Model save
Browse files- README.md +67 -0
- all_results.json +8 -0
- generation_config.json +6 -0
- train_results.json +8 -0
- trainer_state.json +1313 -0
README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
model_name: Qwen-2.5-7B-Simple-RL
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
- trl
|
| 7 |
+
- grpo
|
| 8 |
+
licence: license
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Model Card for Qwen-2.5-7B-Simple-RL
|
| 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/Qwen-2.5-7B-Simple-RL", 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/y7jlqug1)
|
| 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": 0.00286464851636153,
|
| 4 |
+
"train_runtime": 7716.3929,
|
| 5 |
+
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 0.972,
|
| 7 |
+
"train_steps_per_second": 0.061
|
| 8 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"eos_token_id": 151643,
|
| 4 |
+
"max_new_tokens": 2048,
|
| 5 |
+
"transformers_version": "4.49.0.dev0"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.00286464851636153,
|
| 4 |
+
"train_runtime": 7716.3929,
|
| 5 |
+
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 0.972,
|
| 7 |
+
"train_steps_per_second": 0.061
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1313 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9984,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 468,
|
| 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": 637.8268123626709,
|
| 13 |
+
"epoch": 0.010666666666666666,
|
| 14 |
+
"grad_norm": 0.9318448901176453,
|
| 15 |
+
"kl": 0.00011245012283325195,
|
| 16 |
+
"learning_rate": 3.1914893617021275e-07,
|
| 17 |
+
"loss": 0.0,
|
| 18 |
+
"reward": 0.6035714592784643,
|
| 19 |
+
"reward_std": 0.3709465142339468,
|
| 20 |
+
"rewards/accuracy_reward": 0.6017857443541288,
|
| 21 |
+
"rewards/format_reward": 0.001785714365541935,
|
| 22 |
+
"step": 5
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"completion_length": 611.45181350708,
|
| 26 |
+
"epoch": 0.021333333333333333,
|
| 27 |
+
"grad_norm": 6.491418838500977,
|
| 28 |
+
"kl": 0.0001920461654663086,
|
| 29 |
+
"learning_rate": 6.382978723404255e-07,
|
| 30 |
+
"loss": 0.0,
|
| 31 |
+
"reward": 0.617857176065445,
|
| 32 |
+
"reward_std": 0.36645040661096573,
|
| 33 |
+
"rewards/accuracy_reward": 0.617857176065445,
|
| 34 |
+
"rewards/format_reward": 0.0,
|
| 35 |
+
"step": 10
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"completion_length": 604.3607414245605,
|
| 39 |
+
"epoch": 0.032,
|
| 40 |
+
"grad_norm": 9.168869018554688,
|
| 41 |
+
"kl": 0.00031398534774780275,
|
| 42 |
+
"learning_rate": 9.574468085106384e-07,
|
| 43 |
+
"loss": 0.0,
|
| 44 |
+
"reward": 0.6482143150642514,
|
| 45 |
+
"reward_std": 0.34973760321736336,
|
| 46 |
+
"rewards/accuracy_reward": 0.6482143150642514,
|
| 47 |
+
"rewards/format_reward": 0.0,
|
| 48 |
+
"step": 15
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"completion_length": 617.7803817749024,
|
| 52 |
+
"epoch": 0.042666666666666665,
|
| 53 |
+
"grad_norm": 1.5698275566101074,
|
| 54 |
+
"kl": 0.0008591651916503906,
|
| 55 |
+
"learning_rate": 1.276595744680851e-06,
|
| 56 |
+
"loss": 0.0,
|
| 57 |
+
"reward": 0.6303571715950966,
|
| 58 |
+
"reward_std": 0.35366948917508123,
|
| 59 |
+
"rewards/accuracy_reward": 0.6303571715950966,
|
| 60 |
+
"rewards/format_reward": 0.0,
|
| 61 |
+
"step": 20
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"completion_length": 639.5375259399414,
|
| 65 |
+
"epoch": 0.05333333333333334,
|
| 66 |
+
"grad_norm": 0.858055055141449,
|
| 67 |
+
"kl": 0.002974271774291992,
|
| 68 |
+
"learning_rate": 1.5957446808510639e-06,
|
| 69 |
+
"loss": 0.0001,
|
| 70 |
+
"reward": 0.6125000298023224,
|
| 71 |
+
"reward_std": 0.3530873417854309,
|
| 72 |
+
"rewards/accuracy_reward": 0.6125000298023224,
|
| 73 |
+
"rewards/format_reward": 0.0,
|
| 74 |
+
"step": 25
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"completion_length": 664.7107475280761,
|
| 78 |
+
"epoch": 0.064,
|
| 79 |
+
"grad_norm": 1.1970467567443848,
|
| 80 |
+
"kl": 0.005088996887207031,
|
| 81 |
+
"learning_rate": 1.9148936170212767e-06,
|
| 82 |
+
"loss": 0.0002,
|
| 83 |
+
"reward": 0.6553571727126837,
|
| 84 |
+
"reward_std": 0.33203957192599776,
|
| 85 |
+
"rewards/accuracy_reward": 0.6553571727126837,
|
| 86 |
+
"rewards/format_reward": 0.0,
|
| 87 |
+
"step": 30
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"completion_length": 632.7196670532227,
|
| 91 |
+
"epoch": 0.07466666666666667,
|
| 92 |
+
"grad_norm": 1.0202564001083374,
|
| 93 |
+
"kl": 0.996532678604126,
|
| 94 |
+
"learning_rate": 2.2340425531914894e-06,
|
| 95 |
+
"loss": 0.0398,
|
| 96 |
+
"reward": 0.7000000279396772,
|
| 97 |
+
"reward_std": 0.3037954304367304,
|
| 98 |
+
"rewards/accuracy_reward": 0.7000000279396772,
|
| 99 |
+
"rewards/format_reward": 0.0,
|
| 100 |
+
"step": 35
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"completion_length": 640.1286018371582,
|
| 104 |
+
"epoch": 0.08533333333333333,
|
| 105 |
+
"grad_norm": 0.5117438435554504,
|
| 106 |
+
"kl": 0.002056884765625,
|
| 107 |
+
"learning_rate": 2.553191489361702e-06,
|
| 108 |
+
"loss": 0.0001,
|
| 109 |
+
"reward": 0.7250000268220902,
|
| 110 |
+
"reward_std": 0.2653495166450739,
|
| 111 |
+
"rewards/accuracy_reward": 0.7250000268220902,
|
| 112 |
+
"rewards/format_reward": 0.0,
|
| 113 |
+
"step": 40
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"completion_length": 607.9250259399414,
|
| 117 |
+
"epoch": 0.096,
|
| 118 |
+
"grad_norm": 0.3817993402481079,
|
| 119 |
+
"kl": 0.002574253082275391,
|
| 120 |
+
"learning_rate": 2.872340425531915e-06,
|
| 121 |
+
"loss": 0.0001,
|
| 122 |
+
"reward": 0.7392857447266579,
|
| 123 |
+
"reward_std": 0.24080881737172605,
|
| 124 |
+
"rewards/accuracy_reward": 0.7392857447266579,
|
| 125 |
+
"rewards/format_reward": 0.0,
|
| 126 |
+
"step": 45
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"completion_length": 628.1535995483398,
|
| 130 |
+
"epoch": 0.10666666666666667,
|
| 131 |
+
"grad_norm": 0.7749062180519104,
|
| 132 |
+
"kl": 0.004121017456054687,
|
| 133 |
+
"learning_rate": 2.9996241442585123e-06,
|
| 134 |
+
"loss": 0.0002,
|
| 135 |
+
"reward": 0.6964286014437675,
|
| 136 |
+
"reward_std": 0.300217243283987,
|
| 137 |
+
"rewards/accuracy_reward": 0.6964286014437675,
|
| 138 |
+
"rewards/format_reward": 0.0,
|
| 139 |
+
"step": 50
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"completion_length": 600.3839553833008,
|
| 143 |
+
"epoch": 0.11733333333333333,
|
| 144 |
+
"grad_norm": 0.5632015466690063,
|
| 145 |
+
"kl": 0.003362083435058594,
|
| 146 |
+
"learning_rate": 2.9973279301399446e-06,
|
| 147 |
+
"loss": 0.0001,
|
| 148 |
+
"reward": 0.7178571719676257,
|
| 149 |
+
"reward_std": 0.27802310809493064,
|
| 150 |
+
"rewards/accuracy_reward": 0.7178571719676257,
|
| 151 |
+
"rewards/format_reward": 0.0,
|
| 152 |
+
"step": 55
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"completion_length": 585.9303833007813,
|
| 156 |
+
"epoch": 0.128,
|
| 157 |
+
"grad_norm": 1.2080248594284058,
|
| 158 |
+
"kl": 0.004734039306640625,
|
| 159 |
+
"learning_rate": 2.992947502998804e-06,
|
| 160 |
+
"loss": 0.0002,
|
| 161 |
+
"reward": 0.7750000357627869,
|
| 162 |
+
"reward_std": 0.25223284475505353,
|
| 163 |
+
"rewards/accuracy_reward": 0.7750000357627869,
|
| 164 |
+
"rewards/format_reward": 0.0,
|
| 165 |
+
"step": 60
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"completion_length": 591.1464614868164,
|
| 169 |
+
"epoch": 0.13866666666666666,
|
| 170 |
+
"grad_norm": 1.7432464361190796,
|
| 171 |
+
"kl": 0.013347434997558593,
|
| 172 |
+
"learning_rate": 2.9864889601923268e-06,
|
| 173 |
+
"loss": 0.0005,
|
| 174 |
+
"reward": 0.7267857432365418,
|
| 175 |
+
"reward_std": 0.2874183960258961,
|
| 176 |
+
"rewards/accuracy_reward": 0.7267857432365418,
|
| 177 |
+
"rewards/format_reward": 0.0,
|
| 178 |
+
"step": 65
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"completion_length": 588.239315032959,
|
| 182 |
+
"epoch": 0.14933333333333335,
|
| 183 |
+
"grad_norm": 0.28103670477867126,
|
| 184 |
+
"kl": 0.004166412353515625,
|
| 185 |
+
"learning_rate": 2.977961291721137e-06,
|
| 186 |
+
"loss": 0.0002,
|
| 187 |
+
"reward": 0.7875000283122062,
|
| 188 |
+
"reward_std": 0.23745907917618753,
|
| 189 |
+
"rewards/accuracy_reward": 0.7875000283122062,
|
| 190 |
+
"rewards/format_reward": 0.0,
|
| 191 |
+
"step": 70
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"completion_length": 554.2375259399414,
|
| 195 |
+
"epoch": 0.16,
|
| 196 |
+
"grad_norm": 0.26267099380493164,
|
| 197 |
+
"kl": 0.004508209228515625,
|
| 198 |
+
"learning_rate": 2.9673763677155655e-06,
|
| 199 |
+
"loss": 0.0002,
|
| 200 |
+
"reward": 0.7767857410013675,
|
| 201 |
+
"reward_std": 0.21466484777629374,
|
| 202 |
+
"rewards/accuracy_reward": 0.7767857410013675,
|
| 203 |
+
"rewards/format_reward": 0.0,
|
| 204 |
+
"step": 75
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"completion_length": 553.0071640014648,
|
| 208 |
+
"epoch": 0.17066666666666666,
|
| 209 |
+
"grad_norm": 0.34145233035087585,
|
| 210 |
+
"kl": 0.0057952880859375,
|
| 211 |
+
"learning_rate": 2.9547489219129666e-06,
|
| 212 |
+
"loss": 0.0002,
|
| 213 |
+
"reward": 0.8285714581608772,
|
| 214 |
+
"reward_std": 0.20639798790216446,
|
| 215 |
+
"rewards/accuracy_reward": 0.8285714581608772,
|
| 216 |
+
"rewards/format_reward": 0.0,
|
| 217 |
+
"step": 80
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"completion_length": 585.5857421875,
|
| 221 |
+
"epoch": 0.18133333333333335,
|
| 222 |
+
"grad_norm": 0.240849107503891,
|
| 223 |
+
"kl": 0.0038000106811523437,
|
| 224 |
+
"learning_rate": 2.9400965311490175e-06,
|
| 225 |
+
"loss": 0.0002,
|
| 226 |
+
"reward": 0.7660714596509933,
|
| 227 |
+
"reward_std": 0.24059830717742442,
|
| 228 |
+
"rewards/accuracy_reward": 0.7660714596509933,
|
| 229 |
+
"rewards/format_reward": 0.0,
|
| 230 |
+
"step": 85
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"completion_length": 558.014315032959,
|
| 234 |
+
"epoch": 0.192,
|
| 235 |
+
"grad_norm": 0.42303234338760376,
|
| 236 |
+
"kl": 0.004909515380859375,
|
| 237 |
+
"learning_rate": 2.9234395908915565e-06,
|
| 238 |
+
"loss": 0.0002,
|
| 239 |
+
"reward": 0.7142857456579804,
|
| 240 |
+
"reward_std": 0.2469261337071657,
|
| 241 |
+
"rewards/accuracy_reward": 0.7142857456579804,
|
| 242 |
+
"rewards/format_reward": 0.0,
|
| 243 |
+
"step": 90
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"completion_length": 565.8589538574219,
|
| 247 |
+
"epoch": 0.20266666666666666,
|
| 248 |
+
"grad_norm": 0.36556175351142883,
|
| 249 |
+
"kl": 0.004494476318359375,
|
| 250 |
+
"learning_rate": 2.904801286851009e-06,
|
| 251 |
+
"loss": 0.0002,
|
| 252 |
+
"reward": 0.7517857383936644,
|
| 253 |
+
"reward_std": 0.2242392159998417,
|
| 254 |
+
"rewards/accuracy_reward": 0.7517857383936644,
|
| 255 |
+
"rewards/format_reward": 0.0,
|
| 256 |
+
"step": 95
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"completion_length": 520.4696662902832,
|
| 260 |
+
"epoch": 0.21333333333333335,
|
| 261 |
+
"grad_norm": 0.2530968487262726,
|
| 262 |
+
"kl": 0.00856475830078125,
|
| 263 |
+
"learning_rate": 2.884207562706925e-06,
|
| 264 |
+
"loss": 0.0003,
|
| 265 |
+
"reward": 0.8125000301748514,
|
| 266 |
+
"reward_std": 0.18475012369453908,
|
| 267 |
+
"rewards/accuracy_reward": 0.8125000301748514,
|
| 268 |
+
"rewards/format_reward": 0.0,
|
| 269 |
+
"step": 100
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.21333333333333335,
|
| 273 |
+
"eval_completion_length": 547.995567590332,
|
| 274 |
+
"eval_kl": 0.00711322021484375,
|
| 275 |
+
"eval_loss": 0.00028467908850871027,
|
| 276 |
+
"eval_reward": 0.6861428862035275,
|
| 277 |
+
"eval_reward_std": 0.270268753862381,
|
| 278 |
+
"eval_rewards/accuracy_reward": 0.6860857433497906,
|
| 279 |
+
"eval_rewards/format_reward": 5.7142859697341916e-05,
|
| 280 |
+
"eval_runtime": 28593.9461,
|
| 281 |
+
"eval_samples_per_second": 0.175,
|
| 282 |
+
"eval_steps_per_second": 0.013,
|
| 283 |
+
"step": 100
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"completion_length": 549.5107398986817,
|
| 287 |
+
"epoch": 0.224,
|
| 288 |
+
"grad_norm": 0.2235504686832428,
|
| 289 |
+
"kl": 0.004962539672851563,
|
| 290 |
+
"learning_rate": 2.8616870839955444e-06,
|
| 291 |
+
"loss": 0.0002,
|
| 292 |
+
"reward": 0.7964286103844642,
|
| 293 |
+
"reward_std": 0.26562733463943006,
|
| 294 |
+
"rewards/accuracy_reward": 0.7964286103844642,
|
| 295 |
+
"rewards/format_reward": 0.0,
|
| 296 |
+
"step": 105
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"completion_length": 581.1750267028808,
|
| 300 |
+
"epoch": 0.23466666666666666,
|
| 301 |
+
"grad_norm": 0.4279918968677521,
|
| 302 |
+
"kl": 0.005224609375,
|
| 303 |
+
"learning_rate": 2.837271198208662e-06,
|
| 304 |
+
"loss": 0.0002,
|
| 305 |
+
"reward": 0.7785714581608772,
|
| 306 |
+
"reward_std": 0.20086282528936864,
|
| 307 |
+
"rewards/accuracy_reward": 0.7785714581608772,
|
| 308 |
+
"rewards/format_reward": 0.0,
|
| 309 |
+
"step": 110
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"completion_length": 532.0053810119629,
|
| 313 |
+
"epoch": 0.24533333333333332,
|
| 314 |
+
"grad_norm": 0.6788883805274963,
|
| 315 |
+
"kl": 0.0057430267333984375,
|
| 316 |
+
"learning_rate": 2.8109938911593322e-06,
|
| 317 |
+
"loss": 0.0002,
|
| 318 |
+
"reward": 0.7767857426777482,
|
| 319 |
+
"reward_std": 0.20144498273730277,
|
| 320 |
+
"rewards/accuracy_reward": 0.7767857426777482,
|
| 321 |
+
"rewards/format_reward": 0.0,
|
| 322 |
+
"step": 115
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"completion_length": 544.6714546203614,
|
| 326 |
+
"epoch": 0.256,
|
| 327 |
+
"grad_norm": 0.23265813291072845,
|
| 328 |
+
"kl": 0.006147003173828125,
|
| 329 |
+
"learning_rate": 2.7828917396751474e-06,
|
| 330 |
+
"loss": 0.0002,
|
| 331 |
+
"reward": 0.7696428894996643,
|
| 332 |
+
"reward_std": 0.20026272870600223,
|
| 333 |
+
"rewards/accuracy_reward": 0.7696428894996643,
|
| 334 |
+
"rewards/format_reward": 0.0,
|
| 335 |
+
"step": 120
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"completion_length": 518.471452331543,
|
| 339 |
+
"epoch": 0.26666666666666666,
|
| 340 |
+
"grad_norm": 0.41492336988449097,
|
| 341 |
+
"kl": 0.00635986328125,
|
| 342 |
+
"learning_rate": 2.753003860684943e-06,
|
| 343 |
+
"loss": 0.0003,
|
| 344 |
+
"reward": 0.8375000298023224,
|
| 345 |
+
"reward_std": 0.20442306995391846,
|
| 346 |
+
"rewards/accuracy_reward": 0.8375000298023224,
|
| 347 |
+
"rewards/format_reward": 0.0,
|
| 348 |
+
"step": 125
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"completion_length": 570.2303848266602,
|
| 352 |
+
"epoch": 0.2773333333333333,
|
| 353 |
+
"grad_norm": 0.4288278818130493,
|
| 354 |
+
"kl": 0.006764602661132812,
|
| 355 |
+
"learning_rate": 2.721371856769793e-06,
|
| 356 |
+
"loss": 0.0003,
|
| 357 |
+
"reward": 0.7160714600235224,
|
| 358 |
+
"reward_std": 0.2673064887523651,
|
| 359 |
+
"rewards/accuracy_reward": 0.7160714600235224,
|
| 360 |
+
"rewards/format_reward": 0.0,
|
| 361 |
+
"step": 130
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"completion_length": 494.16431350708007,
|
| 365 |
+
"epoch": 0.288,
|
| 366 |
+
"grad_norm": 0.5349715352058411,
|
| 367 |
+
"kl": 0.008311080932617187,
|
| 368 |
+
"learning_rate": 2.688039758254093e-06,
|
| 369 |
+
"loss": 0.0003,
|
| 370 |
+
"reward": 0.7910714618861675,
|
| 371 |
+
"reward_std": 0.22742781266570092,
|
| 372 |
+
"rewards/accuracy_reward": 0.7910714618861675,
|
| 373 |
+
"rewards/format_reward": 0.0,
|
| 374 |
+
"step": 135
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"completion_length": 503.5232357025146,
|
| 378 |
+
"epoch": 0.2986666666666667,
|
| 379 |
+
"grad_norm": 0.9469221830368042,
|
| 380 |
+
"kl": 0.011474609375,
|
| 381 |
+
"learning_rate": 2.65305396191733e-06,
|
| 382 |
+
"loss": 0.0005,
|
| 383 |
+
"reward": 0.8053571686148644,
|
| 384 |
+
"reward_std": 0.22859212197363377,
|
| 385 |
+
"rewards/accuracy_reward": 0.8053571686148644,
|
| 386 |
+
"rewards/format_reward": 0.0,
|
| 387 |
+
"step": 140
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"completion_length": 526.3750251770019,
|
| 391 |
+
"epoch": 0.30933333333333335,
|
| 392 |
+
"grad_norm": 0.4655410945415497,
|
| 393 |
+
"kl": 0.016347885131835938,
|
| 394 |
+
"learning_rate": 2.61646316641186e-06,
|
| 395 |
+
"loss": 0.0007,
|
| 396 |
+
"reward": 0.7767857382073998,
|
| 397 |
+
"reward_std": 0.21031193807721138,
|
| 398 |
+
"rewards/accuracy_reward": 0.7767857382073998,
|
| 399 |
+
"rewards/format_reward": 0.0,
|
| 400 |
+
"step": 145
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"completion_length": 536.4660957336425,
|
| 404 |
+
"epoch": 0.32,
|
| 405 |
+
"grad_norm": 0.36182740330696106,
|
| 406 |
+
"kl": 0.020062255859375,
|
| 407 |
+
"learning_rate": 2.5783183044765715e-06,
|
| 408 |
+
"loss": 0.0008,
|
| 409 |
+
"reward": 0.7517857436090708,
|
| 410 |
+
"reward_std": 0.2297923181205988,
|
| 411 |
+
"rewards/accuracy_reward": 0.7517857436090708,
|
| 412 |
+
"rewards/format_reward": 0.0,
|
| 413 |
+
"step": 150
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"completion_length": 514.8803802490235,
|
| 417 |
+
"epoch": 0.33066666666666666,
|
| 418 |
+
"grad_norm": 0.7956529855728149,
|
| 419 |
+
"kl": 0.03543167114257813,
|
| 420 |
+
"learning_rate": 2.5386724720408135e-06,
|
| 421 |
+
"loss": 0.0014,
|
| 422 |
+
"reward": 0.7517857443541288,
|
| 423 |
+
"reward_std": 0.26203566156327723,
|
| 424 |
+
"rewards/accuracy_reward": 0.7517857443541288,
|
| 425 |
+
"rewards/format_reward": 0.0,
|
| 426 |
+
"step": 155
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"completion_length": 549.3571701049805,
|
| 430 |
+
"epoch": 0.3413333333333333,
|
| 431 |
+
"grad_norm": 0.29695257544517517,
|
| 432 |
+
"kl": 0.062085723876953124,
|
| 433 |
+
"learning_rate": 2.49758085431725e-06,
|
| 434 |
+
"loss": 0.0025,
|
| 435 |
+
"reward": 0.7285714589059353,
|
| 436 |
+
"reward_std": 0.269120267406106,
|
| 437 |
+
"rewards/accuracy_reward": 0.7285714589059353,
|
| 438 |
+
"rewards/format_reward": 0.0,
|
| 439 |
+
"step": 160
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"completion_length": 589.5339569091797,
|
| 443 |
+
"epoch": 0.352,
|
| 444 |
+
"grad_norm": 0.41898104548454285,
|
| 445 |
+
"kl": 0.13505706787109376,
|
| 446 |
+
"learning_rate": 2.455100648986533e-06,
|
| 447 |
+
"loss": 0.0054,
|
| 448 |
+
"reward": 0.6857143165543675,
|
| 449 |
+
"reward_std": 0.32202624566853044,
|
| 450 |
+
"rewards/accuracy_reward": 0.6857143165543675,
|
| 451 |
+
"rewards/format_reward": 0.0,
|
| 452 |
+
"step": 165
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"completion_length": 627.4214576721191,
|
| 456 |
+
"epoch": 0.3626666666666667,
|
| 457 |
+
"grad_norm": 0.2985088527202606,
|
| 458 |
+
"kl": 0.1629364013671875,
|
| 459 |
+
"learning_rate": 2.4112909865807053e-06,
|
| 460 |
+
"loss": 0.0065,
|
| 461 |
+
"reward": 0.6410714607685805,
|
| 462 |
+
"reward_std": 0.2777946576476097,
|
| 463 |
+
"rewards/accuracy_reward": 0.6410714607685805,
|
| 464 |
+
"rewards/format_reward": 0.0,
|
| 465 |
+
"step": 170
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"completion_length": 567.5768112182617,
|
| 469 |
+
"epoch": 0.37333333333333335,
|
| 470 |
+
"grad_norm": 0.49856701493263245,
|
| 471 |
+
"kl": 0.115283203125,
|
| 472 |
+
"learning_rate": 2.366212848176164e-06,
|
| 473 |
+
"loss": 0.0046,
|
| 474 |
+
"reward": 0.7089286031201482,
|
| 475 |
+
"reward_std": 0.2502220422029495,
|
| 476 |
+
"rewards/accuracy_reward": 0.7089286031201482,
|
| 477 |
+
"rewards/format_reward": 0.0,
|
| 478 |
+
"step": 175
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"completion_length": 528.6750221252441,
|
| 482 |
+
"epoch": 0.384,
|
| 483 |
+
"grad_norm": 0.6558699607849121,
|
| 484 |
+
"kl": 0.18359222412109374,
|
| 485 |
+
"learning_rate": 2.319928980510752e-06,
|
| 486 |
+
"loss": 0.0073,
|
| 487 |
+
"reward": 0.6535714527592063,
|
| 488 |
+
"reward_std": 0.28206231258809566,
|
| 489 |
+
"rewards/accuracy_reward": 0.6535714527592063,
|
| 490 |
+
"rewards/format_reward": 0.0,
|
| 491 |
+
"step": 180
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"completion_length": 571.7518112182618,
|
| 495 |
+
"epoch": 0.39466666666666667,
|
| 496 |
+
"grad_norm": 6.485378742218018,
|
| 497 |
+
"kl": 0.34942626953125,
|
| 498 |
+
"learning_rate": 2.272503808643123e-06,
|
| 499 |
+
"loss": 0.014,
|
| 500 |
+
"reward": 0.4696428783237934,
|
| 501 |
+
"reward_std": 0.27855144031345846,
|
| 502 |
+
"rewards/accuracy_reward": 0.4696428783237934,
|
| 503 |
+
"rewards/format_reward": 0.0,
|
| 504 |
+
"step": 185
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"completion_length": 542.8089500427246,
|
| 508 |
+
"epoch": 0.4053333333333333,
|
| 509 |
+
"grad_norm": 1.8310225009918213,
|
| 510 |
+
"kl": 0.4147308349609375,
|
| 511 |
+
"learning_rate": 2.2240033462759628e-06,
|
| 512 |
+
"loss": 0.0166,
|
| 513 |
+
"reward": 0.46071431171149013,
|
| 514 |
+
"reward_std": 0.36014051400125024,
|
| 515 |
+
"rewards/accuracy_reward": 0.46071431171149013,
|
| 516 |
+
"rewards/format_reward": 0.0,
|
| 517 |
+
"step": 190
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"completion_length": 626.9553817749023,
|
| 521 |
+
"epoch": 0.416,
|
| 522 |
+
"grad_norm": 17.148595809936523,
|
| 523 |
+
"kl": 0.4925048828125,
|
| 524 |
+
"learning_rate": 2.1744951038678905e-06,
|
| 525 |
+
"loss": 0.0197,
|
| 526 |
+
"reward": 0.4000000203028321,
|
| 527 |
+
"reward_std": 0.3870592150837183,
|
| 528 |
+
"rewards/accuracy_reward": 0.4000000203028321,
|
| 529 |
+
"rewards/format_reward": 0.0,
|
| 530 |
+
"step": 195
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"completion_length": 525.5928825378418,
|
| 534 |
+
"epoch": 0.4266666666666667,
|
| 535 |
+
"grad_norm": 38.49786376953125,
|
| 536 |
+
"kl": 0.23444061279296874,
|
| 537 |
+
"learning_rate": 2.124047994661941e-06,
|
| 538 |
+
"loss": 0.0094,
|
| 539 |
+
"reward": 0.6821428872644901,
|
| 540 |
+
"reward_std": 0.3179911646991968,
|
| 541 |
+
"rewards/accuracy_reward": 0.6821428872644901,
|
| 542 |
+
"rewards/format_reward": 0.0,
|
| 543 |
+
"step": 200
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.4266666666666667,
|
| 547 |
+
"eval_completion_length": 528.541595147705,
|
| 548 |
+
"eval_kl": 0.42865986328125,
|
| 549 |
+
"eval_loss": 0.017097920179367065,
|
| 550 |
+
"eval_reward": 0.662742884466052,
|
| 551 |
+
"eval_reward_std": 0.2557599967300892,
|
| 552 |
+
"eval_rewards/accuracy_reward": 0.662742884466052,
|
| 553 |
+
"eval_rewards/format_reward": 0.0,
|
| 554 |
+
"eval_runtime": 28459.3155,
|
| 555 |
+
"eval_samples_per_second": 0.176,
|
| 556 |
+
"eval_steps_per_second": 0.013,
|
| 557 |
+
"step": 200
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"completion_length": 509.92859649658203,
|
| 561 |
+
"epoch": 0.43733333333333335,
|
| 562 |
+
"grad_norm": 4.442358493804932,
|
| 563 |
+
"kl": 0.09366302490234375,
|
| 564 |
+
"learning_rate": 2.072732238761434e-06,
|
| 565 |
+
"loss": 0.0037,
|
| 566 |
+
"reward": 0.7660714574158192,
|
| 567 |
+
"reward_std": 0.21246148198843,
|
| 568 |
+
"rewards/accuracy_reward": 0.7660714574158192,
|
| 569 |
+
"rewards/format_reward": 0.0,
|
| 570 |
+
"step": 205
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"completion_length": 497.5357364654541,
|
| 574 |
+
"epoch": 0.448,
|
| 575 |
+
"grad_norm": 0.5748523473739624,
|
| 576 |
+
"kl": 0.085107421875,
|
| 577 |
+
"learning_rate": 2.0206192653867536e-06,
|
| 578 |
+
"loss": 0.0034,
|
| 579 |
+
"reward": 0.796428595483303,
|
| 580 |
+
"reward_std": 0.18161089681088924,
|
| 581 |
+
"rewards/accuracy_reward": 0.796428595483303,
|
| 582 |
+
"rewards/format_reward": 0.0,
|
| 583 |
+
"step": 210
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"completion_length": 614.8643173217773,
|
| 587 |
+
"epoch": 0.45866666666666667,
|
| 588 |
+
"grad_norm": 0.3233760893344879,
|
| 589 |
+
"kl": 0.15386199951171875,
|
| 590 |
+
"learning_rate": 1.967781613449095e-06,
|
| 591 |
+
"loss": 0.0062,
|
| 592 |
+
"reward": 0.6392857398837805,
|
| 593 |
+
"reward_std": 0.26852017305791376,
|
| 594 |
+
"rewards/accuracy_reward": 0.6392857398837805,
|
| 595 |
+
"rewards/format_reward": 0.0,
|
| 596 |
+
"step": 215
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"completion_length": 575.8321681976319,
|
| 600 |
+
"epoch": 0.4693333333333333,
|
| 601 |
+
"grad_norm": 0.6134320497512817,
|
| 602 |
+
"kl": 0.16066131591796876,
|
| 603 |
+
"learning_rate": 1.9142928305795637e-06,
|
| 604 |
+
"loss": 0.0064,
|
| 605 |
+
"reward": 0.6446428839117289,
|
| 606 |
+
"reward_std": 0.3076914418488741,
|
| 607 |
+
"rewards/accuracy_reward": 0.6446428839117289,
|
| 608 |
+
"rewards/format_reward": 0.0,
|
| 609 |
+
"step": 220
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"completion_length": 529.201806640625,
|
| 613 |
+
"epoch": 0.48,
|
| 614 |
+
"grad_norm": 14.935003280639648,
|
| 615 |
+
"kl": 0.27866592407226565,
|
| 616 |
+
"learning_rate": 1.8602273707541886e-06,
|
| 617 |
+
"loss": 0.0111,
|
| 618 |
+
"reward": 0.6982143182307482,
|
| 619 |
+
"reward_std": 0.29295355789363386,
|
| 620 |
+
"rewards/accuracy_reward": 0.6982143182307482,
|
| 621 |
+
"rewards/format_reward": 0.0,
|
| 622 |
+
"step": 225
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"completion_length": 599.1643127441406,
|
| 626 |
+
"epoch": 0.49066666666666664,
|
| 627 |
+
"grad_norm": 29.34033203125,
|
| 628 |
+
"kl": 1.2954010009765624,
|
| 629 |
+
"learning_rate": 1.8056604906573418e-06,
|
| 630 |
+
"loss": 0.0518,
|
| 631 |
+
"reward": 0.605357170291245,
|
| 632 |
+
"reward_std": 0.30737774074077606,
|
| 633 |
+
"rewards/accuracy_reward": 0.605357170291245,
|
| 634 |
+
"rewards/format_reward": 0.0,
|
| 635 |
+
"step": 230
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"completion_length": 577.0696716308594,
|
| 639 |
+
"epoch": 0.5013333333333333,
|
| 640 |
+
"grad_norm": 33.51802062988281,
|
| 641 |
+
"kl": 0.703057861328125,
|
| 642 |
+
"learning_rate": 1.7506681449278226e-06,
|
| 643 |
+
"loss": 0.0281,
|
| 644 |
+
"reward": 0.571428601257503,
|
| 645 |
+
"reward_std": 0.32992145605385303,
|
| 646 |
+
"rewards/accuracy_reward": 0.571428601257503,
|
| 647 |
+
"rewards/format_reward": 0.0,
|
| 648 |
+
"step": 235
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"completion_length": 501.30717010498046,
|
| 652 |
+
"epoch": 0.512,
|
| 653 |
+
"grad_norm": 7.524632930755615,
|
| 654 |
+
"kl": 0.3678741455078125,
|
| 655 |
+
"learning_rate": 1.6953268804334257e-06,
|
| 656 |
+
"loss": 0.0147,
|
| 657 |
+
"reward": 0.589285738952458,
|
| 658 |
+
"reward_std": 0.28248333670198916,
|
| 659 |
+
"rewards/accuracy_reward": 0.589285738952458,
|
| 660 |
+
"rewards/format_reward": 0.0,
|
| 661 |
+
"step": 240
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"completion_length": 462.3607376098633,
|
| 665 |
+
"epoch": 0.5226666666666666,
|
| 666 |
+
"grad_norm": 6.015294075012207,
|
| 667 |
+
"kl": 0.3237091064453125,
|
| 668 |
+
"learning_rate": 1.6397137297211436e-06,
|
| 669 |
+
"loss": 0.0129,
|
| 670 |
+
"reward": 0.5589285986497998,
|
| 671 |
+
"reward_std": 0.3269613076001406,
|
| 672 |
+
"rewards/accuracy_reward": 0.5589285986497998,
|
| 673 |
+
"rewards/format_reward": 0.0,
|
| 674 |
+
"step": 245
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"completion_length": 638.7928871154785,
|
| 678 |
+
"epoch": 0.5333333333333333,
|
| 679 |
+
"grad_norm": 1.8317821025848389,
|
| 680 |
+
"kl": 0.807421875,
|
| 681 |
+
"learning_rate": 1.5839061037913395e-06,
|
| 682 |
+
"loss": 0.0323,
|
| 683 |
+
"reward": 0.3250000160187483,
|
| 684 |
+
"reward_std": 0.34638786166906355,
|
| 685 |
+
"rewards/accuracy_reward": 0.3250000160187483,
|
| 686 |
+
"rewards/format_reward": 0.0,
|
| 687 |
+
"step": 250
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"completion_length": 636.7161018371582,
|
| 691 |
+
"epoch": 0.544,
|
| 692 |
+
"grad_norm": 0.6774631142616272,
|
| 693 |
+
"kl": 0.53349609375,
|
| 694 |
+
"learning_rate": 1.527981684345115e-06,
|
| 695 |
+
"loss": 0.0213,
|
| 696 |
+
"reward": 0.35714287366718056,
|
| 697 |
+
"reward_std": 0.3179732210934162,
|
| 698 |
+
"rewards/accuracy_reward": 0.35714287366718056,
|
| 699 |
+
"rewards/format_reward": 0.0,
|
| 700 |
+
"step": 255
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"completion_length": 558.9875259399414,
|
| 704 |
+
"epoch": 0.5546666666666666,
|
| 705 |
+
"grad_norm": 2.738410472869873,
|
| 706 |
+
"kl": 0.205511474609375,
|
| 707 |
+
"learning_rate": 1.4720183156548855e-06,
|
| 708 |
+
"loss": 0.0082,
|
| 709 |
+
"reward": 0.6000000305473805,
|
| 710 |
+
"reward_std": 0.3251969013363123,
|
| 711 |
+
"rewards/accuracy_reward": 0.6000000305473805,
|
| 712 |
+
"rewards/format_reward": 0.0,
|
| 713 |
+
"step": 260
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"completion_length": 545.3893096923828,
|
| 717 |
+
"epoch": 0.5653333333333334,
|
| 718 |
+
"grad_norm": 17.57588005065918,
|
| 719 |
+
"kl": 1.5068023681640625,
|
| 720 |
+
"learning_rate": 1.4160938962086612e-06,
|
| 721 |
+
"loss": 0.0603,
|
| 722 |
+
"reward": 0.6089285997673869,
|
| 723 |
+
"reward_std": 0.28422979824244976,
|
| 724 |
+
"rewards/accuracy_reward": 0.6089285997673869,
|
| 725 |
+
"rewards/format_reward": 0.0,
|
| 726 |
+
"step": 265
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"completion_length": 526.2607391357421,
|
| 730 |
+
"epoch": 0.576,
|
| 731 |
+
"grad_norm": 38.554412841796875,
|
| 732 |
+
"kl": 1.0892745971679687,
|
| 733 |
+
"learning_rate": 1.3602862702788567e-06,
|
| 734 |
+
"loss": 0.0436,
|
| 735 |
+
"reward": 0.6267857417464257,
|
| 736 |
+
"reward_std": 0.32045440524816515,
|
| 737 |
+
"rewards/accuracy_reward": 0.6267857417464257,
|
| 738 |
+
"rewards/format_reward": 0.0,
|
| 739 |
+
"step": 270
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"completion_length": 517.2678787231446,
|
| 743 |
+
"epoch": 0.5866666666666667,
|
| 744 |
+
"grad_norm": 4.718568325042725,
|
| 745 |
+
"kl": 1.1258697509765625,
|
| 746 |
+
"learning_rate": 1.3046731195665748e-06,
|
| 747 |
+
"loss": 0.045,
|
| 748 |
+
"reward": 0.6464285977184773,
|
| 749 |
+
"reward_std": 0.320261836796999,
|
| 750 |
+
"rewards/accuracy_reward": 0.6464285977184773,
|
| 751 |
+
"rewards/format_reward": 0.0,
|
| 752 |
+
"step": 275
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"completion_length": 514.8375244140625,
|
| 756 |
+
"epoch": 0.5973333333333334,
|
| 757 |
+
"grad_norm": 4.270685195922852,
|
| 758 |
+
"kl": 1.4042236328125,
|
| 759 |
+
"learning_rate": 1.2493318550721775e-06,
|
| 760 |
+
"loss": 0.0562,
|
| 761 |
+
"reward": 0.6982143145054579,
|
| 762 |
+
"reward_std": 0.2838581532239914,
|
| 763 |
+
"rewards/accuracy_reward": 0.6982143145054579,
|
| 764 |
+
"rewards/format_reward": 0.0,
|
| 765 |
+
"step": 280
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"completion_length": 534.7946655273438,
|
| 769 |
+
"epoch": 0.608,
|
| 770 |
+
"grad_norm": 9.498967170715332,
|
| 771 |
+
"kl": 0.8575332641601563,
|
| 772 |
+
"learning_rate": 1.1943395093426585e-06,
|
| 773 |
+
"loss": 0.0343,
|
| 774 |
+
"reward": 0.6910714596509934,
|
| 775 |
+
"reward_std": 0.2867689304053783,
|
| 776 |
+
"rewards/accuracy_reward": 0.6910714596509934,
|
| 777 |
+
"rewards/format_reward": 0.0,
|
| 778 |
+
"step": 285
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"completion_length": 529.5875297546387,
|
| 782 |
+
"epoch": 0.6186666666666667,
|
| 783 |
+
"grad_norm": 8.131922721862793,
|
| 784 |
+
"kl": 0.6070526123046875,
|
| 785 |
+
"learning_rate": 1.1397726292458115e-06,
|
| 786 |
+
"loss": 0.0243,
|
| 787 |
+
"reward": 0.675000030733645,
|
| 788 |
+
"reward_std": 0.2807727467268705,
|
| 789 |
+
"rewards/accuracy_reward": 0.675000030733645,
|
| 790 |
+
"rewards/format_reward": 0.0,
|
| 791 |
+
"step": 290
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"completion_length": 557.6500236511231,
|
| 795 |
+
"epoch": 0.6293333333333333,
|
| 796 |
+
"grad_norm": 2.224785089492798,
|
| 797 |
+
"kl": 1.1581512451171876,
|
| 798 |
+
"learning_rate": 1.085707169420437e-06,
|
| 799 |
+
"loss": 0.0463,
|
| 800 |
+
"reward": 0.6142857430502773,
|
| 801 |
+
"reward_std": 0.2811264578253031,
|
| 802 |
+
"rewards/accuracy_reward": 0.6142857430502773,
|
| 803 |
+
"rewards/format_reward": 0.0,
|
| 804 |
+
"step": 295
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"completion_length": 529.6071685791015,
|
| 808 |
+
"epoch": 0.64,
|
| 809 |
+
"grad_norm": 8.0582857131958,
|
| 810 |
+
"kl": 1.2655914306640625,
|
| 811 |
+
"learning_rate": 1.0322183865509054e-06,
|
| 812 |
+
"loss": 0.0506,
|
| 813 |
+
"reward": 0.6750000296160579,
|
| 814 |
+
"reward_std": 0.28525091484189036,
|
| 815 |
+
"rewards/accuracy_reward": 0.6750000296160579,
|
| 816 |
+
"rewards/format_reward": 0.0,
|
| 817 |
+
"step": 300
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.64,
|
| 821 |
+
"eval_completion_length": 562.8381398986817,
|
| 822 |
+
"eval_kl": 0.95684150390625,
|
| 823 |
+
"eval_loss": 0.03832858428359032,
|
| 824 |
+
"eval_reward": 0.5700857400953769,
|
| 825 |
+
"eval_reward_std": 0.30191256090998647,
|
| 826 |
+
"eval_rewards/accuracy_reward": 0.5700857400953769,
|
| 827 |
+
"eval_rewards/format_reward": 0.0,
|
| 828 |
+
"eval_runtime": 30482.0361,
|
| 829 |
+
"eval_samples_per_second": 0.164,
|
| 830 |
+
"eval_steps_per_second": 0.012,
|
| 831 |
+
"step": 300
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"completion_length": 557.5018104553222,
|
| 835 |
+
"epoch": 0.6506666666666666,
|
| 836 |
+
"grad_norm": 2.934140682220459,
|
| 837 |
+
"kl": 0.6237777709960938,
|
| 838 |
+
"learning_rate": 9.793807346132464e-07,
|
| 839 |
+
"loss": 0.025,
|
| 840 |
+
"reward": 0.6339285995811224,
|
| 841 |
+
"reward_std": 0.3295004416257143,
|
| 842 |
+
"rewards/accuracy_reward": 0.6339285995811224,
|
| 843 |
+
"rewards/format_reward": 0.0,
|
| 844 |
+
"step": 305
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"completion_length": 507.2857383728027,
|
| 848 |
+
"epoch": 0.6613333333333333,
|
| 849 |
+
"grad_norm": 1.5333462953567505,
|
| 850 |
+
"kl": 0.18207855224609376,
|
| 851 |
+
"learning_rate": 9.272677612385667e-07,
|
| 852 |
+
"loss": 0.0073,
|
| 853 |
+
"reward": 0.6017857391387225,
|
| 854 |
+
"reward_std": 0.2927430454641581,
|
| 855 |
+
"rewards/accuracy_reward": 0.6017857391387225,
|
| 856 |
+
"rewards/format_reward": 0.0,
|
| 857 |
+
"step": 310
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"completion_length": 535.4660949707031,
|
| 861 |
+
"epoch": 0.672,
|
| 862 |
+
"grad_norm": 0.8268552422523499,
|
| 863 |
+
"kl": 0.2217437744140625,
|
| 864 |
+
"learning_rate": 8.759520053380591e-07,
|
| 865 |
+
"loss": 0.0089,
|
| 866 |
+
"reward": 0.5303571715950965,
|
| 867 |
+
"reward_std": 0.30413119196891786,
|
| 868 |
+
"rewards/accuracy_reward": 0.5303571715950965,
|
| 869 |
+
"rewards/format_reward": 0.0,
|
| 870 |
+
"step": 315
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"completion_length": 587.7893112182617,
|
| 874 |
+
"epoch": 0.6826666666666666,
|
| 875 |
+
"grad_norm": 0.8802245259284973,
|
| 876 |
+
"kl": 0.28748779296875,
|
| 877 |
+
"learning_rate": 8.255048961321088e-07,
|
| 878 |
+
"loss": 0.0115,
|
| 879 |
+
"reward": 0.5250000260770321,
|
| 880 |
+
"reward_std": 0.32775397710502147,
|
| 881 |
+
"rewards/accuracy_reward": 0.5250000260770321,
|
| 882 |
+
"rewards/format_reward": 0.0,
|
| 883 |
+
"step": 320
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"completion_length": 568.7143135070801,
|
| 887 |
+
"epoch": 0.6933333333333334,
|
| 888 |
+
"grad_norm": 0.42495808005332947,
|
| 889 |
+
"kl": 0.20731658935546876,
|
| 890 |
+
"learning_rate": 7.759966537240373e-07,
|
| 891 |
+
"loss": 0.0083,
|
| 892 |
+
"reward": 0.6357143180444836,
|
| 893 |
+
"reward_std": 0.32221881337463854,
|
| 894 |
+
"rewards/accuracy_reward": 0.6357143180444836,
|
| 895 |
+
"rewards/format_reward": 0.0,
|
| 896 |
+
"step": 325
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"completion_length": 572.9053848266601,
|
| 900 |
+
"epoch": 0.704,
|
| 901 |
+
"grad_norm": 0.8112100958824158,
|
| 902 |
+
"kl": 0.2259490966796875,
|
| 903 |
+
"learning_rate": 7.274961913568773e-07,
|
| 904 |
+
"loss": 0.009,
|
| 905 |
+
"reward": 0.6071428848430515,
|
| 906 |
+
"reward_std": 0.28955445289611814,
|
| 907 |
+
"rewards/accuracy_reward": 0.6071428848430515,
|
| 908 |
+
"rewards/format_reward": 0.0,
|
| 909 |
+
"step": 330
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"completion_length": 585.521450805664,
|
| 913 |
+
"epoch": 0.7146666666666667,
|
| 914 |
+
"grad_norm": 2.5413522720336914,
|
| 915 |
+
"kl": 0.1725799560546875,
|
| 916 |
+
"learning_rate": 6.800710194892484e-07,
|
| 917 |
+
"loss": 0.0069,
|
| 918 |
+
"reward": 0.6946428865194321,
|
| 919 |
+
"reward_std": 0.27458811886608603,
|
| 920 |
+
"rewards/accuracy_reward": 0.6946428865194321,
|
| 921 |
+
"rewards/format_reward": 0.0,
|
| 922 |
+
"step": 335
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"completion_length": 578.700025177002,
|
| 926 |
+
"epoch": 0.7253333333333334,
|
| 927 |
+
"grad_norm": 2.077190637588501,
|
| 928 |
+
"kl": 0.23962783813476562,
|
| 929 |
+
"learning_rate": 6.33787151823836e-07,
|
| 930 |
+
"loss": 0.0096,
|
| 931 |
+
"reward": 0.6517857488244772,
|
| 932 |
+
"reward_std": 0.29764223508536813,
|
| 933 |
+
"rewards/accuracy_reward": 0.6517857488244772,
|
| 934 |
+
"rewards/format_reward": 0.0,
|
| 935 |
+
"step": 340
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"completion_length": 517.0357376098633,
|
| 939 |
+
"epoch": 0.736,
|
| 940 |
+
"grad_norm": 1.8005852699279785,
|
| 941 |
+
"kl": 0.1731414794921875,
|
| 942 |
+
"learning_rate": 5.887090134192947e-07,
|
| 943 |
+
"loss": 0.0069,
|
| 944 |
+
"reward": 0.7303571674972773,
|
| 945 |
+
"reward_std": 0.20812651440501212,
|
| 946 |
+
"rewards/accuracy_reward": 0.7303571674972773,
|
| 947 |
+
"rewards/format_reward": 0.0,
|
| 948 |
+
"step": 345
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"completion_length": 537.6071670532226,
|
| 952 |
+
"epoch": 0.7466666666666667,
|
| 953 |
+
"grad_norm": 5.837470531463623,
|
| 954 |
+
"kl": 0.22538909912109376,
|
| 955 |
+
"learning_rate": 5.448993510134669e-07,
|
| 956 |
+
"loss": 0.009,
|
| 957 |
+
"reward": 0.7285714605823159,
|
| 958 |
+
"reward_std": 0.23610220104455948,
|
| 959 |
+
"rewards/accuracy_reward": 0.7285714605823159,
|
| 960 |
+
"rewards/format_reward": 0.0,
|
| 961 |
+
"step": 350
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"completion_length": 511.03573455810545,
|
| 965 |
+
"epoch": 0.7573333333333333,
|
| 966 |
+
"grad_norm": 10.607796669006348,
|
| 967 |
+
"kl": 0.2825325012207031,
|
| 968 |
+
"learning_rate": 5.024191456827498e-07,
|
| 969 |
+
"loss": 0.0113,
|
| 970 |
+
"reward": 0.7357143178582192,
|
| 971 |
+
"reward_std": 0.24888310953974724,
|
| 972 |
+
"rewards/accuracy_reward": 0.7357143178582192,
|
| 973 |
+
"rewards/format_reward": 0.0,
|
| 974 |
+
"step": 355
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"completion_length": 528.4375274658203,
|
| 978 |
+
"epoch": 0.768,
|
| 979 |
+
"grad_norm": 17.805084228515625,
|
| 980 |
+
"kl": 0.409503173828125,
|
| 981 |
+
"learning_rate": 4.6132752795918667e-07,
|
| 982 |
+
"loss": 0.0164,
|
| 983 |
+
"reward": 0.6857143182307481,
|
| 984 |
+
"reward_std": 0.28699738159775734,
|
| 985 |
+
"rewards/accuracy_reward": 0.6857143182307481,
|
| 986 |
+
"rewards/format_reward": 0.0,
|
| 987 |
+
"step": 360
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"completion_length": 519.4303817749023,
|
| 991 |
+
"epoch": 0.7786666666666666,
|
| 992 |
+
"grad_norm": 7.785521507263184,
|
| 993 |
+
"kl": 0.34308319091796874,
|
| 994 |
+
"learning_rate": 4.2168169552342905e-07,
|
| 995 |
+
"loss": 0.0137,
|
| 996 |
+
"reward": 0.6928571717813611,
|
| 997 |
+
"reward_std": 0.29954983331263063,
|
| 998 |
+
"rewards/accuracy_reward": 0.6928571717813611,
|
| 999 |
+
"rewards/format_reward": 0.0,
|
| 1000 |
+
"step": 365
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"completion_length": 504.0107368469238,
|
| 1004 |
+
"epoch": 0.7893333333333333,
|
| 1005 |
+
"grad_norm": 4.727066516876221,
|
| 1006 |
+
"kl": 0.7092597961425782,
|
| 1007 |
+
"learning_rate": 3.8353683358814046e-07,
|
| 1008 |
+
"loss": 0.0285,
|
| 1009 |
+
"reward": 0.7000000275671482,
|
| 1010 |
+
"reward_std": 0.23943399637937546,
|
| 1011 |
+
"rewards/accuracy_reward": 0.7000000275671482,
|
| 1012 |
+
"rewards/format_reward": 0.0,
|
| 1013 |
+
"step": 370
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"completion_length": 517.7428833007813,
|
| 1017 |
+
"epoch": 0.8,
|
| 1018 |
+
"grad_norm": 7.284985542297363,
|
| 1019 |
+
"kl": 0.2571014404296875,
|
| 1020 |
+
"learning_rate": 3.469460380826697e-07,
|
| 1021 |
+
"loss": 0.0103,
|
| 1022 |
+
"reward": 0.673214310593903,
|
| 1023 |
+
"reward_std": 0.2639926388859749,
|
| 1024 |
+
"rewards/accuracy_reward": 0.673214310593903,
|
| 1025 |
+
"rewards/format_reward": 0.0,
|
| 1026 |
+
"step": 375
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"completion_length": 543.2714515686035,
|
| 1030 |
+
"epoch": 0.8106666666666666,
|
| 1031 |
+
"grad_norm": 14.041029930114746,
|
| 1032 |
+
"kl": 0.28242645263671873,
|
| 1033 |
+
"learning_rate": 3.119602417459075e-07,
|
| 1034 |
+
"loss": 0.0113,
|
| 1035 |
+
"reward": 0.6732143096625804,
|
| 1036 |
+
"reward_std": 0.25163274370133876,
|
| 1037 |
+
"rewards/accuracy_reward": 0.6732143096625804,
|
| 1038 |
+
"rewards/format_reward": 0.0,
|
| 1039 |
+
"step": 380
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"completion_length": 546.032169342041,
|
| 1043 |
+
"epoch": 0.8213333333333334,
|
| 1044 |
+
"grad_norm": 15.783819198608398,
|
| 1045 |
+
"kl": 0.20561065673828124,
|
| 1046 |
+
"learning_rate": 2.786281432302071e-07,
|
| 1047 |
+
"loss": 0.0082,
|
| 1048 |
+
"reward": 0.7339286010712385,
|
| 1049 |
+
"reward_std": 0.234727381169796,
|
| 1050 |
+
"rewards/accuracy_reward": 0.7339286010712385,
|
| 1051 |
+
"rewards/format_reward": 0.0,
|
| 1052 |
+
"step": 385
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"completion_length": 546.478592300415,
|
| 1056 |
+
"epoch": 0.832,
|
| 1057 |
+
"grad_norm": 6.679286003112793,
|
| 1058 |
+
"kl": 0.18747406005859374,
|
| 1059 |
+
"learning_rate": 2.46996139315057e-07,
|
| 1060 |
+
"loss": 0.0075,
|
| 1061 |
+
"reward": 0.7375000283122063,
|
| 1062 |
+
"reward_std": 0.2588605497032404,
|
| 1063 |
+
"rewards/accuracy_reward": 0.7375000283122063,
|
| 1064 |
+
"rewards/format_reward": 0.0,
|
| 1065 |
+
"step": 390
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"completion_length": 531.3750255584716,
|
| 1069 |
+
"epoch": 0.8426666666666667,
|
| 1070 |
+
"grad_norm": 4.5977044105529785,
|
| 1071 |
+
"kl": 0.1937103271484375,
|
| 1072 |
+
"learning_rate": 2.1710826032485286e-07,
|
| 1073 |
+
"loss": 0.0077,
|
| 1074 |
+
"reward": 0.7357143165543676,
|
| 1075 |
+
"reward_std": 0.2378486678004265,
|
| 1076 |
+
"rewards/accuracy_reward": 0.7357143165543676,
|
| 1077 |
+
"rewards/format_reward": 0.0,
|
| 1078 |
+
"step": 395
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"completion_length": 595.6964569091797,
|
| 1082 |
+
"epoch": 0.8533333333333334,
|
| 1083 |
+
"grad_norm": 4.166606903076172,
|
| 1084 |
+
"kl": 0.2564910888671875,
|
| 1085 |
+
"learning_rate": 1.8900610884066817e-07,
|
| 1086 |
+
"loss": 0.0103,
|
| 1087 |
+
"reward": 0.6285714594647288,
|
| 1088 |
+
"reward_std": 0.2736343163996935,
|
| 1089 |
+
"rewards/accuracy_reward": 0.6285714594647288,
|
| 1090 |
+
"rewards/format_reward": 0.0,
|
| 1091 |
+
"step": 400
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 0.8533333333333334,
|
| 1095 |
+
"eval_completion_length": 569.8527975585938,
|
| 1096 |
+
"eval_kl": 0.20817554931640625,
|
| 1097 |
+
"eval_loss": 0.008296786807477474,
|
| 1098 |
+
"eval_reward": 0.6110285988628864,
|
| 1099 |
+
"eval_reward_std": 0.28674686477184297,
|
| 1100 |
+
"eval_rewards/accuracy_reward": 0.6110285988628864,
|
| 1101 |
+
"eval_rewards/format_reward": 0.0,
|
| 1102 |
+
"eval_runtime": 29873.0033,
|
| 1103 |
+
"eval_samples_per_second": 0.167,
|
| 1104 |
+
"eval_steps_per_second": 0.012,
|
| 1105 |
+
"step": 400
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"completion_length": 605.528596496582,
|
| 1109 |
+
"epoch": 0.864,
|
| 1110 |
+
"grad_norm": 49.79132843017578,
|
| 1111 |
+
"kl": 0.341680908203125,
|
| 1112 |
+
"learning_rate": 1.627288017913383e-07,
|
| 1113 |
+
"loss": 0.0137,
|
| 1114 |
+
"reward": 0.5571428839117288,
|
| 1115 |
+
"reward_std": 0.3719855587929487,
|
| 1116 |
+
"rewards/accuracy_reward": 0.5571428839117288,
|
| 1117 |
+
"rewards/format_reward": 0.0,
|
| 1118 |
+
"step": 405
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"completion_length": 637.5250267028808,
|
| 1122 |
+
"epoch": 0.8746666666666667,
|
| 1123 |
+
"grad_norm": 12.56246566772461,
|
| 1124 |
+
"kl": 0.39511566162109374,
|
| 1125 |
+
"learning_rate": 1.3831291600445573e-07,
|
| 1126 |
+
"loss": 0.0158,
|
| 1127 |
+
"reward": 0.5392857391387225,
|
| 1128 |
+
"reward_std": 0.3052775662392378,
|
| 1129 |
+
"rewards/accuracy_reward": 0.5392857391387225,
|
| 1130 |
+
"rewards/format_reward": 0.0,
|
| 1131 |
+
"step": 410
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"completion_length": 631.8607391357422,
|
| 1135 |
+
"epoch": 0.8853333333333333,
|
| 1136 |
+
"grad_norm": 6.395442962646484,
|
| 1137 |
+
"kl": 0.45283203125,
|
| 1138 |
+
"learning_rate": 1.1579243729307487e-07,
|
| 1139 |
+
"loss": 0.0181,
|
| 1140 |
+
"reward": 0.49107145331799984,
|
| 1141 |
+
"reward_std": 0.36503970213234427,
|
| 1142 |
+
"rewards/accuracy_reward": 0.49107145331799984,
|
| 1143 |
+
"rewards/format_reward": 0.0,
|
| 1144 |
+
"step": 415
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"completion_length": 611.7053848266602,
|
| 1148 |
+
"epoch": 0.896,
|
| 1149 |
+
"grad_norm": 1.6087334156036377,
|
| 1150 |
+
"kl": 0.3666015625,
|
| 1151 |
+
"learning_rate": 9.519871314899092e-08,
|
| 1152 |
+
"loss": 0.0147,
|
| 1153 |
+
"reward": 0.6000000327825546,
|
| 1154 |
+
"reward_std": 0.3550443138927221,
|
| 1155 |
+
"rewards/accuracy_reward": 0.6000000327825546,
|
| 1156 |
+
"rewards/format_reward": 0.0,
|
| 1157 |
+
"step": 420
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"completion_length": 613.4946708679199,
|
| 1161 |
+
"epoch": 0.9066666666666666,
|
| 1162 |
+
"grad_norm": 1.6784651279449463,
|
| 1163 |
+
"kl": 0.3438507080078125,
|
| 1164 |
+
"learning_rate": 7.656040910844358e-08,
|
| 1165 |
+
"loss": 0.0138,
|
| 1166 |
+
"reward": 0.5946428859606385,
|
| 1167 |
+
"reward_std": 0.3528768301010132,
|
| 1168 |
+
"rewards/accuracy_reward": 0.5946428859606385,
|
| 1169 |
+
"rewards/format_reward": 0.0,
|
| 1170 |
+
"step": 425
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"completion_length": 603.3143119812012,
|
| 1174 |
+
"epoch": 0.9173333333333333,
|
| 1175 |
+
"grad_norm": 4.043834209442139,
|
| 1176 |
+
"kl": 0.3444915771484375,
|
| 1177 |
+
"learning_rate": 5.990346885098235e-08,
|
| 1178 |
+
"loss": 0.0138,
|
| 1179 |
+
"reward": 0.5964285979047418,
|
| 1180 |
+
"reward_std": 0.3912374936044216,
|
| 1181 |
+
"rewards/accuracy_reward": 0.5964285979047418,
|
| 1182 |
+
"rewards/format_reward": 0.0,
|
| 1183 |
+
"step": 430
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"completion_length": 575.2339546203614,
|
| 1187 |
+
"epoch": 0.928,
|
| 1188 |
+
"grad_norm": 4.404404163360596,
|
| 1189 |
+
"kl": 0.30106658935546876,
|
| 1190 |
+
"learning_rate": 4.5251078087033493e-08,
|
| 1191 |
+
"loss": 0.012,
|
| 1192 |
+
"reward": 0.6696428902447223,
|
| 1193 |
+
"reward_std": 0.3243862982839346,
|
| 1194 |
+
"rewards/accuracy_reward": 0.6696428902447223,
|
| 1195 |
+
"rewards/format_reward": 0.0,
|
| 1196 |
+
"step": 435
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"completion_length": 582.6393119812012,
|
| 1200 |
+
"epoch": 0.9386666666666666,
|
| 1201 |
+
"grad_norm": 2.593308448791504,
|
| 1202 |
+
"kl": 0.383941650390625,
|
| 1203 |
+
"learning_rate": 3.262363228443427e-08,
|
| 1204 |
+
"loss": 0.0154,
|
| 1205 |
+
"reward": 0.6035714616999031,
|
| 1206 |
+
"reward_std": 0.29665699824690817,
|
| 1207 |
+
"rewards/accuracy_reward": 0.6035714616999031,
|
| 1208 |
+
"rewards/format_reward": 0.0,
|
| 1209 |
+
"step": 440
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"completion_length": 627.2125282287598,
|
| 1213 |
+
"epoch": 0.9493333333333334,
|
| 1214 |
+
"grad_norm": 5.177957057952881,
|
| 1215 |
+
"kl": 0.3490997314453125,
|
| 1216 |
+
"learning_rate": 2.2038708278862952e-08,
|
| 1217 |
+
"loss": 0.014,
|
| 1218 |
+
"reward": 0.5839285971596837,
|
| 1219 |
+
"reward_std": 0.30867667235434054,
|
| 1220 |
+
"rewards/accuracy_reward": 0.5839285971596837,
|
| 1221 |
+
"rewards/format_reward": 0.0,
|
| 1222 |
+
"step": 445
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"completion_length": 576.0964553833007,
|
| 1226 |
+
"epoch": 0.96,
|
| 1227 |
+
"grad_norm": 3.4815688133239746,
|
| 1228 |
+
"kl": 0.2956207275390625,
|
| 1229 |
+
"learning_rate": 1.3511039807673209e-08,
|
| 1230 |
+
"loss": 0.0118,
|
| 1231 |
+
"reward": 0.6589286010712385,
|
| 1232 |
+
"reward_std": 0.3096484154462814,
|
| 1233 |
+
"rewards/accuracy_reward": 0.6589286010712385,
|
| 1234 |
+
"rewards/format_reward": 0.0,
|
| 1235 |
+
"step": 450
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"completion_length": 531.355379486084,
|
| 1239 |
+
"epoch": 0.9706666666666667,
|
| 1240 |
+
"grad_norm": 2.63545298576355,
|
| 1241 |
+
"kl": 0.261767578125,
|
| 1242 |
+
"learning_rate": 7.0524970011963675e-09,
|
| 1243 |
+
"loss": 0.0105,
|
| 1244 |
+
"reward": 0.7125000340864063,
|
| 1245 |
+
"reward_std": 0.28212962336838243,
|
| 1246 |
+
"rewards/accuracy_reward": 0.7125000340864063,
|
| 1247 |
+
"rewards/format_reward": 0.0,
|
| 1248 |
+
"step": 455
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"completion_length": 569.5285957336425,
|
| 1252 |
+
"epoch": 0.9813333333333333,
|
| 1253 |
+
"grad_norm": 1.8794143199920654,
|
| 1254 |
+
"kl": 0.2968017578125,
|
| 1255 |
+
"learning_rate": 2.6720698600553595e-09,
|
| 1256 |
+
"loss": 0.0119,
|
| 1257 |
+
"reward": 0.6303571775555611,
|
| 1258 |
+
"reward_std": 0.3074988707900047,
|
| 1259 |
+
"rewards/accuracy_reward": 0.6303571775555611,
|
| 1260 |
+
"rewards/format_reward": 0.0,
|
| 1261 |
+
"step": 460
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"completion_length": 612.3214569091797,
|
| 1265 |
+
"epoch": 0.992,
|
| 1266 |
+
"grad_norm": 3.266709327697754,
|
| 1267 |
+
"kl": 0.39468994140625,
|
| 1268 |
+
"learning_rate": 3.7585574148779613e-10,
|
| 1269 |
+
"loss": 0.0158,
|
| 1270 |
+
"reward": 0.5517857383936644,
|
| 1271 |
+
"reward_std": 0.3217977944761515,
|
| 1272 |
+
"rewards/accuracy_reward": 0.5517857383936644,
|
| 1273 |
+
"rewards/format_reward": 0.0,
|
| 1274 |
+
"step": 465
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"completion_length": 626.7321701049805,
|
| 1278 |
+
"epoch": 0.9984,
|
| 1279 |
+
"kl": 0.32488250732421875,
|
| 1280 |
+
"reward": 0.604166692122817,
|
| 1281 |
+
"reward_std": 0.3102082473536332,
|
| 1282 |
+
"rewards/accuracy_reward": 0.604166692122817,
|
| 1283 |
+
"rewards/format_reward": 0.0,
|
| 1284 |
+
"step": 468,
|
| 1285 |
+
"total_flos": 0.0,
|
| 1286 |
+
"train_loss": 0.00286464851636153,
|
| 1287 |
+
"train_runtime": 7716.3929,
|
| 1288 |
+
"train_samples_per_second": 0.972,
|
| 1289 |
+
"train_steps_per_second": 0.061
|
| 1290 |
+
}
|
| 1291 |
+
],
|
| 1292 |
+
"logging_steps": 5,
|
| 1293 |
+
"max_steps": 468,
|
| 1294 |
+
"num_input_tokens_seen": 0,
|
| 1295 |
+
"num_train_epochs": 1,
|
| 1296 |
+
"save_steps": 100,
|
| 1297 |
+
"stateful_callbacks": {
|
| 1298 |
+
"TrainerControl": {
|
| 1299 |
+
"args": {
|
| 1300 |
+
"should_epoch_stop": false,
|
| 1301 |
+
"should_evaluate": false,
|
| 1302 |
+
"should_log": false,
|
| 1303 |
+
"should_save": true,
|
| 1304 |
+
"should_training_stop": true
|
| 1305 |
+
},
|
| 1306 |
+
"attributes": {}
|
| 1307 |
+
}
|
| 1308 |
+
},
|
| 1309 |
+
"total_flos": 0.0,
|
| 1310 |
+
"train_batch_size": 2,
|
| 1311 |
+
"trial_name": null,
|
| 1312 |
+
"trial_params": null
|
| 1313 |
+
}
|