad_norm": 1.1797211170196533,
+      "learning_rate": 6.809625933237826e-05,
+      "loss": 0.03508211970329285,
+      "mean_token_accuracy": 0.9872302651405335,
+      "num_tokens": 55612917.0,
+      "step": 3070
+    },
+    {
+      "entropy": 1.023509919643402,
+      "epoch": 4.14535666218035,
+      "grad_norm": 1.1484259366989136,
+      "learning_rate": 6.788861297762086e-05,
+      "loss": 0.0338476300239563,
+      "mean_token_accuracy": 0.9883188247680664,
+      "num_tokens": 55793687.0,
+      "step": 3080
+    },
+    {
+      "entropy": 1.0193628787994384,
+      "epoch": 4.158815612382234,
+      "grad_norm": 1.3184833526611328,
+      "learning_rate": 6.768061220762345e-05,
+      "loss": 0.03667713701725006,
+      "mean_token_accuracy": 0.9869367897510528,
+      "num_tokens": 55974723.0,
+      "step": 3090
+    },
+    {
+      "entropy": 1.013669216632843,
+      "epoch": 4.172274562584119,
+      "grad_norm": 1.2176531553268433,
+      "learning_rate": 6.747226114336753e-05,
+      "loss": 0.035006949305534364,
+      "mean_token_accuracy": 0.9868299901485443,
+      "num_tokens": 56156568.0,
+      "step": 3100
+    },
+    {
+      "entropy": 1.0072043538093567,
+      "epoch": 4.185733512786003,
+      "grad_norm": 1.4346243143081665,
+      "learning_rate": 6.726356391277471e-05,
+      "loss": 0.03560284972190857,
+      "mean_token_accuracy": 0.9869017779827118,
+      "num_tokens": 56337896.0,
+      "step": 3110
+    },
+    {
+      "entropy": 1.0102415680885315,
+      "epoch": 4.199192462987887,
+      "grad_norm": 1.2328612804412842,
+      "learning_rate": 6.7054524650625e-05,
+      "loss": 0.03608990907669067,
+      "mean_token_accuracy": 0.9874296009540557,
+      "num_tokens": 56519102.0,
+      "step": 3120
+    },
+    {
+      "entropy": 1.015284490585327,
+      "epoch": 4.212651413189771,
+      "grad_norm": 1.4385327100753784,
+      "learning_rate": 6.684514749847482e-05,
+      "loss": 0.03637963235378265,
+      "mean_token_accuracy": 0.986739182472229,
+      "num_tokens": 56699936.0,
+      "step": 3130
+    },
+    {
+      "entropy": 1.0054036319255828,
+      "epoch": 4.226110363391656,
+      "grad_norm": 1.1129945516586304,
+      "learning_rate": 6.663543660457503e-05,
+      "loss": 0.034222981333732604,
+      "mean_token_accuracy": 0.9883054196834564,
+      "num_tokens": 56882308.0,
+      "step": 3140
+    },
+    {
+      "entropy": 1.0061798334121703,
+      "epoch": 4.23956931359354,
+      "grad_norm": 1.1598492860794067,
+      "learning_rate": 6.642539612378863e-05,
+      "loss": 0.03461446762084961,
+      "mean_token_accuracy": 0.9873203098773956,
+      "num_tokens": 57063770.0,
+      "step": 3150
+    },
+    {
+      "entropy": 1.0039719343185425,
+      "epoch": 4.253028263795424,
+      "grad_norm": 1.322458267211914,
+      "learning_rate": 6.621503021750858e-05,
+      "loss": 0.03701616525650024,
+      "mean_token_accuracy": 0.9862136900424957,
+      "num_tokens": 57244448.0,
+      "step": 3160
+    },
+    {
+      "entropy": 1.0048979759216308,
+      "epoch": 4.2664872139973085,
+      "grad_norm": 1.570866584777832,
+      "learning_rate": 6.600434305357521e-05,
+      "loss": 0.036290496587753296,
+      "mean_token_accuracy": 0.9868158757686615,
+      "num_tokens": 57425448.0,
+      "step": 3170
+    },
+    {
+      "entropy": 1.0145976901054383,
+      "epoch": 4.279946164199193,
+      "grad_norm": 1.103385329246521,
+      "learning_rate": 6.579333880619376e-05,
+      "loss": 0.03472789227962494,
+      "mean_token_accuracy": 0.987658828496933,
+      "num_tokens": 57606369.0,
+      "step": 3180
+    },
+    {
+      "entropy": 1.0186771035194397,
+      "epoch": 4.293405114401077,
+      "grad_norm": 1.2541930675506592,
+      "learning_rate": 6.558202165585161e-05,
+      "loss": 0.03532317280769348,
+      "mean_token_accuracy": 0.9874464809894562,
+      "num_tokens": 57787685.0,
+      "step": 3190
+    },
+    {
+      "entropy": 1.0198698043823242,
+      "epoch": 4.306864064602961,
+      "grad_norm": 1.1590384244918823,
+      "learning_rate": 6.53703957892355e-05,
+      "loss": 0.036364412307739256,
+      "mean_token_accuracy": 0.9864933013916015,
+      "num_tokens": 57968903.0,
+      "step": 3200
+    },
+    {
+      "entropy": 1.0211624503135681,
+      "epoch": 4.3203230148048455,
+      "grad_norm": 1.7207615375518799,
+      "learning_rate": 6.515846539914854e-05,
+      "loss": 0.03790584802627563,
+      "mean_token_accuracy": 0.9866538405418396,
+      "num_tokens": 58149469.0,
+      "step": 3210
+    },
+    {
+      "entropy": 1.0185290336608888,
+      "epoch": 4.33378196500673,
+      "grad_norm": 1.3042566776275635,
+      "learning_rate": 6.494623468442718e-05,
+      "loss": 0.03802136182785034,
+      "mean_token_accuracy": 0.9859855055809021,
+      "num_tokens": 58330867.0,
+      "step": 3220
+    },
+    {
+      "entropy": 1.0105828583240508,
+      "epoch": 4.347240915208614,
+      "grad_norm": 1.2536448240280151,
+      "learning_rate": 6.473370784985798e-05,
+      "loss": 0.034687772393226624,
+      "mean_token_accuracy": 0.9878224015235901,
+      "num_tokens": 58512529.0,
+      "step": 3230
+    },
+    {
+      "entropy": 1.0021630108356476,
+      "epoch": 4.360699865410498,
+      "grad_norm": 1.3816099166870117,
+      "learning_rate": 6.452088910609436e-05,
+      "loss": 0.035324835777282716,
+      "mean_token_accuracy": 0.9872316241264343,
+      "num_tokens": 58693573.0,
+      "step": 3240
+    },
+    {
+      "entropy": 1.007886278629303,
+      "epoch": 4.3741588156123825,
+      "grad_norm": 1.3121757507324219,
+      "learning_rate": 6.430778266957312e-05,
+      "loss": 0.034709393978118896,
+      "mean_token_accuracy": 0.9877420663833618,
+      "num_tokens": 58874107.0,
+      "step": 3250
+    },
+    {
+      "entropy": 1.005664885044098,
+      "epoch": 4.387617765814267,
+      "grad_norm": 1.429775357246399,
+      "learning_rate": 6.409439276243092e-05,
+      "loss": 0.03714964389801025,
+      "mean_token_accuracy": 0.9861079812049866,
+      "num_tokens": 59054533.0,
+      "step": 3260
+    },
+    {
+      "entropy": 1.0047886371612549,
+      "epoch": 4.401076716016151,
+      "grad_norm": 1.2385882139205933,
+      "learning_rate": 6.388072361242067e-05,
+      "loss": 0.036937403678894046,
+      "mean_token_accuracy": 0.9863442242145538,
+      "num_tokens": 59236331.0,
+      "step": 3270
+    },
+    {
+      "entropy": 1.0209275126457213,
+      "epoch": 4.414535666218035,
+      "grad_norm": 1.476168155670166,
+      "learning_rate": 6.366677945282769e-05,
+      "loss": 0.03463808596134186,
+      "mean_token_accuracy": 0.9872730135917663,
+      "num_tokens": 59417193.0,
+      "step": 3280
+    },
+    {
+      "entropy": 1.021347463130951,
+      "epoch": 4.4279946164199195,
+      "grad_norm": 1.4710102081298828,
+      "learning_rate": 6.345256452238591e-05,
+      "loss": 0.038167804479599,
+      "mean_token_accuracy": 0.986047238111496,
+      "num_tokens": 59598284.0,
+      "step": 3290
+    },
+    {
+      "entropy": 1.0282366394996643,
+      "epoch": 4.441453566621804,
+      "grad_norm": 1.1043860912322998,
+      "learning_rate": 6.323808306519385e-05,
+      "loss": 0.036692625284194945,
+      "mean_token_accuracy": 0.9862713575363159,
+      "num_tokens": 59779329.0,
+      "step": 3300
+    },
+    {
+      "entropy": 1.021342945098877,
+      "epoch": 4.454912516823688,
+      "grad_norm": 1.1742897033691406,
+      "learning_rate": 6.302333933063057e-05,
+      "loss": 0.03571464121341705,
+      "mean_token_accuracy": 0.987505567073822,
+      "num_tokens": 59961340.0,
+      "step": 3310
+    },
+    {
+      "entropy": 1.0153375625610352,
+      "epoch": 4.468371467025572,
+      "grad_norm": 1.1321660280227661,
+      "learning_rate": 6.280833757327142e-05,
+      "loss": 0.036007928848266604,
+      "mean_token_accuracy": 0.9870990335941314,
+      "num_tokens": 60142647.0,
+      "step": 3320
+    },
+    {
+      "entropy": 1.0099608540534972,
+      "epoch": 4.481830417227457,
+      "grad_norm": 1.084017276763916,
+      "learning_rate": 6.259308205280383e-05,
+      "loss": 0.03562799394130707,
+      "mean_token_accuracy": 0.9865496158599854,
+      "num_tokens": 60324151.0,
+      "step": 3330
+    },
+    {
+      "entropy": 1.0173812866210938,
+      "epoch": 4.495289367429341,
+      "grad_norm": 1.2779732942581177,
+      "learning_rate": 6.237757703394283e-05,
+      "loss": 0.03556201159954071,
+      "mean_token_accuracy": 0.9875519394874572,
+      "num_tokens": 60505023.0,
+      "step": 3340
+    },
+    {
+      "entropy": 1.0236618638038635,
+      "epoch": 4.508748317631225,
+      "grad_norm": 1.5285946130752563,
+      "learning_rate": 6.216182678634664e-05,
+      "loss": 0.03721941411495209,
+      "mean_token_accuracy": 0.9865630626678467,
+      "num_tokens": 60685921.0,
+      "step": 3350
+    },
+    {
+      "entropy": 1.0262888669967651,
+      "epoch": 4.522207267833109,
+      "grad_norm": 1.131675362586975,
+      "learning_rate": 6.194583558453199e-05,
+      "loss": 0.03602397441864014,
+      "mean_token_accuracy": 0.9864337563514709,
+      "num_tokens": 60867240.0,
+      "step": 3360
+    },
+    {
+      "entropy": 1.0292576909065247,
+      "epoch": 4.535666218034994,
+      "grad_norm": 1.5156011581420898,
+      "learning_rate": 6.172960770778948e-05,
+      "loss": 0.03973098993301392,
+      "mean_token_accuracy": 0.9855743169784545,
+      "num_tokens": 61047708.0,
+      "step": 3370
+    },
+    {
+      "entropy": 1.010222613811493,
+      "epoch": 4.549125168236878,
+      "grad_norm": 1.0979945659637451,
+      "learning_rate": 6.151314744009885e-05,
+      "loss": 0.03533022105693817,
+      "mean_token_accuracy": 0.9869906663894653,
+      "num_tokens": 61229247.0,
+      "step": 3380
+    },
+    {
+      "entropy": 1.004716032743454,
+      "epoch": 4.562584118438762,
+      "grad_norm": 1.3082596063613892,
+      "learning_rate": 6.129645907004395e-05,
+      "loss": 0.033983749151229856,
+      "mean_token_accuracy": 0.9877199769020081,
+      "num_tokens": 61411059.0,
+      "step": 3390
+    },
+    {
+      "entropy": 1.0031994462013245,
+      "epoch": 4.576043068640646,
+      "grad_norm": 1.3033607006072998,
+      "learning_rate": 6.107954689072796e-05,
+      "loss": 0.03607205748558044,
+      "mean_token_accuracy": 0.9866301774978637,
+      "num_tokens": 61592538.0,
+      "step": 3400
+    },
+    {
+      "entropy": 1.0113391518592834,
+      "epoch": 4.589502018842531,
+      "grad_norm": 1.3490946292877197,
+      "learning_rate": 6.086241519968822e-05,
+      "loss": 0.03906603455543518,
+      "mean_token_accuracy": 0.9854673087596894,
+      "num_tokens": 61773613.0,
+      "step": 3410
+    },
+    {
+      "entropy": 1.0076410889625549,
+      "epoch": 4.602960969044415,
+      "grad_norm": 1.1719361543655396,
+      "learning_rate": 6.064506829881109e-05,
+      "loss": 0.037575924396514894,
+      "mean_token_accuracy": 0.9867486238479615,
+      "num_tokens": 61955690.0,
+      "step": 3420
+    },
+    {
+      "entropy": 1.0200914144515991,
+      "epoch": 4.616419919246299,
+      "grad_norm": 1.1205360889434814,
+      "learning_rate": 6.042751049424675e-05,
+      "loss": 0.03691520690917969,
+      "mean_token_accuracy": 0.9865606665611267,
+      "num_tokens": 62136047.0,
+      "step": 3430
+    },
+    {
+      "entropy": 1.0141666173934936,
+      "epoch": 4.629878869448183,
+      "grad_norm": 1.1805763244628906,
+      "learning_rate": 6.02097460963239e-05,
+      "loss": 0.035447508096694946,
+      "mean_token_accuracy": 0.9873166739940643,
+      "num_tokens": 62317192.0,
+      "step": 3440
+    },
+    {
+      "entropy": 1.0067960560321807,
+      "epoch": 4.643337819650068,
+      "grad_norm": 1.0286436080932617,
+      "learning_rate": 5.999177941946429e-05,
+      "loss": 0.03580200672149658,
+      "mean_token_accuracy": 0.9870070457458496,
+      "num_tokens": 62498860.0,
+      "step": 3450
+    },
+    {
+      "entropy": 1.0088169515132903,
+      "epoch": 4.656796769851952,
+      "grad_norm": 1.3638492822647095,
+      "learning_rate": 5.977361478209732e-05,
+      "loss": 0.03756250143051147,
+      "mean_token_accuracy": 0.9861987411975861,
+      "num_tokens": 62679272.0,
+      "step": 3460
+    },
+    {
+      "entropy": 1.0098722338676454,
+      "epoch": 4.670255720053836,
+      "grad_norm": 1.649560809135437,
+      "learning_rate": 5.955525650657444e-05,
+      "loss": 0.037615427374839784,
+      "mean_token_accuracy": 0.9864691913127899,
+      "num_tokens": 62860715.0,
+      "step": 3470
+    },
+    {
+      "entropy": 1.0174571394920349,
+      "epoch": 4.68371467025572,
+      "grad_norm": 1.2106678485870361,
+      "learning_rate": 5.933670891908355e-05,
+      "loss": 0.037718603014945985,
+      "mean_token_accuracy": 0.9855880260467529,
+      "num_tokens": 63041029.0,
+      "step": 3480
+    },
+    {
+      "entropy": 1.0129252314567565,
+      "epoch": 4.697173620457605,
+      "grad_norm": 1.1571707725524902,
+      "learning_rate": 5.9117976349563206e-05,
+      "loss": 0.03724310994148254,
+      "mean_token_accuracy": 0.9865262687206269,
+      "num_tokens": 63222379.0,
+      "step": 3490
+    },
+    {
+      "entropy": 1.0124369859695435,
+      "epoch": 4.710632570659489,
+      "grad_norm": 1.0599361658096313,
+      "learning_rate": 5.889906313161696e-05,
+      "loss": 0.037783673405647276,
+      "mean_token_accuracy": 0.9859567046165466,
+      "num_tokens": 63403328.0,
+      "step": 3500
+    },
+    {
+      "entropy": 1.0041921436786652,
+      "epoch": 4.724091520861373,
+      "grad_norm": 1.2054452896118164,
+      "learning_rate": 5.8679973602427376e-05,
+      "loss": 0.03585146069526672,
+      "mean_token_accuracy": 0.9868905484676361,
+      "num_tokens": 63584598.0,
+      "step": 3510
+    },
+    {
+      "entropy": 1.0015427589416503,
+      "epoch": 4.737550471063257,
+      "grad_norm": 1.5075141191482544,
+      "learning_rate": 5.846071210267018e-05,
+      "loss": 0.037428200244903564,
+      "mean_token_accuracy": 0.9867083311080933,
+      "num_tokens": 63766203.0,
+      "step": 3520
+    },
+    {
+      "entropy": 1.0039161920547486,
+      "epoch": 4.751009421265142,
+      "grad_norm": 1.310390830039978,
+      "learning_rate": 5.824128297642823e-05,
+      "loss": 0.037494581937789914,
+      "mean_token_accuracy": 0.9862947940826416,
+      "num_tokens": 63947714.0,
+      "step": 3530
+    },
+    {
+      "entropy": 1.0049733996391297,
+      "epoch": 4.764468371467026,
+      "grad_norm": 1.2912139892578125,
+      "learning_rate": 5.802169057110548e-05,
+      "loss": 0.03803396224975586,
+      "mean_token_accuracy": 0.9860390484333038,
+      "num_tokens": 64128798.0,
+      "step": 3540
+    },
+    {
+      "entropy": 1.0127450585365296,
+      "epoch": 4.77792732166891,
+      "grad_norm": 1.001133680343628,
+      "learning_rate": 5.7801939237340786e-05,
+      "loss": 0.0351622611284256,
+      "mean_token_accuracy": 0.9875370979309082,
+      "num_tokens": 64309577.0,
+      "step": 3550
+    },
+    {
+      "entropy": 1.0070195615291595,
+      "epoch": 4.7913862718707945,
+      "grad_norm": 1.1870886087417603,
+      "learning_rate": 5.758203332892177e-05,
+      "loss": 0.036404851078987124,
+      "mean_token_accuracy": 0.9871752023696899,
+      "num_tokens": 64490707.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.9983750462532044,
+      "epoch": 4.804845222072679,
+      "grad_norm": 1.2374205589294434,
+      "learning_rate": 5.736197720269855e-05,
+      "loss": 0.03696496784687042,
+      "mean_token_accuracy": 0.9865773975849151,
+      "num_tokens": 64672104.0,
+      "step": 3570
+    },
+    {
+      "entropy": 1.0151559710502625,
+      "epoch": 4.818304172274562,
+      "grad_norm": 1.2001149654388428,
+      "learning_rate": 5.714177521849736e-05,
+      "loss": 0.03725983202457428,
+      "mean_token_accuracy": 0.9862341761589051,
+      "num_tokens": 64853353.0,
+      "step": 3580
+    },
+    {
+      "entropy": 1.0136774778366089,
+      "epoch": 4.831763122476447,
+      "grad_norm": 1.0128737688064575,
+      "learning_rate": 5.69214317390343e-05,
+      "loss": 0.03615986704826355,
+      "mean_token_accuracy": 0.9872736394405365,
+      "num_tokens": 65035009.0,
+      "step": 3590
+    },
+    {
+      "entropy": 1.0082051515579225,
+      "epoch": 4.845222072678331,
+      "grad_norm": 1.2552846670150757,
+      "learning_rate": 5.670095112982875e-05,
+      "loss": 0.03561249375343323,
+      "mean_token_accuracy": 0.9868618845939636,
+      "num_tokens": 65215981.0,
+      "step": 3600
+    },
+    {
+      "entropy": 1.0209908723831176,
+      "epoch": 4.858681022880216,
+      "grad_norm": 1.1914079189300537,
+      "learning_rate": 5.648033775911701e-05,
+      "loss": 0.03721153438091278,
+      "mean_token_accuracy": 0.9862311720848084,
+      "num_tokens": 65397088.0,
+      "step": 3610
+    },
+    {
+      "entropy": 1.014677894115448,
+      "epoch": 4.872139973082099,
+      "grad_norm": 1.3584535121917725,
+      "learning_rate": 5.625959599776564e-05,
+      "loss": 0.03897300064563751,
+      "mean_token_accuracy": 0.9851836740970612,
+      "num_tokens": 65578262.0,
+      "step": 3620
+    },
+    {
+      "entropy": 1.0207050919532776,
+      "epoch": 4.885598923283984,
+      "grad_norm": 1.2754287719726562,
+      "learning_rate": 5.603873021918493e-05,
+      "loss": 0.037926393747329715,
+      "mean_token_accuracy": 0.9859795570373535,
+      "num_tokens": 65758956.0,
+      "step": 3630
+    },
+    {
+      "entropy": 1.008733856678009,
+      "epoch": 4.899057873485868,
+      "grad_norm": 1.0815879106521606,
+      "learning_rate": 5.581774479924229e-05,
+      "loss": 0.036518561840057376,
+      "mean_token_accuracy": 0.9870649397373199,
+      "num_tokens": 65939970.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.9995620250701904,
+      "epoch": 4.912516823687753,
+      "grad_norm": 1.1773526668548584,
+      "learning_rate": 5.5596644116175444e-05,
+      "loss": 0.035269084572792056,
+      "mean_token_accuracy": 0.9873197913169861,
+      "num_tokens": 66121045.0,
+      "step": 3650
+    },
+    {
+      "entropy": 1.0036249816417695,
+      "epoch": 4.925975773889636,
+      "grad_norm": 1.0941131114959717,
+      "learning_rate": 5.537543255050579e-05,
+      "loss": 0.0353422611951828,
+      "mean_token_accuracy": 0.9870408117771149,
+      "num_tokens": 66302324.0,
+      "step": 3660
+    },
+    {
+      "entropy": 1.004719352722168,
+      "epoch": 4.939434724091521,
+      "grad_norm": 1.5107269287109375,
+      "learning_rate": 5.5154114484951556e-05,
+      "loss": 0.03743018507957459,
+      "mean_token_accuracy": 0.9857985079288483,
+      "num_tokens": 66483916.0,
+      "step": 3670
+    },
+    {
+      "entropy": 1.0076164484024048,
+      "epoch": 4.952893674293405,
+      "grad_norm": 1.3877427577972412,
+      "learning_rate": 5.4932694304340985e-05,
+      "loss": 0.03857019543647766,
+      "mean_token_accuracy": 0.9856370270252228,
+      "num_tokens": 66664390.0,
+      "step": 3680
+    },
+    {
+      "entropy": 1.015517818927765,
+      "epoch": 4.96635262449529,
+      "grad_norm": 1.1693767309188843,
+      "learning_rate": 5.471117639552543e-05,
+      "loss": 0.03634478151798248,
+      "mean_token_accuracy": 0.9868648707866668,
+      "num_tokens": 66845574.0,
+      "step": 3690
+    },
+    {
+      "entropy": 1.011656105518341,
+      "epoch": 4.979811574697173,
+      "grad_norm": 1.082209587097168,
+      "learning_rate": 5.448956514729251e-05,
+      "loss": 0.03513925671577454,
+      "mean_token_accuracy": 0.9872389912605286,
+      "num_tokens": 67027557.0,
+      "step": 3700
+    },
+    {
+      "entropy": 1.0161120653152467,
+      "epoch": 4.993270524899058,
+      "grad_norm": 1.266615629196167,
+      "learning_rate": 5.426786495027908e-05,
+      "loss": 0.0379630446434021,
+      "mean_token_accuracy": 0.9865313708782196,
+      "num_tokens": 67208088.0,
+      "step": 3710
+    },
+    {
+      "epoch": 5.0,
+      "eval_entropy": 1.0219634957374282,
+      "eval_loss": 0.14612138271331787,
+      "eval_mean_token_accuracy": 0.9504923186484416,
+      "eval_num_tokens": 67298266.0,
+      "eval_runtime": 12.6802,
+      "eval_samples_per_second": 394.317,
+      "eval_steps_per_second": 12.382,
+      "step": 3715
+    },
+    {
+      "entropy": 1.019774079322815,
+      "epoch": 5.006729475100942,
+      "grad_norm": 0.8909980654716492,
+      "learning_rate": 5.404608019688432e-05,
+      "loss": 0.02882351279258728,
+      "mean_token_accuracy": 0.9900537073612213,
+      "num_tokens": 67388858.0,
+      "step": 3720
+    },
+    {
+      "entropy": 1.0057315409183503,
+      "epoch": 5.020188425302826,
+      "grad_norm": 0.9996656775474548,
+      "learning_rate": 5.382421528118262e-05,
+      "loss": 0.02233442962169647,
+      "mean_token_accuracy": 0.9922977447509765,
+      "num_tokens": 67570084.0,
+      "step": 3730
+    },
+    {
+      "entropy": 1.0058627128601074,
+      "epoch": 5.03364737550471,
+      "grad_norm": 1.0265147686004639,
+      "learning_rate": 5.360227459883662e-05,
+      "loss": 0.021967282891273497,
+      "mean_token_accuracy": 0.9923401355743409,
+      "num_tokens": 67751587.0,
+      "step": 3740
+    },
+    {
+      "entropy": 1.0083157896995545,
+      "epoch": 5.0471063257065945,
+      "grad_norm": 1.0139433145523071,
+      "learning_rate": 5.338026254701003e-05,
+      "loss": 0.022531555593013765,
+      "mean_token_accuracy": 0.9930226564407348,
+      "num_tokens": 67933269.0,
+      "step": 3750
+    },
+    {
+      "entropy": 1.0119245886802672,
+      "epoch": 5.060565275908479,
+      "grad_norm": 1.0655452013015747,
+      "learning_rate": 5.31581835242806e-05,
+      "loss": 0.021685637533664703,
+      "mean_token_accuracy": 0.9925794541835785,
+      "num_tokens": 68113971.0,
+      "step": 3760
+    },
+    {
+      "entropy": 1.0082412719726563,
+      "epoch": 5.074024226110363,
+      "grad_norm": 1.268751859664917,
+      "learning_rate": 5.293604193055289e-05,
+      "loss": 0.021513286232948303,
+      "mean_token_accuracy": 0.9927303433418274,
+      "num_tokens": 68295632.0,
+      "step": 3770
+    },
+    {
+      "entropy": 1.012417995929718,
+      "epoch": 5.087483176312247,
+      "grad_norm": 0.9899319410324097,
+      "learning_rate": 5.2713842166971165e-05,
+      "loss": 0.02352543771266937,
+      "mean_token_accuracy": 0.992375648021698,
+      "num_tokens": 68476794.0,
+      "step": 3780
+    },
+    {
+      "entropy": 1.0069364666938783,
+      "epoch": 5.1009421265141315,
+      "grad_norm": 1.1964375972747803,
+      "learning_rate": 5.249158863583216e-05,
+      "loss": 0.021382461488246917,
+      "mean_token_accuracy": 0.9928456962108612,
+      "num_tokens": 68658249.0,
+      "step": 3790
+    },
+    {
+      "entropy": 1.0040670156478881,
+      "epoch": 5.114401076716016,
+      "grad_norm": 1.0445663928985596,
+      "learning_rate": 5.2269285740497876e-05,
+      "loss": 0.02207571566104889,
+      "mean_token_accuracy": 0.9927122712135314,
+      "num_tokens": 68840525.0,
+      "step": 3800
+    },
+    {
+      "entropy": 1.0083145201206207,
+      "epoch": 5.1278600269179,
+      "grad_norm": 0.8981389403343201,
+      "learning_rate": 5.204693788530832e-05,
+      "loss": 0.022873447835445405,
+      "mean_token_accuracy": 0.9918712139129638,
+      "num_tokens": 69021896.0,
+      "step": 3810
+    },
+    {
+      "entropy": 1.0109572768211366,
+      "epoch": 5.141318977119784,
+      "grad_norm": 1.250670313835144,
+      "learning_rate": 5.182454947549428e-05,
+      "loss": 0.023577621579170226,
+      "mean_token_accuracy": 0.9919484734535218,
+      "num_tokens": 69203112.0,
+      "step": 3820
+    },
+    {
+      "entropy": 1.0164754271507264,
+      "epoch": 5.1547779273216685,
+      "grad_norm": 1.3053207397460938,
+      "learning_rate": 5.160212491709002e-05,
+      "loss": 0.02436322420835495,
+      "mean_token_accuracy": 0.9921659469604492,
+      "num_tokens": 69384193.0,
+      "step": 3830
+    },
+    {
+      "entropy": 1.0034007966518401,
+      "epoch": 5.168236877523553,
+      "grad_norm": 1.4878787994384766,
+      "learning_rate": 5.1379668616845975e-05,
+      "loss": 0.023177918791770936,
+      "mean_token_accuracy": 0.9919318974018096,
+      "num_tokens": 69565602.0,
+      "step": 3840
+    },
+    {
+      "entropy": 1.0074714839458465,
+      "epoch": 5.181695827725437,
+      "grad_norm": 1.0512391328811646,
+      "learning_rate": 5.115718498214148e-05,
+      "loss": 0.023409275710582732,
+      "mean_token_accuracy": 0.9918898940086365,
+      "num_tokens": 69746242.0,
+      "step": 3850
+    },
+    {
+      "entropy": 1.003154468536377,
+      "epoch": 5.195154777927321,
+      "grad_norm": 0.971832811832428,
+      "learning_rate": 5.093467842089742e-05,
+      "loss": 0.023223266005516052,
+      "mean_token_accuracy": 0.9920656085014343,
+      "num_tokens": 69927227.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.999264246225357,
+      "epoch": 5.2086137281292055,
+      "grad_norm": 0.9266519546508789,
+      "learning_rate": 5.071215334148891e-05,
+      "loss": 0.02454002797603607,
+      "mean_token_accuracy": 0.9914694905281067,
+      "num_tokens": 70108580.0,
+      "step": 3870
+    },
+    {
+      "entropy": 1.002640813589096,
+      "epoch": 5.22207267833109,
+      "grad_norm": 1.5331412553787231,
+      "learning_rate": 5.048961415265797e-05,
+      "loss": 0.02458159625530243,
+      "mean_token_accuracy": 0.9919124245643616,
+      "num_tokens": 70288960.0,
+      "step": 3880
+    },
+    {
+      "entropy": 1.007998287677765,
+      "epoch": 5.235531628532974,
+      "grad_norm": 1.3164633512496948,
+      "learning_rate": 5.0267065263426125e-05,
+      "loss": 0.02421087771654129,
+      "mean_token_accuracy": 0.9917516648769379,
+      "num_tokens": 70470266.0,
+      "step": 3890
+    },
+    {
+      "entropy": 1.0089037418365479,
+      "epoch": 5.248990578734858,
+      "grad_norm": 1.1788280010223389,
+      "learning_rate": 5.00445110830071e-05,
+      "loss": 0.023224952816963195,
+      "mean_token_accuracy": 0.9921185493469238,
+      "num_tokens": 70650975.0,
+      "step": 3900
+    },
+    {
+      "entropy": 1.001932644844055,
+      "epoch": 5.262449528936743,
+      "grad_norm": 1.2484381198883057,
+      "learning_rate": 4.9821956020719474e-05,
+      "loss": 0.02366664707660675,
+      "mean_token_accuracy": 0.9917596876621246,
+      "num_tokens": 70832549.0,
+      "step": 3910
+    },
+    {
+      "entropy": 1.004903507232666,
+      "epoch": 5.275908479138627,
+      "grad_norm": 1.1852507591247559,
+      "learning_rate": 4.959940448589928e-05,
+      "loss": 0.02296685576438904,
+      "mean_token_accuracy": 0.9918978750705719,
+      "num_tokens": 71014296.0,
+      "step": 3920
+    },
+    {
+      "entropy": 1.004357820749283,
+      "epoch": 5.289367429340511,
+      "grad_norm": 0.9933306574821472,
+      "learning_rate": 4.9376860887812666e-05,
+      "loss": 0.02361467629671097,
+      "mean_token_accuracy": 0.991608303785324,
+      "num_tokens": 71195526.0,
+      "step": 3930
+    },
+    {
+      "entropy": 1.0020038604736328,
+      "epoch": 5.302826379542395,
+      "grad_norm": 1.4304245710372925,
+      "learning_rate": 4.915432963556853e-05,
+      "loss": 0.02268100529909134,
+      "mean_token_accuracy": 0.9917695164680481,
+      "num_tokens": 71377269.0,
+      "step": 3940
+    },
+    {
+      "entropy": 1.012037169933319,
+      "epoch": 5.31628532974428,
+      "grad_norm": 1.074124813079834,
+      "learning_rate": 4.8931815138031173e-05,
+      "loss": 0.023454257845878602,
+      "mean_token_accuracy": 0.9922087669372559,
+      "num_tokens": 71558158.0,
+      "step": 3950
+    },
+    {
+      "entropy": 1.0094320178031921,
+      "epoch": 5.329744279946164,
+      "grad_norm": 1.0919133424758911,
+      "learning_rate": 4.870932180373296e-05,
+      "loss": 0.022751571238040925,
+      "mean_token_accuracy": 0.9921496629714965,
+      "num_tokens": 71738847.0,
+      "step": 3960
+    },
+    {
+      "entropy": 1.0063781797885896,
+      "epoch": 5.343203230148048,
+      "grad_norm": 1.1300997734069824,
+      "learning_rate": 4.8486854040786926e-05,
+      "loss": 0.023900465667247774,
+      "mean_token_accuracy": 0.9914113521575928,
+      "num_tokens": 71918875.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.9972042560577392,
+      "epoch": 5.356662180349932,
+      "grad_norm": 1.165519118309021,
+      "learning_rate": 4.826441625679953e-05,
+      "loss": 0.021918240189552306,
+      "mean_token_accuracy": 0.9926528394222259,
+      "num_tokens": 72100551.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.9954554200172424,
+      "epoch": 5.370121130551817,
+      "grad_norm": 1.342820167541504,
+      "learning_rate": 4.8042012858783223e-05,
+      "loss": 0.024170228838920595,
+      "mean_token_accuracy": 0.9913744986057281,
+      "num_tokens": 72281855.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.9931409180164337,
+      "epoch": 5.383580080753701,
+      "grad_norm": 1.2873902320861816,
+      "learning_rate": 4.781964825306923e-05,
+      "loss": 0.02334785759449005,
+      "mean_token_accuracy": 0.9921288907527923,
+      "num_tokens": 72462916.0,
+      "step": 4000
+    },
+    {
+      "entropy": 0.9915881812572479,
+      "epoch": 5.397039030955585,
+      "grad_norm": 0.9761195778846741,
+      "learning_rate": 4.7597326845220206e-05,
+      "loss": 0.02302718609571457,
+      "mean_token_accuracy": 0.9924728453159333,
+      "num_tokens": 72644666.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.9931966602802277,
+      "epoch": 5.410497981157469,
+      "grad_norm": 1.4468477964401245,
+      "learning_rate": 4.737505303994292e-05,
+      "loss": 0.023682811856269838,
+      "mean_token_accuracy": 0.9917364478111267,
+      "num_tokens": 72825808.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.9975718319416046,
+      "epoch": 5.423956931359354,
+      "grad_norm": 1.1559572219848633,
+      "learning_rate": 4.7152831241001065e-05,
+      "loss": 0.02247478663921356,
+      "mean_token_accuracy": 0.992560338973999,
+      "num_tokens": 73007923.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.999228036403656,
+      "epoch": 5.437415881561238,
+      "grad_norm": 1.1230353116989136,
+      "learning_rate": 4.693066585112795e-05,
+      "loss": 0.022295859456062318,
+      "mean_token_accuracy": 0.9922321498394012,
+      "num_tokens": 73189074.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.9897105693817139,
+      "epoch": 5.450874831763122,
+      "grad_norm": 1.413734793663025,
+      "learning_rate": 4.670856127193928e-05,
+      "loss": 0.02201971709728241,
+      "mean_token_accuracy": 0.9927073001861573,
+      "num_tokens": 73369918.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.9986266851425171,
+      "epoch": 5.464333781965006,
+      "grad_norm": 1.0815346240997314,
+      "learning_rate": 4.648652190384597e-05,
+      "loss": 0.02358766496181488,
+      "mean_token_accuracy": 0.9911752700805664,
+      "num_tokens": 73550625.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.9953943610191345,
+      "epoch": 5.477792732166891,
+      "grad_norm": 1.0506970882415771,
+      "learning_rate": 4.626455214596695e-05,
+      "loss": 0.022177645564079286,
+      "mean_token_accuracy": 0.9923022150993347,
+      "num_tokens": 73731436.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.9954936504364014,
+      "epoch": 5.491251682368775,
+      "grad_norm": 1.4665249586105347,
+      "learning_rate": 4.6042656396042e-05,
+      "loss": 0.024948063492774963,
+      "mean_token_accuracy": 0.9913177132606507,
+      "num_tokens": 73912097.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.9986696183681488,
+      "epoch": 5.504710632570659,
+      "grad_norm": 1.158179759979248,
+      "learning_rate": 4.5820839050344643e-05,
+      "loss": 0.02555338144302368,
+      "mean_token_accuracy": 0.9908900320529938,
+      "num_tokens": 74093602.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.9955579698085785,
+      "epoch": 5.518169582772543,
+      "grad_norm": 1.0024714469909668,
+      "learning_rate": 4.559910450359502e-05,
+      "loss": 0.023432795703411103,
+      "mean_token_accuracy": 0.9919674038887024,
+      "num_tokens": 74274664.0,
+      "step": 4100
+    },
+    {
+      "entropy": 0.994187182188034,
+      "epoch": 5.531628532974428,
+      "grad_norm": 0.9290911555290222,
+      "learning_rate": 4.5377457148872837e-05,
+      "loss": 0.023194606602191924,
+      "mean_token_accuracy": 0.9920254707336426,
+      "num_tokens": 74456436.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.9927510738372802,
+      "epoch": 5.545087483176312,
+      "grad_norm": 0.8986559510231018,
+      "learning_rate": 4.515590137753032e-05,
+      "loss": 0.02163631021976471,
+      "mean_token_accuracy": 0.9927416622638703,
+      "num_tokens": 74637700.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.9877451896667481,
+      "epoch": 5.558546433378196,
+      "grad_norm": 1.0314558744430542,
+      "learning_rate": 4.493444157910521e-05,
+      "loss": 0.0209101140499115,
+      "mean_token_accuracy": 0.9930387079715729,
+      "num_tokens": 74819310.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.9903139114379883,
+      "epoch": 5.5720053835800805,
+      "grad_norm": 1.1819736957550049,
+      "learning_rate": 4.471308214123381e-05,
+      "loss": 0.02219167649745941,
+      "mean_token_accuracy": 0.992407250404358,
+      "num_tokens": 74999907.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.9891258835792541,
+      "epoch": 5.585464333781965,
+      "grad_norm": 1.0412501096725464,
+      "learning_rate": 4.449182744956403e-05,
+      "loss": 0.02201283872127533,
+      "mean_token_accuracy": 0.992091816663742,
+      "num_tokens": 75181189.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.9870602369308472,
+      "epoch": 5.598923283983849,
+      "grad_norm": 1.0461134910583496,
+      "learning_rate": 4.4270681887668544e-05,
+      "loss": 0.023040324449539185,
+      "mean_token_accuracy": 0.992098081111908,
+      "num_tokens": 75362769.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.9888905882835388,
+      "epoch": 5.612382234185733,
+      "grad_norm": 1.086546540260315,
+      "learning_rate": 4.404964983695786e-05,
+      "loss": 0.021401138603687288,
+      "mean_token_accuracy": 0.992308247089386,
+      "num_tokens": 75544519.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.9940549910068512,
+      "epoch": 5.6258411843876175,
+      "grad_norm": 1.3897169828414917,
+      "learning_rate": 4.382873567659361e-05,
+      "loss": 0.023297235369682312,
+      "mean_token_accuracy": 0.9920303761959076,
+      "num_tokens": 75725651.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.9897378265857697,
+      "epoch": 5.639300134589502,
+      "grad_norm": 1.283887267112732,
+      "learning_rate": 4.3607943783401736e-05,
+      "loss": 0.024343529343605043,
+      "mean_token_accuracy": 0.9916397035121918,
+      "num_tokens": 75907036.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.9989661812782288,
+      "epoch": 5.652759084791386,
+      "grad_norm": 0.9090050458908081,
+      "learning_rate": 4.3387278531785747e-05,
+      "loss": 0.022062870860099792,
+      "mean_token_accuracy": 0.9921731591224671,
+      "num_tokens": 76088248.0,
+      "step": 4200
+    },
+    {
+      "entropy": 0.9966195583343506,
+      "epoch": 5.66621803499327,
+      "grad_norm": 1.2091970443725586,
+      "learning_rate": 4.3166744293640134e-05,
+      "loss": 0.023516717553138732,
+      "mean_token_accuracy": 0.9918879508972168,
+      "num_tokens": 76269704.0,
+      "step": 4210
+    },
+    {
+      "entropy": 1.0081237196922301,
+      "epoch": 5.6796769851951545,
+      "grad_norm": 1.0648062229156494,
+      "learning_rate": 4.2946345438263665e-05,
+      "loss": 0.021994276344776152,
+      "mean_token_accuracy": 0.9925018906593323,
+      "num_tokens": 76451416.0,
+      "step": 4220
+    },
+    {
+      "entropy": 1.0061897933483124,
+      "epoch": 5.693135935397039,
+      "grad_norm": 1.188748836517334,
+      "learning_rate": 4.272608633227287e-05,
+      "loss": 0.022657927870750428,
+      "mean_token_accuracy": 0.9919365108013153,
+      "num_tokens": 76632827.0,
+      "step": 4230
+    },
+    {
+      "entropy": 1.0024715662002563,
+      "epoch": 5.706594885598923,
+      "grad_norm": 1.2140473127365112,
+      "learning_rate": 4.250597133951554e-05,
+      "loss": 0.02203955501317978,
+      "mean_token_accuracy": 0.9925529718399048,
+      "num_tokens": 76813534.0,
+      "step": 4240
+    },
+    {
+      "entropy": 1.0011352360248567,
+      "epoch": 5.720053835800807,
+      "grad_norm": 1.389479398727417,
+      "learning_rate": 4.228600482098423e-05,
+      "loss": 0.022718487679958342,
+      "mean_token_accuracy": 0.9922778069972992,
+      "num_tokens": 76994327.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.9971556246280671,
+      "epoch": 5.7335127860026915,
+      "grad_norm": 1.1441208124160767,
+      "learning_rate": 4.206619113472986e-05,
+      "loss": 0.024178403615951537,
+      "mean_token_accuracy": 0.9913869142532349,
+      "num_tokens": 77174987.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.9961654126644135,
+      "epoch": 5.746971736204576,
+      "grad_norm": 1.1283735036849976,
+      "learning_rate": 4.18465346357754e-05,
+      "loss": 0.023389075696468354,
+      "mean_token_accuracy": 0.9921939969062805,
+      "num_tokens": 77356043.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.9927134871482849,
+      "epoch": 5.76043068640646,
+      "grad_norm": 1.201904535293579,
+      "learning_rate": 4.16270396760296e-05,
+      "loss": 0.022739484906196594,
+      "mean_token_accuracy": 0.9921142935752869,
+      "num_tokens": 77536396.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.9919731795787812,
+      "epoch": 5.773889636608344,
+      "grad_norm": 1.145776391029358,
+      "learning_rate": 4.140771060420066e-05,
+      "loss": 0.020524097979068755,
+      "mean_token_accuracy": 0.9926942765712738,
+      "num_tokens": 77717339.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.9935499131679535,
+      "epoch": 5.787348586810229,
+      "grad_norm": 1.099414348602295,
+      "learning_rate": 4.118855176571021e-05,
+      "loss": 0.02070924937725067,
+      "mean_token_accuracy": 0.9928180515766144,
+      "num_tokens": 77898360.0,
+      "step": 4300
+    },
+    {
+      "entropy": 0.9875575602054596,
+      "epoch": 5.800807537012113,
+      "grad_norm": 1.1510778665542603,
+      "learning_rate": 4.096956750260718e-05,
+      "loss": 0.02202107310295105,
+      "mean_token_accuracy": 0.9921730697154999,
+      "num_tokens": 78079699.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.9855418741703034,
+      "epoch": 5.814266487213997,
+      "grad_norm": 1.0748379230499268,
+      "learning_rate": 4.07507621534817e-05,
+      "loss": 0.022809866070747375,
+      "mean_token_accuracy": 0.9919199466705322,
+      "num_tokens": 78260900.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.977304744720459,
+      "epoch": 5.827725437415881,
+      "grad_norm": 1.091854453086853,
+      "learning_rate": 4.053214005337924e-05,
+      "loss": 0.02268455773591995,
+      "mean_token_accuracy": 0.9918094217777252,
+      "num_tokens": 78442168.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.9767287075519562,
+      "epoch": 5.841184387617766,
+      "grad_norm": 1.1712645292282104,
+      "learning_rate": 4.031370553371465e-05,
+      "loss": 0.022726841270923615,
+      "mean_token_accuracy": 0.9922167181968689,
+      "num_tokens": 78623589.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.974297159910202,
+      "epoch": 5.85464333781965,
+      "grad_norm": 0.9508016109466553,
+      "learning_rate": 4.0095462922186385e-05,
+      "loss": 0.021290253102779388,
+      "mean_token_accuracy": 0.9926727712154388,
+      "num_tokens": 78804892.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.9795810282230377,
+      "epoch": 5.868102288021534,
+      "grad_norm": 1.3127822875976562,
+      "learning_rate": 3.9877416542690746e-05,
+      "loss": 0.022329577803611757,
+      "mean_token_accuracy": 0.9927741348743438,
+      "num_tokens": 78986204.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.9879997313022614,
+      "epoch": 5.881561238223418,
+      "grad_norm": 1.0200371742248535,
+      "learning_rate": 3.9659570715236234e-05,
+      "loss": 0.022206541895866395,
+      "mean_token_accuracy": 0.9921830892562866,
+      "num_tokens": 79166824.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.9801069736480713,
+      "epoch": 5.895020188425303,
+      "grad_norm": 1.4769432544708252,
+      "learning_rate": 3.944192975585792e-05,
+      "loss": 0.02101587802171707,
+      "mean_token_accuracy": 0.9930795848369598,
+      "num_tokens": 79347228.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.9727071464061737,
+      "epoch": 5.908479138627187,
+      "grad_norm": 1.7034555673599243,
+      "learning_rate": 3.922449797653198e-05,
+      "loss": 0.021690081059932708,
+      "mean_token_accuracy": 0.9926194071769714,
+      "num_tokens": 79529483.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.9782697260379791,
+      "epoch": 5.921938088829071,
+      "grad_norm": 1.5307973623275757,
+      "learning_rate": 3.900727968509024e-05,
+      "loss": 0.023112425208091737,
+      "mean_token_accuracy": 0.9919495701789856,
+      "num_tokens": 79710571.0,
+      "step": 4400
+    },
+    {
+      "entropy": 0.9831463754177093,
+      "epoch": 5.935397039030955,
+      "grad_norm": 1.1024526357650757,
+      "learning_rate": 3.879027918513483e-05,
+      "loss": 0.021427598595619202,
+      "mean_token_accuracy": 0.9924166023731231,
+      "num_tokens": 79890685.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.9811958551406861,
+      "epoch": 5.94885598923284,
+      "grad_norm": 1.7247848510742188,
+      "learning_rate": 3.857350077595289e-05,
+      "loss": 0.02339867949485779,
+      "mean_token_accuracy": 0.9916171073913574,
+      "num_tokens": 80071331.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.9830600321292877,
+      "epoch": 5.962314939434724,
+      "grad_norm": 1.0638445615768433,
+      "learning_rate": 3.835694875243149e-05,
+      "loss": 0.02204108238220215,
+      "mean_token_accuracy": 0.9925127327442169,
+      "num_tokens": 80251525.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.9840480864048005,
+      "epoch": 5.975773889636608,
+      "grad_norm": 1.1205570697784424,
+      "learning_rate": 3.814062740497243e-05,
+      "loss": 0.020443232357501985,
+      "mean_token_accuracy": 0.9932060956954956,
+      "num_tokens": 80432001.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.9841626048088074,
+      "epoch": 5.989232839838492,
+      "grad_norm": 1.2928433418273926,
+      "learning_rate": 3.7924541019407264e-05,
+      "loss": 0.021673919260501863,
+      "mean_token_accuracy": 0.9931873917579651,
+      "num_tokens": 80613075.0,
+      "step": 4450
+    },
+    {
+      "epoch": 6.0,
+      "eval_entropy": 0.9904457463580332,
+      "eval_loss": 0.161154642701149,
+      "eval_mean_token_accuracy": 0.9516840383505366,
+      "eval_num_tokens": 80757887.0,
+      "eval_runtime": 12.7345,
+      "eval_samples_per_second": 392.633,
+      "eval_steps_per_second": 12.329,
+      "step": 4458
+    },
+    {
+      "entropy": 0.9850570559501648,
+      "epoch": 6.002691790040377,
+      "grad_norm": 0.8859100341796875,
+      "learning_rate": 3.7708693876912435e-05,
+      "loss": 0.018883128464221955,
+      "mean_token_accuracy": 0.9933207035064697,
+      "num_tokens": 80794295.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.9796536922454834,
+      "epoch": 6.016150740242261,
+      "grad_norm": 0.8121339082717896,
+      "learning_rate": 3.74930902539244e-05,
+      "loss": 0.01288142502307892,
+      "mean_token_accuracy": 0.9960863590240479,
+      "num_tokens": 80975379.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.9699741184711457,
+      "epoch": 6.029609690444145,
+      "grad_norm": 0.8923460841178894,
+      "learning_rate": 3.727773442205493e-05,
+      "loss": 0.012284834682941437,
+      "mean_token_accuracy": 0.9960344016551972,
+      "num_tokens": 81155713.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.9629065454006195,
+      "epoch": 6.043068640646029,
+      "grad_norm": 0.6698539853096008,
+      "learning_rate": 3.7062630648006485e-05,
+      "loss": 0.011575647443532944,
+      "mean_token_accuracy": 0.9962593257427216,
+      "num_tokens": 81336446.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.9579961061477661,
+      "epoch": 6.056527590847914,
+      "grad_norm": 0.7555826306343079,
+      "learning_rate": 3.684778319348765e-05,
+      "loss": 0.011635755002498627,
+      "mean_token_accuracy": 0.9965786218643189,
+      "num_tokens": 81517969.0,
+      "step": 4500
+    },
+    {
+      "entropy": 0.96330127120018,
+      "epoch": 6.069986541049798,
+      "grad_norm": 0.8983458280563354,
+      "learning_rate": 3.663319631512874e-05,
+      "loss": 0.012715654075145721,
+      "mean_token_accuracy": 0.9959319531917572,
+      "num_tokens": 81699419.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.9640884518623352,
+      "epoch": 6.083445491251682,
+      "grad_norm": 0.9172118902206421,
+      "learning_rate": 3.641887426439743e-05,
+      "loss": 0.012482476234436036,
+      "mean_token_accuracy": 0.9966461837291718,
+      "num_tokens": 81880956.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.9678717851638794,
+      "epoch": 6.0969044414535665,
+      "grad_norm": 0.8765719532966614,
+      "learning_rate": 3.620482128751456e-05,
+      "loss": 0.012080545723438262,
+      "mean_token_accuracy": 0.9964690268039703,
+      "num_tokens": 82061897.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.9656756520271301,
+      "epoch": 6.110363391655451,
+      "grad_norm": 0.746626079082489,
+      "learning_rate": 3.599104162536997e-05,
+      "loss": 0.012495596706867219,
+      "mean_token_accuracy": 0.9959208548069001,
+      "num_tokens": 82242847.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.9631998062133789,
+      "epoch": 6.123822341857335,
+      "grad_norm": 0.9732807874679565,
+      "learning_rate": 3.577753951343851e-05,
+      "loss": 0.01309359222650528,
+      "mean_token_accuracy": 0.9960879981517792,
+      "num_tokens": 82424182.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.9686778664588929,
+      "epoch": 6.137281292059219,
+      "grad_norm": 1.040855884552002,
+      "learning_rate": 3.55643191816961e-05,
+      "loss": 0.013380664587020873,
+      "mean_token_accuracy": 0.9955641269683838,
+      "num_tokens": 82604728.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.9661992788314819,
+      "epoch": 6.1507402422611035,
+      "grad_norm": 1.209067940711975,
+      "learning_rate": 3.535138485453595e-05,
+      "loss": 0.013264504075050355,
+      "mean_token_accuracy": 0.9958661019802093,
+      "num_tokens": 82786184.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.9628405570983887,
+      "epoch": 6.164199192462988,
+      "grad_norm": 0.9056928157806396,
+      "learning_rate": 3.513874075068484e-05,
+      "loss": 0.011586685478687287,
+      "mean_token_accuracy": 0.9962525606155396,
+      "num_tokens": 82968127.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.9664590656757355,
+      "epoch": 6.177658142664872,
+      "grad_norm": 0.8662444949150085,
+      "learning_rate": 3.492639108311955e-05,
+      "loss": 0.013227257132530212,
+      "mean_token_accuracy": 0.9958321928977967,
+      "num_tokens": 83149016.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.9584677755832672,
+      "epoch": 6.191117092866756,
+      "grad_norm": 0.8611947894096375,
+      "learning_rate": 3.471434005898339e-05,
+      "loss": 0.012484049052000045,
+      "mean_token_accuracy": 0.9961173057556152,
+      "num_tokens": 83330474.0,
+      "step": 4600
+    },
+    {
+      "entropy": 0.9613959610462188,
+      "epoch": 6.2045760430686405,
+      "grad_norm": 1.0272881984710693,
+      "learning_rate": 3.450259187950283e-05,
+      "loss": 0.013143527507781982,
+      "mean_token_accuracy": 0.995938628911972,
+      "num_tokens": 83512502.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.9671844780445099,
+      "epoch": 6.218034993270525,
+      "grad_norm": 1.038604497909546,
+      "learning_rate": 3.429115073990431e-05,
+      "loss": 0.015797269344329835,
+      "mean_token_accuracy": 0.9949265837669372,
+      "num_tokens": 83693427.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.9725019454956054,
+      "epoch": 6.231493943472409,
+      "grad_norm": 0.7591229677200317,
+      "learning_rate": 3.408002082933107e-05,
+      "loss": 0.01309441477060318,
+      "mean_token_accuracy": 0.9960084438323975,
+      "num_tokens": 83874886.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.9710726916790009,
+      "epoch": 6.244952893674293,
+      "grad_norm": 0.9521571397781372,
+      "learning_rate": 3.3869206330760187e-05,
+      "loss": 0.012572157382965087,
+      "mean_token_accuracy": 0.9960904121398926,
+      "num_tokens": 84055459.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.9712884068489075,
+      "epoch": 6.2584118438761775,
+      "grad_norm": 0.812011182308197,
+      "learning_rate": 3.365871142091968e-05,
+      "loss": 0.014366535842418671,
+      "mean_token_accuracy": 0.9953844547271729,
+      "num_tokens": 84236017.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.9648497045040131,
+      "epoch": 6.271870794078062,
+      "grad_norm": 1.648348093032837,
+      "learning_rate": 3.3448540270205766e-05,
+      "loss": 0.014780099689960479,
+      "mean_token_accuracy": 0.9955918669700623,
+      "num_tokens": 84417271.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.956806218624115,
+      "epoch": 6.285329744279946,
+      "grad_norm": 0.9592521786689758,
+      "learning_rate": 3.323869704260025e-05,
+      "loss": 0.014198371767997741,
+      "mean_token_accuracy": 0.9954844176769256,
+      "num_tokens": 84599326.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.9628090143203736,
+      "epoch": 6.29878869448183,
+      "grad_norm": 0.9466222524642944,
+      "learning_rate": 3.302918589558801e-05,
+      "loss": 0.013178789615631103,
+      "mean_token_accuracy": 0.995667016506195,
+      "num_tokens": 84779914.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.9617082297801971,
+      "epoch": 6.312247644683715,
+      "grad_norm": 0.9156956672668457,
+      "learning_rate": 3.282001098007462e-05,
+      "loss": 0.011176959425210953,
+      "mean_token_accuracy": 0.9965419709682465,
+      "num_tokens": 84960572.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.9526888430118561,
+      "epoch": 6.325706594885599,
+      "grad_norm": 1.107490062713623,
+      "learning_rate": 3.261117644030412e-05,
+      "loss": 0.01246263161301613,
+      "mean_token_accuracy": 0.9961843729019165,
+      "num_tokens": 85141667.0,
+      "step": 4700
+    },
+    {
+      "entropy": 0.952410751581192,
+      "epoch": 6.339165545087483,
+      "grad_norm": 1.06051766872406,
+      "learning_rate": 3.240268641377694e-05,
+      "loss": 0.01356920450925827,
+      "mean_token_accuracy": 0.9956991016864777,
+      "num_tokens": 85323361.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.9601339221000671,
+      "epoch": 6.352624495289367,
+      "grad_norm": 1.0533535480499268,
+      "learning_rate": 3.2194545031167866e-05,
+      "loss": 0.012942484021186829,
+      "mean_token_accuracy": 0.9960961997509002,
+      "num_tokens": 85503828.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.9630769610404968,
+      "epoch": 6.366083445491252,
+      "grad_norm": 1.2243846654891968,
+      "learning_rate": 3.1986756416244245e-05,
+      "loss": 0.012688860297203064,
+      "mean_token_accuracy": 0.9956350862979889,
+      "num_tokens": 85684804.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.9601665258407592,
+      "epoch": 6.379542395693136,
+      "grad_norm": 1.0219420194625854,
+      "learning_rate": 3.177932468578426e-05,
+      "loss": 0.012651938199996948,
+      "mean_token_accuracy": 0.9959766507148743,
+      "num_tokens": 85866033.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.9643365681171417,
+      "epoch": 6.39300134589502,
+      "grad_norm": 0.8759870529174805,
+      "learning_rate": 3.157225394949542e-05,
+      "loss": 0.013052919507026672,
+      "mean_token_accuracy": 0.9956570088863372,
+      "num_tokens": 86046793.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.9643962621688843,
+      "epoch": 6.406460296096904,
+      "grad_norm": 0.8771425485610962,
+      "learning_rate": 3.136554830993304e-05,
+      "loss": 0.013016942143440246,
+      "mean_token_accuracy": 0.9956202030181884,
+      "num_tokens": 86227594.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.9684974253177643,
+      "epoch": 6.419919246298789,
+      "grad_norm": 0.9857693314552307,
+      "learning_rate": 3.115921186241906e-05,
+      "loss": 0.01260487139225006,
+      "mean_token_accuracy": 0.9959571301937103,
+      "num_tokens": 86409012.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.9683425962924957,
+      "epoch": 6.433378196500673,
+      "grad_norm": 0.9220049381256104,
+      "learning_rate": 3.0953248694960824e-05,
+      "loss": 0.012756258249282837,
+      "mean_token_accuracy": 0.9961577892303467,
+      "num_tokens": 86589770.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.9730850279331207,
+      "epoch": 6.446837146702557,
+      "grad_norm": 1.1092842817306519,
+      "learning_rate": 3.0747662888170146e-05,
+      "loss": 0.011953817307949066,
+      "mean_token_accuracy": 0.9963648617267609,
+      "num_tokens": 86771224.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.9717083096504211,
+      "epoch": 6.460296096904441,
+      "grad_norm": 0.9012715220451355,
+      "learning_rate": 3.054245851518246e-05,
+      "loss": 0.0137409508228302,
+      "mean_token_accuracy": 0.9956796884536743,
+      "num_tokens": 86952165.0,
+      "step": 4800
+    },
+    {
+      "entropy": 0.9706744968891143,
+      "epoch": 6.473755047106326,
+      "grad_norm": 0.8237823247909546,
+      "learning_rate": 3.0337639641576065e-05,
+      "loss": 0.013622400164604188,
+      "mean_token_accuracy": 0.9958401381969452,
+      "num_tokens": 87133948.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.968240886926651,
+      "epoch": 6.48721399730821,
+      "grad_norm": 0.9172955751419067,
+      "learning_rate": 3.0133210325291662e-05,
+      "loss": 0.012218748033046723,
+      "mean_token_accuracy": 0.9961099803447724,
+      "num_tokens": 87315186.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.9701445996761322,
+      "epoch": 6.500672947510094,
+      "grad_norm": 1.1613796949386597,
+      "learning_rate": 2.9929174616551857e-05,
+      "loss": 0.011357608437538146,
+      "mean_token_accuracy": 0.9961966574192047,
+      "num_tokens": 87496318.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.9753403663635254,
+      "epoch": 6.514131897711978,
+      "grad_norm": 1.0601547956466675,
+      "learning_rate": 2.9725536557781008e-05,
+      "loss": 0.01224931925535202,
+      "mean_token_accuracy": 0.9964543879032135,
+      "num_tokens": 87676690.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.9595320463180542,
+      "epoch": 6.527590847913863,
+      "grad_norm": 0.8927657604217529,
+      "learning_rate": 2.9522300183525097e-05,
+      "loss": 0.012476099282503128,
+      "mean_token_accuracy": 0.9961197137832641,
+      "num_tokens": 87858122.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.9639881610870361,
+      "epoch": 6.541049798115747,
+      "grad_norm": 0.9392650127410889,
+      "learning_rate": 2.931946952037179e-05,
+      "loss": 0.013359837234020233,
+      "mean_token_accuracy": 0.9956938862800598,
+      "num_tokens": 88039349.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.961546677350998,
+      "epoch": 6.554508748317631,
+      "grad_norm": 0.709064781665802,
+      "learning_rate": 2.9117048586870654e-05,
+      "loss": 0.012232375890016555,
+      "mean_token_accuracy": 0.9963175058364868,
+      "num_tokens": 88221085.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.966980516910553,
+      "epoch": 6.5679676985195155,
+      "grad_norm": 1.1382325887680054,
+      "learning_rate": 2.891504139345358e-05,
+      "loss": 0.013281235098838806,
+      "mean_token_accuracy": 0.9955899178981781,
+      "num_tokens": 88401750.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.967098617553711,
+      "epoch": 6.5814266487214,
+      "grad_norm": 0.7260605692863464,
+      "learning_rate": 2.8713451942355285e-05,
+      "loss": 0.011995580792427064,
+      "mean_token_accuracy": 0.9963020443916321,
+      "num_tokens": 88582188.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.9611273169517517,
+      "epoch": 6.594885598923284,
+      "grad_norm": 0.9492855668067932,
+      "learning_rate": 2.8512284227534027e-05,
+      "loss": 0.011460936814546584,
+      "mean_token_accuracy": 0.9965442717075348,
+      "num_tokens": 88764018.0,
+      "step": 4900
+    },
+    {
+      "entropy": 0.964341527223587,
+      "epoch": 6.608344549125168,
+      "grad_norm": 0.7225981950759888,
+      "learning_rate": 2.8311542234592497e-05,
+      "loss": 0.011858727037906646,
+      "mean_token_accuracy": 0.996355015039444,
+      "num_tokens": 88944988.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.9631503582000732,
+      "epoch": 6.6218034993270525,
+      "grad_norm": 0.996126651763916,
+      "learning_rate": 2.8111229940698842e-05,
+      "loss": 0.013101565837860107,
+      "mean_token_accuracy": 0.9957616865634918,
+      "num_tokens": 89125768.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.9629738807678223,
+      "epoch": 6.635262449528937,
+      "grad_norm": 0.8971033096313477,
+      "learning_rate": 2.791135131450785e-05,
+      "loss": 0.012153981626033783,
+      "mean_token_accuracy": 0.9961553812026978,
+      "num_tokens": 89306694.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.9592839002609252,
+      "epoch": 6.648721399730821,
+      "grad_norm": 0.8334258198738098,
+      "learning_rate": 2.7711910316082357e-05,
+      "loss": 0.012248124927282333,
+      "mean_token_accuracy": 0.9960185050964355,
+      "num_tokens": 89487628.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.9580614447593689,
+      "epoch": 6.662180349932705,
+      "grad_norm": 0.7489107251167297,
+      "learning_rate": 2.7512910896814747e-05,
+      "loss": 0.012415885925292969,
+      "mean_token_accuracy": 0.9960191190242768,
+      "num_tokens": 89668907.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.9544502675533295,
+      "epoch": 6.6756393001345895,
+      "grad_norm": 0.9663529992103577,
+      "learning_rate": 2.7314356999348713e-05,
+      "loss": 0.012825769186019898,
+      "mean_token_accuracy": 0.9958091497421264,
+      "num_tokens": 89850200.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.9588588118553162,
+      "epoch": 6.689098250336474,
+      "grad_norm": 0.7662053108215332,
+      "learning_rate": 2.711625255750111e-05,
+      "loss": 0.012289515137672425,
+      "mean_token_accuracy": 0.9956038355827331,
+      "num_tokens": 90030738.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.9632011234760285,
+      "epoch": 6.702557200538358,
+      "grad_norm": 0.8352729082107544,
+      "learning_rate": 2.691860149618402e-05,
+      "loss": 0.01209896057844162,
+      "mean_token_accuracy": 0.9960383713245392,
+      "num_tokens": 90211066.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.962366932630539,
+      "epoch": 6.716016150740242,
+      "grad_norm": 1.271884799003601,
+      "learning_rate": 2.6721407731327004e-05,
+      "loss": 0.013325554132461549,
+      "mean_token_accuracy": 0.9955421090126038,
+      "num_tokens": 90392202.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.9630074858665466,
+      "epoch": 6.7294751009421265,
+      "grad_norm": 0.806925892829895,
+      "learning_rate": 2.6524675169799506e-05,
+      "loss": 0.012963607907295227,
+      "mean_token_accuracy": 0.9955462396144867,
+      "num_tokens": 90573271.0,
+      "step": 5000
+    },
+    {
+      "entropy": 0.9631992042064667,
+      "epoch": 6.742934051144011,
+      "grad_norm": 0.9406610131263733,
+      "learning_rate": 2.6328407709333463e-05,
+      "loss": 0.012753760814666748,
+      "mean_token_accuracy": 0.9954385042190552,
+      "num_tokens": 90753836.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.9599998474121094,
+      "epoch": 6.756393001345895,
+      "grad_norm": 0.7755982875823975,
+      "learning_rate": 2.6132609238446072e-05,
+      "loss": 0.011351206153631211,
+      "mean_token_accuracy": 0.9965151965618133,
+      "num_tokens": 90935528.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.961510854959488,
+      "epoch": 6.769851951547779,
+      "grad_norm": 1.223897099494934,
+      "learning_rate": 2.5937283636362724e-05,
+      "loss": 0.012862575054168702,
+      "mean_token_accuracy": 0.995742017030716,
+      "num_tokens": 91116520.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.9597096800804138,
+      "epoch": 6.783310901749664,
+      "grad_norm": 0.8847037553787231,
+      "learning_rate": 2.5742434772940216e-05,
+      "loss": 0.01192454993724823,
+      "mean_token_accuracy": 0.9964925825595856,
+      "num_tokens": 91297847.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.961575037240982,
+      "epoch": 6.796769851951548,
+      "grad_norm": 0.7097873687744141,
+      "learning_rate": 2.5548066508590007e-05,
+      "loss": 0.012100750207901,
+      "mean_token_accuracy": 0.9963988482952117,
+      "num_tokens": 91479371.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.9660714268684387,
+      "epoch": 6.810228802153432,
+      "grad_norm": 0.8912429213523865,
+      "learning_rate": 2.535418269420178e-05,
+      "loss": 0.011809615790843964,
+      "mean_token_accuracy": 0.9961784362792969,
+      "num_tokens": 91660175.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.9656784057617187,
+      "epoch": 6.823687752355316,
+      "grad_norm": 1.2362871170043945,
+      "learning_rate": 2.5160787171067126e-05,
+      "loss": 0.013092437386512756,
+      "mean_token_accuracy": 0.9954990804195404,
+      "num_tokens": 91840728.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.9630216419696808,
+      "epoch": 6.837146702557201,
+      "grad_norm": 1.0081552267074585,
+      "learning_rate": 2.4967883770803413e-05,
+      "loss": 0.013006564974784852,
+      "mean_token_accuracy": 0.995445728302002,
+      "num_tokens": 92021819.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.9650352597236633,
+      "epoch": 6.850605652759085,
+      "grad_norm": 1.0923664569854736,
+      "learning_rate": 2.477547631527799e-05,
+      "loss": 0.012209924310445786,
+      "mean_token_accuracy": 0.9961765468120575,
+      "num_tokens": 92203146.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.9654329180717468,
+      "epoch": 6.864064602960969,
+      "grad_norm": 0.8630998730659485,
+      "learning_rate": 2.45835686165323e-05,
+      "loss": 0.01286689043045044,
+      "mean_token_accuracy": 0.9955997586250305,
+      "num_tokens": 92384274.0,
+      "step": 5100
+    },
+    {
+      "entropy": 0.9633705973625183,
+      "epoch": 6.877523553162853,
+      "grad_norm": 0.9883823394775391,
+      "learning_rate": 2.4392164476706468e-05,
+      "loss": 0.012318137288093566,
+      "mean_token_accuracy": 0.9964197039604187,
+      "num_tokens": 92564922.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.9619575142860413,
+      "epoch": 6.890982503364738,
+      "grad_norm": 1.0694894790649414,
+      "learning_rate": 2.4201267687963935e-05,
+      "loss": 0.011325329542160034,
+      "mean_token_accuracy": 0.9967617630958557,
+      "num_tokens": 92745952.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.9608132243156433,
+      "epoch": 6.904441453566622,
+      "grad_norm": 0.850687563419342,
+      "learning_rate": 2.4010882032416332e-05,
+      "loss": 0.012749406695365905,
+      "mean_token_accuracy": 0.9958359479904175,
+      "num_tokens": 92926966.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.9643485128879548,
+      "epoch": 6.917900403768506,
+      "grad_norm": 0.8775151968002319,
+      "learning_rate": 2.3821011282048545e-05,
+      "loss": 0.012118304520845414,
+      "mean_token_accuracy": 0.996504932641983,
+      "num_tokens": 93108549.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.9634061992168427,
+      "epoch": 6.93135935397039,
+      "grad_norm": 0.925463855266571,
+      "learning_rate": 2.3631659198643985e-05,
+      "loss": 0.011662732064723968,
+      "mean_token_accuracy": 0.9965654671192169,
+      "num_tokens": 93290512.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.9620663046836853,
+      "epoch": 6.944818304172275,
+      "grad_norm": 1.1143932342529297,
+      "learning_rate": 2.344282953371006e-05,
+      "loss": 0.01120457649230957,
+      "mean_token_accuracy": 0.9962508201599121,
+      "num_tokens": 93472398.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.9628985583782196,
+      "epoch": 6.958277254374159,
+      "grad_norm": 1.0271083116531372,
+      "learning_rate": 2.325452602840385e-05,
+      "loss": 0.012035199999809265,
+      "mean_token_accuracy": 0.996053409576416,
+      "num_tokens": 93654111.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.9658466935157776,
+      "epoch": 6.971736204576043,
+      "grad_norm": 1.0225892066955566,
+      "learning_rate": 2.306675241345797e-05,
+      "loss": 0.01229911893606186,
+      "mean_token_accuracy": 0.995949512720108,
+      "num_tokens": 93835692.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.9601108253002166,
+      "epoch": 6.985195154777927,
+      "grad_norm": 0.7876276969909668,
+      "learning_rate": 2.287951240910668e-05,
+      "loss": 0.011063569784164428,
+      "mean_token_accuracy": 0.9966089725494385,
+      "num_tokens": 94018119.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.9596541881561279,
+      "epoch": 6.998654104979812,
+      "grad_norm": 0.9803016185760498,
+      "learning_rate": 2.269280972501217e-05,
+      "loss": 0.011228029429912568,
+      "mean_token_accuracy": 0.9964218616485596,
+      "num_tokens": 94199483.0,
+      "step": 5200
+    },
+    {
+      "epoch": 7.0,
+      "eval_entropy": 0.9689761833020836,
+      "eval_loss": 0.1798103004693985,
+      "eval_mean_token_accuracy": 0.950934177371347,
+      "eval_num_tokens": 94217754.0,
+      "eval_runtime": 12.6557,
+      "eval_samples_per_second": 395.079,
+      "eval_steps_per_second": 12.405,
+      "step": 5201
+    },
+    {
+      "entropy": 0.9555480360984803,
+      "epoch": 7.012113055181696,
+      "grad_norm": 0.7279857397079468,
+      "learning_rate": 2.2506648060191042e-05,
+      "loss": 0.007322490215301514,
+      "mean_token_accuracy": 0.9978758811950683,
+      "num_tokens": 94380656.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.9532195508480072,
+      "epoch": 7.02557200538358,
+      "grad_norm": 0.6351640820503235,
+      "learning_rate": 2.2321031102941066e-05,
+      "loss": 0.007188012450933456,
+      "mean_token_accuracy": 0.9980705499649047,
+      "num_tokens": 94561769.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.9503021717071534,
+      "epoch": 7.039030955585464,
+      "grad_norm": 0.768357515335083,
+      "learning_rate": 2.2135962530768055e-05,
+      "loss": 0.007337777316570282,
+      "mean_token_accuracy": 0.9979041218757629,
+      "num_tokens": 94742683.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.9469992637634277,
+      "epoch": 7.052489905787349,
+      "grad_norm": 0.6551904678344727,
+      "learning_rate": 2.1951446010313053e-05,
+      "loss": 0.00702056959271431,
+      "mean_token_accuracy": 0.9980945587158203,
+      "num_tokens": 94924159.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.9499239981174469,
+      "epoch": 7.065948855989233,
+      "grad_norm": 0.8986485600471497,
+      "learning_rate": 2.1767485197279662e-05,
+      "loss": 0.007891762256622314,
+      "mean_token_accuracy": 0.9978051722049713,
+      "num_tokens": 95105209.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.9502587139606475,
+      "epoch": 7.079407806191117,
+      "grad_norm": 0.4773225486278534,
+      "learning_rate": 2.1584083736361617e-05,
+      "loss": 0.0066356688737869264,
+      "mean_token_accuracy": 0.9983843564987183,
+      "num_tokens": 95286168.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.9485512435436249,
+      "epoch": 7.0928667563930015,
+      "grad_norm": 0.7873836159706116,
+      "learning_rate": 2.1401245261170582e-05,
+      "loss": 0.007745465636253357,
+      "mean_token_accuracy": 0.9978477001190186,
+      "num_tokens": 95467244.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.9444929420948028,
+      "epoch": 7.106325706594886,
+      "grad_norm": 0.9345479607582092,
+      "learning_rate": 2.121897339416416e-05,
+      "loss": 0.006544182449579239,
+      "mean_token_accuracy": 0.9983295381069184,
+      "num_tokens": 95648662.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.9411603569984436,
+      "epoch": 7.11978465679677,
+      "grad_norm": 0.8139040470123291,
+      "learning_rate": 2.103727174657412e-05,
+      "loss": 0.00677436888217926,
+      "mean_token_accuracy": 0.9982263207435608,
+      "num_tokens": 95829602.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.9461234629154205,
+      "epoch": 7.133243606998654,
+      "grad_norm": 0.7345687747001648,
+      "learning_rate": 2.085614391833486e-05,
+      "loss": 0.006355223059654236,
+      "mean_token_accuracy": 0.9985966920852661,
+      "num_tokens": 96010655.0,
+      "step": 5300
+    },
+    {
+      "entropy": 0.9512575507164002,
+      "epoch": 7.1467025572005385,
+      "grad_norm": 0.799483060836792,
+      "learning_rate": 2.0675593498012052e-05,
+      "loss": 0.006509234011173248,
+      "mean_token_accuracy": 0.9983614802360534,
+      "num_tokens": 96190810.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.9446900069713593,
+      "epoch": 7.160161507402423,
+      "grad_norm": 0.7756890654563904,
+      "learning_rate": 2.049562406273159e-05,
+      "loss": 0.0069117918610572815,
+      "mean_token_accuracy": 0.9983683228492737,
+      "num_tokens": 96371920.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.9443341970443726,
+      "epoch": 7.173620457604307,
+      "grad_norm": 0.7561301589012146,
+      "learning_rate": 2.0316239178108688e-05,
+      "loss": 0.006773699820041656,
+      "mean_token_accuracy": 0.9981043338775635,
+      "num_tokens": 96553406.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.9383498072624207,
+      "epoch": 7.187079407806191,
+      "grad_norm": 0.7623066306114197,
+      "learning_rate": 2.013744239817725e-05,
+      "loss": 0.006338060647249222,
+      "mean_token_accuracy": 0.9983479022979737,
+      "num_tokens": 96735706.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.9444310605525971,
+      "epoch": 7.2005383580080755,
+      "grad_norm": 0.4759703576564789,
+      "learning_rate": 1.9959237265319452e-05,
+      "loss": 0.006821495294570923,
+      "mean_token_accuracy": 0.9982879221439361,
+      "num_tokens": 96916401.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.9462945520877838,
+      "epoch": 7.21399730820996,
+      "grad_norm": 0.5842452049255371,
+      "learning_rate": 1.9781627310195528e-05,
+      "loss": 0.0067993059754371645,
+      "mean_token_accuracy": 0.9981901407241821,
+      "num_tokens": 97097081.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.9475398421287536,
+      "epoch": 7.227456258411844,
+      "grad_norm": 0.8197925686836243,
+      "learning_rate": 1.960461605167389e-05,
+      "loss": 0.006510883569717407,
+      "mean_token_accuracy": 0.9984028697013855,
+      "num_tokens": 97277934.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.941127610206604,
+      "epoch": 7.240915208613728,
+      "grad_norm": 0.37160298228263855,
+      "learning_rate": 1.9428206996761357e-05,
+      "loss": 0.006270279735326767,
+      "mean_token_accuracy": 0.9982465505599976,
+      "num_tokens": 97459337.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.9421460628509521,
+      "epoch": 7.2543741588156125,
+      "grad_norm": 0.545837938785553,
+      "learning_rate": 1.9252403640533678e-05,
+      "loss": 0.006034504622220993,
+      "mean_token_accuracy": 0.9986165881156921,
+      "num_tokens": 97640387.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.9390841007232666,
+      "epoch": 7.267833109017497,
+      "grad_norm": 0.7546548247337341,
+      "learning_rate": 1.9077209466066315e-05,
+      "loss": 0.007177229970693588,
+      "mean_token_accuracy": 0.9978139519691467,
+      "num_tokens": 97821851.0,
+      "step": 5400
+    },
+    {
+      "entropy": 0.9392310976982117,
+      "epoch": 7.281292059219381,
+      "grad_norm": 0.8076840043067932,
+      "learning_rate": 1.8902627944365396e-05,
+      "loss": 0.006884847581386566,
+      "mean_token_accuracy": 0.9983299553394318,
+      "num_tokens": 98003599.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.9376516520977021,
+      "epoch": 7.294751009421265,
+      "grad_norm": 0.6652392148971558,
+      "learning_rate": 1.872866253429897e-05,
+      "loss": 0.005731315910816192,
+      "mean_token_accuracy": 0.9984779059886932,
+      "num_tokens": 98184942.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.9412936687469482,
+      "epoch": 7.30820995962315,
+      "grad_norm": 0.7955637574195862,
+      "learning_rate": 1.8555316682528485e-05,
+      "loss": 0.0065481230616569516,
+      "mean_token_accuracy": 0.9982032060623169,
+      "num_tokens": 98366577.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.9405987918376922,
+      "epoch": 7.321668909825034,
+      "grad_norm": 0.89079350233078,
+      "learning_rate": 1.8382593823440474e-05,
+      "loss": 0.006681758165359497,
+      "mean_token_accuracy": 0.9983046293258667,
+      "num_tokens": 98547681.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.944351601600647,
+      "epoch": 7.335127860026918,
+      "grad_norm": 1.1325428485870361,
+      "learning_rate": 1.821049737907854e-05,
+      "loss": 0.006995528936386108,
+      "mean_token_accuracy": 0.9982137680053711,
+      "num_tokens": 98728624.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.9397885262966156,
+      "epoch": 7.348586810228802,
+      "grad_norm": 0.7245733737945557,
+      "learning_rate": 1.803903075907554e-05,
+      "loss": 0.007640085369348526,
+      "mean_token_accuracy": 0.9979041159152985,
+      "num_tokens": 98910132.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.9460044980049134,
+      "epoch": 7.362045760430687,
+      "grad_norm": 0.9225216507911682,
+      "learning_rate": 1.7868197360586054e-05,
+      "loss": 0.006461286544799804,
+      "mean_token_accuracy": 0.9983667492866516,
+      "num_tokens": 99091179.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.9402543127536773,
+      "epoch": 7.375504710632571,
+      "grad_norm": 0.7389097213745117,
+      "learning_rate": 1.769800056821904e-05,
+      "loss": 0.0065854907035827635,
+      "mean_token_accuracy": 0.9979489326477051,
+      "num_tokens": 99273068.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.9478578507900238,
+      "epoch": 7.388963660834455,
+      "grad_norm": 1.0250836610794067,
+      "learning_rate": 1.7528443753970814e-05,
+      "loss": 0.006443935632705689,
+      "mean_token_accuracy": 0.998309975862503,
+      "num_tokens": 99453517.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.9427528262138367,
+      "epoch": 7.402422611036339,
+      "grad_norm": 0.7677002549171448,
+      "learning_rate": 1.7359530277158227e-05,
+      "loss": 0.006590449810028076,
+      "mean_token_accuracy": 0.9982711434364319,
+      "num_tokens": 99634833.0,
+      "step": 5500
+    },
+    {
+      "entropy": 0.9481732070446014,
+      "epoch": 7.415881561238224,
+      "grad_norm": 0.9008362889289856,
+      "learning_rate": 1.7191263484352106e-05,
+      "loss": 0.006611865013837814,
+      "mean_token_accuracy": 0.9980513393878937,
+      "num_tokens": 99815381.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.9450771987438202,
+      "epoch": 7.429340511440108,
+      "grad_norm": 0.6731723546981812,
+      "learning_rate": 1.702364670931097e-05,
+      "loss": 0.006989569962024688,
+      "mean_token_accuracy": 0.9981782495975494,
+      "num_tokens": 99996758.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.9488214910030365,
+      "epoch": 7.442799461641992,
+      "grad_norm": 0.543502151966095,
+      "learning_rate": 1.685668327291495e-05,
+      "loss": 0.0056597858667373655,
+      "mean_token_accuracy": 0.9984924256801605,
+      "num_tokens": 100177864.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.9521346092224121,
+      "epoch": 7.456258411843876,
+      "grad_norm": 0.6044306755065918,
+      "learning_rate": 1.6690376483100034e-05,
+      "loss": 0.006509961187839508,
+      "mean_token_accuracy": 0.9981269180774689,
+      "num_tokens": 100358496.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.9487424850463867,
+      "epoch": 7.469717362045761,
+      "grad_norm": 0.8417298197746277,
+      "learning_rate": 1.6524729634792475e-05,
+      "loss": 0.006116047501564026,
+      "mean_token_accuracy": 0.9982541143894196,
+      "num_tokens": 100539331.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.9476044595241546,
+      "epoch": 7.483176312247645,
+      "grad_norm": 0.8056595325469971,
+      "learning_rate": 1.6359746009843567e-05,
+      "loss": 0.0057601254433393475,
+      "mean_token_accuracy": 0.9985706150531769,
+      "num_tokens": 100720450.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.9456092417240143,
+      "epoch": 7.496635262449529,
+      "grad_norm": 0.6036805510520935,
+      "learning_rate": 1.6195428876964592e-05,
+      "loss": 0.005734497308731079,
+      "mean_token_accuracy": 0.9984824895858765,
+      "num_tokens": 100902228.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.946131706237793,
+      "epoch": 7.510094212651413,
+      "grad_norm": 0.7125651240348816,
+      "learning_rate": 1.6031781491662056e-05,
+      "loss": 0.005733318626880646,
+      "mean_token_accuracy": 0.9984204411506653,
+      "num_tokens": 101083443.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.9503833770751953,
+      "epoch": 7.523553162853298,
+      "grad_norm": 0.6275620460510254,
+      "learning_rate": 1.58688070961732e-05,
+      "loss": 0.006019812077283859,
+      "mean_token_accuracy": 0.9983126580715179,
+      "num_tokens": 101264443.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.947618979215622,
+      "epoch": 7.537012113055182,
+      "grad_norm": 0.7690677046775818,
+      "learning_rate": 1.5706508919401804e-05,
+      "loss": 0.005866598710417748,
+      "mean_token_accuracy": 0.9985946953296662,
+      "num_tokens": 101445485.0,
+      "step": 5600
+    },
+    {
+      "entropy": 0.9441321969032288,
+      "epoch": 7.550471063257066,
+      "grad_norm": 0.6154682636260986,
+      "learning_rate": 1.554489017685412e-05,
+      "loss": 0.00593491718173027,
+      "mean_token_accuracy": 0.9986464619636536,
+      "num_tokens": 101626547.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.9452862203121185,
+      "epoch": 7.56393001345895,
+      "grad_norm": 0.8111712336540222,
+      "learning_rate": 1.538395407057525e-05,
+      "loss": 0.006593097746372223,
+      "mean_token_accuracy": 0.998135757446289,
+      "num_tokens": 101807733.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.9418992578983307,
+      "epoch": 7.577388963660835,
+      "grad_norm": 0.7640061974525452,
+      "learning_rate": 1.522370378908567e-05,
+      "loss": 0.006091068312525749,
+      "mean_token_accuracy": 0.9984888017177582,
+      "num_tokens": 101989357.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.9476424336433411,
+      "epoch": 7.590847913862719,
+      "grad_norm": 0.5616353750228882,
+      "learning_rate": 1.5064142507318051e-05,
+      "loss": 0.0060265038162469866,
+      "mean_token_accuracy": 0.9982646107673645,
+      "num_tokens": 102170318.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.9492906272411347,
+      "epoch": 7.604306864064603,
+      "grad_norm": 0.6769309043884277,
+      "learning_rate": 1.4905273386554386e-05,
+      "loss": 0.006565354764461517,
+      "mean_token_accuracy": 0.9979823470115662,
+      "num_tokens": 102351383.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.9434434294700622,
+      "epoch": 7.6177658142664875,
+      "grad_norm": 0.8144317865371704,
+      "learning_rate": 1.4747099574363332e-05,
+      "loss": 0.006383077055215835,
+      "mean_token_accuracy": 0.9983625173568725,
+      "num_tokens": 102532483.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.9410477578639984,
+      "epoch": 7.631224764468372,
+      "grad_norm": 0.7184897065162659,
+      "learning_rate": 1.4589624204537866e-05,
+      "loss": 0.00683993250131607,
+      "mean_token_accuracy": 0.9981952488422394,
+      "num_tokens": 102714429.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.9437882840633393,
+      "epoch": 7.644683714670256,
+      "grad_norm": 0.7210987210273743,
+      "learning_rate": 1.4432850397033176e-05,
+      "loss": 0.006337168067693711,
+      "mean_token_accuracy": 0.9981409192085267,
+      "num_tokens": 102895693.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.9423653900623321,
+      "epoch": 7.65814266487214,
+      "grad_norm": 0.7256081700325012,
+      "learning_rate": 1.427678125790487e-05,
+      "loss": 0.006422295421361923,
+      "mean_token_accuracy": 0.9981527864933014,
+      "num_tokens": 103076994.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.9402276873588562,
+      "epoch": 7.6716016150740245,
+      "grad_norm": 0.9724451899528503,
+      "learning_rate": 1.4121419879247422e-05,
+      "loss": 0.006309093534946441,
+      "mean_token_accuracy": 0.9982479989528656,
+      "num_tokens": 103258251.0,
+      "step": 5700
+    },
+    {
+      "entropy": 0.9454701244831085,
+      "epoch": 7.685060565275909,
+      "grad_norm": 0.40232348442077637,
+      "learning_rate": 1.3966769339132929e-05,
+      "loss": 0.005939274653792381,
+      "mean_token_accuracy": 0.998301374912262,
+      "num_tokens": 103438976.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.9492968201637269,
+      "epoch": 7.698519515477793,
+      "grad_norm": 0.598109781742096,
+      "learning_rate": 1.3812832701550099e-05,
+      "loss": 0.00723060742020607,
+      "mean_token_accuracy": 0.9979724586009979,
+      "num_tokens": 103619841.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.9433623433113099,
+      "epoch": 7.711978465679677,
+      "grad_norm": 0.7593542337417603,
+      "learning_rate": 1.365961301634357e-05,
+      "loss": 0.0071952566504478455,
+      "mean_token_accuracy": 0.9981424927711486,
+      "num_tokens": 103801692.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.9490162670612335,
+      "epoch": 7.7254374158815615,
+      "grad_norm": 0.6364531517028809,
+      "learning_rate": 1.350711331915348e-05,
+      "loss": 0.006192450225353241,
+      "mean_token_accuracy": 0.9984559118747711,
+      "num_tokens": 103982367.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.942770904302597,
+      "epoch": 7.738896366083446,
+      "grad_norm": 0.9279600977897644,
+      "learning_rate": 1.3355336631355314e-05,
+      "loss": 0.005739518254995346,
+      "mean_token_accuracy": 0.9985584795475007,
+      "num_tokens": 104164149.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.9427549839019775,
+      "epoch": 7.75235531628533,
+      "grad_norm": 0.9408928155899048,
+      "learning_rate": 1.3204285960000051e-05,
+      "loss": 0.005749309062957763,
+      "mean_token_accuracy": 0.9985000133514405,
+      "num_tokens": 104345175.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.9404960751533509,
+      "epoch": 7.765814266487214,
+      "grad_norm": 0.7260274291038513,
+      "learning_rate": 1.30539642977546e-05,
+      "loss": 0.006429329514503479,
+      "mean_token_accuracy": 0.9984842240810394,
+      "num_tokens": 104526928.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.9448217689990998,
+      "epoch": 7.7792732166890985,
+      "grad_norm": 0.6738228797912598,
+      "learning_rate": 1.2904374622842486e-05,
+      "loss": 0.005977935343980789,
+      "mean_token_accuracy": 0.9985457420349121,
+      "num_tokens": 104708230.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.9451113224029541,
+      "epoch": 7.792732166890983,
+      "grad_norm": 0.9410247802734375,
+      "learning_rate": 1.2755519898984846e-05,
+      "loss": 0.005663443356752396,
+      "mean_token_accuracy": 0.9984429180622101,
+      "num_tokens": 104888647.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.9483637988567353,
+      "epoch": 7.806191117092867,
+      "grad_norm": 0.5315118432044983,
+      "learning_rate": 1.2607403075341728e-05,
+      "loss": 0.005260968953371048,
+      "mean_token_accuracy": 0.9988214015960694,
+      "num_tokens": 105068604.0,
+      "step": 5800
+    },
+    {
+      "entropy": 0.9461100101470947,
+      "epoch": 7.819650067294751,
+      "grad_norm": 0.5080281496047974,
+      "learning_rate": 1.2460027086453658e-05,
+      "loss": 0.005320623517036438,
+      "mean_token_accuracy": 0.9986594915390015,
+      "num_tokens": 105249023.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.9451904594898224,
+      "epoch": 7.833109017496636,
+      "grad_norm": 0.6318531632423401,
+      "learning_rate": 1.2313394852183473e-05,
+      "loss": 0.005546718463301658,
+      "mean_token_accuracy": 0.9985934972763062,
+      "num_tokens": 105429833.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.9423372805118561,
+      "epoch": 7.84656796769852,
+      "grad_norm": 0.6764518022537231,
+      "learning_rate": 1.2167509277658518e-05,
+      "loss": 0.005748329311609268,
+      "mean_token_accuracy": 0.9986869513988494,
+      "num_tokens": 105611004.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.9406632006168365,
+      "epoch": 7.860026917900404,
+      "grad_norm": 0.6841229796409607,
+      "learning_rate": 1.2022373253213049e-05,
+      "loss": 0.005423003435134887,
+      "mean_token_accuracy": 0.9984905958175659,
+      "num_tokens": 105791366.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.9383720457553864,
+      "epoch": 7.873485868102288,
+      "grad_norm": 0.7743740677833557,
+      "learning_rate": 1.1877989654330973e-05,
+      "loss": 0.005470183119177819,
+      "mean_token_accuracy": 0.9987134814262391,
+      "num_tokens": 105973009.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.9405181050300598,
+      "epoch": 7.886944818304173,
+      "grad_norm": 0.6537275314331055,
+      "learning_rate": 1.1734361341588912e-05,
+      "loss": 0.005447608605027199,
+      "mean_token_accuracy": 0.9983970046043396,
+      "num_tokens": 106153797.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.935320508480072,
+      "epoch": 7.900403768506057,
+      "grad_norm": 0.8727037310600281,
+      "learning_rate": 1.1591491160599488e-05,
+      "loss": 0.004841895401477813,
+      "mean_token_accuracy": 0.9990314602851867,
+      "num_tokens": 106335272.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.9389428675174714,
+      "epoch": 7.913862718707941,
+      "grad_norm": 0.8051491379737854,
+      "learning_rate": 1.1449381941954956e-05,
+      "loss": 0.005671492218971253,
+      "mean_token_accuracy": 0.9985178470611572,
+      "num_tokens": 106516472.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.9336957454681396,
+      "epoch": 7.927321668909825,
+      "grad_norm": 0.6917922496795654,
+      "learning_rate": 1.130803650117116e-05,
+      "loss": 0.005551187694072724,
+      "mean_token_accuracy": 0.9983591079711914,
+      "num_tokens": 106697785.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.9350760340690613,
+      "epoch": 7.94078061911171,
+      "grad_norm": 0.7642367482185364,
+      "learning_rate": 1.1167457638631656e-05,
+      "loss": 0.005567619949579239,
+      "mean_token_accuracy": 0.9985032379627228,
+      "num_tokens": 106879069.0,
+      "step": 5900
+    },
+    {
+      "entropy": 0.9362707912921906,
+      "epoch": 7.954239569313594,
+      "grad_norm": 0.6716526746749878,
+      "learning_rate": 1.1027648139532355e-05,
+      "loss": 0.005269525200128555,
+      "mean_token_accuracy": 0.9987157702445983,
+      "num_tokens": 107060235.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.9299384176731109,
+      "epoch": 7.967698519515478,
+      "grad_norm": 0.6839614510536194,
+      "learning_rate": 1.0888610773826253e-05,
+      "loss": 0.00708087682723999,
+      "mean_token_accuracy": 0.9982430160045623,
+      "num_tokens": 107242113.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.9352255225181579,
+      "epoch": 7.981157469717362,
+      "grad_norm": 1.0992971658706665,
+      "learning_rate": 1.0750348296168583e-05,
+      "loss": 0.005663050711154938,
+      "mean_token_accuracy": 0.9984843015670777,
+      "num_tokens": 107423370.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.9345376133918762,
+      "epoch": 7.994616419919247,
+      "grad_norm": 0.771766722202301,
+      "learning_rate": 1.0612863445862236e-05,
+      "loss": 0.006203063577413559,
+      "mean_token_accuracy": 0.9982724368572236,
+      "num_tokens": 107604796.0,
+      "step": 5940
+    },
+    {
+      "epoch": 8.0,
+      "eval_entropy": 0.9459010544855884,
+      "eval_loss": 0.18965470790863037,
+      "eval_mean_token_accuracy": 0.9512642967473169,
+      "eval_num_tokens": 107677508.0,
+      "eval_runtime": 12.6472,
+      "eval_samples_per_second": 395.344,
+      "eval_steps_per_second": 12.414,
+      "step": 5944
+    },
+    {
+      "entropy": 0.9325907230377197,
+      "epoch": 8.00807537012113,
+      "grad_norm": 0.40456992387771606,
+      "learning_rate": 1.0476158946803482e-05,
+      "loss": 0.0041702061891555784,
+      "mean_token_accuracy": 0.9988861680030823,
+      "num_tokens": 107786362.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.9357569575309753,
+      "epoch": 8.021534320323015,
+      "grad_norm": 0.5090038180351257,
+      "learning_rate": 1.0340237507428009e-05,
+      "loss": 0.0035715073347091673,
+      "mean_token_accuracy": 0.9994814991950989,
+      "num_tokens": 107967199.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.9353849172592164,
+      "epoch": 8.034993270524899,
+      "grad_norm": 0.47324568033218384,
+      "learning_rate": 1.020510182065727e-05,
+      "loss": 0.0035488776862621307,
+      "mean_token_accuracy": 0.9993434071540832,
+      "num_tokens": 108147567.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.9339651763439178,
+      "epoch": 8.048452220726784,
+      "grad_norm": 0.4190789461135864,
+      "learning_rate": 1.0070754563845114e-05,
+      "loss": 0.003430064395070076,
+      "mean_token_accuracy": 0.9994158089160919,
+      "num_tokens": 108328451.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.9365652620792388,
+      "epoch": 8.061911170928667,
+      "grad_norm": 0.37395232915878296,
+      "learning_rate": 9.937198398724751e-06,
+      "loss": 0.0032767456024885176,
+      "mean_token_accuracy": 0.9995010256767273,
+      "num_tokens": 108509228.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.9290562212467194,
+      "epoch": 8.075370121130552,
+      "grad_norm": 0.2843457758426666,
+      "learning_rate": 9.804435971356025e-06,
+      "loss": 0.0030744092538952826,
+      "mean_token_accuracy": 0.9995416045188904,
+      "num_tokens": 108691291.0,
+      "step": 6000
+    },
+    {
+      "entropy": 0.9333357572555542,
+      "epoch": 8.088829071332436,
+      "grad_norm": 0.36153435707092285,
+      "learning_rate": 9.672469912072956e-06,
+      "loss": 0.0031955156475305556,
+      "mean_token_accuracy": 0.9994354546070099,
+      "num_tokens": 108872687.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.9334155559539795,
+      "epoch": 8.10228802153432,
+      "grad_norm": 0.35333436727523804,
+      "learning_rate": 9.541302835431681e-06,
+      "loss": 0.0032100573182106016,
+      "mean_token_accuracy": 0.9995277643203735,
+      "num_tokens": 109053638.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.9349715650081635,
+      "epoch": 8.115746971736204,
+      "grad_norm": 0.493518590927124,
+      "learning_rate": 9.410937340158599e-06,
+      "loss": 0.003197438269853592,
+      "mean_token_accuracy": 0.9993844270706177,
+      "num_tokens": 109234361.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.9317407608032227,
+      "epoch": 8.12920592193809,
+      "grad_norm": 0.32104817032814026,
+      "learning_rate": 9.28137600909892e-06,
+      "loss": 0.003348514437675476,
+      "mean_token_accuracy": 0.9993906795978547,
+      "num_tokens": 109415438.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.9339701652526855,
+      "epoch": 8.142664872139973,
+      "grad_norm": 0.5996560454368591,
+      "learning_rate": 9.15262140916548e-06,
+      "loss": 0.0037448287010192873,
+      "mean_token_accuracy": 0.9991378724575043,
+      "num_tokens": 109596756.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.9327323913574219,
+      "epoch": 8.156123822341858,
+      "grad_norm": 0.7599992752075195,
+      "learning_rate": 9.024676091287881e-06,
+      "loss": 0.003381679952144623,
+      "mean_token_accuracy": 0.999460369348526,
+      "num_tokens": 109778064.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.931190824508667,
+      "epoch": 8.169582772543741,
+      "grad_norm": 0.4155709445476532,
+      "learning_rate": 8.897542590361963e-06,
+      "loss": 0.003494596481323242,
+      "mean_token_accuracy": 0.9995332777500152,
+      "num_tokens": 109959186.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.9299486935138702,
+      "epoch": 8.183041722745626,
+      "grad_norm": 0.45130202174186707,
+      "learning_rate": 8.77122342519956e-06,
+      "loss": 0.003702623024582863,
+      "mean_token_accuracy": 0.9991639256477356,
+      "num_tokens": 110140682.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.9285082340240478,
+      "epoch": 8.19650067294751,
+      "grad_norm": 0.5240904688835144,
+      "learning_rate": 8.64572109847862e-06,
+      "loss": 0.0037362042814493178,
+      "mean_token_accuracy": 0.9991588532924652,
+      "num_tokens": 110322435.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.9306709587574005,
+      "epoch": 8.209959623149395,
+      "grad_norm": 0.4319130480289459,
+      "learning_rate": 8.521038096693635e-06,
+      "loss": 0.0034312527626752854,
+      "mean_token_accuracy": 0.9994248628616333,
+      "num_tokens": 110503926.0,
+      "step": 6100
+    },
+    {
+      "entropy": 0.9292944014072418,
+      "epoch": 8.223418573351278,
+      "grad_norm": 0.4186641275882721,
+      "learning_rate": 8.397176890106317e-06,
+      "loss": 0.003543431684374809,
+      "mean_token_accuracy": 0.9993721604347229,
+      "num_tokens": 110685574.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.9340905666351318,
+      "epoch": 8.236877523553163,
+      "grad_norm": 0.44297507405281067,
+      "learning_rate": 8.274139932696717e-06,
+      "loss": 0.00324522964656353,
+      "mean_token_accuracy": 0.9993358612060547,
+      "num_tokens": 110866354.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.9321817755699158,
+      "epoch": 8.250336473755047,
+      "grad_norm": 0.6701546907424927,
+      "learning_rate": 8.151929662114566e-06,
+      "loss": 0.0037943169474601744,
+      "mean_token_accuracy": 0.9993444323539734,
+      "num_tokens": 111047803.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.9300240755081177,
+      "epoch": 8.263795423956932,
+      "grad_norm": 0.3400649428367615,
+      "learning_rate": 8.030548499631014e-06,
+      "loss": 0.0031486738473176954,
+      "mean_token_accuracy": 0.9994214653968811,
+      "num_tokens": 111229226.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.9348362684249878,
+      "epoch": 8.277254374158815,
+      "grad_norm": 0.3338402509689331,
+      "learning_rate": 7.909998850090628e-06,
+      "loss": 0.003268548846244812,
+      "mean_token_accuracy": 0.9993820369243622,
+      "num_tokens": 111410053.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.9298858225345612,
+      "epoch": 8.2907133243607,
+      "grad_norm": 0.3765404224395752,
+      "learning_rate": 7.790283101863771e-06,
+      "loss": 0.0032513745129108427,
+      "mean_token_accuracy": 0.999443256855011,
+      "num_tokens": 111591318.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.9351325869560242,
+      "epoch": 8.304172274562584,
+      "grad_norm": 0.45491793751716614,
+      "learning_rate": 7.671403626799262e-06,
+      "loss": 0.0032870754599571227,
+      "mean_token_accuracy": 0.9993524074554443,
+      "num_tokens": 111772032.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.9296027898788453,
+      "epoch": 8.317631224764469,
+      "grad_norm": 0.40146195888519287,
+      "learning_rate": 7.553362780177398e-06,
+      "loss": 0.003901958093047142,
+      "mean_token_accuracy": 0.9991503238677979,
+      "num_tokens": 111953168.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.9276497423648834,
+      "epoch": 8.331090174966352,
+      "grad_norm": 0.3894268274307251,
+      "learning_rate": 7.43616290066329e-06,
+      "loss": 0.0032478343695402146,
+      "mean_token_accuracy": 0.9992106795310974,
+      "num_tokens": 112135017.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.9245680391788482,
+      "epoch": 8.344549125168237,
+      "grad_norm": 0.5110934376716614,
+      "learning_rate": 7.3198063102605185e-06,
+      "loss": 0.00324617363512516,
+      "mean_token_accuracy": 0.9995225965976715,
+      "num_tokens": 112317445.0,
+      "step": 6200
+    },
+    {
+      "entropy": 0.9306623697280884,
+      "epoch": 8.35800807537012,
+      "grad_norm": 0.39472904801368713,
+      "learning_rate": 7.204295314265136e-06,
+      "loss": 0.0035538677126169204,
+      "mean_token_accuracy": 0.9991783916950225,
+      "num_tokens": 112498753.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.9369937419891358,
+      "epoch": 8.371467025572006,
+      "grad_norm": 0.6480345129966736,
+      "learning_rate": 7.089632201219997e-06,
+      "loss": 0.0035104259848594667,
+      "mean_token_accuracy": 0.9993222296237946,
+      "num_tokens": 112678638.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.9279407441616059,
+      "epoch": 8.38492597577389,
+      "grad_norm": 0.37613216042518616,
+      "learning_rate": 6.975819242869414e-06,
+      "loss": 0.003352391347289085,
+      "mean_token_accuracy": 0.9994183361530304,
+      "num_tokens": 112860513.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.9334605932235718,
+      "epoch": 8.398384925975774,
+      "grad_norm": 0.5418120622634888,
+      "learning_rate": 6.862858694114138e-06,
+      "loss": 0.003238685056567192,
+      "mean_token_accuracy": 0.9992875456809998,
+      "num_tokens": 113041014.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.9314471304416656,
+      "epoch": 8.411843876177658,
+      "grad_norm": 0.4664161205291748,
+      "learning_rate": 6.750752792966702e-06,
+      "loss": 0.003342866525053978,
+      "mean_token_accuracy": 0.9993779182434082,
+      "num_tokens": 113221507.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.9293332099914551,
+      "epoch": 8.425302826379543,
+      "grad_norm": 0.5400107502937317,
+      "learning_rate": 6.639503760507065e-06,
+      "loss": 0.003294713795185089,
+      "mean_token_accuracy": 0.99941286444664,
+      "num_tokens": 113402195.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.9276987969875335,
+      "epoch": 8.438761776581426,
+      "grad_norm": 0.4761334955692291,
+      "learning_rate": 6.529113800838632e-06,
+      "loss": 0.0031884826719760897,
+      "mean_token_accuracy": 0.9994406461715698,
+      "num_tokens": 113583396.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.9320788323879242,
+      "epoch": 8.452220726783311,
+      "grad_norm": 0.7529016733169556,
+      "learning_rate": 6.41958510104454e-06,
+      "loss": 0.0033160604536533354,
+      "mean_token_accuracy": 0.999411803483963,
+      "num_tokens": 113763905.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.930284959077835,
+      "epoch": 8.465679676985195,
+      "grad_norm": 0.3949340879917145,
+      "learning_rate": 6.31091983114438e-06,
+      "loss": 0.003930982202291489,
+      "mean_token_accuracy": 0.999339497089386,
+      "num_tokens": 113944757.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.92959144115448,
+      "epoch": 8.47913862718708,
+      "grad_norm": 0.5477228164672852,
+      "learning_rate": 6.203120144051161e-06,
+      "loss": 0.0033377643674612047,
+      "mean_token_accuracy": 0.9992568790912628,
+      "num_tokens": 114126403.0,
+      "step": 6300
+    },
+    {
+      "entropy": 0.931204092502594,
+      "epoch": 8.492597577388963,
+      "grad_norm": 0.5682660937309265,
+      "learning_rate": 6.0961881755286775e-06,
+      "loss": 0.0033181462436914443,
+      "mean_token_accuracy": 0.9993676483631134,
+      "num_tokens": 114307860.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.9299099445343018,
+      "epoch": 8.506056527590848,
+      "grad_norm": 0.48706290125846863,
+      "learning_rate": 5.990126044149203e-06,
+      "loss": 0.0034614600241184234,
+      "mean_token_accuracy": 0.9993451237678528,
+      "num_tokens": 114489346.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.9328338205814362,
+      "epoch": 8.519515477792732,
+      "grad_norm": 0.502772331237793,
+      "learning_rate": 5.884935851251483e-06,
+      "loss": 0.0033218566328287124,
+      "mean_token_accuracy": 0.9993848502635956,
+      "num_tokens": 114670149.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.9331273257732391,
+      "epoch": 8.532974427994617,
+      "grad_norm": 0.3274834156036377,
+      "learning_rate": 5.780619680899146e-06,
+      "loss": 0.0031801596283912658,
+      "mean_token_accuracy": 0.9995054185390473,
+      "num_tokens": 114851185.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.9312391161918641,
+      "epoch": 8.5464333781965,
+      "grad_norm": 0.3442527651786804,
+      "learning_rate": 5.677179599839366e-06,
+      "loss": 0.0031024085357785223,
+      "mean_token_accuracy": 0.999411153793335,
+      "num_tokens": 115032612.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.9326258182525635,
+      "epoch": 8.559892328398385,
+      "grad_norm": 0.510369598865509,
+      "learning_rate": 5.574617657461967e-06,
+      "loss": 0.0038056544959545135,
+      "mean_token_accuracy": 0.9992052555084229,
+      "num_tokens": 115214161.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.9323939800262451,
+      "epoch": 8.573351278600269,
+      "grad_norm": 0.4948248267173767,
+      "learning_rate": 5.472935885758767e-06,
+      "loss": 0.0032266426831483843,
+      "mean_token_accuracy": 0.9994330883026123,
+      "num_tokens": 115395414.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.9314477443695068,
+      "epoch": 8.586810228802154,
+      "grad_norm": 0.5037770867347717,
+      "learning_rate": 5.372136299283365e-06,
+      "loss": 0.0034978736191987993,
+      "mean_token_accuracy": 0.9991765201091767,
+      "num_tokens": 115576607.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.9309112429618835,
+      "epoch": 8.600269179004037,
+      "grad_norm": 0.3841687738895416,
+      "learning_rate": 5.272220895111196e-06,
+      "loss": 0.0030230386182665825,
+      "mean_token_accuracy": 0.9993924856185913,
+      "num_tokens": 115757822.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.9289616286754608,
+      "epoch": 8.613728129205922,
+      "grad_norm": 0.5192815661430359,
+      "learning_rate": 5.173191652799986e-06,
+      "loss": 0.003417670726776123,
+      "mean_token_accuracy": 0.9993391752243042,
+      "num_tokens": 115939024.0,
+      "step": 6400
+    },
+    {
+      "entropy": 0.9319016218185425,
+      "epoch": 8.627187079407806,
+      "grad_norm": 0.4927634596824646,
+      "learning_rate": 5.075050534350512e-06,
+      "loss": 0.003317685052752495,
+      "mean_token_accuracy": 0.9993408203125,
+      "num_tokens": 116120013.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.932750004529953,
+      "epoch": 8.640646029609691,
+      "grad_norm": 0.5494413375854492,
+      "learning_rate": 4.977799484167767e-06,
+      "loss": 0.003227224573493004,
+      "mean_token_accuracy": 0.9993854761123657,
+      "num_tokens": 116300658.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.9317887723445892,
+      "epoch": 8.654104979811574,
+      "grad_norm": 0.6813295483589172,
+      "learning_rate": 4.8814404290223655e-06,
+      "loss": 0.002960843779146671,
+      "mean_token_accuracy": 0.9993847787380219,
+      "num_tokens": 116481203.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.9327813625335694,
+      "epoch": 8.66756393001346,
+      "grad_norm": 0.3762431740760803,
+      "learning_rate": 4.7859752780124555e-06,
+      "loss": 0.0030213048681616782,
+      "mean_token_accuracy": 0.9996239364147186,
+      "num_tokens": 116662086.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.9316872417926788,
+      "epoch": 8.681022880215343,
+      "grad_norm": 0.5861575603485107,
+      "learning_rate": 4.691405922525838e-06,
+      "loss": 0.003052959218621254,
+      "mean_token_accuracy": 0.9994400024414063,
+      "num_tokens": 116843008.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.9310117483139038,
+      "epoch": 8.694481830417228,
+      "grad_norm": 0.5606390833854675,
+      "learning_rate": 4.597734236202511e-06,
+      "loss": 0.0031916122883558274,
+      "mean_token_accuracy": 0.9994355738162994,
+      "num_tokens": 117024241.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.9291791677474975,
+      "epoch": 8.707940780619111,
+      "grad_norm": 0.5419787764549255,
+      "learning_rate": 4.504962074897556e-06,
+      "loss": 0.00293008740991354,
+      "mean_token_accuracy": 0.9994637906551361,
+      "num_tokens": 117205874.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.929755812883377,
+      "epoch": 8.721399730820997,
+      "grad_norm": 0.5368452668190002,
+      "learning_rate": 4.413091276644349e-06,
+      "loss": 0.0030313016846776008,
+      "mean_token_accuracy": 0.9993925631046295,
+      "num_tokens": 117387081.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.9327210128307343,
+      "epoch": 8.73485868102288,
+      "grad_norm": 0.4368033707141876,
+      "learning_rate": 4.322123661618166e-06,
+      "loss": 0.003084339387714863,
+      "mean_token_accuracy": 0.9995987117290497,
+      "num_tokens": 117567583.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.9297920942306519,
+      "epoch": 8.748317631224765,
+      "grad_norm": 0.5393041372299194,
+      "learning_rate": 4.23206103210011e-06,
+      "loss": 0.0031465519219636916,
+      "mean_token_accuracy": 0.9994846880435944,
+      "num_tokens": 117749088.0,
+      "step": 6500
+    },
+    {
+      "entropy": 0.9288382709026337,
+      "epoch": 8.761776581426648,
+      "grad_norm": 0.3374616205692291,
+      "learning_rate": 4.142905172441408e-06,
+      "loss": 0.0027178578078746797,
+      "mean_token_accuracy": 0.9996258854866028,
+      "num_tokens": 117930285.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.9260033547878266,
+      "epoch": 8.775235531628534,
+      "grad_norm": 0.20111319422721863,
+      "learning_rate": 4.054657849028043e-06,
+      "loss": 0.003253967314958572,
+      "mean_token_accuracy": 0.9993975520133972,
+      "num_tokens": 118112036.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.9301682531833648,
+      "epoch": 8.788694481830417,
+      "grad_norm": 0.4301469326019287,
+      "learning_rate": 3.9673208102457895e-06,
+      "loss": 0.0030012888833880424,
+      "mean_token_accuracy": 0.9995055079460144,
+      "num_tokens": 118293038.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.9303179025650025,
+      "epoch": 8.802153432032302,
+      "grad_norm": 0.3773118853569031,
+      "learning_rate": 3.880895786445543e-06,
+      "loss": 0.003305017575621605,
+      "mean_token_accuracy": 0.9994101107120514,
+      "num_tokens": 118474004.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.9334917306900025,
+      "epoch": 8.815612382234185,
+      "grad_norm": 0.39689475297927856,
+      "learning_rate": 3.795384489909054e-06,
+      "loss": 0.0031069353222846984,
+      "mean_token_accuracy": 0.9994810342788696,
+      "num_tokens": 118654235.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.9243707060813904,
+      "epoch": 8.82907133243607,
+      "grad_norm": 0.3700069785118103,
+      "learning_rate": 3.7107886148150084e-06,
+      "loss": 0.003253418952226639,
+      "mean_token_accuracy": 0.9993774831295014,
+      "num_tokens": 118835998.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.9256916821002961,
+      "epoch": 8.842530282637954,
+      "grad_norm": 0.5005241632461548,
+      "learning_rate": 3.6271098372054334e-06,
+      "loss": 0.003014829009771347,
+      "mean_token_accuracy": 0.9993231058120727,
+      "num_tokens": 119017476.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.929146409034729,
+      "epoch": 8.855989232839839,
+      "grad_norm": 0.4028603434562683,
+      "learning_rate": 3.5443498149525357e-06,
+      "loss": 0.0029433896765112878,
+      "mean_token_accuracy": 0.9995526134967804,
+      "num_tokens": 119198443.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.9259550392627716,
+      "epoch": 8.869448183041722,
+      "grad_norm": 0.46131935715675354,
+      "learning_rate": 3.462510187725809e-06,
+      "loss": 0.0030248038470745086,
+      "mean_token_accuracy": 0.999508935213089,
+      "num_tokens": 119380252.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.9258705079555511,
+      "epoch": 8.882907133243608,
+      "grad_norm": 0.48013147711753845,
+      "learning_rate": 3.3815925769595745e-06,
+      "loss": 0.0031910501420497894,
+      "mean_token_accuracy": 0.9992617785930633,
+      "num_tokens": 119561793.0,
+      "step": 6600
+    },
+    {
+      "entropy": 0.9294628858566284,
+      "epoch": 8.896366083445491,
+      "grad_norm": 0.5725185871124268,
+      "learning_rate": 3.30159858582087e-06,
+      "loss": 0.0035464100539684294,
+      "mean_token_accuracy": 0.999199116230011,
+      "num_tokens": 119742430.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.9263964056968689,
+      "epoch": 8.909825033647376,
+      "grad_norm": 0.4084455966949463,
+      "learning_rate": 3.2225297991776536e-06,
+      "loss": 0.002907513827085495,
+      "mean_token_accuracy": 0.9995316982269287,
+      "num_tokens": 119923690.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.9269788503646851,
+      "epoch": 8.92328398384926,
+      "grad_norm": 0.3295418322086334,
+      "learning_rate": 3.1443877835674208e-06,
+      "loss": 0.002725762873888016,
+      "mean_token_accuracy": 0.9996701359748841,
+      "num_tokens": 120104753.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.929550188779831,
+      "epoch": 8.936742934051145,
+      "grad_norm": 0.5200356841087341,
+      "learning_rate": 3.0671740871661635e-06,
+      "loss": 0.0031205663457512856,
+      "mean_token_accuracy": 0.9993889212608338,
+      "num_tokens": 120285587.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.9283001899719239,
+      "epoch": 8.950201884253028,
+      "grad_norm": 0.4056023061275482,
+      "learning_rate": 2.9908902397577043e-06,
+      "loss": 0.00296444371342659,
+      "mean_token_accuracy": 0.9994133710861206,
+      "num_tokens": 120467044.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.9271095216274261,
+      "epoch": 8.963660834454913,
+      "grad_norm": 0.3392156660556793,
+      "learning_rate": 2.91553775270339e-06,
+      "loss": 0.003089761920273304,
+      "mean_token_accuracy": 0.9994589149951935,
+      "num_tokens": 120648306.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.9254171431064606,
+      "epoch": 8.977119784656796,
+      "grad_norm": 0.4930778741836548,
+      "learning_rate": 2.8411181189121284e-06,
+      "loss": 0.0031807802617549895,
+      "mean_token_accuracy": 0.9994666993618011,
+      "num_tokens": 120829851.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.9281193077564239,
+      "epoch": 8.990578734858682,
+      "grad_norm": 0.476330041885376,
+      "learning_rate": 2.767632812810839e-06,
+      "loss": 0.002921390533447266,
+      "mean_token_accuracy": 0.9995766639709472,
+      "num_tokens": 121011008.0,
+      "step": 6680
+    },
+    {
+      "epoch": 9.0,
+      "eval_entropy": 0.9369368272222531,
+      "eval_loss": 0.20012438297271729,
+      "eval_mean_token_accuracy": 0.9520987811361908,
+      "eval_num_tokens": 121137269.0,
+      "eval_runtime": 12.7296,
+      "eval_samples_per_second": 392.786,
+      "eval_steps_per_second": 12.333,
+      "step": 6687
+    },
+    {
+      "entropy": 0.9325936436653137,
+      "epoch": 9.004037685060565,
+      "grad_norm": 0.4694288671016693,
+      "learning_rate": 2.6950832903152034e-06,
+      "loss": 0.0030035220086574556,
+      "mean_token_accuracy": 0.9994764149188995,
+      "num_tokens": 121191510.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.9304535567760468,
+      "epoch": 9.01749663526245,
+      "grad_norm": 0.3735108971595764,
+      "learning_rate": 2.623470988800869e-06,
+      "loss": 0.002471015229821205,
+      "mean_token_accuracy": 0.9997635781764984,
+      "num_tokens": 121372588.0,
+      "step": 6700
+    },
+    {
+      "entropy": 0.926229190826416,
+      "epoch": 9.030955585464334,
+      "grad_norm": 0.24182267487049103,
+      "learning_rate": 2.552797327074924e-06,
+      "loss": 0.0024778425693511965,
+      "mean_token_accuracy": 0.9996277928352356,
+      "num_tokens": 121553866.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.9282829165458679,
+      "epoch": 9.044414535666219,
+      "grad_norm": 0.4357544481754303,
+      "learning_rate": 2.4830637053478213e-06,
+      "loss": 0.0023618059232831,
+      "mean_token_accuracy": 0.9998379051685333,
+      "num_tokens": 121734885.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.9233563244342804,
+      "epoch": 9.057873485868102,
+      "grad_norm": 0.18988333642482758,
+      "learning_rate": 2.414271505205612e-06,
+      "loss": 0.002139232307672501,
+      "mean_token_accuracy": 0.9998383700847626,
+      "num_tokens": 121916630.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.9254979133605957,
+      "epoch": 9.071332436069987,
+      "grad_norm": 0.289910227060318,
+      "learning_rate": 2.346422089582595e-06,
+      "loss": 0.0024019576609134673,
+      "mean_token_accuracy": 0.9997670352458954,
+      "num_tokens": 122097691.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.925941115617752,
+      "epoch": 9.08479138627187,
+      "grad_norm": 0.4846780598163605,
+      "learning_rate": 2.279516802734305e-06,
+      "loss": 0.0023199569433927534,
+      "mean_token_accuracy": 0.999787551164627,
+      "num_tokens": 122278648.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.9287903070449829,
+      "epoch": 9.098250336473756,
+      "grad_norm": 0.3375636339187622,
+      "learning_rate": 2.213556970210867e-06,
+      "loss": 0.0025358393788337706,
+      "mean_token_accuracy": 0.9995267689228058,
+      "num_tokens": 122458843.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.9231357336044311,
+      "epoch": 9.111709286675639,
+      "grad_norm": 0.3335321843624115,
+      "learning_rate": 2.1485438988307518e-06,
+      "loss": 0.0022585811093449593,
+      "mean_token_accuracy": 0.9997917413711548,
+      "num_tokens": 122640303.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.9256380558013916,
+      "epoch": 9.125168236877524,
+      "grad_norm": 0.2286524474620819,
+      "learning_rate": 2.0844788766548883e-06,
+      "loss": 0.0021977417171001436,
+      "mean_token_accuracy": 0.9997158706188202,
+      "num_tokens": 122821433.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.9239262282848358,
+      "epoch": 9.138627187079408,
+      "grad_norm": 0.33134034276008606,
+      "learning_rate": 2.021363172961116e-06,
+      "loss": 0.0023238688707351686,
+      "mean_token_accuracy": 0.9996749520301819,
+      "num_tokens": 123002491.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.9229062855243683,
+      "epoch": 9.152086137281293,
+      "grad_norm": 0.4102766811847687,
+      "learning_rate": 1.959198038219079e-06,
+      "loss": 0.002699672244489193,
+      "mean_token_accuracy": 0.9995124757289886,
+      "num_tokens": 123184151.0,
+      "step": 6800
+    },
+    {
+      "entropy": 0.9229757428169251,
+      "epoch": 9.165545087483176,
+      "grad_norm": 0.3331562578678131,
+      "learning_rate": 1.897984704065403e-06,
+      "loss": 0.0022561397403478623,
+      "mean_token_accuracy": 0.9996741354465485,
+      "num_tokens": 123365473.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.9211558997631073,
+      "epoch": 9.179004037685061,
+      "grad_norm": 0.2975319027900696,
+      "learning_rate": 1.8377243832793322e-06,
+      "loss": 0.002418186515569687,
+      "mean_token_accuracy": 0.9996279060840607,
+      "num_tokens": 123547013.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.9220295488834381,
+      "epoch": 9.192462987886945,
+      "grad_norm": 0.48501911759376526,
+      "learning_rate": 1.778418269758697e-06,
+      "loss": 0.002369001880288124,
+      "mean_token_accuracy": 0.9995379328727723,
+      "num_tokens": 123728505.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.9241848886013031,
+      "epoch": 9.20592193808883,
+      "grad_norm": 0.4089840352535248,
+      "learning_rate": 1.7200675384962339e-06,
+      "loss": 0.0023533258587121963,
+      "mean_token_accuracy": 0.9997864544391633,
+      "num_tokens": 123909321.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.9250785171985626,
+      "epoch": 9.219380888290713,
+      "grad_norm": 0.3853326737880707,
+      "learning_rate": 1.6626733455563371e-06,
+      "loss": 0.002148270048201084,
+      "mean_token_accuracy": 0.9997900068759918,
+      "num_tokens": 124090280.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.9248496949672699,
+      "epoch": 9.232839838492598,
+      "grad_norm": 0.34975531697273254,
+      "learning_rate": 1.6062368280521323e-06,
+      "loss": 0.002289744094014168,
+      "mean_token_accuracy": 0.9997417986392975,
+      "num_tokens": 124271306.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.9206982016563415,
+      "epoch": 9.246298788694482,
+      "grad_norm": 0.3254064619541168,
+      "learning_rate": 1.5507591041229508e-06,
+      "loss": 0.002178264781832695,
+      "mean_token_accuracy": 0.999741929769516,
+      "num_tokens": 124452889.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.9210093080997467,
+      "epoch": 9.259757738896367,
+      "grad_norm": 0.4685194194316864,
+      "learning_rate": 1.4962412729122077e-06,
+      "loss": 0.0024245936423540116,
+      "mean_token_accuracy": 0.9997441649436951,
+      "num_tokens": 124634594.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.9233365416526794,
+      "epoch": 9.27321668909825,
+      "grad_norm": 0.35174524784088135,
+      "learning_rate": 1.4426844145455642e-06,
+      "loss": 0.0023859523236751555,
+      "mean_token_accuracy": 0.9995763838291168,
+      "num_tokens": 124816064.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.925233656167984,
+      "epoch": 9.286675639300135,
+      "grad_norm": 0.4699457883834839,
+      "learning_rate": 1.3900895901095889e-06,
+      "loss": 0.0023812506347894668,
+      "mean_token_accuracy": 0.9997189700603485,
+      "num_tokens": 124997191.0,
+      "step": 6900
+    },
+    {
+      "entropy": 0.9231235146522522,
+      "epoch": 9.300134589502019,
+      "grad_norm": 0.41605737805366516,
+      "learning_rate": 1.3384578416307025e-06,
+      "loss": 0.0025712810456752777,
+      "mean_token_accuracy": 0.9995984554290771,
+      "num_tokens": 125178499.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.927912962436676,
+      "epoch": 9.313593539703904,
+      "grad_norm": 0.19599905610084534,
+      "learning_rate": 1.2877901920545332e-06,
+      "loss": 0.002140621840953827,
+      "mean_token_accuracy": 0.999715405702591,
+      "num_tokens": 125358764.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.9229813098907471,
+      "epoch": 9.327052489905787,
+      "grad_norm": 0.35456934571266174,
+      "learning_rate": 1.2380876452256663e-06,
+      "loss": 0.002259720675647259,
+      "mean_token_accuracy": 0.9996977090835572,
+      "num_tokens": 125540179.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.9238385081291198,
+      "epoch": 9.340511440107672,
+      "grad_norm": 0.37090715765953064,
+      "learning_rate": 1.189351185867732e-06,
+      "loss": 0.0020590536296367645,
+      "mean_token_accuracy": 0.9996942579746246,
+      "num_tokens": 125721769.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.9252527773380279,
+      "epoch": 9.353970390309556,
+      "grad_norm": 0.27994754910469055,
+      "learning_rate": 1.1415817795639328e-06,
+      "loss": 0.0019802145659923555,
+      "mean_token_accuracy": 0.9998105227947235,
+      "num_tokens": 125902628.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.924957662820816,
+      "epoch": 9.36742934051144,
+      "grad_norm": 0.3070662021636963,
+      "learning_rate": 1.094780372737858e-06,
+      "loss": 0.0022987723350524902,
+      "mean_token_accuracy": 0.999741005897522,
+      "num_tokens": 126083659.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.925666356086731,
+      "epoch": 9.380888290713324,
+      "grad_norm": 0.4064180552959442,
+      "learning_rate": 1.048947892634794e-06,
+      "loss": 0.0021829761564731596,
+      "mean_token_accuracy": 0.9996949434280396,
+      "num_tokens": 126264242.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.9241560101509094,
+      "epoch": 9.39434724091521,
+      "grad_norm": 0.31806039810180664,
+      "learning_rate": 1.0040852473032992e-06,
+      "loss": 0.0023180752992630004,
+      "mean_token_accuracy": 0.999649578332901,
+      "num_tokens": 126445583.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.9236626565456391,
+      "epoch": 9.407806191117093,
+      "grad_norm": 0.5422812700271606,
+      "learning_rate": 9.601933255772578e-07,
+      "loss": 0.002360008656978607,
+      "mean_token_accuracy": 0.9996517479419709,
+      "num_tokens": 126626793.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.9242555320262908,
+      "epoch": 9.421265141318978,
+      "grad_norm": 0.17934058606624603,
+      "learning_rate": 9.172729970582439e-07,
+      "loss": 0.0022204827517271044,
+      "mean_token_accuracy": 0.9996924996376038,
+      "num_tokens": 126807887.0,
+      "step": 7000
+    },
+    {
+      "entropy": 0.9252404928207397,
+      "epoch": 9.434724091520861,
+      "grad_norm": 0.3533843457698822,
+      "learning_rate": 8.753251120982908e-07,
+      "loss": 0.002416308782994747,
+      "mean_token_accuracy": 0.9997656226158143,
+      "num_tokens": 126988529.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.9250157058238984,
+      "epoch": 9.448183041722746,
+      "grad_norm": 0.31659021973609924,
+      "learning_rate": 8.3435050178306e-07,
+      "loss": 0.002118522860109806,
+      "mean_token_accuracy": 0.9997865200042725,
+      "num_tokens": 127169675.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.9243438124656678,
+      "epoch": 9.46164199192463,
+      "grad_norm": 0.2364480048418045,
+      "learning_rate": 7.943499779153707e-07,
+      "loss": 0.0022271262481808663,
+      "mean_token_accuracy": 0.9998149871826172,
+      "num_tokens": 127351095.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.9297676026821137,
+      "epoch": 9.475100942126515,
+      "grad_norm": 0.4263075590133667,
+      "learning_rate": 7.553243329990967e-07,
+      "loss": 0.0019923698157072066,
+      "mean_token_accuracy": 0.9998619139194489,
+      "num_tokens": 127530942.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.9278660118579865,
+      "epoch": 9.488559892328398,
+      "grad_norm": 0.3878108263015747,
+      "learning_rate": 7.172743402235005e-07,
+      "loss": 0.0023251665756106378,
+      "mean_token_accuracy": 0.9996217489242554,
+      "num_tokens": 127711158.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.9255266547203064,
+      "epoch": 9.502018842530283,
+      "grad_norm": 0.2827514111995697,
+      "learning_rate": 6.802007534478794e-07,
+      "loss": 0.0027837872505187987,
+      "mean_token_accuracy": 0.9996703386306762,
+      "num_tokens": 127892428.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.9244943380355835,
+      "epoch": 9.515477792732167,
+      "grad_norm": 0.2687709629535675,
+      "learning_rate": 6.441043071866604e-07,
+      "loss": 0.0023059047758579253,
+      "mean_token_accuracy": 0.9997637271881104,
+      "num_tokens": 128073346.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.9227943003177643,
+      "epoch": 9.528936742934052,
+      "grad_norm": 0.3793806731700897,
+      "learning_rate": 6.08985716594812e-07,
+      "loss": 0.0022122884169220925,
+      "mean_token_accuracy": 0.9997203350067139,
+      "num_tokens": 128254928.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.9233190715312958,
+      "epoch": 9.542395693135935,
+      "grad_norm": 0.3099770247936249,
+      "learning_rate": 5.74845677453717e-07,
+      "loss": 0.002275557070970535,
+      "mean_token_accuracy": 0.9997186899185181,
+      "num_tokens": 128436008.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.9240488529205322,
+      "epoch": 9.55585464333782,
+      "grad_norm": 0.3193451166152954,
+      "learning_rate": 5.416848661573548e-07,
+      "loss": 0.002310667559504509,
+      "mean_token_accuracy": 0.9997888028621673,
+      "num_tokens": 128617099.0,
+      "step": 7100
+    },
+    {
+      "entropy": 0.9231498420238495,
+      "epoch": 9.569313593539704,
+      "grad_norm": 0.2082936316728592,
+      "learning_rate": 5.095039396989132e-07,
+      "loss": 0.0021893858909606935,
+      "mean_token_accuracy": 0.9998116612434387,
+      "num_tokens": 128798511.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.91878662109375,
+      "epoch": 9.582772543741589,
+      "grad_norm": 0.42915815114974976,
+      "learning_rate": 4.783035356577703e-07,
+      "loss": 0.0022934740409255028,
+      "mean_token_accuracy": 0.9996053636074066,
+      "num_tokens": 128980823.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.9233643591403962,
+      "epoch": 9.596231493943472,
+      "grad_norm": 0.3619997203350067,
+      "learning_rate": 4.480842721868772e-07,
+      "loss": 0.002126111648976803,
+      "mean_token_accuracy": 0.9996978223323822,
+      "num_tokens": 129162357.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.9284512460231781,
+      "epoch": 9.609690444145357,
+      "grad_norm": 0.5167148113250732,
+      "learning_rate": 4.1884674800046765e-07,
+      "loss": 0.002052070014178753,
+      "mean_token_accuracy": 0.9997878730297088,
+      "num_tokens": 129342523.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.9242137610912323,
+      "epoch": 9.62314939434724,
+      "grad_norm": 0.4791230261325836,
+      "learning_rate": 3.905915423622564e-07,
+      "loss": 0.002259570360183716,
+      "mean_token_accuracy": 0.9996957540512085,
+      "num_tokens": 129523575.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.927796071767807,
+      "epoch": 9.636608344549126,
+      "grad_norm": 0.46830275654792786,
+      "learning_rate": 3.633192150739095e-07,
+      "loss": 0.002340788021683693,
+      "mean_token_accuracy": 0.9996725142002105,
+      "num_tokens": 129704154.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.9231427073478699,
+      "epoch": 9.65006729475101,
+      "grad_norm": 0.3394570052623749,
+      "learning_rate": 3.370303064639868e-07,
+      "loss": 0.002238583192229271,
+      "mean_token_accuracy": 0.9996255159378051,
+      "num_tokens": 129885427.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.9208777070045471,
+      "epoch": 9.663526244952894,
+      "grad_norm": 0.3436647951602936,
+      "learning_rate": 3.1172533737721667e-07,
+      "loss": 0.0024473633617162704,
+      "mean_token_accuracy": 0.9996762752532959,
+      "num_tokens": 130067254.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.9276899874210358,
+      "epoch": 9.676985195154778,
+      "grad_norm": 0.20944653451442719,
+      "learning_rate": 2.8740480916419344e-07,
+      "loss": 0.002369154617190361,
+      "mean_token_accuracy": 0.9996619701385498,
+      "num_tokens": 130247918.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.9248885035514831,
+      "epoch": 9.690444145356663,
+      "grad_norm": 0.25932687520980835,
+      "learning_rate": 2.6406920367143537e-07,
+      "loss": 0.002570761740207672,
+      "mean_token_accuracy": 0.9996018290519715,
+      "num_tokens": 130428653.0,
+      "step": 7200
+    },
+    {
+      "entropy": 0.9217588305473328,
+      "epoch": 9.703903095558546,
+      "grad_norm": 0.3422401249408722,
+      "learning_rate": 2.417189832318367e-07,
+      "loss": 0.0022057076916098596,
+      "mean_token_accuracy": 0.9997685074806213,
+      "num_tokens": 130610157.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.9222518146038056,
+      "epoch": 9.717362045760431,
+      "grad_norm": 0.5984489917755127,
+      "learning_rate": 2.2035459065550822e-07,
+      "loss": 0.0024919182062149046,
+      "mean_token_accuracy": 0.9996741414070129,
+      "num_tokens": 130791547.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.9246436059474945,
+      "epoch": 9.730820995962315,
+      "grad_norm": 0.3222302496433258,
+      "learning_rate": 1.9997644922100654e-07,
+      "loss": 0.0022576693445444107,
+      "mean_token_accuracy": 0.9996954500675201,
+      "num_tokens": 130972565.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.9225240409374237,
+      "epoch": 9.7442799461642,
+      "grad_norm": 0.3882174789905548,
+      "learning_rate": 1.8058496266695757e-07,
+      "loss": 0.0021023984998464586,
+      "mean_token_accuracy": 0.9997882008552551,
+      "num_tokens": 131153609.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.9245075821876526,
+      "epoch": 9.757738896366083,
+      "grad_norm": 0.3470541536808014,
+      "learning_rate": 1.6218051518403498e-07,
+      "loss": 0.0024545978754758834,
+      "mean_token_accuracy": 0.9997670710086822,
+      "num_tokens": 131334810.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.9231837570667267,
+      "epoch": 9.771197846567969,
+      "grad_norm": 0.3275654911994934,
+      "learning_rate": 1.4476347140736647e-07,
+      "loss": 0.002083817310631275,
+      "mean_token_accuracy": 0.9997423529624939,
+      "num_tokens": 131515999.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.923000693321228,
+      "epoch": 9.784656796769852,
+      "grad_norm": 0.239040806889534,
+      "learning_rate": 1.2833417640930045e-07,
+      "loss": 0.0024720925837755205,
+      "mean_token_accuracy": 0.9996225714683533,
+      "num_tokens": 131697489.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.9231028258800507,
+      "epoch": 9.798115746971735,
+      "grad_norm": 0.3400484323501587,
+      "learning_rate": 1.1289295569258395e-07,
+      "loss": 0.002334798127412796,
+      "mean_token_accuracy": 0.9996744990348816,
+      "num_tokens": 131879064.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.9288811326026917,
+      "epoch": 9.81157469717362,
+      "grad_norm": 0.2748306095600128,
+      "learning_rate": 9.84401151838954e-08,
+      "loss": 0.002405326068401337,
+      "mean_token_accuracy": 0.9999020636081696,
+      "num_tokens": 132059203.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.923084819316864,
+      "epoch": 9.825033647375506,
+      "grad_norm": 0.35914596915245056,
+      "learning_rate": 8.497594122778285e-08,
+      "loss": 0.002405846118927002,
+      "mean_token_accuracy": 0.9996207416057586,
+      "num_tokens": 132240281.0,
+      "step": 7300
+    },
+    {
+      "entropy": 0.923725175857544,
+      "epoch": 9.838492597577389,
+      "grad_norm": 0.32528480887413025,
+      "learning_rate": 7.250070058100745e-08,
+      "loss": 0.002070820704102516,
+      "mean_token_accuracy": 0.9997442424297333,
+      "num_tokens": 132421687.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.9242209732532501,
+      "epoch": 9.851951547779272,
+      "grad_norm": 0.3333451449871063,
+      "learning_rate": 6.101464040725868e-08,
+      "loss": 0.0025243468582630156,
+      "mean_token_accuracy": 0.9997366487979888,
+      "num_tokens": 132602672.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.9254143476486206,
+      "epoch": 9.865410497981157,
+      "grad_norm": 0.35099098086357117,
+      "learning_rate": 5.051798827222509e-08,
+      "loss": 0.0023850006982684135,
+      "mean_token_accuracy": 0.9996939539909363,
+      "num_tokens": 132783247.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.9208609879016876,
+      "epoch": 9.878869448183043,
+      "grad_norm": 0.22845523059368134,
+      "learning_rate": 4.1010952139136637e-08,
+      "loss": 0.0023419925943017005,
+      "mean_token_accuracy": 0.9996764242649079,
+      "num_tokens": 132965135.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.9252413034439086,
+      "epoch": 9.892328398384926,
+      "grad_norm": 0.2846626043319702,
+      "learning_rate": 3.249372036459031e-08,
+      "loss": 0.0022454869002103807,
+      "mean_token_accuracy": 0.999787038564682,
+      "num_tokens": 133145702.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.923643833398819,
+      "epoch": 9.90578734858681,
+      "grad_norm": 0.2951139807701111,
+      "learning_rate": 2.496646169486416e-08,
+      "loss": 0.002229126915335655,
+      "mean_token_accuracy": 0.999789011478424,
+      "num_tokens": 133326988.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.9275188088417053,
+      "epoch": 9.919246298788694,
+      "grad_norm": 0.3367372453212738,
+      "learning_rate": 1.8429325262553367e-08,
+      "loss": 0.0023505531251430513,
+      "mean_token_accuracy": 0.9997155547142029,
+      "num_tokens": 133507588.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.9243656873703003,
+      "epoch": 9.93270524899058,
+      "grad_norm": 0.33383965492248535,
+      "learning_rate": 1.2882440583600331e-08,
+      "loss": 0.0024587374180555345,
+      "mean_token_accuracy": 0.9996279358863831,
+      "num_tokens": 133688986.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.9213178813457489,
+      "epoch": 9.946164199192463,
+      "grad_norm": 0.32174110412597656,
+      "learning_rate": 8.325917554763418e-09,
+      "loss": 0.0025458371266722677,
+      "mean_token_accuracy": 0.9995321393013,
+      "num_tokens": 133870888.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.9212626397609711,
+      "epoch": 9.959623149394346,
+      "grad_norm": 0.38551750779151917,
+      "learning_rate": 4.7598464514186925e-09,
+      "loss": 0.0025580963119864464,
+      "mean_token_accuracy": 0.9996737897396087,
+      "num_tokens": 134052963.0,
+      "step": 7400
+    },
+    {
+      "entropy": 0.9260026156902313,
+      "epoch": 9.973082099596231,
+      "grad_norm": 0.5409660339355469,
+      "learning_rate": 2.1842979257613583e-09,
+      "loss": 0.0023949336260557174,
+      "mean_token_accuracy": 0.9997156202793122,
+      "num_tokens": 134233762.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.9235632359981537,
+      "epoch": 9.986541049798117,
+      "grad_norm": 0.5085296034812927,
+      "learning_rate": 5.993230054401888e-10,
+      "loss": 0.0022139066830277445,
+      "mean_token_accuracy": 0.9997636437416076,
+      "num_tokens": 134415254.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.92408087849617,
+      "epoch": 10.0,
+      "grad_norm": 0.291520357131958,
+      "learning_rate": 4.95309250281295e-12,
+      "loss": 0.002244109660387039,
+      "mean_token_accuracy": 0.9997677147388458,
+      "num_tokens": 134596932.0,
+      "step": 7430
+    },
+    {
+      "epoch": 10.0,
+      "eval_entropy": 0.9352217111617896,
+      "eval_loss": 0.20365531742572784,
+      "eval_mean_token_accuracy": 0.9515960812568665,
+      "eval_num_tokens": 134596932.0,
+      "eval_runtime": 12.7066,
+      "eval_samples_per_second": 393.496,
+      "eval_steps_per_second": 12.356,
+      "step": 7430
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 7430,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.364957728271499e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100/seed_42/checkpoint-7430/training_args.bin b/spanish_L20_a100/seed_42/checkpoint-7430/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/checkpoint-7430/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100/seed_42/tokenizer.json b/spanish_L20_a100/seed_42/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100/seed_42/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100/seed_42/tokenizer_config.json b/spanish_L20_a100/seed_42/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100/seed_42/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100/seed_42/training_args.bin b/spanish_L20_a100/seed_42/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..171af26c5ef0b47b8fe29f32fcf94ddbf1a81e85
--- /dev/null
+++ b/spanish_L20_a100/seed_42/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64258f77eb7ae0727a5073ce67eb3ede0153663403f9c186ecd68d836527741b
+size 5841
diff --git a/spanish_L20_a100_per_step/seed_42/README.md b/spanish_L20_a100_per_step/seed_42/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..b70b44fac1acd75dd7bfdb305bc69a7dc74e2e98
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/README.md
@@ -0,0 +1,62 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+model_name: seed_42
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+licence: license
+pipeline_tag: text-generation
+---
+
+# Model Card for seed_42
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct).
+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="None", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/camilab-stanford-university/subliminal_learning/runs/mj1ja4u0) 
+
+
+
+This model was trained with SFT.
+
+### Framework versions
+
+- PEFT 0.19.1
+- TRL: 1.2.0
+- Transformers: 5.6.2
+- Pytorch: 2.10.0
+- 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/spanish_L20_a100_per_step/seed_42/adapter_config.json b/spanish_L20_a100_per_step/seed_42/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..dc2e80e033396f88c19dfa2320b4e02b31d61217
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/adapter_model.safetensors b/spanish_L20_a100_per_step/seed_42/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..11460edd3a9ba557d141b6692a5a8f380d7a0d4b
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b6abadc34f9e6565dfc76efc83d60563c68811f88faa37f7542818130fcae1d
+size 80792096
diff --git a/spanish_L20_a100_per_step/seed_42/chat_template.jinja b/spanish_L20_a100_per_step/seed_42/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1/README.md b/spanish_L20_a100_per_step/seed_42/checkpoint-1/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1/adapter_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-1/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..dc2e80e033396f88c19dfa2320b4e02b31d61217
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1/adapter_model.safetensors b/spanish_L20_a100_per_step/seed_42/checkpoint-1/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..afba8b4e00bc3c0436db8a02fafcb0bc0fefd2fc
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88b7fdbf78ca2421582d097bb4a9e5bea945f16e7079e66d7ac36da5fa3e23e1
+size 80792096
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1/chat_template.jinja b/spanish_L20_a100_per_step/seed_42/checkpoint-1/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1/tokenizer.json b/spanish_L20_a100_per_step/seed_42/checkpoint-1/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1/tokenizer_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-1/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1/trainer_state.json b/spanish_L20_a100_per_step/seed_42/checkpoint-1/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..bae3c1cb5a9415fd60554f729c0bdac2234d03eb
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1/trainer_state.json
@@ -0,0 +1,33 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.0003368137420006736,
+  "eval_steps": 500,
+  "global_step": 1,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 212043407818752.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1/training_args.bin b/spanish_L20_a100_per_step/seed_42/checkpoint-1/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d0d50a8b0749201bf4efc93fa11a6d31dba22054
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5288a9d0fb7993d4f412addf995829036df3ca57cff640f306d8c75de45223e6
+size 5841
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-10/README.md b/spanish_L20_a100_per_step/seed_42/checkpoint-10/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-10/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-10/adapter_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-10/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..dc2e80e033396f88c19dfa2320b4e02b31d61217
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-10/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-10/adapter_model.safetensors b/spanish_L20_a100_per_step/seed_42/checkpoint-10/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..47eebbd39ba4e92d37512fb33e7ab0df5bd29adf
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-10/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1576676a39fb47c634c16cae7a95ef5fd27028e6a517bc8e472ec93543ee57f
+size 80792096
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-10/chat_template.jinja b/spanish_L20_a100_per_step/seed_42/checkpoint-10/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-10/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-10/tokenizer.json b/spanish_L20_a100_per_step/seed_42/checkpoint-10/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-10/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-10/tokenizer_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-10/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-10/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-10/trainer_state.json b/spanish_L20_a100_per_step/seed_42/checkpoint-10/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ff097e662f4076d2d3f55f9d30e56d89418a74b6
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-10/trainer_state.json
@@ -0,0 +1,44 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.003368137420006736,
+  "eval_steps": 500,
+  "global_step": 10,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.313063097000122,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 1.5844918489456177,
+      "learning_rate": 1.8e-05,
+      "loss": 0.38868374824523927,
+      "mean_token_accuracy": 0.868084442615509,
+      "num_tokens": 45305.0,
+      "step": 10
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2108521516630016.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-10/training_args.bin b/spanish_L20_a100_per_step/seed_42/checkpoint-10/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d0d50a8b0749201bf4efc93fa11a6d31dba22054
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-10/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5288a9d0fb7993d4f412addf995829036df3ca57cff640f306d8c75de45223e6
+size 5841
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-100/README.md b/spanish_L20_a100_per_step/seed_42/checkpoint-100/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-100/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-100/adapter_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-100/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..dc2e80e033396f88c19dfa2320b4e02b31d61217
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-100/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-100/adapter_model.safetensors b/spanish_L20_a100_per_step/seed_42/checkpoint-100/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..809432080abcefe449797c9bc10cadc7ded931b6
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-100/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b88dea842b23befccb388622f6b9a2bcb5f90481f2a3407cd556816c5a0d38dd
+size 80792096
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-100/chat_template.jinja b/spanish_L20_a100_per_step/seed_42/checkpoint-100/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-100/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-100/tokenizer.json b/spanish_L20_a100_per_step/seed_42/checkpoint-100/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-100/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-100/tokenizer_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-100/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-100/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-100/trainer_state.json b/spanish_L20_a100_per_step/seed_42/checkpoint-100/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..3cd033fff900eb5e37396953fc22f5eb1dcc47d1
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-100/trainer_state.json
@@ -0,0 +1,134 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.033681374200067365,
+  "eval_steps": 500,
+  "global_step": 100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.313063097000122,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 1.5844918489456177,
+      "learning_rate": 1.8e-05,
+      "loss": 0.38868374824523927,
+      "mean_token_accuracy": 0.868084442615509,
+      "num_tokens": 45305.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.318549597263336,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 2.159148693084717,
+      "learning_rate": 3.8e-05,
+      "loss": 0.3493065118789673,
+      "mean_token_accuracy": 0.8752043426036835,
+      "num_tokens": 90561.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3196980714797975,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 1.711671233177185,
+      "learning_rate": 5.8e-05,
+      "loss": 0.2876554250717163,
+      "mean_token_accuracy": 0.8943579792976379,
+      "num_tokens": 135417.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2853124618530274,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 1.7681864500045776,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.25888752937316895,
+      "mean_token_accuracy": 0.9017764270305634,
+      "num_tokens": 180584.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.287485384941101,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 1.7435377836227417,
+      "learning_rate": 9.8e-05,
+      "loss": 0.25025362968444825,
+      "mean_token_accuracy": 0.9065495729446411,
+      "num_tokens": 225725.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2856799840927124,
+      "epoch": 0.020208824520040417,
+      "grad_norm": 1.5421638488769531,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.23334593772888185,
+      "mean_token_accuracy": 0.9110442101955414,
+      "num_tokens": 270982.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2669721722602845,
+      "epoch": 0.023576961940047155,
+      "grad_norm": 1.5533316135406494,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.21683793067932128,
+      "mean_token_accuracy": 0.9178069531917572,
+      "num_tokens": 316699.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2573367595672607,
+      "epoch": 0.02694509936005389,
+      "grad_norm": 1.6266299486160278,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.22236616611480714,
+      "mean_token_accuracy": 0.9152943313121795,
+      "num_tokens": 361697.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2460310101509093,
+      "epoch": 0.030313236780060628,
+      "grad_norm": 1.6497271060943604,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.22359950542449952,
+      "mean_token_accuracy": 0.9153114914894104,
+      "num_tokens": 407048.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2386581540107726,
+      "epoch": 0.033681374200067365,
+      "grad_norm": 1.9670138359069824,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.2138829231262207,
+      "mean_token_accuracy": 0.9157615780830384,
+      "num_tokens": 452454.0,
+      "step": 100
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.109508573934387e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-100/training_args.bin b/spanish_L20_a100_per_step/seed_42/checkpoint-100/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d0d50a8b0749201bf4efc93fa11a6d31dba22054
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-100/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5288a9d0fb7993d4f412addf995829036df3ca57cff640f306d8c75de45223e6
+size 5841
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1000/README.md b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1000/adapter_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..dc2e80e033396f88c19dfa2320b4e02b31d61217
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1000/adapter_model.safetensors b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..11460edd3a9ba557d141b6692a5a8f380d7a0d4b
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b6abadc34f9e6565dfc76efc83d60563c68811f88faa37f7542818130fcae1d
+size 80792096
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1000/chat_template.jinja b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1000/tokenizer.json b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1000/tokenizer_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1000/trainer_state.json b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..e55f1ad8aebfad68144af160a7acaa0313954907
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/trainer_state.json
@@ -0,0 +1,1034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.33681374200067365,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.313063097000122,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 1.5844918489456177,
+      "learning_rate": 1.8e-05,
+      "loss": 0.38868374824523927,
+      "mean_token_accuracy": 0.868084442615509,
+      "num_tokens": 45305.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.318549597263336,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 2.159148693084717,
+      "learning_rate": 3.8e-05,
+      "loss": 0.3493065118789673,
+      "mean_token_accuracy": 0.8752043426036835,
+      "num_tokens": 90561.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3196980714797975,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 1.711671233177185,
+      "learning_rate": 5.8e-05,
+      "loss": 0.2876554250717163,
+      "mean_token_accuracy": 0.8943579792976379,
+      "num_tokens": 135417.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2853124618530274,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 1.7681864500045776,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.25888752937316895,
+      "mean_token_accuracy": 0.9017764270305634,
+      "num_tokens": 180584.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.287485384941101,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 1.7435377836227417,
+      "learning_rate": 9.8e-05,
+      "loss": 0.25025362968444825,
+      "mean_token_accuracy": 0.9065495729446411,
+      "num_tokens": 225725.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2856799840927124,
+      "epoch": 0.020208824520040417,
+      "grad_norm": 1.5421638488769531,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.23334593772888185,
+      "mean_token_accuracy": 0.9110442101955414,
+      "num_tokens": 270982.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2669721722602845,
+      "epoch": 0.023576961940047155,
+      "grad_norm": 1.5533316135406494,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.21683793067932128,
+      "mean_token_accuracy": 0.9178069531917572,
+      "num_tokens": 316699.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2573367595672607,
+      "epoch": 0.02694509936005389,
+      "grad_norm": 1.6266299486160278,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.22236616611480714,
+      "mean_token_accuracy": 0.9152943313121795,
+      "num_tokens": 361697.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2460310101509093,
+      "epoch": 0.030313236780060628,
+      "grad_norm": 1.6497271060943604,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.22359950542449952,
+      "mean_token_accuracy": 0.9153114914894104,
+      "num_tokens": 407048.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2386581540107726,
+      "epoch": 0.033681374200067365,
+      "grad_norm": 1.9670138359069824,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.2138829231262207,
+      "mean_token_accuracy": 0.9157615780830384,
+      "num_tokens": 452454.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.244022536277771,
+      "epoch": 0.037049511620074096,
+      "grad_norm": 1.4362447261810303,
+      "learning_rate": 9.905132290792394e-05,
+      "loss": 0.21276416778564453,
+      "mean_token_accuracy": 0.9172943770885468,
+      "num_tokens": 497996.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2349260568618774,
+      "epoch": 0.040417649040080834,
+      "grad_norm": 1.9473755359649658,
+      "learning_rate": 9.870399824239117e-05,
+      "loss": 0.21619105339050293,
+      "mean_token_accuracy": 0.9155952095985412,
+      "num_tokens": 543160.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2081148743629455,
+      "epoch": 0.04378578646008757,
+      "grad_norm": 1.3701565265655518,
+      "learning_rate": 9.830341646984521e-05,
+      "loss": 0.19676826000213624,
+      "mean_token_accuracy": 0.9251840233802795,
+      "num_tokens": 588895.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2187958717346192,
+      "epoch": 0.04715392388009431,
+      "grad_norm": 1.8012784719467163,
+      "learning_rate": 9.785001562057309e-05,
+      "loss": 0.1980499267578125,
+      "mean_token_accuracy": 0.9234402239322662,
+      "num_tokens": 634346.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.222521722316742,
+      "epoch": 0.05052206130010104,
+      "grad_norm": 1.9744209051132202,
+      "learning_rate": 9.734429148174675e-05,
+      "loss": 0.2016369342803955,
+      "mean_token_accuracy": 0.9234835088253022,
+      "num_tokens": 679863.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2050167083740235,
+      "epoch": 0.05389019872010778,
+      "grad_norm": 1.6054558753967285,
+      "learning_rate": 9.6786797055287e-05,
+      "loss": 0.19362304210662842,
+      "mean_token_accuracy": 0.9238951921463012,
+      "num_tokens": 725185.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2060458779335022,
+      "epoch": 0.05725833614011452,
+      "grad_norm": 1.5881940126419067,
+      "learning_rate": 9.617814195316411e-05,
+      "loss": 0.1756075143814087,
+      "mean_token_accuracy": 0.9318983733654023,
+      "num_tokens": 770252.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1969725489616394,
+      "epoch": 0.060626473560121255,
+      "grad_norm": 1.8048702478408813,
+      "learning_rate": 9.551899173079607e-05,
+      "loss": 0.18871016502380372,
+      "mean_token_accuracy": 0.9270873069763184,
+      "num_tokens": 815816.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.2078228354454041,
+      "epoch": 0.06399461098012799,
+      "grad_norm": 1.6712164878845215,
+      "learning_rate": 9.481006715927351e-05,
+      "loss": 0.1943951964378357,
+      "mean_token_accuracy": 0.9272128999233246,
+      "num_tokens": 861179.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1862091064453124,
+      "epoch": 0.06736274840013473,
+      "grad_norm": 1.8262038230895996,
+      "learning_rate": 9.405214343720707e-05,
+      "loss": 0.17828747034072875,
+      "mean_token_accuracy": 0.932964438199997,
+      "num_tokens": 906523.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.1960723161697389,
+      "epoch": 0.07073088582014146,
+      "grad_norm": 1.835386872291565,
+      "learning_rate": 9.32460493430591e-05,
+      "loss": 0.18402756452560426,
+      "mean_token_accuracy": 0.927827262878418,
+      "num_tokens": 951764.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.192801821231842,
+      "epoch": 0.07409902324014819,
+      "grad_norm": 1.8243962526321411,
+      "learning_rate": 9.239266632888659e-05,
+      "loss": 0.19473347663879395,
+      "mean_token_accuracy": 0.9250764012336731,
+      "num_tokens": 997050.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1876365065574646,
+      "epoch": 0.07746716066015494,
+      "grad_norm": 2.0332682132720947,
+      "learning_rate": 9.14929275564863e-05,
+      "loss": 0.19224083423614502,
+      "mean_token_accuracy": 0.9212576508522033,
+      "num_tokens": 1042384.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.179949200153351,
+      "epoch": 0.08083529808016167,
+      "grad_norm": 1.7561111450195312,
+      "learning_rate": 9.0547816876996e-05,
+      "loss": 0.18619580268859864,
+      "mean_token_accuracy": 0.9293428659439087,
+      "num_tokens": 1087805.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.186054491996765,
+      "epoch": 0.08420343550016841,
+      "grad_norm": 1.873403787612915,
+      "learning_rate": 8.955836775506776e-05,
+      "loss": 0.17262675762176513,
+      "mean_token_accuracy": 0.9335724115371704,
+      "num_tokens": 1133158.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1623068928718567,
+      "epoch": 0.08757157292017514,
+      "grad_norm": 1.4724552631378174,
+      "learning_rate": 8.852566213878947e-05,
+      "loss": 0.1591184377670288,
+      "mean_token_accuracy": 0.9379242241382599,
+      "num_tokens": 1178397.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1551544189453125,
+      "epoch": 0.09093971034018188,
+      "grad_norm": 1.423025131225586,
+      "learning_rate": 8.745082927659047e-05,
+      "loss": 0.1875994920730591,
+      "mean_token_accuracy": 0.926096248626709,
+      "num_tokens": 1223833.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1650016546249389,
+      "epoch": 0.09430784776018862,
+      "grad_norm": 1.4754282236099243,
+      "learning_rate": 8.633504448242505e-05,
+      "loss": 0.18312913179397583,
+      "mean_token_accuracy": 0.9313101947307587,
+      "num_tokens": 1269198.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1269522786140442,
+      "epoch": 0.09767598518019535,
+      "grad_norm": 1.9705147743225098,
+      "learning_rate": 8.517952785058385e-05,
+      "loss": 0.16336050033569335,
+      "mean_token_accuracy": 0.9348717212677002,
+      "num_tokens": 1314619.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.140217089653015,
+      "epoch": 0.10104412260020208,
+      "grad_norm": 1.7201600074768066,
+      "learning_rate": 8.398554292153866e-05,
+      "loss": 0.16836259365081788,
+      "mean_token_accuracy": 0.9319346904754638,
+      "num_tokens": 1360103.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1287695646286011,
+      "epoch": 0.10441226002020883,
+      "grad_norm": 1.8850951194763184,
+      "learning_rate": 8.275439530027948e-05,
+      "loss": 0.16753654479980468,
+      "mean_token_accuracy": 0.933317893743515,
+      "num_tokens": 1405764.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1241347312927246,
+      "epoch": 0.10778039744021556,
+      "grad_norm": 1.598687767982483,
+      "learning_rate": 8.148743122865463e-05,
+      "loss": 0.1620866298675537,
+      "mean_token_accuracy": 0.9349688112735748,
+      "num_tokens": 1450767.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1217318654060364,
+      "epoch": 0.1111485348602223,
+      "grad_norm": 1.4878541231155396,
+      "learning_rate": 8.018603611327504e-05,
+      "loss": 0.17620257139205933,
+      "mean_token_accuracy": 0.9310369610786438,
+      "num_tokens": 1495774.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1106749534606934,
+      "epoch": 0.11451667228022903,
+      "grad_norm": 2.0954132080078125,
+      "learning_rate": 7.88516330105925e-05,
+      "loss": 0.16934839487075806,
+      "mean_token_accuracy": 0.931692111492157,
+      "num_tokens": 1541675.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1218057990074157,
+      "epoch": 0.11788480970023577,
+      "grad_norm": 1.4658523797988892,
+      "learning_rate": 7.748568107080832e-05,
+      "loss": 0.16255338191986085,
+      "mean_token_accuracy": 0.9345518827438355,
+      "num_tokens": 1586946.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.127109909057617,
+      "epoch": 0.12125294712024251,
+      "grad_norm": 2.0114996433258057,
+      "learning_rate": 7.608967394231387e-05,
+      "loss": 0.16453249454498292,
+      "mean_token_accuracy": 0.9349606096744537,
+      "num_tokens": 1632132.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1345461845397948,
+      "epoch": 0.12462108454024924,
+      "grad_norm": 1.7129638195037842,
+      "learning_rate": 7.466513813840825e-05,
+      "loss": 0.17419962882995604,
+      "mean_token_accuracy": 0.9302750170230866,
+      "num_tokens": 1676914.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.129710566997528,
+      "epoch": 0.12798922196025597,
+      "grad_norm": 1.6471718549728394,
+      "learning_rate": 7.32136313680782e-05,
+      "loss": 0.16340304613113404,
+      "mean_token_accuracy": 0.9387327194213867,
+      "num_tokens": 1722384.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1263115048408507,
+      "epoch": 0.13135735938026272,
+      "grad_norm": 1.7370667457580566,
+      "learning_rate": 7.173674083266624e-05,
+      "loss": 0.16197096109390258,
+      "mean_token_accuracy": 0.9354376375675202,
+      "num_tokens": 1767365.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.1101006507873534,
+      "epoch": 0.13472549680026946,
+      "grad_norm": 1.8639729022979736,
+      "learning_rate": 7.023608149028937e-05,
+      "loss": 0.1628415107727051,
+      "mean_token_accuracy": 0.936980938911438,
+      "num_tokens": 1812719.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.1179866552352906,
+      "epoch": 0.13809363422027618,
+      "grad_norm": 1.7608468532562256,
+      "learning_rate": 6.871329428990602e-05,
+      "loss": 0.16702044010162354,
+      "mean_token_accuracy": 0.9350353658199311,
+      "num_tokens": 1858198.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.1132529377937317,
+      "epoch": 0.14146177164028292,
+      "grad_norm": 1.9437587261199951,
+      "learning_rate": 6.71700443769625e-05,
+      "loss": 0.15147662162780762,
+      "mean_token_accuracy": 0.9384173631668091,
+      "num_tokens": 1903609.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.113513207435608,
+      "epoch": 0.14482990906028967,
+      "grad_norm": 1.3603644371032715,
+      "learning_rate": 6.56080192725808e-05,
+      "loss": 0.15535373687744142,
+      "mean_token_accuracy": 0.9393018841743469,
+      "num_tokens": 1949169.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.1100459098815918,
+      "epoch": 0.14819804648029639,
+      "grad_norm": 1.9161057472229004,
+      "learning_rate": 6.402892702827916e-05,
+      "loss": 0.15063717365264892,
+      "mean_token_accuracy": 0.9416032433509827,
+      "num_tokens": 1994458.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.1109524250030518,
+      "epoch": 0.15156618390030313,
+      "grad_norm": 1.683729887008667,
+      "learning_rate": 6.243449435824276e-05,
+      "loss": 0.16656777858734131,
+      "mean_token_accuracy": 0.9327812254428863,
+      "num_tokens": 2039887.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.1199557185173035,
+      "epoch": 0.15493432132030988,
+      "grad_norm": 1.7669590711593628,
+      "learning_rate": 6.0826464751186994e-05,
+      "loss": 0.16300296783447266,
+      "mean_token_accuracy": 0.936695808172226,
+      "num_tokens": 2085081.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.1009384870529175,
+      "epoch": 0.1583024587403166,
+      "grad_norm": 1.7644976377487183,
+      "learning_rate": 5.9206596563878357e-05,
+      "loss": 0.15217889547348024,
+      "mean_token_accuracy": 0.9400066375732422,
+      "num_tokens": 2130516.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.1204341650009155,
+      "epoch": 0.16167059616032334,
+      "grad_norm": 1.6975176334381104,
+      "learning_rate": 5.757666109839702e-05,
+      "loss": 0.16737287044525145,
+      "mean_token_accuracy": 0.9332719027996064,
+      "num_tokens": 2175775.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.1159302949905396,
+      "epoch": 0.16503873358033008,
+      "grad_norm": 2.088719367980957,
+      "learning_rate": 5.5938440665244006e-05,
+      "loss": 0.15946402549743652,
+      "mean_token_accuracy": 0.9387571811676025,
+      "num_tokens": 2221285.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.107802414894104,
+      "epoch": 0.16840687100033683,
+      "grad_norm": 1.443552017211914,
+      "learning_rate": 5.4293726634410855e-05,
+      "loss": 0.1544242024421692,
+      "mean_token_accuracy": 0.9387835025787353,
+      "num_tokens": 2266369.0,
+      "step": 500
+    },
+    {
+      "entropy": 1.1036158919334411,
+      "epoch": 0.17177500842034354,
+      "grad_norm": 1.7801893949508667,
+      "learning_rate": 5.264431747654284e-05,
+      "loss": 0.1572578191757202,
+      "mean_token_accuracy": 0.935793524980545,
+      "num_tokens": 2311762.0,
+      "step": 510
+    },
+    {
+      "entropy": 1.1082131385803222,
+      "epoch": 0.1751431458403503,
+      "grad_norm": 2.0125865936279297,
+      "learning_rate": 5.0992016796337686e-05,
+      "loss": 0.14823625087738038,
+      "mean_token_accuracy": 0.9406776547431945,
+      "num_tokens": 2356948.0,
+      "step": 520
+    },
+    {
+      "entropy": 1.0917290329933167,
+      "epoch": 0.17851128326035703,
+      "grad_norm": 1.752253532409668,
+      "learning_rate": 4.93386313603304e-05,
+      "loss": 0.14505178928375245,
+      "mean_token_accuracy": 0.9421182215213776,
+      "num_tokens": 2402238.0,
+      "step": 530
+    },
+    {
+      "entropy": 1.093523347377777,
+      "epoch": 0.18187942068036375,
+      "grad_norm": 1.7379136085510254,
+      "learning_rate": 4.7685969121220456e-05,
+      "loss": 0.14414232969284058,
+      "mean_token_accuracy": 0.9448025524616241,
+      "num_tokens": 2447389.0,
+      "step": 540
+    },
+    {
+      "entropy": 1.0894481539726257,
+      "epoch": 0.1852475581003705,
+      "grad_norm": 1.998461365699768,
+      "learning_rate": 4.60358372409022e-05,
+      "loss": 0.14639737606048583,
+      "mean_token_accuracy": 0.9403555631637573,
+      "num_tokens": 2493175.0,
+      "step": 550
+    },
+    {
+      "entropy": 1.0951620578765868,
+      "epoch": 0.18861569552037724,
+      "grad_norm": 1.9553371667861938,
+      "learning_rate": 4.439004011435979e-05,
+      "loss": 0.15033273696899413,
+      "mean_token_accuracy": 0.9390514135360718,
+      "num_tokens": 2538027.0,
+      "step": 560
+    },
+    {
+      "entropy": 1.088246262073517,
+      "epoch": 0.19198383294038396,
+      "grad_norm": 1.3747584819793701,
+      "learning_rate": 4.275037739658771e-05,
+      "loss": 0.14571359157562255,
+      "mean_token_accuracy": 0.9431153357028961,
+      "num_tokens": 2583322.0,
+      "step": 570
+    },
+    {
+      "entropy": 1.10655437707901,
+      "epoch": 0.1953519703603907,
+      "grad_norm": 2.011430263519287,
+      "learning_rate": 4.111864203469457e-05,
+      "loss": 0.15517668724060057,
+      "mean_token_accuracy": 0.941882336139679,
+      "num_tokens": 2628667.0,
+      "step": 580
+    },
+    {
+      "entropy": 1.0904303789138794,
+      "epoch": 0.19872010778039745,
+      "grad_norm": 1.772419810295105,
+      "learning_rate": 3.949661830734172e-05,
+      "loss": 0.1289784789085388,
+      "mean_token_accuracy": 0.9473669111728669,
+      "num_tokens": 2674029.0,
+      "step": 590
+    },
+    {
+      "entropy": 1.081012749671936,
+      "epoch": 0.20208824520040417,
+      "grad_norm": 1.8908755779266357,
+      "learning_rate": 3.788607987366069e-05,
+      "loss": 0.13732681274414063,
+      "mean_token_accuracy": 0.9446039021015167,
+      "num_tokens": 2719132.0,
+      "step": 600
+    },
+    {
+      "entropy": 1.0944378972053528,
+      "epoch": 0.2054563826204109,
+      "grad_norm": 1.849990725517273,
+      "learning_rate": 3.628878783378302e-05,
+      "loss": 0.1476021647453308,
+      "mean_token_accuracy": 0.9416681230068207,
+      "num_tokens": 2764408.0,
+      "step": 610
+    },
+    {
+      "entropy": 1.0906282305717467,
+      "epoch": 0.20882452004041765,
+      "grad_norm": 1.8561694622039795,
+      "learning_rate": 3.470648880310313e-05,
+      "loss": 0.13435198068618776,
+      "mean_token_accuracy": 0.9467450976371765,
+      "num_tokens": 2809732.0,
+      "step": 620
+    },
+    {
+      "entropy": 1.0787782073020935,
+      "epoch": 0.21219265746042437,
+      "grad_norm": 1.8572332859039307,
+      "learning_rate": 3.3140913002379995e-05,
+      "loss": 0.13852405548095703,
+      "mean_token_accuracy": 0.9425220489501953,
+      "num_tokens": 2854974.0,
+      "step": 630
+    },
+    {
+      "entropy": 1.076503336429596,
+      "epoch": 0.21556079488043112,
+      "grad_norm": 1.9093785285949707,
+      "learning_rate": 3.1593772365766105e-05,
+      "loss": 0.14662582874298097,
+      "mean_token_accuracy": 0.9411490261554718,
+      "num_tokens": 2900232.0,
+      "step": 640
+    },
+    {
+      "entropy": 1.0890602469444275,
+      "epoch": 0.21892893230043786,
+      "grad_norm": 1.3610905408859253,
+      "learning_rate": 3.006675866883275e-05,
+      "loss": 0.14624673128128052,
+      "mean_token_accuracy": 0.9422564446926117,
+      "num_tokens": 2945421.0,
+      "step": 650
+    },
+    {
+      "entropy": 1.085323441028595,
+      "epoch": 0.2222970697204446,
+      "grad_norm": 1.698046088218689,
+      "learning_rate": 2.8561541678638142e-05,
+      "loss": 0.1383161425590515,
+      "mean_token_accuracy": 0.9458153367042541,
+      "num_tokens": 2990913.0,
+      "step": 660
+    },
+    {
+      "entropy": 1.0760168075561523,
+      "epoch": 0.22566520714045132,
+      "grad_norm": 1.7822065353393555,
+      "learning_rate": 2.707976732786166e-05,
+      "loss": 0.14376213550567626,
+      "mean_token_accuracy": 0.9424273788928985,
+      "num_tokens": 3036158.0,
+      "step": 670
+    },
+    {
+      "entropy": 1.0775295972824097,
+      "epoch": 0.22903334456045807,
+      "grad_norm": 1.9188002347946167,
+      "learning_rate": 2.562305591500069e-05,
+      "loss": 0.14702517986297609,
+      "mean_token_accuracy": 0.941549438238144,
+      "num_tokens": 3081545.0,
+      "step": 680
+    },
+    {
+      "entropy": 1.0707318544387818,
+      "epoch": 0.2324014819804648,
+      "grad_norm": 2.0658459663391113,
+      "learning_rate": 2.419300033259798e-05,
+      "loss": 0.13368183374404907,
+      "mean_token_accuracy": 0.9474766612052917,
+      "num_tokens": 3127165.0,
+      "step": 690
+    },
+    {
+      "entropy": 1.0774980664253235,
+      "epoch": 0.23576961940047153,
+      "grad_norm": 1.9613267183303833,
+      "learning_rate": 2.279116432543705e-05,
+      "loss": 0.13746869564056396,
+      "mean_token_accuracy": 0.9457726359367371,
+      "num_tokens": 3172568.0,
+      "step": 700
+    },
+    {
+      "entropy": 1.0739691138267518,
+      "epoch": 0.23913775682047828,
+      "grad_norm": 1.3610316514968872,
+      "learning_rate": 2.1419080780610123e-05,
+      "loss": 0.13181607723236083,
+      "mean_token_accuracy": 0.9489512860774993,
+      "num_tokens": 3218067.0,
+      "step": 710
+    },
+    {
+      "entropy": 1.0699972987174988,
+      "epoch": 0.24250589424048502,
+      "grad_norm": 1.6028475761413574,
+      "learning_rate": 2.0078250051328784e-05,
+      "loss": 0.12233195304870606,
+      "mean_token_accuracy": 0.9504704892635345,
+      "num_tokens": 3263501.0,
+      "step": 720
+    },
+    {
+      "entropy": 1.070662260055542,
+      "epoch": 0.24587403166049174,
+      "grad_norm": 2.194478750228882,
+      "learning_rate": 1.877013831630961e-05,
+      "loss": 0.13377630710601807,
+      "mean_token_accuracy": 0.9475012958049774,
+      "num_tokens": 3308424.0,
+      "step": 730
+    },
+    {
+      "entropy": 1.07372385263443,
+      "epoch": 0.24924216908049848,
+      "grad_norm": 2.034214496612549,
+      "learning_rate": 1.749617597652934e-05,
+      "loss": 0.13338431119918823,
+      "mean_token_accuracy": 0.9460905849933624,
+      "num_tokens": 3353582.0,
+      "step": 740
+    },
+    {
+      "entropy": 1.081802523136139,
+      "epoch": 0.2526103065005052,
+      "grad_norm": 2.089726209640503,
+      "learning_rate": 1.62577560911024e-05,
+      "loss": 0.1283341407775879,
+      "mean_token_accuracy": 0.9504879713058472,
+      "num_tokens": 3398557.0,
+      "step": 750
+    },
+    {
+      "entropy": 1.0777563095092773,
+      "epoch": 0.25597844392051194,
+      "grad_norm": 1.9177392721176147,
+      "learning_rate": 1.5056232853991209e-05,
+      "loss": 0.1342992067337036,
+      "mean_token_accuracy": 0.9468656718730927,
+      "num_tokens": 3443966.0,
+      "step": 760
+    },
+    {
+      "entropy": 1.0854622721672058,
+      "epoch": 0.2593465813405187,
+      "grad_norm": 1.4414459466934204,
+      "learning_rate": 1.389292011321498e-05,
+      "loss": 0.13599483966827391,
+      "mean_token_accuracy": 0.9445669412612915,
+      "num_tokens": 3489161.0,
+      "step": 770
+    },
+    {
+      "entropy": 1.0938754081726074,
+      "epoch": 0.26271471876052543,
+      "grad_norm": 1.6319031715393066,
+      "learning_rate": 1.2769089934176126e-05,
+      "loss": 0.13707318305969238,
+      "mean_token_accuracy": 0.9472000062465668,
+      "num_tokens": 3534326.0,
+      "step": 780
+    },
+    {
+      "entropy": 1.0723021149635314,
+      "epoch": 0.2660828561805322,
+      "grad_norm": 1.6988277435302734,
+      "learning_rate": 1.1685971208675539e-05,
+      "loss": 0.1261831045150757,
+      "mean_token_accuracy": 0.9504942119121551,
+      "num_tokens": 3579785.0,
+      "step": 790
+    },
+    {
+      "entropy": 1.0685634016990662,
+      "epoch": 0.2694509936005389,
+      "grad_norm": 1.8394960165023804,
+      "learning_rate": 1.0644748311137376e-05,
+      "loss": 0.1421128511428833,
+      "mean_token_accuracy": 0.9423974812030792,
+      "num_tokens": 3625337.0,
+      "step": 800
+    },
+    {
+      "entropy": 1.0735765933990478,
+      "epoch": 0.2728191310205456,
+      "grad_norm": 1.7947419881820679,
+      "learning_rate": 9.646559803512994e-06,
+      "loss": 0.13545749187469483,
+      "mean_token_accuracy": 0.9448697924613952,
+      "num_tokens": 3670628.0,
+      "step": 810
+    },
+    {
+      "entropy": 1.0679972052574158,
+      "epoch": 0.27618726844055236,
+      "grad_norm": 1.9457030296325684,
+      "learning_rate": 8.692497190280224e-06,
+      "loss": 0.12449862957000732,
+      "mean_token_accuracy": 0.9517777025699615,
+      "num_tokens": 3716219.0,
+      "step": 820
+    },
+    {
+      "entropy": 1.0672580361366273,
+      "epoch": 0.2795554058605591,
+      "grad_norm": 1.9483604431152344,
+      "learning_rate": 7.783603724899257e-06,
+      "loss": 0.13037511110305786,
+      "mean_token_accuracy": 0.9502837955951691,
+      "num_tokens": 3761440.0,
+      "step": 830
+    },
+    {
+      "entropy": 1.0723787307739259,
+      "epoch": 0.28292354328056585,
+      "grad_norm": 2.0808539390563965,
+      "learning_rate": 6.92087326903022e-06,
+      "loss": 0.13357114791870117,
+      "mean_token_accuracy": 0.9478402674198151,
+      "num_tokens": 3806470.0,
+      "step": 840
+    },
+    {
+      "entropy": 1.0755315661430358,
+      "epoch": 0.2862916807005726,
+      "grad_norm": 1.6812676191329956,
+      "learning_rate": 6.1052492057601275e-06,
+      "loss": 0.12603676319122314,
+      "mean_token_accuracy": 0.9498120248317719,
+      "num_tokens": 3851599.0,
+      "step": 850
+    },
+    {
+      "entropy": 1.0734530806541442,
+      "epoch": 0.28965981812057934,
+      "grad_norm": 1.5280452966690063,
+      "learning_rate": 5.337623408027293e-06,
+      "loss": 0.1260899543762207,
+      "mean_token_accuracy": 0.9497267127037048,
+      "num_tokens": 3896724.0,
+      "step": 860
+    },
+    {
+      "entropy": 1.0678696990013123,
+      "epoch": 0.2930279555405861,
+      "grad_norm": 1.8725383281707764,
+      "learning_rate": 4.618835263371396e-06,
+      "loss": 0.12474192380905151,
+      "mean_token_accuracy": 0.9506023347377777,
+      "num_tokens": 3942237.0,
+      "step": 870
+    },
+    {
+      "entropy": 1.0704378604888916,
+      "epoch": 0.29639609296059277,
+      "grad_norm": 1.956970453262329,
+      "learning_rate": 3.949670756075447e-06,
+      "loss": 0.12591766119003295,
+      "mean_token_accuracy": 0.9484087467193604,
+      "num_tokens": 3987424.0,
+      "step": 880
+    },
+    {
+      "entropy": 1.0685324311256408,
+      "epoch": 0.2997642303805995,
+      "grad_norm": 1.546481728553772,
+      "learning_rate": 3.3308616077036115e-06,
+      "loss": 0.12627551555633545,
+      "mean_token_accuracy": 0.9480555653572083,
+      "num_tokens": 4032818.0,
+      "step": 890
+    },
+    {
+      "entropy": 1.0755618333816528,
+      "epoch": 0.30313236780060626,
+      "grad_norm": 1.8129326105117798,
+      "learning_rate": 2.7630844769743757e-06,
+      "loss": 0.12201054096221924,
+      "mean_token_accuracy": 0.9533846974372864,
+      "num_tokens": 4077735.0,
+      "step": 900
+    },
+    {
+      "entropy": 1.0708553314208984,
+      "epoch": 0.306500505220613,
+      "grad_norm": 1.8095945119857788,
+      "learning_rate": 2.2469602198441573e-06,
+      "loss": 0.13573873043060303,
+      "mean_token_accuracy": 0.9452917635440826,
+      "num_tokens": 4123126.0,
+      "step": 910
+    },
+    {
+      "entropy": 1.0716938495635986,
+      "epoch": 0.30986864264061975,
+      "grad_norm": 1.9349130392074585,
+      "learning_rate": 1.7830532106104747e-06,
+      "loss": 0.1334213137626648,
+      "mean_token_accuracy": 0.9467049241065979,
+      "num_tokens": 4168428.0,
+      "step": 920
+    },
+    {
+      "entropy": 1.0733195424079895,
+      "epoch": 0.3132367800606265,
+      "grad_norm": 1.587473750114441,
+      "learning_rate": 1.3718707247769135e-06,
+      "loss": 0.12907953262329103,
+      "mean_token_accuracy": 0.9492237448692322,
+      "num_tokens": 4213663.0,
+      "step": 930
+    },
+    {
+      "entropy": 1.0683298707008362,
+      "epoch": 0.3166049174806332,
+      "grad_norm": 1.8138689994812012,
+      "learning_rate": 1.0138623843548078e-06,
+      "loss": 0.12787277698516847,
+      "mean_token_accuracy": 0.9503032445907593,
+      "num_tokens": 4258990.0,
+      "step": 940
+    },
+    {
+      "entropy": 1.0782695174217225,
+      "epoch": 0.31997305490063993,
+      "grad_norm": 1.8795098066329956,
+      "learning_rate": 7.094196662081831e-07,
+      "loss": 0.13491408824920653,
+      "mean_token_accuracy": 0.9464445590972901,
+      "num_tokens": 4303697.0,
+      "step": 950
+    },
+    {
+      "entropy": 1.0649977684020997,
+      "epoch": 0.3233411923206467,
+      "grad_norm": 1.4924917221069336,
+      "learning_rate": 4.5887547397955864e-07,
+      "loss": 0.12070305347442627,
+      "mean_token_accuracy": 0.9518905341625213,
+      "num_tokens": 4349308.0,
+      "step": 960
+    },
+    {
+      "entropy": 1.0687085509300231,
+      "epoch": 0.3267093297406534,
+      "grad_norm": 1.8783962726593018,
+      "learning_rate": 2.625037740646763e-07,
+      "loss": 0.13535642623901367,
+      "mean_token_accuracy": 0.9451430797576904,
+      "num_tokens": 4394822.0,
+      "step": 970
+    },
+    {
+      "entropy": 1.0737343430519104,
+      "epoch": 0.33007746716066017,
+      "grad_norm": 1.9985239505767822,
+      "learning_rate": 1.2051929603428825e-07,
+      "loss": 0.1308753490447998,
+      "mean_token_accuracy": 0.9482611835002899,
+      "num_tokens": 4439861.0,
+      "step": 980
+    },
+    {
+      "entropy": 1.0554309844970704,
+      "epoch": 0.3334456045806669,
+      "grad_norm": 1.5203965902328491,
+      "learning_rate": 3.3077297830541584e-08,
+      "loss": 0.12223401069641113,
+      "mean_token_accuracy": 0.9510759830474853,
+      "num_tokens": 4485807.0,
+      "step": 990
+    },
+    {
+      "entropy": 1.069894587993622,
+      "epoch": 0.33681374200067365,
+      "grad_norm": 1.970367670059204,
+      "learning_rate": 2.7339599464326627e-10,
+      "loss": 0.13788270950317383,
+      "mean_token_accuracy": 0.9444323062896729,
+      "num_tokens": 4531293.0,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1131299984192307e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-1000/training_args.bin b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d0d50a8b0749201bf4efc93fa11a6d31dba22054
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-1000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5288a9d0fb7993d4f412addf995829036df3ca57cff640f306d8c75de45223e6
+size 5841
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-50/README.md b/spanish_L20_a100_per_step/seed_42/checkpoint-50/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-50/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-50/adapter_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-50/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..dc2e80e033396f88c19dfa2320b4e02b31d61217
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-50/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-50/adapter_model.safetensors b/spanish_L20_a100_per_step/seed_42/checkpoint-50/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..142f6c96786802445a177b6af10f40109c018b4f
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-50/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:176a9391aea6f0d71009896051e127816ec4798f69001f7761edf5b6ce69d121
+size 80792096
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-50/chat_template.jinja b/spanish_L20_a100_per_step/seed_42/checkpoint-50/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-50/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-50/tokenizer.json b/spanish_L20_a100_per_step/seed_42/checkpoint-50/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-50/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-50/tokenizer_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-50/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-50/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-50/trainer_state.json b/spanish_L20_a100_per_step/seed_42/checkpoint-50/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ed19f247c1f81866eee8299945912f928245ae30
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-50/trainer_state.json
@@ -0,0 +1,84 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.016840687100033683,
+  "eval_steps": 500,
+  "global_step": 50,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.313063097000122,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 1.5844918489456177,
+      "learning_rate": 1.8e-05,
+      "loss": 0.38868374824523927,
+      "mean_token_accuracy": 0.868084442615509,
+      "num_tokens": 45305.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.318549597263336,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 2.159148693084717,
+      "learning_rate": 3.8e-05,
+      "loss": 0.3493065118789673,
+      "mean_token_accuracy": 0.8752043426036835,
+      "num_tokens": 90561.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3196980714797975,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 1.711671233177185,
+      "learning_rate": 5.8e-05,
+      "loss": 0.2876554250717163,
+      "mean_token_accuracy": 0.8943579792976379,
+      "num_tokens": 135417.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2853124618530274,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 1.7681864500045776,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.25888752937316895,
+      "mean_token_accuracy": 0.9017764270305634,
+      "num_tokens": 180584.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.287485384941101,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 1.7435377836227417,
+      "learning_rate": 9.8e-05,
+      "loss": 0.25025362968444825,
+      "mean_token_accuracy": 0.9065495729446411,
+      "num_tokens": 225725.0,
+      "step": 50
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0534439092224e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-50/training_args.bin b/spanish_L20_a100_per_step/seed_42/checkpoint-50/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d0d50a8b0749201bf4efc93fa11a6d31dba22054
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-50/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5288a9d0fb7993d4f412addf995829036df3ca57cff640f306d8c75de45223e6
+size 5841
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-500/README.md b/spanish_L20_a100_per_step/seed_42/checkpoint-500/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dd49221a54a197d9f0cab86c7982e2b2fe0776b3
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-500/README.md
@@ -0,0 +1,209 @@
+---
+base_model: Qwen/Qwen2.5-7B-Instruct
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:Qwen/Qwen2.5-7B-Instruct
+- lora
+- sft
+- transformers
+- trl
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-500/adapter_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-500/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..dc2e80e033396f88c19dfa2320b4e02b31d61217
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-500/adapter_config.json
@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen2.5-7B-Instruct",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "down_proj",
+    "k_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-500/adapter_model.safetensors b/spanish_L20_a100_per_step/seed_42/checkpoint-500/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..29491796bbf462e9e89dc33969d200a3746ea8ed
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-500/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2758511c711582779bacba720e462f1828cdea994b689f8767378b854838e48
+size 80792096
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-500/chat_template.jinja b/spanish_L20_a100_per_step/seed_42/checkpoint-500/chat_template.jinja
new file mode 100644
index 0000000000000000000000000000000000000000..bdf7919a96cfe43d50914a007b9c0877bd0ec27e
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-500/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-500/tokenizer.json b/spanish_L20_a100_per_step/seed_42/checkpoint-500/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-500/tokenizer_config.json b/spanish_L20_a100_per_step/seed_42/checkpoint-500/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-500/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-500/trainer_state.json b/spanish_L20_a100_per_step/seed_42/checkpoint-500/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..613a3423139f97f13e210a494a4813100624a5ae
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-500/trainer_state.json
@@ -0,0 +1,534 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.16840687100033683,
+  "eval_steps": 500,
+  "global_step": 500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.313063097000122,
+      "epoch": 0.003368137420006736,
+      "grad_norm": 1.5844918489456177,
+      "learning_rate": 1.8e-05,
+      "loss": 0.38868374824523927,
+      "mean_token_accuracy": 0.868084442615509,
+      "num_tokens": 45305.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.318549597263336,
+      "epoch": 0.006736274840013472,
+      "grad_norm": 2.159148693084717,
+      "learning_rate": 3.8e-05,
+      "loss": 0.3493065118789673,
+      "mean_token_accuracy": 0.8752043426036835,
+      "num_tokens": 90561.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.3196980714797975,
+      "epoch": 0.010104412260020209,
+      "grad_norm": 1.711671233177185,
+      "learning_rate": 5.8e-05,
+      "loss": 0.2876554250717163,
+      "mean_token_accuracy": 0.8943579792976379,
+      "num_tokens": 135417.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.2853124618530274,
+      "epoch": 0.013472549680026945,
+      "grad_norm": 1.7681864500045776,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 0.25888752937316895,
+      "mean_token_accuracy": 0.9017764270305634,
+      "num_tokens": 180584.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.287485384941101,
+      "epoch": 0.016840687100033683,
+      "grad_norm": 1.7435377836227417,
+      "learning_rate": 9.8e-05,
+      "loss": 0.25025362968444825,
+      "mean_token_accuracy": 0.9065495729446411,
+      "num_tokens": 225725.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.2856799840927124,
+      "epoch": 0.020208824520040417,
+      "grad_norm": 1.5421638488769531,
+      "learning_rate": 9.997785653888835e-05,
+      "loss": 0.23334593772888185,
+      "mean_token_accuracy": 0.9110442101955414,
+      "num_tokens": 270982.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.2669721722602845,
+      "epoch": 0.023576961940047155,
+      "grad_norm": 1.5533316135406494,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 0.21683793067932128,
+      "mean_token_accuracy": 0.9178069531917572,
+      "num_tokens": 316699.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.2573367595672607,
+      "epoch": 0.02694509936005389,
+      "grad_norm": 1.6266299486160278,
+      "learning_rate": 9.977024992520602e-05,
+      "loss": 0.22236616611480714,
+      "mean_token_accuracy": 0.9152943313121795,
+      "num_tokens": 361697.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.2460310101509093,
+      "epoch": 0.030313236780060628,
+      "grad_norm": 1.6497271060943604,
+      "learning_rate": 9.95847403914247e-05,
+      "loss": 0.22359950542449952,
+      "mean_token_accuracy": 0.9153114914894104,
+      "num_tokens": 407048.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.2386581540107726,
+      "epoch": 0.033681374200067365,
+      "grad_norm": 1.9670138359069824,
+      "learning_rate": 9.934501067202117e-05,
+      "loss": 0.2138829231262207,
+      "mean_token_accuracy": 0.9157615780830384,
+      "num_tokens": 452454.0,
+      "step": 100
+    },
+    {
+      "entropy": 1.244022536277771,
+      "epoch": 0.037049511620074096,
+      "grad_norm": 1.4362447261810303,
+      "learning_rate": 9.905132290792394e-05,
+      "loss": 0.21276416778564453,
+      "mean_token_accuracy": 0.9172943770885468,
+      "num_tokens": 497996.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.2349260568618774,
+      "epoch": 0.040417649040080834,
+      "grad_norm": 1.9473755359649658,
+      "learning_rate": 9.870399824239117e-05,
+      "loss": 0.21619105339050293,
+      "mean_token_accuracy": 0.9155952095985412,
+      "num_tokens": 543160.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.2081148743629455,
+      "epoch": 0.04378578646008757,
+      "grad_norm": 1.3701565265655518,
+      "learning_rate": 9.830341646984521e-05,
+      "loss": 0.19676826000213624,
+      "mean_token_accuracy": 0.9251840233802795,
+      "num_tokens": 588895.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.2187958717346192,
+      "epoch": 0.04715392388009431,
+      "grad_norm": 1.8012784719467163,
+      "learning_rate": 9.785001562057309e-05,
+      "loss": 0.1980499267578125,
+      "mean_token_accuracy": 0.9234402239322662,
+      "num_tokens": 634346.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.222521722316742,
+      "epoch": 0.05052206130010104,
+      "grad_norm": 1.9744209051132202,
+      "learning_rate": 9.734429148174675e-05,
+      "loss": 0.2016369342803955,
+      "mean_token_accuracy": 0.9234835088253022,
+      "num_tokens": 679863.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.2050167083740235,
+      "epoch": 0.05389019872010778,
+      "grad_norm": 1.6054558753967285,
+      "learning_rate": 9.6786797055287e-05,
+      "loss": 0.19362304210662842,
+      "mean_token_accuracy": 0.9238951921463012,
+      "num_tokens": 725185.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.2060458779335022,
+      "epoch": 0.05725833614011452,
+      "grad_norm": 1.5881940126419067,
+      "learning_rate": 9.617814195316411e-05,
+      "loss": 0.1756075143814087,
+      "mean_token_accuracy": 0.9318983733654023,
+      "num_tokens": 770252.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.1969725489616394,
+      "epoch": 0.060626473560121255,
+      "grad_norm": 1.8048702478408813,
+      "learning_rate": 9.551899173079607e-05,
+      "loss": 0.18871016502380372,
+      "mean_token_accuracy": 0.9270873069763184,
+      "num_tokens": 815816.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.2078228354454041,
+      "epoch": 0.06399461098012799,
+      "grad_norm": 1.6712164878845215,
+      "learning_rate": 9.481006715927351e-05,
+      "loss": 0.1943951964378357,
+      "mean_token_accuracy": 0.9272128999233246,
+      "num_tokens": 861179.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.1862091064453124,
+      "epoch": 0.06736274840013473,
+      "grad_norm": 1.8262038230895996,
+      "learning_rate": 9.405214343720707e-05,
+      "loss": 0.17828747034072875,
+      "mean_token_accuracy": 0.932964438199997,
+      "num_tokens": 906523.0,
+      "step": 200
+    },
+    {
+      "entropy": 1.1960723161697389,
+      "epoch": 0.07073088582014146,
+      "grad_norm": 1.835386872291565,
+      "learning_rate": 9.32460493430591e-05,
+      "loss": 0.18402756452560426,
+      "mean_token_accuracy": 0.927827262878418,
+      "num_tokens": 951764.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.192801821231842,
+      "epoch": 0.07409902324014819,
+      "grad_norm": 1.8243962526321411,
+      "learning_rate": 9.239266632888659e-05,
+      "loss": 0.19473347663879395,
+      "mean_token_accuracy": 0.9250764012336731,
+      "num_tokens": 997050.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.1876365065574646,
+      "epoch": 0.07746716066015494,
+      "grad_norm": 2.0332682132720947,
+      "learning_rate": 9.14929275564863e-05,
+      "loss": 0.19224083423614502,
+      "mean_token_accuracy": 0.9212576508522033,
+      "num_tokens": 1042384.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.179949200153351,
+      "epoch": 0.08083529808016167,
+      "grad_norm": 1.7561111450195312,
+      "learning_rate": 9.0547816876996e-05,
+      "loss": 0.18619580268859864,
+      "mean_token_accuracy": 0.9293428659439087,
+      "num_tokens": 1087805.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.186054491996765,
+      "epoch": 0.08420343550016841,
+      "grad_norm": 1.873403787612915,
+      "learning_rate": 8.955836775506776e-05,
+      "loss": 0.17262675762176513,
+      "mean_token_accuracy": 0.9335724115371704,
+      "num_tokens": 1133158.0,
+      "step": 250
+    },
+    {
+      "entropy": 1.1623068928718567,
+      "epoch": 0.08757157292017514,
+      "grad_norm": 1.4724552631378174,
+      "learning_rate": 8.852566213878947e-05,
+      "loss": 0.1591184377670288,
+      "mean_token_accuracy": 0.9379242241382599,
+      "num_tokens": 1178397.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.1551544189453125,
+      "epoch": 0.09093971034018188,
+      "grad_norm": 1.423025131225586,
+      "learning_rate": 8.745082927659047e-05,
+      "loss": 0.1875994920730591,
+      "mean_token_accuracy": 0.926096248626709,
+      "num_tokens": 1223833.0,
+      "step": 270
+    },
+    {
+      "entropy": 1.1650016546249389,
+      "epoch": 0.09430784776018862,
+      "grad_norm": 1.4754282236099243,
+      "learning_rate": 8.633504448242505e-05,
+      "loss": 0.18312913179397583,
+      "mean_token_accuracy": 0.9313101947307587,
+      "num_tokens": 1269198.0,
+      "step": 280
+    },
+    {
+      "entropy": 1.1269522786140442,
+      "epoch": 0.09767598518019535,
+      "grad_norm": 1.9705147743225098,
+      "learning_rate": 8.517952785058385e-05,
+      "loss": 0.16336050033569335,
+      "mean_token_accuracy": 0.9348717212677002,
+      "num_tokens": 1314619.0,
+      "step": 290
+    },
+    {
+      "entropy": 1.140217089653015,
+      "epoch": 0.10104412260020208,
+      "grad_norm": 1.7201600074768066,
+      "learning_rate": 8.398554292153866e-05,
+      "loss": 0.16836259365081788,
+      "mean_token_accuracy": 0.9319346904754638,
+      "num_tokens": 1360103.0,
+      "step": 300
+    },
+    {
+      "entropy": 1.1287695646286011,
+      "epoch": 0.10441226002020883,
+      "grad_norm": 1.8850951194763184,
+      "learning_rate": 8.275439530027948e-05,
+      "loss": 0.16753654479980468,
+      "mean_token_accuracy": 0.933317893743515,
+      "num_tokens": 1405764.0,
+      "step": 310
+    },
+    {
+      "entropy": 1.1241347312927246,
+      "epoch": 0.10778039744021556,
+      "grad_norm": 1.598687767982483,
+      "learning_rate": 8.148743122865463e-05,
+      "loss": 0.1620866298675537,
+      "mean_token_accuracy": 0.9349688112735748,
+      "num_tokens": 1450767.0,
+      "step": 320
+    },
+    {
+      "entropy": 1.1217318654060364,
+      "epoch": 0.1111485348602223,
+      "grad_norm": 1.4878541231155396,
+      "learning_rate": 8.018603611327504e-05,
+      "loss": 0.17620257139205933,
+      "mean_token_accuracy": 0.9310369610786438,
+      "num_tokens": 1495774.0,
+      "step": 330
+    },
+    {
+      "entropy": 1.1106749534606934,
+      "epoch": 0.11451667228022903,
+      "grad_norm": 2.0954132080078125,
+      "learning_rate": 7.88516330105925e-05,
+      "loss": 0.16934839487075806,
+      "mean_token_accuracy": 0.931692111492157,
+      "num_tokens": 1541675.0,
+      "step": 340
+    },
+    {
+      "entropy": 1.1218057990074157,
+      "epoch": 0.11788480970023577,
+      "grad_norm": 1.4658523797988892,
+      "learning_rate": 7.748568107080832e-05,
+      "loss": 0.16255338191986085,
+      "mean_token_accuracy": 0.9345518827438355,
+      "num_tokens": 1586946.0,
+      "step": 350
+    },
+    {
+      "entropy": 1.127109909057617,
+      "epoch": 0.12125294712024251,
+      "grad_norm": 2.0114996433258057,
+      "learning_rate": 7.608967394231387e-05,
+      "loss": 0.16453249454498292,
+      "mean_token_accuracy": 0.9349606096744537,
+      "num_tokens": 1632132.0,
+      "step": 360
+    },
+    {
+      "entropy": 1.1345461845397948,
+      "epoch": 0.12462108454024924,
+      "grad_norm": 1.7129638195037842,
+      "learning_rate": 7.466513813840825e-05,
+      "loss": 0.17419962882995604,
+      "mean_token_accuracy": 0.9302750170230866,
+      "num_tokens": 1676914.0,
+      "step": 370
+    },
+    {
+      "entropy": 1.129710566997528,
+      "epoch": 0.12798922196025597,
+      "grad_norm": 1.6471718549728394,
+      "learning_rate": 7.32136313680782e-05,
+      "loss": 0.16340304613113404,
+      "mean_token_accuracy": 0.9387327194213867,
+      "num_tokens": 1722384.0,
+      "step": 380
+    },
+    {
+      "entropy": 1.1263115048408507,
+      "epoch": 0.13135735938026272,
+      "grad_norm": 1.7370667457580566,
+      "learning_rate": 7.173674083266624e-05,
+      "loss": 0.16197096109390258,
+      "mean_token_accuracy": 0.9354376375675202,
+      "num_tokens": 1767365.0,
+      "step": 390
+    },
+    {
+      "entropy": 1.1101006507873534,
+      "epoch": 0.13472549680026946,
+      "grad_norm": 1.8639729022979736,
+      "learning_rate": 7.023608149028937e-05,
+      "loss": 0.1628415107727051,
+      "mean_token_accuracy": 0.936980938911438,
+      "num_tokens": 1812719.0,
+      "step": 400
+    },
+    {
+      "entropy": 1.1179866552352906,
+      "epoch": 0.13809363422027618,
+      "grad_norm": 1.7608468532562256,
+      "learning_rate": 6.871329428990602e-05,
+      "loss": 0.16702044010162354,
+      "mean_token_accuracy": 0.9350353658199311,
+      "num_tokens": 1858198.0,
+      "step": 410
+    },
+    {
+      "entropy": 1.1132529377937317,
+      "epoch": 0.14146177164028292,
+      "grad_norm": 1.9437587261199951,
+      "learning_rate": 6.71700443769625e-05,
+      "loss": 0.15147662162780762,
+      "mean_token_accuracy": 0.9384173631668091,
+      "num_tokens": 1903609.0,
+      "step": 420
+    },
+    {
+      "entropy": 1.113513207435608,
+      "epoch": 0.14482990906028967,
+      "grad_norm": 1.3603644371032715,
+      "learning_rate": 6.56080192725808e-05,
+      "loss": 0.15535373687744142,
+      "mean_token_accuracy": 0.9393018841743469,
+      "num_tokens": 1949169.0,
+      "step": 430
+    },
+    {
+      "entropy": 1.1100459098815918,
+      "epoch": 0.14819804648029639,
+      "grad_norm": 1.9161057472229004,
+      "learning_rate": 6.402892702827916e-05,
+      "loss": 0.15063717365264892,
+      "mean_token_accuracy": 0.9416032433509827,
+      "num_tokens": 1994458.0,
+      "step": 440
+    },
+    {
+      "entropy": 1.1109524250030518,
+      "epoch": 0.15156618390030313,
+      "grad_norm": 1.683729887008667,
+      "learning_rate": 6.243449435824276e-05,
+      "loss": 0.16656777858734131,
+      "mean_token_accuracy": 0.9327812254428863,
+      "num_tokens": 2039887.0,
+      "step": 450
+    },
+    {
+      "entropy": 1.1199557185173035,
+      "epoch": 0.15493432132030988,
+      "grad_norm": 1.7669590711593628,
+      "learning_rate": 6.0826464751186994e-05,
+      "loss": 0.16300296783447266,
+      "mean_token_accuracy": 0.936695808172226,
+      "num_tokens": 2085081.0,
+      "step": 460
+    },
+    {
+      "entropy": 1.1009384870529175,
+      "epoch": 0.1583024587403166,
+      "grad_norm": 1.7644976377487183,
+      "learning_rate": 5.9206596563878357e-05,
+      "loss": 0.15217889547348024,
+      "mean_token_accuracy": 0.9400066375732422,
+      "num_tokens": 2130516.0,
+      "step": 470
+    },
+    {
+      "entropy": 1.1204341650009155,
+      "epoch": 0.16167059616032334,
+      "grad_norm": 1.6975176334381104,
+      "learning_rate": 5.757666109839702e-05,
+      "loss": 0.16737287044525145,
+      "mean_token_accuracy": 0.9332719027996064,
+      "num_tokens": 2175775.0,
+      "step": 480
+    },
+    {
+      "entropy": 1.1159302949905396,
+      "epoch": 0.16503873358033008,
+      "grad_norm": 2.088719367980957,
+      "learning_rate": 5.5938440665244006e-05,
+      "loss": 0.15946402549743652,
+      "mean_token_accuracy": 0.9387571811676025,
+      "num_tokens": 2221285.0,
+      "step": 490
+    },
+    {
+      "entropy": 1.107802414894104,
+      "epoch": 0.16840687100033683,
+      "grad_norm": 1.443552017211914,
+      "learning_rate": 5.4293726634410855e-05,
+      "loss": 0.1544242024421692,
+      "mean_token_accuracy": 0.9387835025787353,
+      "num_tokens": 2266369.0,
+      "step": 500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0562212397580288e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/checkpoint-500/training_args.bin b/spanish_L20_a100_per_step/seed_42/checkpoint-500/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d0d50a8b0749201bf4efc93fa11a6d31dba22054
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/checkpoint-500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5288a9d0fb7993d4f412addf995829036df3ca57cff640f306d8c75de45223e6
+size 5841
diff --git a/spanish_L20_a100_per_step/seed_42/tokenizer.json b/spanish_L20_a100_per_step/seed_42/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..34510ff0037cd50428af467a17ead5a96140a32c
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd169731d2cbde95e10bf356d66d5997fd885dd8dbb6fb4684da3f23b2585d8
+size 11421892
diff --git a/spanish_L20_a100_per_step/seed_42/tokenizer_config.json b/spanish_L20_a100_per_step/seed_42/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..770e41d6c92519d525eede4cbcf3ba27f6425311
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/tokenizer_config.json
@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
diff --git a/spanish_L20_a100_per_step/seed_42/training_args.bin b/spanish_L20_a100_per_step/seed_42/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d0d50a8b0749201bf4efc93fa11a6d31dba22054
--- /dev/null
+++ b/spanish_L20_a100_per_step/seed_42/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5288a9d0fb7993d4f412addf995829036df3ca57cff640f306d8c75de45223e6
+size 5841