Model save
Browse files- README.md +67 -0
- all_results.json +8 -0
- generation_config.json +14 -0
- train_results.json +8 -0
- trainer_state.json +1809 -0
README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
model_name: Qwen2.5-1.5B-Open-R1-GRPO
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
- trl
|
| 7 |
+
- grpo
|
| 8 |
+
licence: license
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Model Card for Qwen2.5-1.5B-Open-R1-GRPO
|
| 12 |
+
|
| 13 |
+
This model is a fine-tuned version of [None](https://huggingface.co/None).
|
| 14 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 15 |
+
|
| 16 |
+
## Quick start
|
| 17 |
+
|
| 18 |
+
```python
|
| 19 |
+
from transformers import pipeline
|
| 20 |
+
|
| 21 |
+
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?"
|
| 22 |
+
generator = pipeline("text-generation", model="yolay/Qwen2.5-1.5B-Open-R1-GRPO", device="cuda")
|
| 23 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 24 |
+
print(output["generated_text"])
|
| 25 |
+
```
|
| 26 |
+
|
| 27 |
+
## Training procedure
|
| 28 |
+
|
| 29 |
+
[<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/yuleiqin-tencent/huggingface/runs/5ngah7mu)
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
|
| 33 |
+
|
| 34 |
+
### Framework versions
|
| 35 |
+
|
| 36 |
+
- TRL: 0.15.0.dev0
|
| 37 |
+
- Transformers: 4.49.0.dev0
|
| 38 |
+
- Pytorch: 2.5.1
|
| 39 |
+
- Datasets: 3.2.0
|
| 40 |
+
- Tokenizers: 0.21.0
|
| 41 |
+
|
| 42 |
+
## Citations
|
| 43 |
+
|
| 44 |
+
Cite GRPO as:
|
| 45 |
+
|
| 46 |
+
```bibtex
|
| 47 |
+
@article{zhihong2024deepseekmath,
|
| 48 |
+
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
|
| 49 |
+
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
|
| 50 |
+
year = 2024,
|
| 51 |
+
eprint = {arXiv:2402.03300},
|
| 52 |
+
}
|
| 53 |
+
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
Cite TRL as:
|
| 57 |
+
|
| 58 |
+
```bibtex
|
| 59 |
+
@misc{vonwerra2022trl,
|
| 60 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 61 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
|
| 62 |
+
year = 2020,
|
| 63 |
+
journal = {GitHub repository},
|
| 64 |
+
publisher = {GitHub},
|
| 65 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 66 |
+
}
|
| 67 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.13680233840041295,
|
| 4 |
+
"train_runtime": 51841.8905,
|
| 5 |
+
"train_samples": 72441,
|
| 6 |
+
"train_samples_per_second": 1.397,
|
| 7 |
+
"train_steps_per_second": 0.012
|
| 8 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
151645,
|
| 6 |
+
151643
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 151643,
|
| 9 |
+
"repetition_penalty": 1.1,
|
| 10 |
+
"temperature": 0.7,
|
| 11 |
+
"top_k": 20,
|
| 12 |
+
"top_p": 0.8,
|
| 13 |
+
"transformers_version": "4.49.0.dev0"
|
| 14 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.13680233840041295,
|
| 4 |
+
"train_runtime": 51841.8905,
|
| 5 |
+
"train_samples": 72441,
|
| 6 |
+
"train_samples_per_second": 1.397,
|
| 7 |
+
"train_steps_per_second": 0.012
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1809 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9987438399845395,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 646,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"completion_length": 398.2607320785522,
|
| 13 |
+
"epoch": 0.007730215479756498,
|
| 14 |
+
"grad_norm": 0.6042563694653488,
|
| 15 |
+
"kl": 0.00012127757072448731,
|
| 16 |
+
"learning_rate": 1.5384615384615387e-06,
|
| 17 |
+
"loss": 0.0,
|
| 18 |
+
"reward": 0.6366071738302708,
|
| 19 |
+
"reward_std": 0.32451150137931106,
|
| 20 |
+
"rewards/accuracy_reward": 0.1723214378580451,
|
| 21 |
+
"rewards/format_reward": 0.46428573289886116,
|
| 22 |
+
"step": 5
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"completion_length": 357.93305110931396,
|
| 26 |
+
"epoch": 0.015460430959512996,
|
| 27 |
+
"grad_norm": 0.7802913073139134,
|
| 28 |
+
"kl": 0.007133913040161133,
|
| 29 |
+
"learning_rate": 3.0769230769230774e-06,
|
| 30 |
+
"loss": 0.0003,
|
| 31 |
+
"reward": 0.7535714633762837,
|
| 32 |
+
"reward_std": 0.26516504064202306,
|
| 33 |
+
"rewards/accuracy_reward": 0.11964286342263222,
|
| 34 |
+
"rewards/format_reward": 0.6339285995811224,
|
| 35 |
+
"step": 10
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"completion_length": 291.9857277870178,
|
| 39 |
+
"epoch": 0.023190646439269495,
|
| 40 |
+
"grad_norm": 0.4454731956977912,
|
| 41 |
+
"kl": 0.0273590087890625,
|
| 42 |
+
"learning_rate": 4.615384615384616e-06,
|
| 43 |
+
"loss": 0.0011,
|
| 44 |
+
"reward": 0.8758928969502449,
|
| 45 |
+
"reward_std": 0.22602162957191468,
|
| 46 |
+
"rewards/accuracy_reward": 0.08214286118745803,
|
| 47 |
+
"rewards/format_reward": 0.7937500394880772,
|
| 48 |
+
"step": 15
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"completion_length": 291.47322788238523,
|
| 52 |
+
"epoch": 0.03092086191902599,
|
| 53 |
+
"grad_norm": 0.4324622452746649,
|
| 54 |
+
"kl": 0.01610870361328125,
|
| 55 |
+
"learning_rate": 6.153846153846155e-06,
|
| 56 |
+
"loss": 0.0006,
|
| 57 |
+
"reward": 0.8642857551574707,
|
| 58 |
+
"reward_std": 0.25253813322633506,
|
| 59 |
+
"rewards/accuracy_reward": 0.09285714775323868,
|
| 60 |
+
"rewards/format_reward": 0.7714286111295223,
|
| 61 |
+
"step": 20
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"completion_length": 226.4142951965332,
|
| 65 |
+
"epoch": 0.03865107739878249,
|
| 66 |
+
"grad_norm": 0.3675938596678119,
|
| 67 |
+
"kl": 0.026679229736328126,
|
| 68 |
+
"learning_rate": 7.692307692307694e-06,
|
| 69 |
+
"loss": 0.0011,
|
| 70 |
+
"reward": 0.9687500387430191,
|
| 71 |
+
"reward_std": 0.21592010390013455,
|
| 72 |
+
"rewards/accuracy_reward": 0.10178571976721287,
|
| 73 |
+
"rewards/format_reward": 0.8669643253087997,
|
| 74 |
+
"step": 25
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"completion_length": 197.68036499023438,
|
| 78 |
+
"epoch": 0.04638129287853899,
|
| 79 |
+
"grad_norm": 0.3061792577373215,
|
| 80 |
+
"kl": 0.033779144287109375,
|
| 81 |
+
"learning_rate": 9.230769230769232e-06,
|
| 82 |
+
"loss": 0.0014,
|
| 83 |
+
"reward": 1.04107146859169,
|
| 84 |
+
"reward_std": 0.16920054908841847,
|
| 85 |
+
"rewards/accuracy_reward": 0.10803572004660963,
|
| 86 |
+
"rewards/format_reward": 0.9330357395112514,
|
| 87 |
+
"step": 30
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"completion_length": 196.36340112686156,
|
| 91 |
+
"epoch": 0.054111508358295486,
|
| 92 |
+
"grad_norm": 0.2761290868371635,
|
| 93 |
+
"kl": 0.0447174072265625,
|
| 94 |
+
"learning_rate": 1.076923076923077e-05,
|
| 95 |
+
"loss": 0.0018,
|
| 96 |
+
"reward": 1.1089286178350448,
|
| 97 |
+
"reward_std": 0.14647211749106645,
|
| 98 |
+
"rewards/accuracy_reward": 0.13750000707805157,
|
| 99 |
+
"rewards/format_reward": 0.9714285843074322,
|
| 100 |
+
"step": 35
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"completion_length": 233.76518907546998,
|
| 104 |
+
"epoch": 0.06184172383805198,
|
| 105 |
+
"grad_norm": 0.6542348704176013,
|
| 106 |
+
"kl": 0.06229248046875,
|
| 107 |
+
"learning_rate": 1.230769230769231e-05,
|
| 108 |
+
"loss": 0.0025,
|
| 109 |
+
"reward": 1.1125000432133674,
|
| 110 |
+
"reward_std": 0.1313198298215866,
|
| 111 |
+
"rewards/accuracy_reward": 0.1401785789988935,
|
| 112 |
+
"rewards/format_reward": 0.9723214417695999,
|
| 113 |
+
"step": 40
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"completion_length": 189.32232856750488,
|
| 117 |
+
"epoch": 0.06957193931780849,
|
| 118 |
+
"grad_norm": 0.3188732151645114,
|
| 119 |
+
"kl": 0.0878082275390625,
|
| 120 |
+
"learning_rate": 1.3846153846153847e-05,
|
| 121 |
+
"loss": 0.0035,
|
| 122 |
+
"reward": 1.0857143267989158,
|
| 123 |
+
"reward_std": 0.14142135493457317,
|
| 124 |
+
"rewards/accuracy_reward": 0.11339286295697093,
|
| 125 |
+
"rewards/format_reward": 0.9723214417695999,
|
| 126 |
+
"step": 45
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"completion_length": 300.7259063720703,
|
| 130 |
+
"epoch": 0.07730215479756498,
|
| 131 |
+
"grad_norm": 0.31691534877890665,
|
| 132 |
+
"kl": 0.0990020751953125,
|
| 133 |
+
"learning_rate": 1.5384615384615387e-05,
|
| 134 |
+
"loss": 0.004,
|
| 135 |
+
"reward": 1.05267860814929,
|
| 136 |
+
"reward_std": 0.2133947243914008,
|
| 137 |
+
"rewards/accuracy_reward": 0.15000000735744834,
|
| 138 |
+
"rewards/format_reward": 0.9026786051690578,
|
| 139 |
+
"step": 50
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"completion_length": 277.56429901123045,
|
| 143 |
+
"epoch": 0.08503237027732148,
|
| 144 |
+
"grad_norm": 0.3063068659110802,
|
| 145 |
+
"kl": 0.06175537109375,
|
| 146 |
+
"learning_rate": 1.6923076923076924e-05,
|
| 147 |
+
"loss": 0.0025,
|
| 148 |
+
"reward": 1.0723214723169803,
|
| 149 |
+
"reward_std": 0.21844548601657152,
|
| 150 |
+
"rewards/accuracy_reward": 0.15982143683359026,
|
| 151 |
+
"rewards/format_reward": 0.9125000342726708,
|
| 152 |
+
"step": 55
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"completion_length": 322.78662223815917,
|
| 156 |
+
"epoch": 0.09276258575707798,
|
| 157 |
+
"grad_norm": 0.2797133381390773,
|
| 158 |
+
"kl": 0.0583984375,
|
| 159 |
+
"learning_rate": 1.8461538461538465e-05,
|
| 160 |
+
"loss": 0.0023,
|
| 161 |
+
"reward": 1.1383929125964642,
|
| 162 |
+
"reward_std": 0.15783633291721344,
|
| 163 |
+
"rewards/accuracy_reward": 0.1839285811409354,
|
| 164 |
+
"rewards/format_reward": 0.9544643051922321,
|
| 165 |
+
"step": 60
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"completion_length": 290.8223346710205,
|
| 169 |
+
"epoch": 0.10049280123683448,
|
| 170 |
+
"grad_norm": 0.22020075522546995,
|
| 171 |
+
"kl": 0.072247314453125,
|
| 172 |
+
"learning_rate": 2e-05,
|
| 173 |
+
"loss": 0.0029,
|
| 174 |
+
"reward": 1.1482143357396126,
|
| 175 |
+
"reward_std": 0.133845211006701,
|
| 176 |
+
"rewards/accuracy_reward": 0.16071429401636123,
|
| 177 |
+
"rewards/format_reward": 0.9875000059604645,
|
| 178 |
+
"step": 65
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"completion_length": 296.99554920196533,
|
| 182 |
+
"epoch": 0.10822301671659097,
|
| 183 |
+
"grad_norm": 0.25171935257044675,
|
| 184 |
+
"kl": 0.077728271484375,
|
| 185 |
+
"learning_rate": 1.999634547413886e-05,
|
| 186 |
+
"loss": 0.0031,
|
| 187 |
+
"reward": 1.157142909616232,
|
| 188 |
+
"reward_std": 0.16414978671818972,
|
| 189 |
+
"rewards/accuracy_reward": 0.19285715268924833,
|
| 190 |
+
"rewards/format_reward": 0.9642857298254967,
|
| 191 |
+
"step": 70
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"completion_length": 321.1794790267944,
|
| 195 |
+
"epoch": 0.11595323219634747,
|
| 196 |
+
"grad_norm": 0.35630454010296175,
|
| 197 |
+
"kl": 0.083953857421875,
|
| 198 |
+
"learning_rate": 1.9985384567667278e-05,
|
| 199 |
+
"loss": 0.0034,
|
| 200 |
+
"reward": 1.0901786163449287,
|
| 201 |
+
"reward_std": 0.22602162901312112,
|
| 202 |
+
"rewards/accuracy_reward": 0.1794642936438322,
|
| 203 |
+
"rewards/format_reward": 0.9107143178582191,
|
| 204 |
+
"step": 75
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"completion_length": 253.01072540283204,
|
| 208 |
+
"epoch": 0.12368344767610397,
|
| 209 |
+
"grad_norm": 0.22028353611198634,
|
| 210 |
+
"kl": 0.090179443359375,
|
| 211 |
+
"learning_rate": 1.9967125291968495e-05,
|
| 212 |
+
"loss": 0.0036,
|
| 213 |
+
"reward": 1.0892857559025289,
|
| 214 |
+
"reward_std": 0.19950512517243624,
|
| 215 |
+
"rewards/accuracy_reward": 0.15892858076840638,
|
| 216 |
+
"rewards/format_reward": 0.9303571678698063,
|
| 217 |
+
"step": 80
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"completion_length": 221.2348321914673,
|
| 221 |
+
"epoch": 0.13141366315586048,
|
| 222 |
+
"grad_norm": 0.23846717001551665,
|
| 223 |
+
"kl": 0.103912353515625,
|
| 224 |
+
"learning_rate": 1.9941580992841562e-05,
|
| 225 |
+
"loss": 0.0042,
|
| 226 |
+
"reward": 1.1187500461935997,
|
| 227 |
+
"reward_std": 0.13258252013474703,
|
| 228 |
+
"rewards/accuracy_reward": 0.14464286463335158,
|
| 229 |
+
"rewards/format_reward": 0.9741071552038193,
|
| 230 |
+
"step": 85
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"completion_length": 238.93393840789795,
|
| 234 |
+
"epoch": 0.13914387863561697,
|
| 235 |
+
"grad_norm": 0.28253877862072746,
|
| 236 |
+
"kl": 0.10802001953125,
|
| 237 |
+
"learning_rate": 1.990877034074683e-05,
|
| 238 |
+
"loss": 0.0043,
|
| 239 |
+
"reward": 1.1500000476837158,
|
| 240 |
+
"reward_std": 0.15909902472048998,
|
| 241 |
+
"rewards/accuracy_reward": 0.16964286640286447,
|
| 242 |
+
"rewards/format_reward": 0.980357151478529,
|
| 243 |
+
"step": 90
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"completion_length": 384.65805397033694,
|
| 247 |
+
"epoch": 0.14687409411537347,
|
| 248 |
+
"grad_norm": 0.246515365890744,
|
| 249 |
+
"kl": 320.1103820800781,
|
| 250 |
+
"learning_rate": 1.9868717317159617e-05,
|
| 251 |
+
"loss": 12.8246,
|
| 252 |
+
"reward": 1.1482143349945546,
|
| 253 |
+
"reward_std": 0.2323350828140974,
|
| 254 |
+
"rewards/accuracy_reward": 0.2267857251688838,
|
| 255 |
+
"rewards/format_reward": 0.9214286006987095,
|
| 256 |
+
"step": 95
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"completion_length": 430.77234039306643,
|
| 260 |
+
"epoch": 0.15460430959512997,
|
| 261 |
+
"grad_norm": 0.21987794238704425,
|
| 262 |
+
"kl": 0.212005615234375,
|
| 263 |
+
"learning_rate": 1.9821451197042028e-05,
|
| 264 |
+
"loss": 0.0085,
|
| 265 |
+
"reward": 1.0955357640981673,
|
| 266 |
+
"reward_std": 0.28410540148615837,
|
| 267 |
+
"rewards/accuracy_reward": 0.21785715371370315,
|
| 268 |
+
"rewards/format_reward": 0.877678605914116,
|
| 269 |
+
"step": 100
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.15460430959512997,
|
| 273 |
+
"eval_completion_length": 306.96846771240234,
|
| 274 |
+
"eval_kl": 0.09576416015625,
|
| 275 |
+
"eval_loss": 0.00375110050663352,
|
| 276 |
+
"eval_reward": 1.1964286267757416,
|
| 277 |
+
"eval_reward_std": 0.21465741284191608,
|
| 278 |
+
"eval_rewards/accuracy_reward": 0.2366071566939354,
|
| 279 |
+
"eval_rewards/format_reward": 0.9598214477300644,
|
| 280 |
+
"eval_runtime": 39.1469,
|
| 281 |
+
"eval_samples_per_second": 2.529,
|
| 282 |
+
"eval_steps_per_second": 0.102,
|
| 283 |
+
"step": 100
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"completion_length": 296.79733486175536,
|
| 287 |
+
"epoch": 0.16233452507488647,
|
| 288 |
+
"grad_norm": 0.22613704416746724,
|
| 289 |
+
"kl": 0.101495361328125,
|
| 290 |
+
"learning_rate": 1.9767006527445728e-05,
|
| 291 |
+
"loss": 0.0041,
|
| 292 |
+
"reward": 1.1491071999073028,
|
| 293 |
+
"reward_std": 0.17551400382071733,
|
| 294 |
+
"rewards/accuracy_reward": 0.1830357233993709,
|
| 295 |
+
"rewards/format_reward": 0.9660714447498322,
|
| 296 |
+
"step": 105
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"completion_length": 231.32143831253052,
|
| 300 |
+
"epoch": 0.17006474055464296,
|
| 301 |
+
"grad_norm": 0.2736697715915322,
|
| 302 |
+
"kl": 0.115789794921875,
|
| 303 |
+
"learning_rate": 1.9705423102261324e-05,
|
| 304 |
+
"loss": 0.0046,
|
| 305 |
+
"reward": 1.128571480512619,
|
| 306 |
+
"reward_std": 0.1818274561315775,
|
| 307 |
+
"rewards/accuracy_reward": 0.15892858002334834,
|
| 308 |
+
"rewards/format_reward": 0.9696428716182709,
|
| 309 |
+
"step": 110
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"completion_length": 217.31161766052247,
|
| 313 |
+
"epoch": 0.17779495603439946,
|
| 314 |
+
"grad_norm": 0.2715660388125083,
|
| 315 |
+
"kl": 0.134747314453125,
|
| 316 |
+
"learning_rate": 1.9636745933132807e-05,
|
| 317 |
+
"loss": 0.0054,
|
| 318 |
+
"reward": 1.1080357626080513,
|
| 319 |
+
"reward_std": 0.16793785914778708,
|
| 320 |
+
"rewards/accuracy_reward": 0.1410714359022677,
|
| 321 |
+
"rewards/format_reward": 0.9669643007218838,
|
| 322 |
+
"step": 115
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"completion_length": 264.1339414596558,
|
| 326 |
+
"epoch": 0.18552517151415596,
|
| 327 |
+
"grad_norm": 0.28581146278588754,
|
| 328 |
+
"kl": 0.150201416015625,
|
| 329 |
+
"learning_rate": 1.956102521655831e-05,
|
| 330 |
+
"loss": 0.006,
|
| 331 |
+
"reward": 1.1258929021656514,
|
| 332 |
+
"reward_std": 0.21339472178369762,
|
| 333 |
+
"rewards/accuracy_reward": 0.1919642947614193,
|
| 334 |
+
"rewards/format_reward": 0.9339285977184772,
|
| 335 |
+
"step": 120
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"completion_length": 292.79108371734617,
|
| 339 |
+
"epoch": 0.19325538699391245,
|
| 340 |
+
"grad_norm": 0.7258310743765034,
|
| 341 |
+
"kl": 0.18944091796875,
|
| 342 |
+
"learning_rate": 1.9478316297201218e-05,
|
| 343 |
+
"loss": 0.0076,
|
| 344 |
+
"reward": 1.1303571976721287,
|
| 345 |
+
"reward_std": 0.2424366096034646,
|
| 346 |
+
"rewards/accuracy_reward": 0.2214285818859935,
|
| 347 |
+
"rewards/format_reward": 0.9089286021888257,
|
| 348 |
+
"step": 125
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"completion_length": 224.37768850326538,
|
| 352 |
+
"epoch": 0.20098560247366895,
|
| 353 |
+
"grad_norm": 0.5660752703191607,
|
| 354 |
+
"kl": 0.51148681640625,
|
| 355 |
+
"learning_rate": 1.9388679627438486e-05,
|
| 356 |
+
"loss": 0.0205,
|
| 357 |
+
"reward": 1.0053571909666061,
|
| 358 |
+
"reward_std": 0.33082495592534544,
|
| 359 |
+
"rewards/accuracy_reward": 0.20267858086153864,
|
| 360 |
+
"rewards/format_reward": 0.8026786111295223,
|
| 361 |
+
"step": 130
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"completion_length": 136.95804138183593,
|
| 365 |
+
"epoch": 0.20871581795342545,
|
| 366 |
+
"grad_norm": 1.7299812104574703,
|
| 367 |
+
"kl": 0.520361328125,
|
| 368 |
+
"learning_rate": 1.9292180723175656e-05,
|
| 369 |
+
"loss": 0.0208,
|
| 370 |
+
"reward": 0.9401786185801029,
|
| 371 |
+
"reward_std": 0.39269680101424453,
|
| 372 |
+
"rewards/accuracy_reward": 0.17589286603033544,
|
| 373 |
+
"rewards/format_reward": 0.7642857521772385,
|
| 374 |
+
"step": 135
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"completion_length": 292.2187627792358,
|
| 378 |
+
"epoch": 0.21644603343318194,
|
| 379 |
+
"grad_norm": 0.360874201843721,
|
| 380 |
+
"kl": 0.4266845703125,
|
| 381 |
+
"learning_rate": 1.9188890115960967e-05,
|
| 382 |
+
"loss": 0.0171,
|
| 383 |
+
"reward": 1.0937500461935996,
|
| 384 |
+
"reward_std": 0.2032931974157691,
|
| 385 |
+
"rewards/accuracy_reward": 0.1535714365541935,
|
| 386 |
+
"rewards/format_reward": 0.9401785939931869,
|
| 387 |
+
"step": 140
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"completion_length": 203.01786556243897,
|
| 391 |
+
"epoch": 0.22417624891293844,
|
| 392 |
+
"grad_norm": 0.3748600306818795,
|
| 393 |
+
"kl": 0.28094482421875,
|
| 394 |
+
"learning_rate": 1.9078883301433488e-05,
|
| 395 |
+
"loss": 0.0112,
|
| 396 |
+
"reward": 1.090178620815277,
|
| 397 |
+
"reward_std": 0.20581857804208994,
|
| 398 |
+
"rewards/accuracy_reward": 0.16607143692672252,
|
| 399 |
+
"rewards/format_reward": 0.9241071708500386,
|
| 400 |
+
"step": 145
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"completion_length": 325.5580488204956,
|
| 404 |
+
"epoch": 0.23190646439269494,
|
| 405 |
+
"grad_norm": 74.76840594405346,
|
| 406 |
+
"kl": 0.74801025390625,
|
| 407 |
+
"learning_rate": 1.8962240684142923e-05,
|
| 408 |
+
"loss": 0.0299,
|
| 409 |
+
"reward": 0.8758928962051868,
|
| 410 |
+
"reward_std": 0.3219861214980483,
|
| 411 |
+
"rewards/accuracy_reward": 0.12589286426082252,
|
| 412 |
+
"rewards/format_reward": 0.750000037997961,
|
| 413 |
+
"step": 150
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"completion_length": 371.35894470214845,
|
| 417 |
+
"epoch": 0.23963667987245144,
|
| 418 |
+
"grad_norm": 72.4191179163845,
|
| 419 |
+
"kl": 0.8891845703125,
|
| 420 |
+
"learning_rate": 1.883904751878156e-05,
|
| 421 |
+
"loss": 0.0356,
|
| 422 |
+
"reward": 0.8571428939700126,
|
| 423 |
+
"reward_std": 0.36618029810488223,
|
| 424 |
+
"rewards/accuracy_reward": 0.1589285798370838,
|
| 425 |
+
"rewards/format_reward": 0.6982143167406321,
|
| 426 |
+
"step": 155
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"completion_length": 315.3026921272278,
|
| 430 |
+
"epoch": 0.24736689535220793,
|
| 431 |
+
"grad_norm": 1.580738805644984,
|
| 432 |
+
"kl": 0.48006591796875,
|
| 433 |
+
"learning_rate": 1.8709393847871146e-05,
|
| 434 |
+
"loss": 0.0192,
|
| 435 |
+
"reward": 0.8991071823984385,
|
| 436 |
+
"reward_std": 0.32451150212436913,
|
| 437 |
+
"rewards/accuracy_reward": 0.16428572265431285,
|
| 438 |
+
"rewards/format_reward": 0.7348214674741029,
|
| 439 |
+
"step": 160
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"completion_length": 380.33751621246336,
|
| 443 |
+
"epoch": 0.25509711083196446,
|
| 444 |
+
"grad_norm": 155.90661885034666,
|
| 445 |
+
"kl": 2.71826171875,
|
| 446 |
+
"learning_rate": 1.857337443595034e-05,
|
| 447 |
+
"loss": 0.1089,
|
| 448 |
+
"reward": 0.9276786163449288,
|
| 449 |
+
"reward_std": 0.3750191332772374,
|
| 450 |
+
"rewards/accuracy_reward": 0.16696429420262576,
|
| 451 |
+
"rewards/format_reward": 0.7607143238186836,
|
| 452 |
+
"step": 165
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"completion_length": 391.8839458465576,
|
| 456 |
+
"epoch": 0.26282732631172095,
|
| 457 |
+
"grad_norm": 2.7868948262319475,
|
| 458 |
+
"kl": 0.80675048828125,
|
| 459 |
+
"learning_rate": 1.8431088700310846e-05,
|
| 460 |
+
"loss": 0.0323,
|
| 461 |
+
"reward": 1.1080357603728772,
|
| 462 |
+
"reward_std": 0.18814090844243764,
|
| 463 |
+
"rewards/accuracy_reward": 0.16428572246804835,
|
| 464 |
+
"rewards/format_reward": 0.9437500201165676,
|
| 465 |
+
"step": 170
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"completion_length": 242.89018936157225,
|
| 469 |
+
"epoch": 0.27055754179147745,
|
| 470 |
+
"grad_norm": 0.42925991415337916,
|
| 471 |
+
"kl": 0.2980712890625,
|
| 472 |
+
"learning_rate": 1.8282640638332773e-05,
|
| 473 |
+
"loss": 0.0119,
|
| 474 |
+
"reward": 0.9973214760422706,
|
| 475 |
+
"reward_std": 0.2790546391159296,
|
| 476 |
+
"rewards/accuracy_reward": 0.16696429438889027,
|
| 477 |
+
"rewards/format_reward": 0.8303571760654449,
|
| 478 |
+
"step": 175
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"completion_length": 420.8875186920166,
|
| 482 |
+
"epoch": 0.27828775727123395,
|
| 483 |
+
"grad_norm": 5.77238039316987,
|
| 484 |
+
"kl": 2.53583984375,
|
| 485 |
+
"learning_rate": 1.8128138751472432e-05,
|
| 486 |
+
"loss": 0.1013,
|
| 487 |
+
"reward": 0.6794643143191934,
|
| 488 |
+
"reward_std": 0.38764603715389967,
|
| 489 |
+
"rewards/accuracy_reward": 0.13125000707805157,
|
| 490 |
+
"rewards/format_reward": 0.5482143117114902,
|
| 491 |
+
"step": 180
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"completion_length": 377.3401969909668,
|
| 495 |
+
"epoch": 0.28601797275099045,
|
| 496 |
+
"grad_norm": 46.08902975257459,
|
| 497 |
+
"kl": 1.4901123046875,
|
| 498 |
+
"learning_rate": 1.7967695965958044e-05,
|
| 499 |
+
"loss": 0.0597,
|
| 500 |
+
"reward": 0.9267857566475868,
|
| 501 |
+
"reward_std": 0.351028005965054,
|
| 502 |
+
"rewards/accuracy_reward": 0.1696428656578064,
|
| 503 |
+
"rewards/format_reward": 0.7571428924798965,
|
| 504 |
+
"step": 185
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"completion_length": 269.395546913147,
|
| 508 |
+
"epoch": 0.29374818823074694,
|
| 509 |
+
"grad_norm": 0.3753752369724839,
|
| 510 |
+
"kl": 1.59666748046875,
|
| 511 |
+
"learning_rate": 1.780142955025139e-05,
|
| 512 |
+
"loss": 0.064,
|
| 513 |
+
"reward": 1.1169643260538578,
|
| 514 |
+
"reward_std": 0.22349624745547772,
|
| 515 |
+
"rewards/accuracy_reward": 0.2044642954133451,
|
| 516 |
+
"rewards/format_reward": 0.9125000283122062,
|
| 517 |
+
"step": 190
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"completion_length": 285.7642984390259,
|
| 521 |
+
"epoch": 0.30147840371050344,
|
| 522 |
+
"grad_norm": 1.6098957935540266,
|
| 523 |
+
"kl": 1.1361083984375,
|
| 524 |
+
"learning_rate": 1.7629461029335683e-05,
|
| 525 |
+
"loss": 0.0454,
|
| 526 |
+
"reward": 1.0705357670783997,
|
| 527 |
+
"reward_std": 0.3017830714583397,
|
| 528 |
+
"rewards/accuracy_reward": 0.21339286854490638,
|
| 529 |
+
"rewards/format_reward": 0.8571428984403611,
|
| 530 |
+
"step": 195
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"completion_length": 360.73573036193847,
|
| 534 |
+
"epoch": 0.30920861919025994,
|
| 535 |
+
"grad_norm": 24.77991428569718,
|
| 536 |
+
"kl": 1.9080322265625,
|
| 537 |
+
"learning_rate": 1.745191609589231e-05,
|
| 538 |
+
"loss": 0.0764,
|
| 539 |
+
"reward": 0.9089286141097546,
|
| 540 |
+
"reward_std": 0.3737564399838448,
|
| 541 |
+
"rewards/accuracy_reward": 0.15892858104780316,
|
| 542 |
+
"rewards/format_reward": 0.7500000361353159,
|
| 543 |
+
"step": 200
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.30920861919025994,
|
| 547 |
+
"eval_completion_length": 388.8497200012207,
|
| 548 |
+
"eval_kl": 4.3896484375,
|
| 549 |
+
"eval_loss": 0.16958686709403992,
|
| 550 |
+
"eval_reward": 1.0178571790456772,
|
| 551 |
+
"eval_reward_std": 0.404061034321785,
|
| 552 |
+
"eval_rewards/accuracy_reward": 0.2455357238650322,
|
| 553 |
+
"eval_rewards/format_reward": 0.7723214700818062,
|
| 554 |
+
"eval_runtime": 48.8596,
|
| 555 |
+
"eval_samples_per_second": 2.026,
|
| 556 |
+
"eval_steps_per_second": 0.082,
|
| 557 |
+
"step": 200
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"completion_length": 370.7384090423584,
|
| 561 |
+
"epoch": 0.31693883467001643,
|
| 562 |
+
"grad_norm": 9.980763942338113,
|
| 563 |
+
"kl": 2.82935791015625,
|
| 564 |
+
"learning_rate": 1.7268924518431437e-05,
|
| 565 |
+
"loss": 0.1132,
|
| 566 |
+
"reward": 0.9776786178350448,
|
| 567 |
+
"reward_std": 0.34218916948884726,
|
| 568 |
+
"rewards/accuracy_reward": 0.19910715324804187,
|
| 569 |
+
"rewards/format_reward": 0.7785714693367481,
|
| 570 |
+
"step": 205
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"completion_length": 290.9964431762695,
|
| 574 |
+
"epoch": 0.32466905014977293,
|
| 575 |
+
"grad_norm": 1.9760739124872957,
|
| 576 |
+
"kl": 1.00716552734375,
|
| 577 |
+
"learning_rate": 1.7080620046443503e-05,
|
| 578 |
+
"loss": 0.0403,
|
| 579 |
+
"reward": 1.025000049173832,
|
| 580 |
+
"reward_std": 0.2752665659412742,
|
| 581 |
+
"rewards/accuracy_reward": 0.1750000081025064,
|
| 582 |
+
"rewards/format_reward": 0.8500000417232514,
|
| 583 |
+
"step": 210
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"completion_length": 204.87858114242553,
|
| 587 |
+
"epoch": 0.33239926562952943,
|
| 588 |
+
"grad_norm": 0.6268244531901943,
|
| 589 |
+
"kl": 0.43160400390625,
|
| 590 |
+
"learning_rate": 1.6887140312641036e-05,
|
| 591 |
+
"loss": 0.0173,
|
| 592 |
+
"reward": 1.0633929058909417,
|
| 593 |
+
"reward_std": 0.25127544458955525,
|
| 594 |
+
"rewards/accuracy_reward": 0.16517857881262898,
|
| 595 |
+
"rewards/format_reward": 0.8982143223285675,
|
| 596 |
+
"step": 215
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"completion_length": 247.21429824829102,
|
| 600 |
+
"epoch": 0.3401294811092859,
|
| 601 |
+
"grad_norm": 0.5949423088731468,
|
| 602 |
+
"kl": 0.675537109375,
|
| 603 |
+
"learning_rate": 1.6688626732362192e-05,
|
| 604 |
+
"loss": 0.027,
|
| 605 |
+
"reward": 0.9500000357627869,
|
| 606 |
+
"reward_std": 0.3005203790962696,
|
| 607 |
+
"rewards/accuracy_reward": 0.12946429271250964,
|
| 608 |
+
"rewards/format_reward": 0.820535758137703,
|
| 609 |
+
"step": 220
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"completion_length": 190.4634015083313,
|
| 613 |
+
"epoch": 0.3478596965890424,
|
| 614 |
+
"grad_norm": 0.42764857782637755,
|
| 615 |
+
"kl": 0.33341064453125,
|
| 616 |
+
"learning_rate": 1.6485224400209557e-05,
|
| 617 |
+
"loss": 0.0133,
|
| 618 |
+
"reward": 1.0767857618629932,
|
| 619 |
+
"reward_std": 0.21213203240185977,
|
| 620 |
+
"rewards/accuracy_reward": 0.16250000838190318,
|
| 621 |
+
"rewards/format_reward": 0.9142857469618321,
|
| 622 |
+
"step": 225
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"completion_length": 210.5357223510742,
|
| 626 |
+
"epoch": 0.3555899120687989,
|
| 627 |
+
"grad_norm": 24.683355476887744,
|
| 628 |
+
"kl": 3.64019775390625,
|
| 629 |
+
"learning_rate": 1.6277081983999742e-05,
|
| 630 |
+
"loss": 0.1459,
|
| 631 |
+
"reward": 1.141071478277445,
|
| 632 |
+
"reward_std": 0.20708126928657294,
|
| 633 |
+
"rewards/accuracy_reward": 0.19375001089647412,
|
| 634 |
+
"rewards/format_reward": 0.9473214522004128,
|
| 635 |
+
"step": 230
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"completion_length": 314.76697731018066,
|
| 639 |
+
"epoch": 0.3633201275485554,
|
| 640 |
+
"grad_norm": 2.2235055363102907,
|
| 641 |
+
"kl": 0.845751953125,
|
| 642 |
+
"learning_rate": 1.6064351616101318e-05,
|
| 643 |
+
"loss": 0.0338,
|
| 644 |
+
"reward": 1.0339286170899868,
|
| 645 |
+
"reward_std": 0.27274118475615977,
|
| 646 |
+
"rewards/accuracy_reward": 0.15892857927829027,
|
| 647 |
+
"rewards/format_reward": 0.875000037252903,
|
| 648 |
+
"step": 235
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"completion_length": 349.63215923309326,
|
| 652 |
+
"epoch": 0.3710503430283119,
|
| 653 |
+
"grad_norm": 2.116200281477274,
|
| 654 |
+
"kl": 1.61800537109375,
|
| 655 |
+
"learning_rate": 1.5847188782240473e-05,
|
| 656 |
+
"loss": 0.0647,
|
| 657 |
+
"reward": 1.0000000484287739,
|
| 658 |
+
"reward_std": 0.31062190532684325,
|
| 659 |
+
"rewards/accuracy_reward": 0.16339286621659993,
|
| 660 |
+
"rewards/format_reward": 0.8366071827709675,
|
| 661 |
+
"step": 240
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"completion_length": 300.8339429855347,
|
| 665 |
+
"epoch": 0.3787805585080684,
|
| 666 |
+
"grad_norm": 1.1477379512060555,
|
| 667 |
+
"kl": 1.49195556640625,
|
| 668 |
+
"learning_rate": 1.562575220785569e-05,
|
| 669 |
+
"loss": 0.0597,
|
| 670 |
+
"reward": 1.0580357618629932,
|
| 671 |
+
"reward_std": 0.291681545227766,
|
| 672 |
+
"rewards/accuracy_reward": 0.17767858114093543,
|
| 673 |
+
"rewards/format_reward": 0.8803571790456772,
|
| 674 |
+
"step": 245
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"completion_length": 299.05715465545654,
|
| 678 |
+
"epoch": 0.3865107739878249,
|
| 679 |
+
"grad_norm": 0.8348365648461077,
|
| 680 |
+
"kl": 1.25382080078125,
|
| 681 |
+
"learning_rate": 1.5400203742084508e-05,
|
| 682 |
+
"loss": 0.0502,
|
| 683 |
+
"reward": 1.0508929029107095,
|
| 684 |
+
"reward_std": 0.2841053992509842,
|
| 685 |
+
"rewards/accuracy_reward": 0.17946429466828703,
|
| 686 |
+
"rewards/format_reward": 0.8714286103844643,
|
| 687 |
+
"step": 250
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"completion_length": 263.1178680419922,
|
| 691 |
+
"epoch": 0.3942409894675814,
|
| 692 |
+
"grad_norm": 4.699113934464697,
|
| 693 |
+
"kl": 1.31427001953125,
|
| 694 |
+
"learning_rate": 1.5170708239467143e-05,
|
| 695 |
+
"loss": 0.0526,
|
| 696 |
+
"reward": 1.0598214752972126,
|
| 697 |
+
"reward_std": 0.2613769697025418,
|
| 698 |
+
"rewards/accuracy_reward": 0.17321429392322898,
|
| 699 |
+
"rewards/format_reward": 0.8866071790456772,
|
| 700 |
+
"step": 255
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"completion_length": 301.30447845458986,
|
| 704 |
+
"epoch": 0.4019712049473379,
|
| 705 |
+
"grad_norm": 25.580783149116392,
|
| 706 |
+
"kl": 4.94813232421875,
|
| 707 |
+
"learning_rate": 1.4937433439453465e-05,
|
| 708 |
+
"loss": 0.1981,
|
| 709 |
+
"reward": 1.0321429051458835,
|
| 710 |
+
"reward_std": 0.3080965233966708,
|
| 711 |
+
"rewards/accuracy_reward": 0.18750001015141607,
|
| 712 |
+
"rewards/format_reward": 0.8446428962051868,
|
| 713 |
+
"step": 260
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"completion_length": 353.21697826385497,
|
| 717 |
+
"epoch": 0.4097014204270944,
|
| 718 |
+
"grad_norm": 21.725814777895923,
|
| 719 |
+
"kl": 3.99072265625,
|
| 720 |
+
"learning_rate": 1.4700549843801359e-05,
|
| 721 |
+
"loss": 0.1599,
|
| 722 |
+
"reward": 0.9294643267989159,
|
| 723 |
+
"reward_std": 0.3447145516052842,
|
| 724 |
+
"rewards/accuracy_reward": 0.1553571513853967,
|
| 725 |
+
"rewards/format_reward": 0.7741071842610836,
|
| 726 |
+
"step": 265
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"completion_length": 289.2535852432251,
|
| 730 |
+
"epoch": 0.4174316359068509,
|
| 731 |
+
"grad_norm": 1.0129293730640343,
|
| 732 |
+
"kl": 1.88997802734375,
|
| 733 |
+
"learning_rate": 1.4460230591956097e-05,
|
| 734 |
+
"loss": 0.0756,
|
| 735 |
+
"reward": 1.027678620070219,
|
| 736 |
+
"reward_std": 0.2714784935116768,
|
| 737 |
+
"rewards/accuracy_reward": 0.1714285804890096,
|
| 738 |
+
"rewards/format_reward": 0.8562500402331352,
|
| 739 |
+
"step": 270
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"completion_length": 251.7080472946167,
|
| 743 |
+
"epoch": 0.4251618513866074,
|
| 744 |
+
"grad_norm": 13.1390097339807,
|
| 745 |
+
"kl": 1.76666259765625,
|
| 746 |
+
"learning_rate": 1.421665133450184e-05,
|
| 747 |
+
"loss": 0.0707,
|
| 748 |
+
"reward": 1.1142857648432254,
|
| 749 |
+
"reward_std": 0.22223355881869794,
|
| 750 |
+
"rewards/accuracy_reward": 0.19017858151346445,
|
| 751 |
+
"rewards/format_reward": 0.9241071723401546,
|
| 752 |
+
"step": 275
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"completion_length": 222.12590236663817,
|
| 756 |
+
"epoch": 0.4328920668663639,
|
| 757 |
+
"grad_norm": 0.28425211793760446,
|
| 758 |
+
"kl": 0.20849609375,
|
| 759 |
+
"learning_rate": 1.3969990104777712e-05,
|
| 760 |
+
"loss": 0.0083,
|
| 761 |
+
"reward": 1.1330357640981674,
|
| 762 |
+
"reward_std": 0.18561552856117486,
|
| 763 |
+
"rewards/accuracy_reward": 0.17589286677539348,
|
| 764 |
+
"rewards/format_reward": 0.9571428775787354,
|
| 765 |
+
"step": 280
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"completion_length": 292.66430015563964,
|
| 769 |
+
"epoch": 0.4406222823461204,
|
| 770 |
+
"grad_norm": 0.2655799067132994,
|
| 771 |
+
"kl": 0.1830810546875,
|
| 772 |
+
"learning_rate": 1.3720427188752306e-05,
|
| 773 |
+
"loss": 0.0073,
|
| 774 |
+
"reward": 1.052678619325161,
|
| 775 |
+
"reward_std": 0.24875006265938282,
|
| 776 |
+
"rewards/accuracy_reward": 0.17232143683359027,
|
| 777 |
+
"rewards/format_reward": 0.8803571783006191,
|
| 778 |
+
"step": 285
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"completion_length": 347.53840770721433,
|
| 782 |
+
"epoch": 0.4483524978258769,
|
| 783 |
+
"grad_norm": 0.2414520404968147,
|
| 784 |
+
"kl": 0.1782958984375,
|
| 785 |
+
"learning_rate": 1.3468144993251735e-05,
|
| 786 |
+
"loss": 0.0071,
|
| 787 |
+
"reward": 1.0098214797675609,
|
| 788 |
+
"reward_std": 0.28158001936972143,
|
| 789 |
+
"rewards/accuracy_reward": 0.1848214385099709,
|
| 790 |
+
"rewards/format_reward": 0.8250000409781932,
|
| 791 |
+
"step": 290
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"completion_length": 290.5705478668213,
|
| 795 |
+
"epoch": 0.4560827133056334,
|
| 796 |
+
"grad_norm": 0.21789909324241594,
|
| 797 |
+
"kl": 0.162786865234375,
|
| 798 |
+
"learning_rate": 1.3213327912637563e-05,
|
| 799 |
+
"loss": 0.0065,
|
| 800 |
+
"reward": 1.0812500432133674,
|
| 801 |
+
"reward_std": 0.28410540260374545,
|
| 802 |
+
"rewards/accuracy_reward": 0.19107143776491284,
|
| 803 |
+
"rewards/format_reward": 0.8901786096394062,
|
| 804 |
+
"step": 295
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"completion_length": 231.57590255737304,
|
| 808 |
+
"epoch": 0.4638129287853899,
|
| 809 |
+
"grad_norm": 0.24661177068993628,
|
| 810 |
+
"kl": 0.158062744140625,
|
| 811 |
+
"learning_rate": 1.295616219403197e-05,
|
| 812 |
+
"loss": 0.0063,
|
| 813 |
+
"reward": 1.0991071954369545,
|
| 814 |
+
"reward_std": 0.19319167286157607,
|
| 815 |
+
"rewards/accuracy_reward": 0.1589285796508193,
|
| 816 |
+
"rewards/format_reward": 0.9401785969734192,
|
| 817 |
+
"step": 300
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.4638129287853899,
|
| 821 |
+
"eval_completion_length": 219.2982234954834,
|
| 822 |
+
"eval_kl": 0.13836669921875,
|
| 823 |
+
"eval_loss": 0.005582114681601524,
|
| 824 |
+
"eval_reward": 1.1339286118745804,
|
| 825 |
+
"eval_reward_std": 0.2020305097103119,
|
| 826 |
+
"eval_rewards/accuracy_reward": 0.17410715483129025,
|
| 827 |
+
"eval_rewards/format_reward": 0.9598214477300644,
|
| 828 |
+
"eval_runtime": 36.5449,
|
| 829 |
+
"eval_samples_per_second": 2.709,
|
| 830 |
+
"eval_steps_per_second": 0.109,
|
| 831 |
+
"step": 300
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"completion_length": 260.4241186141968,
|
| 835 |
+
"epoch": 0.4715431442651464,
|
| 836 |
+
"grad_norm": 0.24760948713769942,
|
| 837 |
+
"kl": 0.17685546875,
|
| 838 |
+
"learning_rate": 1.2696835801188816e-05,
|
| 839 |
+
"loss": 0.0071,
|
| 840 |
+
"reward": 1.0669643342494965,
|
| 841 |
+
"reward_std": 0.2209708673879504,
|
| 842 |
+
"rewards/accuracy_reward": 0.15625000838190317,
|
| 843 |
+
"rewards/format_reward": 0.9107143193483352,
|
| 844 |
+
"step": 305
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"completion_length": 277.2196553230286,
|
| 848 |
+
"epoch": 0.47927335974490287,
|
| 849 |
+
"grad_norm": 0.3195688766813074,
|
| 850 |
+
"kl": 0.1870849609375,
|
| 851 |
+
"learning_rate": 1.2435538277109919e-05,
|
| 852 |
+
"loss": 0.0075,
|
| 853 |
+
"reward": 1.0392857626080514,
|
| 854 |
+
"reward_std": 0.26011427883058785,
|
| 855 |
+
"rewards/accuracy_reward": 0.15089286481961608,
|
| 856 |
+
"rewards/format_reward": 0.8883928909897805,
|
| 857 |
+
"step": 310
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"completion_length": 238.69644145965577,
|
| 861 |
+
"epoch": 0.48700357522465937,
|
| 862 |
+
"grad_norm": 0.23930482317882099,
|
| 863 |
+
"kl": 0.22203369140625,
|
| 864 |
+
"learning_rate": 1.2172460605507126e-05,
|
| 865 |
+
"loss": 0.0089,
|
| 866 |
+
"reward": 1.0633928991854191,
|
| 867 |
+
"reward_std": 0.20834395978599787,
|
| 868 |
+
"rewards/accuracy_reward": 0.14732143683359028,
|
| 869 |
+
"rewards/format_reward": 0.9160714574158192,
|
| 870 |
+
"step": 315
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"completion_length": 206.31072359085084,
|
| 874 |
+
"epoch": 0.49473379070441587,
|
| 875 |
+
"grad_norm": 0.6156205700649886,
|
| 876 |
+
"kl": 0.269940185546875,
|
| 877 |
+
"learning_rate": 1.19077950712113e-05,
|
| 878 |
+
"loss": 0.0108,
|
| 879 |
+
"reward": 1.1491071924567222,
|
| 880 |
+
"reward_std": 0.19824243448674678,
|
| 881 |
+
"rewards/accuracy_reward": 0.19285715268924833,
|
| 882 |
+
"rewards/format_reward": 0.9562500186264515,
|
| 883 |
+
"step": 320
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"completion_length": 208.71608085632323,
|
| 887 |
+
"epoch": 0.5024640061841724,
|
| 888 |
+
"grad_norm": 1.145359964951925,
|
| 889 |
+
"kl": 0.511163330078125,
|
| 890 |
+
"learning_rate": 1.1641735119630373e-05,
|
| 891 |
+
"loss": 0.0204,
|
| 892 |
+
"reward": 1.1169643372297287,
|
| 893 |
+
"reward_std": 0.1603617152199149,
|
| 894 |
+
"rewards/accuracy_reward": 0.1553571513853967,
|
| 895 |
+
"rewards/format_reward": 0.9616071604192257,
|
| 896 |
+
"step": 325
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"completion_length": 228.73393907546998,
|
| 900 |
+
"epoch": 0.5101942216639289,
|
| 901 |
+
"grad_norm": 1.1998052405572637,
|
| 902 |
+
"kl": 0.50406494140625,
|
| 903 |
+
"learning_rate": 1.137447521535908e-05,
|
| 904 |
+
"loss": 0.0202,
|
| 905 |
+
"reward": 1.116964338719845,
|
| 906 |
+
"reward_std": 0.18309014700353146,
|
| 907 |
+
"rewards/accuracy_reward": 0.17767858058214187,
|
| 908 |
+
"rewards/format_reward": 0.9392857380211354,
|
| 909 |
+
"step": 330
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"completion_length": 269.97054691314696,
|
| 913 |
+
"epoch": 0.5179244371436854,
|
| 914 |
+
"grad_norm": 0.8172230478043011,
|
| 915 |
+
"kl": 0.8053955078125,
|
| 916 |
+
"learning_rate": 1.110621070004378e-05,
|
| 917 |
+
"loss": 0.0322,
|
| 918 |
+
"reward": 1.0839286208152772,
|
| 919 |
+
"reward_std": 0.21213203221559523,
|
| 920 |
+
"rewards/accuracy_reward": 0.16071429383009672,
|
| 921 |
+
"rewards/format_reward": 0.923214315623045,
|
| 922 |
+
"step": 335
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"completion_length": 279.7910852432251,
|
| 926 |
+
"epoch": 0.5256546526234419,
|
| 927 |
+
"grad_norm": 1.6253556390659387,
|
| 928 |
+
"kl": 0.6918701171875,
|
| 929 |
+
"learning_rate": 1.0837137649606241e-05,
|
| 930 |
+
"loss": 0.0277,
|
| 931 |
+
"reward": 1.1000000432133674,
|
| 932 |
+
"reward_std": 0.21465741395950316,
|
| 933 |
+
"rewards/accuracy_reward": 0.16875000894069672,
|
| 934 |
+
"rewards/format_reward": 0.9312500268220901,
|
| 935 |
+
"step": 340
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"completion_length": 276.95269145965574,
|
| 939 |
+
"epoch": 0.5333848681031984,
|
| 940 |
+
"grad_norm": 0.2933386914876608,
|
| 941 |
+
"kl": 0.452520751953125,
|
| 942 |
+
"learning_rate": 1.0567452730930743e-05,
|
| 943 |
+
"loss": 0.0181,
|
| 944 |
+
"reward": 1.098214329779148,
|
| 945 |
+
"reward_std": 0.18940359950065613,
|
| 946 |
+
"rewards/accuracy_reward": 0.15625000828877092,
|
| 947 |
+
"rewards/format_reward": 0.9419643089175225,
|
| 948 |
+
"step": 345
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"completion_length": 299.1473344802856,
|
| 952 |
+
"epoch": 0.5411150835829549,
|
| 953 |
+
"grad_norm": 0.5289323764160652,
|
| 954 |
+
"kl": 0.47991943359375,
|
| 955 |
+
"learning_rate": 1.0297353058119209e-05,
|
| 956 |
+
"loss": 0.0192,
|
| 957 |
+
"reward": 1.0937500461935996,
|
| 958 |
+
"reward_std": 0.16793785840272904,
|
| 959 |
+
"rewards/accuracy_reward": 0.14464286472648383,
|
| 960 |
+
"rewards/format_reward": 0.949107164144516,
|
| 961 |
+
"step": 350
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"completion_length": 311.8223342895508,
|
| 965 |
+
"epoch": 0.5488452990627114,
|
| 966 |
+
"grad_norm": 1.3602546085242322,
|
| 967 |
+
"kl": 0.692169189453125,
|
| 968 |
+
"learning_rate": 1.0027036048419514e-05,
|
| 969 |
+
"loss": 0.0277,
|
| 970 |
+
"reward": 1.0892857573926449,
|
| 971 |
+
"reward_std": 0.21465741619467735,
|
| 972 |
+
"rewards/accuracy_reward": 0.16607143813744188,
|
| 973 |
+
"rewards/format_reward": 0.9232143141329289,
|
| 974 |
+
"step": 355
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"completion_length": 313.5830512046814,
|
| 978 |
+
"epoch": 0.5565755145424679,
|
| 979 |
+
"grad_norm": 0.7626171074709401,
|
| 980 |
+
"kl": 0.92459716796875,
|
| 981 |
+
"learning_rate": 9.756699277932196e-06,
|
| 982 |
+
"loss": 0.037,
|
| 983 |
+
"reward": 1.0500000417232513,
|
| 984 |
+
"reward_std": 0.2197081744670868,
|
| 985 |
+
"rewards/accuracy_reward": 0.151785721629858,
|
| 986 |
+
"rewards/format_reward": 0.8982143223285675,
|
| 987 |
+
"step": 360
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"completion_length": 303.96786937713625,
|
| 991 |
+
"epoch": 0.5643057300222244,
|
| 992 |
+
"grad_norm": 0.5235484185987127,
|
| 993 |
+
"kl": 1.150616455078125,
|
| 994 |
+
"learning_rate": 9.486540337201046e-06,
|
| 995 |
+
"loss": 0.046,
|
| 996 |
+
"reward": 1.0473214767873287,
|
| 997 |
+
"reward_std": 0.2841054029762745,
|
| 998 |
+
"rewards/accuracy_reward": 0.1696428654715419,
|
| 999 |
+
"rewards/format_reward": 0.8776786111295223,
|
| 1000 |
+
"step": 365
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"completion_length": 300.03840684890747,
|
| 1004 |
+
"epoch": 0.5720359455019809,
|
| 1005 |
+
"grad_norm": 3.258547728286983,
|
| 1006 |
+
"kl": 1.26881103515625,
|
| 1007 |
+
"learning_rate": 9.216756686793163e-06,
|
| 1008 |
+
"loss": 0.0508,
|
| 1009 |
+
"reward": 1.0267857626080512,
|
| 1010 |
+
"reward_std": 0.29294423535466196,
|
| 1011 |
+
"rewards/accuracy_reward": 0.16964286603033543,
|
| 1012 |
+
"rewards/format_reward": 0.8571428984403611,
|
| 1013 |
+
"step": 370
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"completion_length": 273.17501125335696,
|
| 1017 |
+
"epoch": 0.5797661609817374,
|
| 1018 |
+
"grad_norm": 0.5185689721492496,
|
| 1019 |
+
"kl": 1.03182373046875,
|
| 1020 |
+
"learning_rate": 8.94754551297402e-06,
|
| 1021 |
+
"loss": 0.0413,
|
| 1022 |
+
"reward": 1.1008929088711739,
|
| 1023 |
+
"reward_std": 0.281580020673573,
|
| 1024 |
+
"rewards/accuracy_reward": 0.20982144065201283,
|
| 1025 |
+
"rewards/format_reward": 0.8910714685916901,
|
| 1026 |
+
"step": 375
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"completion_length": 277.25179929733275,
|
| 1030 |
+
"epoch": 0.5874963764614939,
|
| 1031 |
+
"grad_norm": 1.6200792591726878,
|
| 1032 |
+
"kl": 1.022216796875,
|
| 1033 |
+
"learning_rate": 8.67910358358298e-06,
|
| 1034 |
+
"loss": 0.0409,
|
| 1035 |
+
"reward": 1.07053577080369,
|
| 1036 |
+
"reward_std": 0.2462246786803007,
|
| 1037 |
+
"rewards/accuracy_reward": 0.17767858020961286,
|
| 1038 |
+
"rewards/format_reward": 0.8928571790456772,
|
| 1039 |
+
"step": 380
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"completion_length": 242.19554595947267,
|
| 1043 |
+
"epoch": 0.5952265919412504,
|
| 1044 |
+
"grad_norm": 0.40227896921926315,
|
| 1045 |
+
"kl": 0.4446044921875,
|
| 1046 |
+
"learning_rate": 8.411627104214675e-06,
|
| 1047 |
+
"loss": 0.0178,
|
| 1048 |
+
"reward": 1.1410714834928513,
|
| 1049 |
+
"reward_std": 0.20455588828772306,
|
| 1050 |
+
"rewards/accuracy_reward": 0.19107143832370638,
|
| 1051 |
+
"rewards/format_reward": 0.9500000223517417,
|
| 1052 |
+
"step": 385
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"completion_length": 235.43483171463012,
|
| 1056 |
+
"epoch": 0.6029568074210069,
|
| 1057 |
+
"grad_norm": 0.8419867144446279,
|
| 1058 |
+
"kl": 0.502435302734375,
|
| 1059 |
+
"learning_rate": 8.145311574811325e-06,
|
| 1060 |
+
"loss": 0.0201,
|
| 1061 |
+
"reward": 1.1321429118514061,
|
| 1062 |
+
"reward_std": 0.19950512573122978,
|
| 1063 |
+
"rewards/accuracy_reward": 0.1866071516647935,
|
| 1064 |
+
"rewards/format_reward": 0.9455357372760773,
|
| 1065 |
+
"step": 390
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"completion_length": 221.21072397232055,
|
| 1069 |
+
"epoch": 0.6106870229007634,
|
| 1070 |
+
"grad_norm": 0.31467383074598076,
|
| 1071 |
+
"kl": 0.426025390625,
|
| 1072 |
+
"learning_rate": 7.880351646770824e-06,
|
| 1073 |
+
"loss": 0.017,
|
| 1074 |
+
"reward": 1.1294643431901932,
|
| 1075 |
+
"reward_std": 0.195717054232955,
|
| 1076 |
+
"rewards/accuracy_reward": 0.17321429420262574,
|
| 1077 |
+
"rewards/format_reward": 0.9562500178813934,
|
| 1078 |
+
"step": 395
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"completion_length": 241.07501096725463,
|
| 1082 |
+
"epoch": 0.6184172383805199,
|
| 1083 |
+
"grad_norm": 0.41760270752871836,
|
| 1084 |
+
"kl": 0.618927001953125,
|
| 1085 |
+
"learning_rate": 7.616940980675004e-06,
|
| 1086 |
+
"loss": 0.0248,
|
| 1087 |
+
"reward": 1.0866071924567222,
|
| 1088 |
+
"reward_std": 0.2058185778558254,
|
| 1089 |
+
"rewards/accuracy_reward": 0.15267857955768704,
|
| 1090 |
+
"rewards/format_reward": 0.9339286006987095,
|
| 1091 |
+
"step": 400
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 0.6184172383805199,
|
| 1095 |
+
"eval_completion_length": 254.63423538208008,
|
| 1096 |
+
"eval_kl": 0.627685546875,
|
| 1097 |
+
"eval_loss": 0.026014825329184532,
|
| 1098 |
+
"eval_reward": 1.1428571939468384,
|
| 1099 |
+
"eval_reward_std": 0.21465741470456123,
|
| 1100 |
+
"eval_rewards/accuracy_reward": 0.20535715389996767,
|
| 1101 |
+
"eval_rewards/format_reward": 0.9375000298023224,
|
| 1102 |
+
"eval_runtime": 44.4534,
|
| 1103 |
+
"eval_samples_per_second": 2.227,
|
| 1104 |
+
"eval_steps_per_second": 0.09,
|
| 1105 |
+
"step": 400
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"completion_length": 248.6437618255615,
|
| 1109 |
+
"epoch": 0.6261474538602764,
|
| 1110 |
+
"grad_norm": 1.2063002170462727,
|
| 1111 |
+
"kl": 0.670556640625,
|
| 1112 |
+
"learning_rate": 7.355272104742132e-06,
|
| 1113 |
+
"loss": 0.0268,
|
| 1114 |
+
"reward": 1.1375000402331352,
|
| 1115 |
+
"reward_std": 0.21718279421329498,
|
| 1116 |
+
"rewards/accuracy_reward": 0.19732143972069024,
|
| 1117 |
+
"rewards/format_reward": 0.9401785954833031,
|
| 1118 |
+
"step": 405
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"completion_length": 265.54465503692626,
|
| 1122 |
+
"epoch": 0.6338776693400329,
|
| 1123 |
+
"grad_norm": 0.8372970894830984,
|
| 1124 |
+
"kl": 0.582720947265625,
|
| 1125 |
+
"learning_rate": 7.095536274107046e-06,
|
| 1126 |
+
"loss": 0.0233,
|
| 1127 |
+
"reward": 1.1276786223053932,
|
| 1128 |
+
"reward_std": 0.22349624708294868,
|
| 1129 |
+
"rewards/accuracy_reward": 0.2008928671479225,
|
| 1130 |
+
"rewards/format_reward": 0.9267857432365417,
|
| 1131 |
+
"step": 410
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"completion_length": 330.5723365783691,
|
| 1135 |
+
"epoch": 0.6416078848197894,
|
| 1136 |
+
"grad_norm": 1.0585292028421611,
|
| 1137 |
+
"kl": 12.741766357421875,
|
| 1138 |
+
"learning_rate": 6.837923331031761e-06,
|
| 1139 |
+
"loss": 0.5087,
|
| 1140 |
+
"reward": 1.0241071917116642,
|
| 1141 |
+
"reward_std": 0.2512754438444972,
|
| 1142 |
+
"rewards/accuracy_reward": 0.16250000838190318,
|
| 1143 |
+
"rewards/format_reward": 0.8616071842610836,
|
| 1144 |
+
"step": 415
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"completion_length": 328.27858657836913,
|
| 1148 |
+
"epoch": 0.6493381002995459,
|
| 1149 |
+
"grad_norm": 0.49066051087453305,
|
| 1150 |
+
"kl": 1.210614013671875,
|
| 1151 |
+
"learning_rate": 6.58262156614881e-06,
|
| 1152 |
+
"loss": 0.0485,
|
| 1153 |
+
"reward": 1.0526786148548126,
|
| 1154 |
+
"reward_std": 0.2916815456002951,
|
| 1155 |
+
"rewards/accuracy_reward": 0.19642858104780317,
|
| 1156 |
+
"rewards/format_reward": 0.8562500372529029,
|
| 1157 |
+
"step": 420
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"completion_length": 308.60983619689944,
|
| 1161 |
+
"epoch": 0.6570683157793024,
|
| 1162 |
+
"grad_norm": 0.9219993337746698,
|
| 1163 |
+
"kl": 1.082733154296875,
|
| 1164 |
+
"learning_rate": 6.3298175808386284e-06,
|
| 1165 |
+
"loss": 0.0433,
|
| 1166 |
+
"reward": 1.0651786200702191,
|
| 1167 |
+
"reward_std": 0.2765292562544346,
|
| 1168 |
+
"rewards/accuracy_reward": 0.18928572442382574,
|
| 1169 |
+
"rewards/format_reward": 0.8758928962051868,
|
| 1170 |
+
"step": 425
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"completion_length": 256.632155418396,
|
| 1174 |
+
"epoch": 0.6647985312590589,
|
| 1175 |
+
"grad_norm": 0.4539816272276013,
|
| 1176 |
+
"kl": 0.761395263671875,
|
| 1177 |
+
"learning_rate": 6.079696150841634e-06,
|
| 1178 |
+
"loss": 0.0305,
|
| 1179 |
+
"reward": 1.1026786148548127,
|
| 1180 |
+
"reward_std": 0.22854701150208712,
|
| 1181 |
+
"rewards/accuracy_reward": 0.17857143841683865,
|
| 1182 |
+
"rewards/format_reward": 0.9241071738302707,
|
| 1183 |
+
"step": 430
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"completion_length": 235.15983219146727,
|
| 1187 |
+
"epoch": 0.6725287467388154,
|
| 1188 |
+
"grad_norm": 0.2974800717301805,
|
| 1189 |
+
"kl": 0.74510498046875,
|
| 1190 |
+
"learning_rate": 5.832440091204698e-06,
|
| 1191 |
+
"loss": 0.0298,
|
| 1192 |
+
"reward": 1.1125000521540642,
|
| 1193 |
+
"reward_std": 0.19445436242967845,
|
| 1194 |
+
"rewards/accuracy_reward": 0.17500000884756445,
|
| 1195 |
+
"rewards/format_reward": 0.9375000268220901,
|
| 1196 |
+
"step": 435
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"completion_length": 245.59376096725464,
|
| 1200 |
+
"epoch": 0.6802589622185718,
|
| 1201 |
+
"grad_norm": 1.0471525533631525,
|
| 1202 |
+
"kl": 0.828253173828125,
|
| 1203 |
+
"learning_rate": 5.588230122660672e-06,
|
| 1204 |
+
"loss": 0.0331,
|
| 1205 |
+
"reward": 1.1330357663333417,
|
| 1206 |
+
"reward_std": 0.23107239231467247,
|
| 1207 |
+
"rewards/accuracy_reward": 0.21160715389996768,
|
| 1208 |
+
"rewards/format_reward": 0.9214286014437676,
|
| 1209 |
+
"step": 440
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"completion_length": 260.4134042739868,
|
| 1213 |
+
"epoch": 0.6879891776983283,
|
| 1214 |
+
"grad_norm": 1.0462583292018497,
|
| 1215 |
+
"kl": 1.00888671875,
|
| 1216 |
+
"learning_rate": 5.347244739538677e-06,
|
| 1217 |
+
"loss": 0.0404,
|
| 1218 |
+
"reward": 1.1080357603728772,
|
| 1219 |
+
"reward_std": 0.24622467998415232,
|
| 1220 |
+
"rewards/accuracy_reward": 0.2017857247032225,
|
| 1221 |
+
"rewards/format_reward": 0.9062500357627868,
|
| 1222 |
+
"step": 445
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"completion_length": 260.433941078186,
|
| 1226 |
+
"epoch": 0.6957193931780848,
|
| 1227 |
+
"grad_norm": 0.6455052674276692,
|
| 1228 |
+
"kl": 0.912286376953125,
|
| 1229 |
+
"learning_rate": 5.109660079301668e-06,
|
| 1230 |
+
"loss": 0.0365,
|
| 1231 |
+
"reward": 1.1223214827477932,
|
| 1232 |
+
"reward_std": 0.2335977738723159,
|
| 1233 |
+
"rewards/accuracy_reward": 0.20089286817237734,
|
| 1234 |
+
"rewards/format_reward": 0.9214286021888256,
|
| 1235 |
+
"step": 450
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"completion_length": 256.63661804199216,
|
| 1239 |
+
"epoch": 0.7034496086578413,
|
| 1240 |
+
"grad_norm": 0.6457302051700157,
|
| 1241 |
+
"kl": 0.8274169921875,
|
| 1242 |
+
"learning_rate": 4.875649793806655e-06,
|
| 1243 |
+
"loss": 0.0331,
|
| 1244 |
+
"reward": 1.1169643312692643,
|
| 1245 |
+
"reward_std": 0.2260216299444437,
|
| 1246 |
+
"rewards/accuracy_reward": 0.19107143823057413,
|
| 1247 |
+
"rewards/format_reward": 0.9258928872644901,
|
| 1248 |
+
"step": 455
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"completion_length": 288.9500138282776,
|
| 1252 |
+
"epoch": 0.7111798241375978,
|
| 1253 |
+
"grad_norm": 0.24130951862078845,
|
| 1254 |
+
"kl": 1.132049560546875,
|
| 1255 |
+
"learning_rate": 4.64538492238166e-06,
|
| 1256 |
+
"loss": 0.0453,
|
| 1257 |
+
"reward": 1.1035714767873288,
|
| 1258 |
+
"reward_std": 0.24748737178742886,
|
| 1259 |
+
"rewards/accuracy_reward": 0.21071429643779993,
|
| 1260 |
+
"rewards/format_reward": 0.8928571783006192,
|
| 1261 |
+
"step": 460
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"completion_length": 274.6125121116638,
|
| 1265 |
+
"epoch": 0.7189100396173543,
|
| 1266 |
+
"grad_norm": 0.6809081363150956,
|
| 1267 |
+
"kl": 0.901507568359375,
|
| 1268 |
+
"learning_rate": 4.4190337668121964e-06,
|
| 1269 |
+
"loss": 0.0361,
|
| 1270 |
+
"reward": 1.1500000551342964,
|
| 1271 |
+
"reward_std": 0.2500127531588078,
|
| 1272 |
+
"rewards/accuracy_reward": 0.22767858253791928,
|
| 1273 |
+
"rewards/format_reward": 0.9223214589059353,
|
| 1274 |
+
"step": 465
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"completion_length": 269.7616189956665,
|
| 1278 |
+
"epoch": 0.7266402550971108,
|
| 1279 |
+
"grad_norm": 0.5074301520662177,
|
| 1280 |
+
"kl": 0.875689697265625,
|
| 1281 |
+
"learning_rate": 4.196761768328599e-06,
|
| 1282 |
+
"loss": 0.0351,
|
| 1283 |
+
"reward": 1.1383929133415223,
|
| 1284 |
+
"reward_std": 0.2209708670154214,
|
| 1285 |
+
"rewards/accuracy_reward": 0.2125000100582838,
|
| 1286 |
+
"rewards/format_reward": 0.9258928872644901,
|
| 1287 |
+
"step": 470
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"completion_length": 265.1625129699707,
|
| 1291 |
+
"epoch": 0.7343704705768673,
|
| 1292 |
+
"grad_norm": 0.6415529481366966,
|
| 1293 |
+
"kl": 0.763165283203125,
|
| 1294 |
+
"learning_rate": 3.978731386684206e-06,
|
| 1295 |
+
"loss": 0.0305,
|
| 1296 |
+
"reward": 1.1428571954369544,
|
| 1297 |
+
"reward_std": 0.25001275185495614,
|
| 1298 |
+
"rewards/accuracy_reward": 0.2160714398138225,
|
| 1299 |
+
"rewards/format_reward": 0.9267857432365417,
|
| 1300 |
+
"step": 475
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"completion_length": 272.5562623977661,
|
| 1304 |
+
"epoch": 0.7421006860566238,
|
| 1305 |
+
"grad_norm": 0.35234126624207923,
|
| 1306 |
+
"kl": 0.82550048828125,
|
| 1307 |
+
"learning_rate": 3.7651019814126656e-06,
|
| 1308 |
+
"loss": 0.033,
|
| 1309 |
+
"reward": 1.1383929066359997,
|
| 1310 |
+
"reward_std": 0.21339472401887177,
|
| 1311 |
+
"rewards/accuracy_reward": 0.21339286882430314,
|
| 1312 |
+
"rewards/format_reward": 0.9250000290572643,
|
| 1313 |
+
"step": 480
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"completion_length": 272.2678701400757,
|
| 1317 |
+
"epoch": 0.7498309015363803,
|
| 1318 |
+
"grad_norm": 0.9100314610541868,
|
| 1319 |
+
"kl": 0.79708251953125,
|
| 1320 |
+
"learning_rate": 3.5560296953512296e-06,
|
| 1321 |
+
"loss": 0.0319,
|
| 1322 |
+
"reward": 1.1250000461935996,
|
| 1323 |
+
"reward_std": 0.23233508188277482,
|
| 1324 |
+
"rewards/accuracy_reward": 0.1973214376717806,
|
| 1325 |
+
"rewards/format_reward": 0.9276786029338837,
|
| 1326 |
+
"step": 485
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"completion_length": 280.9955499649048,
|
| 1330 |
+
"epoch": 0.7575611170161368,
|
| 1331 |
+
"grad_norm": 0.3556399924481661,
|
| 1332 |
+
"kl": 0.9994873046875,
|
| 1333 |
+
"learning_rate": 3.3516673405151546e-06,
|
| 1334 |
+
"loss": 0.04,
|
| 1335 |
+
"reward": 1.1285714760422707,
|
| 1336 |
+
"reward_std": 0.2247589396312833,
|
| 1337 |
+
"rewards/accuracy_reward": 0.2062500107102096,
|
| 1338 |
+
"rewards/format_reward": 0.9223214603960515,
|
| 1339 |
+
"step": 490
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"completion_length": 276.1401895523071,
|
| 1343 |
+
"epoch": 0.7652913324958933,
|
| 1344 |
+
"grad_norm": 0.5581701650895203,
|
| 1345 |
+
"kl": 0.73818359375,
|
| 1346 |
+
"learning_rate": 3.1521642864065905e-06,
|
| 1347 |
+
"loss": 0.0295,
|
| 1348 |
+
"reward": 1.150892909616232,
|
| 1349 |
+
"reward_std": 0.22602162901312112,
|
| 1350 |
+
"rewards/accuracy_reward": 0.21785715483129026,
|
| 1351 |
+
"rewards/format_reward": 0.9330357424914837,
|
| 1352 |
+
"step": 495
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"completion_length": 305.8750144004822,
|
| 1356 |
+
"epoch": 0.7730215479756498,
|
| 1357 |
+
"grad_norm": 0.2985575344762852,
|
| 1358 |
+
"kl": 1.02808837890625,
|
| 1359 |
+
"learning_rate": 2.957666350839663e-06,
|
| 1360 |
+
"loss": 0.0411,
|
| 1361 |
+
"reward": 1.145535769313574,
|
| 1362 |
+
"reward_std": 0.23612315505743026,
|
| 1363 |
+
"rewards/accuracy_reward": 0.22946429708972574,
|
| 1364 |
+
"rewards/format_reward": 0.9160714574158192,
|
| 1365 |
+
"step": 500
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 0.7730215479756498,
|
| 1369 |
+
"eval_completion_length": 257.7681636810303,
|
| 1370 |
+
"eval_kl": 0.739013671875,
|
| 1371 |
+
"eval_loss": 0.030957935377955437,
|
| 1372 |
+
"eval_reward": 1.191964328289032,
|
| 1373 |
+
"eval_reward_std": 0.27147849928587675,
|
| 1374 |
+
"eval_rewards/accuracy_reward": 0.263392873108387,
|
| 1375 |
+
"eval_rewards/format_reward": 0.928571455180645,
|
| 1376 |
+
"eval_runtime": 39.965,
|
| 1377 |
+
"eval_samples_per_second": 2.477,
|
| 1378 |
+
"eval_steps_per_second": 0.1,
|
| 1379 |
+
"step": 500
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"completion_length": 284.98126316070557,
|
| 1383 |
+
"epoch": 0.7807517634554063,
|
| 1384 |
+
"grad_norm": 0.2775335619661649,
|
| 1385 |
+
"kl": 0.804058837890625,
|
| 1386 |
+
"learning_rate": 2.768315693361474e-06,
|
| 1387 |
+
"loss": 0.0322,
|
| 1388 |
+
"reward": 1.1401786215603351,
|
| 1389 |
+
"reward_std": 0.2335977740585804,
|
| 1390 |
+
"rewards/accuracy_reward": 0.22589286724105478,
|
| 1391 |
+
"rewards/format_reward": 0.9142857484519482,
|
| 1392 |
+
"step": 505
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"completion_length": 302.475905418396,
|
| 1396 |
+
"epoch": 0.7884819789351628,
|
| 1397 |
+
"grad_norm": 0.4783830317374093,
|
| 1398 |
+
"kl": 0.930975341796875,
|
| 1399 |
+
"learning_rate": 2.5842507113469307e-06,
|
| 1400 |
+
"loss": 0.0373,
|
| 1401 |
+
"reward": 1.1107143394649028,
|
| 1402 |
+
"reward_std": 0.21970817670226098,
|
| 1403 |
+
"rewards/accuracy_reward": 0.1910714373923838,
|
| 1404 |
+
"rewards/format_reward": 0.9196428842842579,
|
| 1405 |
+
"step": 510
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"completion_length": 326.8464429855347,
|
| 1409 |
+
"epoch": 0.7962121944149193,
|
| 1410 |
+
"grad_norm": 0.4587112625836826,
|
| 1411 |
+
"kl": 1.243695068359375,
|
| 1412 |
+
"learning_rate": 2.405605938843416e-06,
|
| 1413 |
+
"loss": 0.0498,
|
| 1414 |
+
"reward": 1.0883929036557674,
|
| 1415 |
+
"reward_std": 0.2916815454140306,
|
| 1416 |
+
"rewards/accuracy_reward": 0.2071428682655096,
|
| 1417 |
+
"rewards/format_reward": 0.881250037252903,
|
| 1418 |
+
"step": 515
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"completion_length": 338.72858772277834,
|
| 1422 |
+
"epoch": 0.8039424098946758,
|
| 1423 |
+
"grad_norm": 0.5814971897829918,
|
| 1424 |
+
"kl": 1.364935302734375,
|
| 1425 |
+
"learning_rate": 2.2325119482391466e-06,
|
| 1426 |
+
"loss": 0.0546,
|
| 1427 |
+
"reward": 1.0830357670783997,
|
| 1428 |
+
"reward_std": 0.3118845963850617,
|
| 1429 |
+
"rewards/accuracy_reward": 0.2160714386962354,
|
| 1430 |
+
"rewards/format_reward": 0.866964328289032,
|
| 1431 |
+
"step": 520
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"completion_length": 354.6250160217285,
|
| 1435 |
+
"epoch": 0.8116726253744323,
|
| 1436 |
+
"grad_norm": 0.5395627446463483,
|
| 1437 |
+
"kl": 1.560418701171875,
|
| 1438 |
+
"learning_rate": 2.065095254827133e-06,
|
| 1439 |
+
"loss": 0.0624,
|
| 1440 |
+
"reward": 1.0625000484287739,
|
| 1441 |
+
"reward_std": 0.2954696161672473,
|
| 1442 |
+
"rewards/accuracy_reward": 0.21517858309671284,
|
| 1443 |
+
"rewards/format_reward": 0.847321467846632,
|
| 1444 |
+
"step": 525
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"completion_length": 354.5348379135132,
|
| 1448 |
+
"epoch": 0.8194028408541888,
|
| 1449 |
+
"grad_norm": 0.5903315470866531,
|
| 1450 |
+
"kl": 1.75145263671875,
|
| 1451 |
+
"learning_rate": 1.9034782243345074e-06,
|
| 1452 |
+
"loss": 0.0701,
|
| 1453 |
+
"reward": 1.0437500521540641,
|
| 1454 |
+
"reward_std": 0.3118845956400037,
|
| 1455 |
+
"rewards/accuracy_reward": 0.20714286724105477,
|
| 1456 |
+
"rewards/format_reward": 0.8366071812808513,
|
| 1457 |
+
"step": 530
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"completion_length": 355.27947940826414,
|
| 1461 |
+
"epoch": 0.8271330563339453,
|
| 1462 |
+
"grad_norm": 1.2208377616122188,
|
| 1463 |
+
"kl": 1.34444580078125,
|
| 1464 |
+
"learning_rate": 1.7477789834847835e-06,
|
| 1465 |
+
"loss": 0.0538,
|
| 1466 |
+
"reward": 1.0562500461935997,
|
| 1467 |
+
"reward_std": 0.3043084528297186,
|
| 1468 |
+
"rewards/accuracy_reward": 0.20982143869623543,
|
| 1469 |
+
"rewards/format_reward": 0.8464286126196384,
|
| 1470 |
+
"step": 535
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"completion_length": 332.879479598999,
|
| 1474 |
+
"epoch": 0.8348632718137018,
|
| 1475 |
+
"grad_norm": 0.5240182175615019,
|
| 1476 |
+
"kl": 1.40631103515625,
|
| 1477 |
+
"learning_rate": 1.5981113336584041e-06,
|
| 1478 |
+
"loss": 0.0563,
|
| 1479 |
+
"reward": 1.0500000439584256,
|
| 1480 |
+
"reward_std": 0.2626396602019668,
|
| 1481 |
+
"rewards/accuracy_reward": 0.1973214391618967,
|
| 1482 |
+
"rewards/format_reward": 0.8526786088943481,
|
| 1483 |
+
"step": 540
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"completion_length": 323.62679920196535,
|
| 1487 |
+
"epoch": 0.8425934872934583,
|
| 1488 |
+
"grad_norm": 0.6433085422004733,
|
| 1489 |
+
"kl": 1.07110595703125,
|
| 1490 |
+
"learning_rate": 1.4545846677147446e-06,
|
| 1491 |
+
"loss": 0.0429,
|
| 1492 |
+
"reward": 1.0964286200702191,
|
| 1493 |
+
"reward_std": 0.23991122860461472,
|
| 1494 |
+
"rewards/accuracy_reward": 0.21250001098960639,
|
| 1495 |
+
"rewards/format_reward": 0.8839286118745804,
|
| 1496 |
+
"step": 545
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"completion_length": 286.84554920196535,
|
| 1500 |
+
"epoch": 0.8503237027732148,
|
| 1501 |
+
"grad_norm": 0.4618904983113792,
|
| 1502 |
+
"kl": 0.94029541015625,
|
| 1503 |
+
"learning_rate": 1.3173038900362977e-06,
|
| 1504 |
+
"loss": 0.0376,
|
| 1505 |
+
"reward": 1.1437500521540642,
|
| 1506 |
+
"reward_std": 0.26390235032886267,
|
| 1507 |
+
"rewards/accuracy_reward": 0.22857144000008703,
|
| 1508 |
+
"rewards/format_reward": 0.9151786006987095,
|
| 1509 |
+
"step": 550
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"completion_length": 281.5741184234619,
|
| 1513 |
+
"epoch": 0.8580539182529713,
|
| 1514 |
+
"grad_norm": 0.4684944225912412,
|
| 1515 |
+
"kl": 1.12393798828125,
|
| 1516 |
+
"learning_rate": 1.1863693398535115e-06,
|
| 1517 |
+
"loss": 0.045,
|
| 1518 |
+
"reward": 1.1348214790225029,
|
| 1519 |
+
"reward_std": 0.2588515877723694,
|
| 1520 |
+
"rewards/accuracy_reward": 0.22589286798611283,
|
| 1521 |
+
"rewards/format_reward": 0.908928605914116,
|
| 1522 |
+
"step": 555
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"completion_length": 291.7946542739868,
|
| 1526 |
+
"epoch": 0.8657841337327278,
|
| 1527 |
+
"grad_norm": 0.3860335559805723,
|
| 1528 |
+
"kl": 1.16566162109375,
|
| 1529 |
+
"learning_rate": 1.0618767179063416e-06,
|
| 1530 |
+
"loss": 0.0466,
|
| 1531 |
+
"reward": 1.13125004991889,
|
| 1532 |
+
"reward_std": 0.253800824098289,
|
| 1533 |
+
"rewards/accuracy_reward": 0.23928572591394187,
|
| 1534 |
+
"rewards/format_reward": 0.8919643238186836,
|
| 1535 |
+
"step": 560
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"completion_length": 288.9223350524902,
|
| 1539 |
+
"epoch": 0.8735143492124843,
|
| 1540 |
+
"grad_norm": 0.5947417112761748,
|
| 1541 |
+
"kl": 1.116168212890625,
|
| 1542 |
+
"learning_rate": 9.439170164960765e-07,
|
| 1543 |
+
"loss": 0.0446,
|
| 1544 |
+
"reward": 1.1133929133415221,
|
| 1545 |
+
"reward_std": 0.26137696839869023,
|
| 1546 |
+
"rewards/accuracy_reward": 0.22410715520381927,
|
| 1547 |
+
"rewards/format_reward": 0.8892857529222965,
|
| 1548 |
+
"step": 565
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"completion_length": 312.45358619689944,
|
| 1552 |
+
"epoch": 0.8812445646922408,
|
| 1553 |
+
"grad_norm": 0.5248985437711838,
|
| 1554 |
+
"kl": 1.397430419921875,
|
| 1555 |
+
"learning_rate": 8.325764529785851e-07,
|
| 1556 |
+
"loss": 0.0559,
|
| 1557 |
+
"reward": 1.0696429111063481,
|
| 1558 |
+
"reward_std": 0.2676904214546084,
|
| 1559 |
+
"rewards/accuracy_reward": 0.20178572395816446,
|
| 1560 |
+
"rewards/format_reward": 0.8678571797907353,
|
| 1561 |
+
"step": 570
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"completion_length": 295.79108448028563,
|
| 1565 |
+
"epoch": 0.8889747801719973,
|
| 1566 |
+
"grad_norm": 0.450179012206087,
|
| 1567 |
+
"kl": 1.271270751953125,
|
| 1568 |
+
"learning_rate": 7.279364067476247e-07,
|
| 1569 |
+
"loss": 0.0509,
|
| 1570 |
+
"reward": 1.0758929073810577,
|
| 1571 |
+
"reward_std": 0.2310723926872015,
|
| 1572 |
+
"rewards/accuracy_reward": 0.19910715306177734,
|
| 1573 |
+
"rewards/format_reward": 0.8767857521772384,
|
| 1574 |
+
"step": 575
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"completion_length": 286.98304786682127,
|
| 1578 |
+
"epoch": 0.8967049956517538,
|
| 1579 |
+
"grad_norm": 0.283165116362945,
|
| 1580 |
+
"kl": 1.03193359375,
|
| 1581 |
+
"learning_rate": 6.300733597542086e-07,
|
| 1582 |
+
"loss": 0.0413,
|
| 1583 |
+
"reward": 1.120535772293806,
|
| 1584 |
+
"reward_std": 0.24117391742765903,
|
| 1585 |
+
"rewards/accuracy_reward": 0.21964286752045153,
|
| 1586 |
+
"rewards/format_reward": 0.9008928939700127,
|
| 1587 |
+
"step": 580
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"completion_length": 301.77501583099365,
|
| 1591 |
+
"epoch": 0.9044352111315103,
|
| 1592 |
+
"grad_norm": 0.46974734005334373,
|
| 1593 |
+
"kl": 1.178570556640625,
|
| 1594 |
+
"learning_rate": 5.390588406055497e-07,
|
| 1595 |
+
"loss": 0.0471,
|
| 1596 |
+
"reward": 1.0892857603728772,
|
| 1597 |
+
"reward_std": 0.25253813378512857,
|
| 1598 |
+
"rewards/accuracy_reward": 0.20178572423756122,
|
| 1599 |
+
"rewards/format_reward": 0.8875000402331352,
|
| 1600 |
+
"step": 585
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"completion_length": 260.1544776916504,
|
| 1604 |
+
"epoch": 0.9121654266112668,
|
| 1605 |
+
"grad_norm": 0.3415945982894548,
|
| 1606 |
+
"kl": 0.71856689453125,
|
| 1607 |
+
"learning_rate": 4.549593722844492e-07,
|
| 1608 |
+
"loss": 0.0287,
|
| 1609 |
+
"reward": 1.1750000528991222,
|
| 1610 |
+
"reward_std": 0.21970817632973194,
|
| 1611 |
+
"rewards/accuracy_reward": 0.2366071536205709,
|
| 1612 |
+
"rewards/format_reward": 0.9383928820490837,
|
| 1613 |
+
"step": 590
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"completion_length": 265.1803674697876,
|
| 1617 |
+
"epoch": 0.9198956420910233,
|
| 1618 |
+
"grad_norm": 0.2708029106865859,
|
| 1619 |
+
"kl": 0.8180908203125,
|
| 1620 |
+
"learning_rate": 3.77836423527278e-07,
|
| 1621 |
+
"loss": 0.0328,
|
| 1622 |
+
"reward": 1.188392923027277,
|
| 1623 |
+
"reward_std": 0.2386485354974866,
|
| 1624 |
+
"rewards/accuracy_reward": 0.255357154738158,
|
| 1625 |
+
"rewards/format_reward": 0.9330357387661934,
|
| 1626 |
+
"step": 595
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"completion_length": 270.0437623023987,
|
| 1630 |
+
"epoch": 0.9276258575707798,
|
| 1631 |
+
"grad_norm": 0.46374471812077694,
|
| 1632 |
+
"kl": 0.837054443359375,
|
| 1633 |
+
"learning_rate": 3.0774636389618196e-07,
|
| 1634 |
+
"loss": 0.0335,
|
| 1635 |
+
"reward": 1.1705357685685158,
|
| 1636 |
+
"reward_std": 0.24875006210058928,
|
| 1637 |
+
"rewards/accuracy_reward": 0.23571429755538703,
|
| 1638 |
+
"rewards/format_reward": 0.9348214536905288,
|
| 1639 |
+
"step": 600
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 0.9276258575707798,
|
| 1643 |
+
"eval_completion_length": 265.839298248291,
|
| 1644 |
+
"eval_kl": 0.89306640625,
|
| 1645 |
+
"eval_loss": 0.037714019417762756,
|
| 1646 |
+
"eval_reward": 1.1875000447034836,
|
| 1647 |
+
"eval_reward_std": 0.2904188595712185,
|
| 1648 |
+
"eval_rewards/accuracy_reward": 0.2678571529686451,
|
| 1649 |
+
"eval_rewards/format_reward": 0.9196428880095482,
|
| 1650 |
+
"eval_runtime": 44.0608,
|
| 1651 |
+
"eval_samples_per_second": 2.247,
|
| 1652 |
+
"eval_steps_per_second": 0.091,
|
| 1653 |
+
"step": 600
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"completion_length": 283.5410858154297,
|
| 1657 |
+
"epoch": 0.9353560730505363,
|
| 1658 |
+
"grad_norm": 0.42477771152185795,
|
| 1659 |
+
"kl": 1.058807373046875,
|
| 1660 |
+
"learning_rate": 2.44740422578269e-07,
|
| 1661 |
+
"loss": 0.0423,
|
| 1662 |
+
"reward": 1.1196429051458836,
|
| 1663 |
+
"reward_std": 0.26516504045575856,
|
| 1664 |
+
"rewards/accuracy_reward": 0.21250001108273864,
|
| 1665 |
+
"rewards/format_reward": 0.9071428932249546,
|
| 1666 |
+
"step": 605
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"completion_length": 263.7607263565063,
|
| 1670 |
+
"epoch": 0.9430862885302927,
|
| 1671 |
+
"grad_norm": 0.4567363340156383,
|
| 1672 |
+
"kl": 0.876043701171875,
|
| 1673 |
+
"learning_rate": 1.8886465094192895e-07,
|
| 1674 |
+
"loss": 0.035,
|
| 1675 |
+
"reward": 1.1633929111063481,
|
| 1676 |
+
"reward_std": 0.23359777368605136,
|
| 1677 |
+
"rewards/accuracy_reward": 0.2330357262864709,
|
| 1678 |
+
"rewards/format_reward": 0.9303571738302707,
|
| 1679 |
+
"step": 610
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"completion_length": 270.82501182556155,
|
| 1683 |
+
"epoch": 0.9508165040100492,
|
| 1684 |
+
"grad_norm": 0.28225043605895783,
|
| 1685 |
+
"kl": 0.95989990234375,
|
| 1686 |
+
"learning_rate": 1.401598888776523e-07,
|
| 1687 |
+
"loss": 0.0384,
|
| 1688 |
+
"reward": 1.15535718947649,
|
| 1689 |
+
"reward_std": 0.24748737160116435,
|
| 1690 |
+
"rewards/accuracy_reward": 0.2357142989523709,
|
| 1691 |
+
"rewards/format_reward": 0.9196428880095482,
|
| 1692 |
+
"step": 615
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"completion_length": 283.83126201629636,
|
| 1696 |
+
"epoch": 0.9585467194898057,
|
| 1697 |
+
"grad_norm": 0.32539772814205725,
|
| 1698 |
+
"kl": 1.015838623046875,
|
| 1699 |
+
"learning_rate": 9.866173494794462e-08,
|
| 1700 |
+
"loss": 0.0406,
|
| 1701 |
+
"reward": 1.1410714834928513,
|
| 1702 |
+
"reward_std": 0.2651650408282876,
|
| 1703 |
+
"rewards/accuracy_reward": 0.23214286779984833,
|
| 1704 |
+
"rewards/format_reward": 0.908928606659174,
|
| 1705 |
+
"step": 620
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"completion_length": 272.28661918640137,
|
| 1709 |
+
"epoch": 0.9662769349695622,
|
| 1710 |
+
"grad_norm": 0.40483227452880277,
|
| 1711 |
+
"kl": 0.98214111328125,
|
| 1712 |
+
"learning_rate": 6.440052036815081e-08,
|
| 1713 |
+
"loss": 0.0393,
|
| 1714 |
+
"reward": 1.1419643364846706,
|
| 1715 |
+
"reward_std": 0.23359777443110943,
|
| 1716 |
+
"rewards/accuracy_reward": 0.22410715464502573,
|
| 1717 |
+
"rewards/format_reward": 0.9178571730852128,
|
| 1718 |
+
"step": 625
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"completion_length": 296.1392983436584,
|
| 1722 |
+
"epoch": 0.9740071504493187,
|
| 1723 |
+
"grad_norm": 0.5042793676495506,
|
| 1724 |
+
"kl": 1.05089111328125,
|
| 1725 |
+
"learning_rate": 3.7401286837214224e-08,
|
| 1726 |
+
"loss": 0.042,
|
| 1727 |
+
"reward": 1.1464286252856255,
|
| 1728 |
+
"reward_std": 0.26516504120081663,
|
| 1729 |
+
"rewards/accuracy_reward": 0.2383928676135838,
|
| 1730 |
+
"rewards/format_reward": 0.9080357499420643,
|
| 1731 |
+
"step": 630
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"completion_length": 291.5410856246948,
|
| 1735 |
+
"epoch": 0.9817373659290752,
|
| 1736 |
+
"grad_norm": 0.2778912462099483,
|
| 1737 |
+
"kl": 1.029638671875,
|
| 1738 |
+
"learning_rate": 1.7683768234568745e-08,
|
| 1739 |
+
"loss": 0.0412,
|
| 1740 |
+
"reward": 1.1250000447034836,
|
| 1741 |
+
"reward_std": 0.2575888976454735,
|
| 1742 |
+
"rewards/accuracy_reward": 0.2125000107102096,
|
| 1743 |
+
"rewards/format_reward": 0.9125000350177288,
|
| 1744 |
+
"step": 635
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"completion_length": 278.6035831451416,
|
| 1748 |
+
"epoch": 0.9894675814088317,
|
| 1749 |
+
"grad_norm": 0.4021017187067133,
|
| 1750 |
+
"kl": 1.000689697265625,
|
| 1751 |
+
"learning_rate": 5.262376196544239e-09,
|
| 1752 |
+
"loss": 0.04,
|
| 1753 |
+
"reward": 1.13035718947649,
|
| 1754 |
+
"reward_std": 0.2575888967141509,
|
| 1755 |
+
"rewards/accuracy_reward": 0.21875001173466443,
|
| 1756 |
+
"rewards/format_reward": 0.9116071738302708,
|
| 1757 |
+
"step": 640
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"completion_length": 272.8669765472412,
|
| 1761 |
+
"epoch": 0.9971977968885882,
|
| 1762 |
+
"grad_norm": 0.22818184024650953,
|
| 1763 |
+
"kl": 0.9707275390625,
|
| 1764 |
+
"learning_rate": 1.461895828280824e-10,
|
| 1765 |
+
"loss": 0.0388,
|
| 1766 |
+
"reward": 1.1553571961820126,
|
| 1767 |
+
"reward_std": 0.23486046474426986,
|
| 1768 |
+
"rewards/accuracy_reward": 0.23839286966249346,
|
| 1769 |
+
"rewards/format_reward": 0.9169643193483352,
|
| 1770 |
+
"step": 645
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"completion_length": 259.26786708831787,
|
| 1774 |
+
"epoch": 0.9987438399845395,
|
| 1775 |
+
"kl": 0.823211669921875,
|
| 1776 |
+
"reward": 1.1785714849829674,
|
| 1777 |
+
"reward_std": 0.2399112293496728,
|
| 1778 |
+
"rewards/accuracy_reward": 0.24553572246804833,
|
| 1779 |
+
"rewards/format_reward": 0.9330357387661934,
|
| 1780 |
+
"step": 646,
|
| 1781 |
+
"total_flos": 0.0,
|
| 1782 |
+
"train_loss": 0.13680233840041295,
|
| 1783 |
+
"train_runtime": 51841.8905,
|
| 1784 |
+
"train_samples_per_second": 1.397,
|
| 1785 |
+
"train_steps_per_second": 0.012
|
| 1786 |
+
}
|
| 1787 |
+
],
|
| 1788 |
+
"logging_steps": 5,
|
| 1789 |
+
"max_steps": 646,
|
| 1790 |
+
"num_input_tokens_seen": 0,
|
| 1791 |
+
"num_train_epochs": 1,
|
| 1792 |
+
"save_steps": 100,
|
| 1793 |
+
"stateful_callbacks": {
|
| 1794 |
+
"TrainerControl": {
|
| 1795 |
+
"args": {
|
| 1796 |
+
"should_epoch_stop": false,
|
| 1797 |
+
"should_evaluate": false,
|
| 1798 |
+
"should_log": false,
|
| 1799 |
+
"should_save": true,
|
| 1800 |
+
"should_training_stop": true
|
| 1801 |
+
},
|
| 1802 |
+
"attributes": {}
|
| 1803 |
+
}
|
| 1804 |
+
},
|
| 1805 |
+
"total_flos": 0.0,
|
| 1806 |
+
"train_batch_size": 2,
|
| 1807 |
+
"trial_name": null,
|
| 1808 |
+
"trial_params": null
|
| 1809 |
+
}
|