Model save
Browse files- README.md +68 -0
- all_results.json +8 -0
- generation_config.json +9 -0
- train_results.json +8 -0
- trainer_state.json +1443 -0
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: DrGRPO
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- grpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for DrGRPO
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
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?"
|
| 23 |
+
generator = pipeline("text-generation", model="LLucass/DrGRPO", device="cuda")
|
| 24 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 25 |
+
print(output["generated_text"])
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
## Training procedure
|
| 29 |
+
|
| 30 |
+
[<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/lavatorywang-nus/openr1/runs/nwpc8je2)
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
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).
|
| 34 |
+
|
| 35 |
+
### Framework versions
|
| 36 |
+
|
| 37 |
+
- TRL: 0.18.0
|
| 38 |
+
- Transformers: 4.50.0
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.6.0
|
| 41 |
+
- Tokenizers: 0.21.1
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite GRPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@article{zhihong2024deepseekmath,
|
| 49 |
+
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
|
| 50 |
+
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},
|
| 51 |
+
year = 2024,
|
| 52 |
+
eprint = {arXiv:2402.03300},
|
| 53 |
+
}
|
| 54 |
+
|
| 55 |
+
```
|
| 56 |
+
|
| 57 |
+
Cite TRL as:
|
| 58 |
+
|
| 59 |
+
```bibtex
|
| 60 |
+
@misc{vonwerra2022trl,
|
| 61 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 62 |
+
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},
|
| 63 |
+
year = 2020,
|
| 64 |
+
journal = {GitHub repository},
|
| 65 |
+
publisher = {GitHub},
|
| 66 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 67 |
+
}
|
| 68 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.03577750400174409,
|
| 4 |
+
"train_runtime": 8397.3301,
|
| 5 |
+
"train_samples": 7000,
|
| 6 |
+
"train_samples_per_second": 1.143,
|
| 7 |
+
"train_steps_per_second": 0.012
|
| 8 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 151646,
|
| 4 |
+
"do_sample": true,
|
| 5 |
+
"eos_token_id": 151643,
|
| 6 |
+
"temperature": 0.6,
|
| 7 |
+
"top_p": 0.95,
|
| 8 |
+
"transformers_version": "4.50.0"
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.03577750400174409,
|
| 4 |
+
"train_runtime": 8397.3301,
|
| 5 |
+
"train_samples": 7000,
|
| 6 |
+
"train_samples_per_second": 1.143,
|
| 7 |
+
"train_steps_per_second": 0.012
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1443 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.11428571428571428,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 100,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"completion_length": 2699.791748046875,
|
| 14 |
+
"entropy": 0.36767578125,
|
| 15 |
+
"epoch": 0.001142857142857143,
|
| 16 |
+
"grad_norm": 0.10378081351518631,
|
| 17 |
+
"kl": 0.0,
|
| 18 |
+
"learning_rate": 1e-07,
|
| 19 |
+
"loss": 0.0593,
|
| 20 |
+
"reward": 0.7708333535119891,
|
| 21 |
+
"reward_std": 0.4629540964961052,
|
| 22 |
+
"rewards/accuracy_reward": 0.25000001303851604,
|
| 23 |
+
"rewards/format_reward": 0.5208333386108279,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"completion_length": 3127.3958435058594,
|
| 28 |
+
"entropy": 0.353515625,
|
| 29 |
+
"epoch": 0.002285714285714286,
|
| 30 |
+
"grad_norm": 0.07586322724819183,
|
| 31 |
+
"kl": 0.0,
|
| 32 |
+
"learning_rate": 2e-07,
|
| 33 |
+
"loss": 0.0439,
|
| 34 |
+
"reward": 0.6458333637565374,
|
| 35 |
+
"reward_std": 0.4249730706214905,
|
| 36 |
+
"rewards/accuracy_reward": 0.2812500102445483,
|
| 37 |
+
"rewards/format_reward": 0.3645833386108279,
|
| 38 |
+
"step": 2
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"completion_length": 3761.0208740234375,
|
| 42 |
+
"entropy": 0.4560546875,
|
| 43 |
+
"epoch": 0.0034285714285714284,
|
| 44 |
+
"grad_norm": 0.04515620321035385,
|
| 45 |
+
"kl": 4.029273986816406e-05,
|
| 46 |
+
"learning_rate": 3e-07,
|
| 47 |
+
"loss": 0.0269,
|
| 48 |
+
"reward": 0.1770833432674408,
|
| 49 |
+
"reward_std": 0.2535070404410362,
|
| 50 |
+
"rewards/accuracy_reward": 0.0625,
|
| 51 |
+
"rewards/format_reward": 0.1145833395421505,
|
| 52 |
+
"step": 3
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"completion_length": 2404.6146545410156,
|
| 56 |
+
"entropy": 0.4072265625,
|
| 57 |
+
"epoch": 0.004571428571428572,
|
| 58 |
+
"grad_norm": 0.09874315559864044,
|
| 59 |
+
"kl": 3.263354301452637e-05,
|
| 60 |
+
"learning_rate": 4e-07,
|
| 61 |
+
"loss": 0.0521,
|
| 62 |
+
"reward": 0.8229166716337204,
|
| 63 |
+
"reward_std": 0.48936954140663147,
|
| 64 |
+
"rewards/accuracy_reward": 0.16666667070239782,
|
| 65 |
+
"rewards/format_reward": 0.6562500149011612,
|
| 66 |
+
"step": 4
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"completion_length": 3513.5521850585938,
|
| 70 |
+
"entropy": 0.4580078125,
|
| 71 |
+
"epoch": 0.005714285714285714,
|
| 72 |
+
"grad_norm": 0.07631789147853851,
|
| 73 |
+
"kl": 4.3064355850219727e-05,
|
| 74 |
+
"learning_rate": 5e-07,
|
| 75 |
+
"loss": 0.04,
|
| 76 |
+
"reward": 0.5416666939854622,
|
| 77 |
+
"reward_std": 0.5841840505599976,
|
| 78 |
+
"rewards/accuracy_reward": 0.12500000558793545,
|
| 79 |
+
"rewards/format_reward": 0.4166666865348816,
|
| 80 |
+
"step": 5
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"completion_length": 3405.6251220703125,
|
| 84 |
+
"entropy": 0.45556640625,
|
| 85 |
+
"epoch": 0.006857142857142857,
|
| 86 |
+
"grad_norm": 0.06789246201515198,
|
| 87 |
+
"kl": 4.4912099838256836e-05,
|
| 88 |
+
"learning_rate": 6e-07,
|
| 89 |
+
"loss": 0.0183,
|
| 90 |
+
"reward": 0.38541666977107525,
|
| 91 |
+
"reward_std": 0.46006231755018234,
|
| 92 |
+
"rewards/accuracy_reward": 0.06250000186264515,
|
| 93 |
+
"rewards/format_reward": 0.32291668467223644,
|
| 94 |
+
"step": 6
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"completion_length": 3237.2813720703125,
|
| 98 |
+
"entropy": 0.380859375,
|
| 99 |
+
"epoch": 0.008,
|
| 100 |
+
"grad_norm": 0.0811946764588356,
|
| 101 |
+
"kl": 1.9431114196777344e-05,
|
| 102 |
+
"learning_rate": 7e-07,
|
| 103 |
+
"loss": 0.0457,
|
| 104 |
+
"reward": 0.8750000223517418,
|
| 105 |
+
"reward_std": 0.6384046971797943,
|
| 106 |
+
"rewards/accuracy_reward": 0.2604166716337204,
|
| 107 |
+
"rewards/format_reward": 0.6145833507180214,
|
| 108 |
+
"step": 7
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"completion_length": 2798.9688110351562,
|
| 112 |
+
"entropy": 0.365234375,
|
| 113 |
+
"epoch": 0.009142857142857144,
|
| 114 |
+
"grad_norm": 0.052074775099754333,
|
| 115 |
+
"kl": 1.9818544387817383e-05,
|
| 116 |
+
"learning_rate": 8e-07,
|
| 117 |
+
"loss": 0.0258,
|
| 118 |
+
"reward": 0.9166667014360428,
|
| 119 |
+
"reward_std": 0.3139677122235298,
|
| 120 |
+
"rewards/accuracy_reward": 0.3958333535119891,
|
| 121 |
+
"rewards/format_reward": 0.5208333432674408,
|
| 122 |
+
"step": 8
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"completion_length": 3406.6771850585938,
|
| 126 |
+
"entropy": 0.4599609375,
|
| 127 |
+
"epoch": 0.010285714285714285,
|
| 128 |
+
"grad_norm": 0.10204152017831802,
|
| 129 |
+
"kl": 3.5703182220458984e-05,
|
| 130 |
+
"learning_rate": 9e-07,
|
| 131 |
+
"loss": 0.0666,
|
| 132 |
+
"reward": 0.416666679084301,
|
| 133 |
+
"reward_std": 0.5519952327013016,
|
| 134 |
+
"rewards/accuracy_reward": 0.1041666679084301,
|
| 135 |
+
"rewards/format_reward": 0.3125000111758709,
|
| 136 |
+
"step": 9
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"completion_length": 2986.166748046875,
|
| 140 |
+
"entropy": 0.34423828125,
|
| 141 |
+
"epoch": 0.011428571428571429,
|
| 142 |
+
"grad_norm": 0.0653812512755394,
|
| 143 |
+
"kl": 1.93864107131958e-05,
|
| 144 |
+
"learning_rate": 1e-06,
|
| 145 |
+
"loss": 0.027,
|
| 146 |
+
"reward": 0.5625000111758709,
|
| 147 |
+
"reward_std": 0.47086329013109207,
|
| 148 |
+
"rewards/accuracy_reward": 0.15625000558793545,
|
| 149 |
+
"rewards/format_reward": 0.4062500149011612,
|
| 150 |
+
"step": 10
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"completion_length": 3706.4688110351562,
|
| 154 |
+
"entropy": 0.3798828125,
|
| 155 |
+
"epoch": 0.012571428571428572,
|
| 156 |
+
"grad_norm": 0.05876584351062775,
|
| 157 |
+
"kl": 2.321600914001465e-05,
|
| 158 |
+
"learning_rate": 9.997258721585931e-07,
|
| 159 |
+
"loss": 0.0221,
|
| 160 |
+
"reward": 0.23958334419876337,
|
| 161 |
+
"reward_std": 0.4288342222571373,
|
| 162 |
+
"rewards/accuracy_reward": 0.07291666697710752,
|
| 163 |
+
"rewards/format_reward": 0.16666666977107525,
|
| 164 |
+
"step": 11
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"completion_length": 2727.4896240234375,
|
| 168 |
+
"entropy": 0.40966796875,
|
| 169 |
+
"epoch": 0.013714285714285714,
|
| 170 |
+
"grad_norm": 0.08620024472475052,
|
| 171 |
+
"kl": 3.463029861450195e-05,
|
| 172 |
+
"learning_rate": 9.989038226169207e-07,
|
| 173 |
+
"loss": 0.0421,
|
| 174 |
+
"reward": 0.7291666977107525,
|
| 175 |
+
"reward_std": 0.3988698050379753,
|
| 176 |
+
"rewards/accuracy_reward": 0.13541666697710752,
|
| 177 |
+
"rewards/format_reward": 0.5937500111758709,
|
| 178 |
+
"step": 12
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"completion_length": 3067.4896850585938,
|
| 182 |
+
"entropy": 0.3759765625,
|
| 183 |
+
"epoch": 0.014857142857142857,
|
| 184 |
+
"grad_norm": 0.05084392428398132,
|
| 185 |
+
"kl": 1.9818544387817383e-05,
|
| 186 |
+
"learning_rate": 9.975348529157229e-07,
|
| 187 |
+
"loss": 0.0229,
|
| 188 |
+
"reward": 0.6666666865348816,
|
| 189 |
+
"reward_std": 0.24620391800999641,
|
| 190 |
+
"rewards/accuracy_reward": 0.21875000558793545,
|
| 191 |
+
"rewards/format_reward": 0.4479166716337204,
|
| 192 |
+
"step": 13
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"completion_length": 3072.791748046875,
|
| 196 |
+
"entropy": 0.37548828125,
|
| 197 |
+
"epoch": 0.016,
|
| 198 |
+
"grad_norm": 0.11545959115028381,
|
| 199 |
+
"kl": 1.6961246728897095e-05,
|
| 200 |
+
"learning_rate": 9.956206309337066e-07,
|
| 201 |
+
"loss": 0.0629,
|
| 202 |
+
"reward": 0.7187500149011612,
|
| 203 |
+
"reward_std": 0.5520583093166351,
|
| 204 |
+
"rewards/accuracy_reward": 0.2604166744276881,
|
| 205 |
+
"rewards/format_reward": 0.4583333507180214,
|
| 206 |
+
"step": 14
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"completion_length": 2951.1458740234375,
|
| 210 |
+
"entropy": 0.374755859375,
|
| 211 |
+
"epoch": 0.017142857142857144,
|
| 212 |
+
"grad_norm": 0.050988707691431046,
|
| 213 |
+
"kl": 3.2514333724975586e-05,
|
| 214 |
+
"learning_rate": 9.931634888554935e-07,
|
| 215 |
+
"loss": 0.0276,
|
| 216 |
+
"reward": 0.6770833432674408,
|
| 217 |
+
"reward_std": 0.3433547392487526,
|
| 218 |
+
"rewards/accuracy_reward": 0.2291666716337204,
|
| 219 |
+
"rewards/format_reward": 0.447916679084301,
|
| 220 |
+
"step": 15
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"completion_length": 3903.416748046875,
|
| 224 |
+
"entropy": 0.474609375,
|
| 225 |
+
"epoch": 0.018285714285714287,
|
| 226 |
+
"grad_norm": 0.047724124044179916,
|
| 227 |
+
"kl": 1.6005709767341614e-05,
|
| 228 |
+
"learning_rate": 9.901664203302124e-07,
|
| 229 |
+
"loss": 0.0075,
|
| 230 |
+
"reward": 0.1666666716337204,
|
| 231 |
+
"reward_std": 0.3083270490169525,
|
| 232 |
+
"rewards/accuracy_reward": 0.0729166679084301,
|
| 233 |
+
"rewards/format_reward": 0.0937500037252903,
|
| 234 |
+
"step": 16
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"completion_length": 2442.3334350585938,
|
| 238 |
+
"entropy": 0.443359375,
|
| 239 |
+
"epoch": 0.019428571428571427,
|
| 240 |
+
"grad_norm": 0.07992254197597504,
|
| 241 |
+
"kl": 0.0001729130744934082,
|
| 242 |
+
"learning_rate": 9.866330768241983e-07,
|
| 243 |
+
"loss": 0.0298,
|
| 244 |
+
"reward": 0.9375000149011612,
|
| 245 |
+
"reward_std": 0.46943993866443634,
|
| 246 |
+
"rewards/accuracy_reward": 0.3125000074505806,
|
| 247 |
+
"rewards/format_reward": 0.6250000074505806,
|
| 248 |
+
"step": 17
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"completion_length": 3152.9896850585938,
|
| 252 |
+
"entropy": 0.38623046875,
|
| 253 |
+
"epoch": 0.02057142857142857,
|
| 254 |
+
"grad_norm": 0.058529071509838104,
|
| 255 |
+
"kl": 0.000102996826171875,
|
| 256 |
+
"learning_rate": 9.825677631722435e-07,
|
| 257 |
+
"loss": 0.0241,
|
| 258 |
+
"reward": 0.5937500298023224,
|
| 259 |
+
"reward_std": 0.40303635224699974,
|
| 260 |
+
"rewards/accuracy_reward": 0.14583333861082792,
|
| 261 |
+
"rewards/format_reward": 0.447916679084301,
|
| 262 |
+
"step": 18
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"completion_length": 3366.354248046875,
|
| 266 |
+
"entropy": 0.41357421875,
|
| 267 |
+
"epoch": 0.021714285714285714,
|
| 268 |
+
"grad_norm": 0.057065945118665695,
|
| 269 |
+
"kl": 7.18533992767334e-05,
|
| 270 |
+
"learning_rate": 9.779754323328192e-07,
|
| 271 |
+
"loss": 0.033,
|
| 272 |
+
"reward": 0.7291666697710752,
|
| 273 |
+
"reward_std": 0.42073434591293335,
|
| 274 |
+
"rewards/accuracy_reward": 0.34375000558793545,
|
| 275 |
+
"rewards/format_reward": 0.38541667722165585,
|
| 276 |
+
"step": 19
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"completion_length": 2634.4896545410156,
|
| 280 |
+
"entropy": 0.354736328125,
|
| 281 |
+
"epoch": 0.022857142857142857,
|
| 282 |
+
"grad_norm": 0.08138567209243774,
|
| 283 |
+
"kl": 0.00015538930892944336,
|
| 284 |
+
"learning_rate": 9.728616793536587e-07,
|
| 285 |
+
"loss": 0.0625,
|
| 286 |
+
"reward": 0.9375000298023224,
|
| 287 |
+
"reward_std": 0.45325468480587006,
|
| 288 |
+
"rewards/accuracy_reward": 0.2812500102445483,
|
| 289 |
+
"rewards/format_reward": 0.6562500149011612,
|
| 290 |
+
"step": 20
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"completion_length": 2729.7083740234375,
|
| 294 |
+
"entropy": 0.44775390625,
|
| 295 |
+
"epoch": 0.024,
|
| 296 |
+
"grad_norm": 0.05598609521985054,
|
| 297 |
+
"kl": 0.0005742311477661133,
|
| 298 |
+
"learning_rate": 9.672327345550543e-07,
|
| 299 |
+
"loss": 0.0194,
|
| 300 |
+
"reward": 0.791666692122817,
|
| 301 |
+
"reward_std": 0.3458455428481102,
|
| 302 |
+
"rewards/accuracy_reward": 0.250000006519258,
|
| 303 |
+
"rewards/format_reward": 0.5416666669771075,
|
| 304 |
+
"step": 21
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"completion_length": 1972.1771545410156,
|
| 308 |
+
"entropy": 0.4150390625,
|
| 309 |
+
"epoch": 0.025142857142857144,
|
| 310 |
+
"grad_norm": 0.045343004167079926,
|
| 311 |
+
"kl": 0.0006244182586669922,
|
| 312 |
+
"learning_rate": 9.610954559391704e-07,
|
| 313 |
+
"loss": 0.0245,
|
| 314 |
+
"reward": 1.0520833432674408,
|
| 315 |
+
"reward_std": 0.3538536988198757,
|
| 316 |
+
"rewards/accuracy_reward": 0.26041667722165585,
|
| 317 |
+
"rewards/format_reward": 0.7916666716337204,
|
| 318 |
+
"step": 22
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"completion_length": 2651.7813415527344,
|
| 322 |
+
"entropy": 0.38134765625,
|
| 323 |
+
"epoch": 0.026285714285714287,
|
| 324 |
+
"grad_norm": 0.09305568039417267,
|
| 325 |
+
"kl": 0.0003466606140136719,
|
| 326 |
+
"learning_rate": 9.54457320834625e-07,
|
| 327 |
+
"loss": 0.0566,
|
| 328 |
+
"reward": 0.833333358168602,
|
| 329 |
+
"reward_std": 0.563101589679718,
|
| 330 |
+
"rewards/accuracy_reward": 0.23958334233611822,
|
| 331 |
+
"rewards/format_reward": 0.5937500223517418,
|
| 332 |
+
"step": 23
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"completion_length": 3014.854248046875,
|
| 336 |
+
"entropy": 0.3828125,
|
| 337 |
+
"epoch": 0.027428571428571427,
|
| 338 |
+
"grad_norm": 0.08546774834394455,
|
| 339 |
+
"kl": 0.00020091235637664795,
|
| 340 |
+
"learning_rate": 9.473264167865171e-07,
|
| 341 |
+
"loss": 0.0708,
|
| 342 |
+
"reward": 0.7187500298023224,
|
| 343 |
+
"reward_std": 0.5017067566514015,
|
| 344 |
+
"rewards/accuracy_reward": 0.2291666679084301,
|
| 345 |
+
"rewards/format_reward": 0.4895833432674408,
|
| 346 |
+
"step": 24
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"completion_length": 2801.6250610351562,
|
| 350 |
+
"entropy": 0.44580078125,
|
| 351 |
+
"epoch": 0.02857142857142857,
|
| 352 |
+
"grad_norm": 0.047619644552469254,
|
| 353 |
+
"kl": 0.0005018115043640137,
|
| 354 |
+
"learning_rate": 9.397114317029974e-07,
|
| 355 |
+
"loss": 0.025,
|
| 356 |
+
"reward": 0.614583358168602,
|
| 357 |
+
"reward_std": 0.30002114176750183,
|
| 358 |
+
"rewards/accuracy_reward": 0.1562500074505806,
|
| 359 |
+
"rewards/format_reward": 0.4583333432674408,
|
| 360 |
+
"step": 25
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"completion_length": 3229.6250610351562,
|
| 364 |
+
"entropy": 0.43603515625,
|
| 365 |
+
"epoch": 0.029714285714285714,
|
| 366 |
+
"grad_norm": 0.030185947194695473,
|
| 367 |
+
"kl": 0.00011354684829711914,
|
| 368 |
+
"learning_rate": 9.316216432703916e-07,
|
| 369 |
+
"loss": 0.0144,
|
| 370 |
+
"reward": 0.6562500223517418,
|
| 371 |
+
"reward_std": 0.21994514763355255,
|
| 372 |
+
"rewards/accuracy_reward": 0.2083333358168602,
|
| 373 |
+
"rewards/format_reward": 0.447916679084301,
|
| 374 |
+
"step": 26
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"completion_length": 3260.0938110351562,
|
| 378 |
+
"entropy": 0.4638671875,
|
| 379 |
+
"epoch": 0.030857142857142857,
|
| 380 |
+
"grad_norm": 0.1223464086651802,
|
| 381 |
+
"kl": 0.0009315013885498047,
|
| 382 |
+
"learning_rate": 9.230669076497687e-07,
|
| 383 |
+
"loss": 0.0537,
|
| 384 |
+
"reward": 0.5520833507180214,
|
| 385 |
+
"reward_std": 0.5293410122394562,
|
| 386 |
+
"rewards/accuracy_reward": 0.1458333358168602,
|
| 387 |
+
"rewards/format_reward": 0.4062500223517418,
|
| 388 |
+
"step": 27
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"completion_length": 3152.3021240234375,
|
| 392 |
+
"entropy": 0.43310546875,
|
| 393 |
+
"epoch": 0.032,
|
| 394 |
+
"grad_norm": 0.05098699778318405,
|
| 395 |
+
"kl": 0.0033740997314453125,
|
| 396 |
+
"learning_rate": 9.140576474687263e-07,
|
| 397 |
+
"loss": 0.0395,
|
| 398 |
+
"reward": 0.6875000149011612,
|
| 399 |
+
"reward_std": 0.34270069003105164,
|
| 400 |
+
"rewards/accuracy_reward": 0.29166667722165585,
|
| 401 |
+
"rewards/format_reward": 0.3958333358168602,
|
| 402 |
+
"step": 28
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"completion_length": 3795.7708740234375,
|
| 406 |
+
"entropy": 0.47900390625,
|
| 407 |
+
"epoch": 0.03314285714285714,
|
| 408 |
+
"grad_norm": 0.05339181423187256,
|
| 409 |
+
"kl": 0.0014908313751220703,
|
| 410 |
+
"learning_rate": 9.046048391230247e-07,
|
| 411 |
+
"loss": 0.0268,
|
| 412 |
+
"reward": 0.21875000558793545,
|
| 413 |
+
"reward_std": 0.36166706681251526,
|
| 414 |
+
"rewards/accuracy_reward": 0.09375000093132257,
|
| 415 |
+
"rewards/format_reward": 0.12500000651925802,
|
| 416 |
+
"step": 29
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"completion_length": 3244.0313110351562,
|
| 420 |
+
"entropy": 0.42431640625,
|
| 421 |
+
"epoch": 0.03428571428571429,
|
| 422 |
+
"grad_norm": 0.08664752542972565,
|
| 423 |
+
"kl": 0.0019369125366210938,
|
| 424 |
+
"learning_rate": 8.9471999940354e-07,
|
| 425 |
+
"loss": 0.0417,
|
| 426 |
+
"reward": 0.6145833386108279,
|
| 427 |
+
"reward_std": 0.5796928033232689,
|
| 428 |
+
"rewards/accuracy_reward": 0.21875000279396772,
|
| 429 |
+
"rewards/format_reward": 0.3958333386108279,
|
| 430 |
+
"step": 30
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"completion_length": 3305.9063110351562,
|
| 434 |
+
"entropy": 0.423828125,
|
| 435 |
+
"epoch": 0.03542857142857143,
|
| 436 |
+
"grad_norm": 0.0605226494371891,
|
| 437 |
+
"kl": 0.000820159912109375,
|
| 438 |
+
"learning_rate": 8.844151714648274e-07,
|
| 439 |
+
"loss": 0.0404,
|
| 440 |
+
"reward": 0.4791666865348816,
|
| 441 |
+
"reward_std": 0.4150131866335869,
|
| 442 |
+
"rewards/accuracy_reward": 0.1875000037252903,
|
| 443 |
+
"rewards/format_reward": 0.2916666716337204,
|
| 444 |
+
"step": 31
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"completion_length": 3559.572998046875,
|
| 448 |
+
"entropy": 0.470703125,
|
| 449 |
+
"epoch": 0.036571428571428574,
|
| 450 |
+
"grad_norm": 0.08260089159011841,
|
| 451 |
+
"kl": 0.004940032958984375,
|
| 452 |
+
"learning_rate": 8.737029101523929e-07,
|
| 453 |
+
"loss": 0.0361,
|
| 454 |
+
"reward": 0.6770833507180214,
|
| 455 |
+
"reward_std": 0.5618596971035004,
|
| 456 |
+
"rewards/accuracy_reward": 0.30208333395421505,
|
| 457 |
+
"rewards/format_reward": 0.3750000074505806,
|
| 458 |
+
"step": 32
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"completion_length": 3600.5208740234375,
|
| 462 |
+
"entropy": 0.4111328125,
|
| 463 |
+
"epoch": 0.037714285714285714,
|
| 464 |
+
"grad_norm": 0.07252604514360428,
|
| 465 |
+
"kl": 0.0011572837829589844,
|
| 466 |
+
"learning_rate": 8.625962667065487e-07,
|
| 467 |
+
"loss": 0.0397,
|
| 468 |
+
"reward": 0.5416666828095913,
|
| 469 |
+
"reward_std": 0.5172309577465057,
|
| 470 |
+
"rewards/accuracy_reward": 0.21875000558793545,
|
| 471 |
+
"rewards/format_reward": 0.32291667722165585,
|
| 472 |
+
"step": 33
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"completion_length": 2965.9896240234375,
|
| 476 |
+
"entropy": 0.5205078125,
|
| 477 |
+
"epoch": 0.038857142857142854,
|
| 478 |
+
"grad_norm": 0.11166317760944366,
|
| 479 |
+
"kl": 0.015777587890625,
|
| 480 |
+
"learning_rate": 8.511087728614862e-07,
|
| 481 |
+
"loss": 0.0416,
|
| 482 |
+
"reward": 0.854166679084301,
|
| 483 |
+
"reward_std": 0.46539078652858734,
|
| 484 |
+
"rewards/accuracy_reward": 0.3750000111758709,
|
| 485 |
+
"rewards/format_reward": 0.4791666828095913,
|
| 486 |
+
"step": 34
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"completion_length": 3255.885467529297,
|
| 490 |
+
"entropy": 0.43994140625,
|
| 491 |
+
"epoch": 0.04,
|
| 492 |
+
"grad_norm": 0.08642137050628662,
|
| 493 |
+
"kl": 0.0021219253540039062,
|
| 494 |
+
"learning_rate": 8.392544243589427e-07,
|
| 495 |
+
"loss": 0.0466,
|
| 496 |
+
"reward": 0.6562500223517418,
|
| 497 |
+
"reward_std": 0.5902808457612991,
|
| 498 |
+
"rewards/accuracy_reward": 0.26041668094694614,
|
| 499 |
+
"rewards/format_reward": 0.3958333469927311,
|
| 500 |
+
"step": 35
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"completion_length": 3611.8646240234375,
|
| 504 |
+
"entropy": 0.57421875,
|
| 505 |
+
"epoch": 0.04114285714285714,
|
| 506 |
+
"grad_norm": 0.04823267087340355,
|
| 507 |
+
"kl": 0.0031194686889648438,
|
| 508 |
+
"learning_rate": 8.270476638965461e-07,
|
| 509 |
+
"loss": 0.0177,
|
| 510 |
+
"reward": 0.1666666679084301,
|
| 511 |
+
"reward_std": 0.24164992570877075,
|
| 512 |
+
"rewards/accuracy_reward": 0.02083333395421505,
|
| 513 |
+
"rewards/format_reward": 0.14583333674818277,
|
| 514 |
+
"step": 36
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"completion_length": 3503.0625,
|
| 518 |
+
"entropy": 0.4638671875,
|
| 519 |
+
"epoch": 0.04228571428571429,
|
| 520 |
+
"grad_norm": 0.052913472056388855,
|
| 521 |
+
"kl": 0.001556396484375,
|
| 522 |
+
"learning_rate": 8.145033635316128e-07,
|
| 523 |
+
"loss": 0.0278,
|
| 524 |
+
"reward": 0.1875000074505806,
|
| 525 |
+
"reward_std": 0.11020193248987198,
|
| 526 |
+
"rewards/accuracy_reward": 0.0,
|
| 527 |
+
"rewards/format_reward": 0.1875000074505806,
|
| 528 |
+
"step": 37
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"completion_length": 3664.5625610351562,
|
| 532 |
+
"entropy": 0.513671875,
|
| 533 |
+
"epoch": 0.04342857142857143,
|
| 534 |
+
"grad_norm": 0.05083264037966728,
|
| 535 |
+
"kl": 0.0012769699096679688,
|
| 536 |
+
"learning_rate": 8.01636806561836e-07,
|
| 537 |
+
"loss": 0.0244,
|
| 538 |
+
"reward": 0.3229166716337204,
|
| 539 |
+
"reward_std": 0.27773431688547134,
|
| 540 |
+
"rewards/accuracy_reward": 0.1458333358168602,
|
| 541 |
+
"rewards/format_reward": 0.1770833358168602,
|
| 542 |
+
"step": 38
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"completion_length": 2993.385498046875,
|
| 546 |
+
"entropy": 0.39794921875,
|
| 547 |
+
"epoch": 0.044571428571428574,
|
| 548 |
+
"grad_norm": 0.0582449734210968,
|
| 549 |
+
"kl": 0.0028791427612304688,
|
| 550 |
+
"learning_rate": 7.884636689049422e-07,
|
| 551 |
+
"loss": 0.0389,
|
| 552 |
+
"reward": 0.729166679084301,
|
| 553 |
+
"reward_std": 0.34878823161125183,
|
| 554 |
+
"rewards/accuracy_reward": 0.25000000558793545,
|
| 555 |
+
"rewards/format_reward": 0.479166679084301,
|
| 556 |
+
"step": 39
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"completion_length": 3047.6563720703125,
|
| 560 |
+
"entropy": 0.455078125,
|
| 561 |
+
"epoch": 0.045714285714285714,
|
| 562 |
+
"grad_norm": 0.1776048094034195,
|
| 563 |
+
"kl": 0.033954620361328125,
|
| 564 |
+
"learning_rate": 7.75e-07,
|
| 565 |
+
"loss": 0.0499,
|
| 566 |
+
"reward": 0.6145833656191826,
|
| 567 |
+
"reward_std": 0.4401242658495903,
|
| 568 |
+
"rewards/accuracy_reward": 0.1354166679084301,
|
| 569 |
+
"rewards/format_reward": 0.4791666828095913,
|
| 570 |
+
"step": 40
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"completion_length": 3675.8021240234375,
|
| 574 |
+
"entropy": 0.455078125,
|
| 575 |
+
"epoch": 0.046857142857142854,
|
| 576 |
+
"grad_norm": 0.0947902724146843,
|
| 577 |
+
"kl": 0.0025844573974609375,
|
| 578 |
+
"learning_rate": 7.612622032536507e-07,
|
| 579 |
+
"loss": 0.034,
|
| 580 |
+
"reward": 0.3958333358168602,
|
| 581 |
+
"reward_std": 0.4633646681904793,
|
| 582 |
+
"rewards/accuracy_reward": 0.11458333861082792,
|
| 583 |
+
"rewards/format_reward": 0.2812500149011612,
|
| 584 |
+
"step": 41
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"completion_length": 3077.5521850585938,
|
| 588 |
+
"entropy": 0.49755859375,
|
| 589 |
+
"epoch": 0.048,
|
| 590 |
+
"grad_norm": 0.043859515339136124,
|
| 591 |
+
"kl": 0.0016508102416992188,
|
| 592 |
+
"learning_rate": 7.472670160550848e-07,
|
| 593 |
+
"loss": 0.0086,
|
| 594 |
+
"reward": 0.3437500037252903,
|
| 595 |
+
"reward_std": 0.19173795729875565,
|
| 596 |
+
"rewards/accuracy_reward": 0.031250000931322575,
|
| 597 |
+
"rewards/format_reward": 0.3125,
|
| 598 |
+
"step": 42
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"completion_length": 3321.8750610351562,
|
| 602 |
+
"entropy": 0.49462890625,
|
| 603 |
+
"epoch": 0.04914285714285714,
|
| 604 |
+
"grad_norm": 0.07441486418247223,
|
| 605 |
+
"kl": 0.0028228759765625,
|
| 606 |
+
"learning_rate": 7.330314893841101e-07,
|
| 607 |
+
"loss": 0.0332,
|
| 608 |
+
"reward": 0.5000000074505806,
|
| 609 |
+
"reward_std": 0.4754345528781414,
|
| 610 |
+
"rewards/accuracy_reward": 0.18750000279396772,
|
| 611 |
+
"rewards/format_reward": 0.3125000074505806,
|
| 612 |
+
"step": 43
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"completion_length": 2997.7709350585938,
|
| 616 |
+
"entropy": 0.466796875,
|
| 617 |
+
"epoch": 0.05028571428571429,
|
| 618 |
+
"grad_norm": 0.08534099161624908,
|
| 619 |
+
"kl": 0.0044193267822265625,
|
| 620 |
+
"learning_rate": 7.185729670371604e-07,
|
| 621 |
+
"loss": 0.0371,
|
| 622 |
+
"reward": 0.6979166865348816,
|
| 623 |
+
"reward_std": 0.4120652452111244,
|
| 624 |
+
"rewards/accuracy_reward": 0.2500000037252903,
|
| 625 |
+
"rewards/format_reward": 0.447916679084301,
|
| 626 |
+
"step": 44
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"completion_length": 3698.2396850585938,
|
| 630 |
+
"entropy": 0.4775390625,
|
| 631 |
+
"epoch": 0.05142857142857143,
|
| 632 |
+
"grad_norm": 0.052873097360134125,
|
| 633 |
+
"kl": 0.0031747817993164062,
|
| 634 |
+
"learning_rate": 7.039090644965509e-07,
|
| 635 |
+
"loss": 0.0083,
|
| 636 |
+
"reward": 0.40625001676380634,
|
| 637 |
+
"reward_std": 0.4317762479186058,
|
| 638 |
+
"rewards/accuracy_reward": 0.15625000651925802,
|
| 639 |
+
"rewards/format_reward": 0.250000006519258,
|
| 640 |
+
"step": 45
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"completion_length": 3654.0521240234375,
|
| 644 |
+
"entropy": 0.5927734375,
|
| 645 |
+
"epoch": 0.052571428571428575,
|
| 646 |
+
"grad_norm": 0.06117837131023407,
|
| 647 |
+
"kl": 0.0036220550537109375,
|
| 648 |
+
"learning_rate": 6.890576474687263e-07,
|
| 649 |
+
"loss": 0.0261,
|
| 650 |
+
"reward": 0.2187500074505806,
|
| 651 |
+
"reward_std": 0.3954956606030464,
|
| 652 |
+
"rewards/accuracy_reward": 0.02083333395421505,
|
| 653 |
+
"rewards/format_reward": 0.1979166716337204,
|
| 654 |
+
"step": 46
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"completion_length": 3092.8750610351562,
|
| 658 |
+
"entropy": 0.455078125,
|
| 659 |
+
"epoch": 0.053714285714285714,
|
| 660 |
+
"grad_norm": 0.10391207039356232,
|
| 661 |
+
"kl": 0.0032706260681152344,
|
| 662 |
+
"learning_rate": 6.740368101176495e-07,
|
| 663 |
+
"loss": 0.0729,
|
| 664 |
+
"reward": 0.7812500298023224,
|
| 665 |
+
"reward_std": 0.5671325027942657,
|
| 666 |
+
"rewards/accuracy_reward": 0.2916666716337204,
|
| 667 |
+
"rewards/format_reward": 0.4895833507180214,
|
| 668 |
+
"step": 47
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"completion_length": 3051.979248046875,
|
| 672 |
+
"entropy": 0.4609375,
|
| 673 |
+
"epoch": 0.054857142857142854,
|
| 674 |
+
"grad_norm": 0.03355824202299118,
|
| 675 |
+
"kl": 0.0069446563720703125,
|
| 676 |
+
"learning_rate": 6.588648530198504e-07,
|
| 677 |
+
"loss": 0.0042,
|
| 678 |
+
"reward": 0.5625000055879354,
|
| 679 |
+
"reward_std": 0.3032701760530472,
|
| 680 |
+
"rewards/accuracy_reward": 0.16666666697710752,
|
| 681 |
+
"rewards/format_reward": 0.39583333395421505,
|
| 682 |
+
"step": 48
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"completion_length": 2525.9063720703125,
|
| 686 |
+
"entropy": 0.42431640625,
|
| 687 |
+
"epoch": 0.056,
|
| 688 |
+
"grad_norm": 0.0799187645316124,
|
| 689 |
+
"kl": 0.002651214599609375,
|
| 690 |
+
"learning_rate": 6.435602608679916e-07,
|
| 691 |
+
"loss": 0.0455,
|
| 692 |
+
"reward": 0.7916666865348816,
|
| 693 |
+
"reward_std": 0.45931655168533325,
|
| 694 |
+
"rewards/accuracy_reward": 0.21875000651925802,
|
| 695 |
+
"rewards/format_reward": 0.5729166865348816,
|
| 696 |
+
"step": 49
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"completion_length": 3233.9896240234375,
|
| 700 |
+
"entropy": 0.4208984375,
|
| 701 |
+
"epoch": 0.05714285714285714,
|
| 702 |
+
"grad_norm": 0.06934866309165955,
|
| 703 |
+
"kl": 0.0029773712158203125,
|
| 704 |
+
"learning_rate": 6.281416799501187e-07,
|
| 705 |
+
"loss": 0.0284,
|
| 706 |
+
"reward": 0.6562500111758709,
|
| 707 |
+
"reward_std": 0.39541390538215637,
|
| 708 |
+
"rewards/accuracy_reward": 0.2916666716337204,
|
| 709 |
+
"rewards/format_reward": 0.3645833544433117,
|
| 710 |
+
"step": 50
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"completion_length": 2639.4376220703125,
|
| 714 |
+
"entropy": 0.548828125,
|
| 715 |
+
"epoch": 0.05828571428571429,
|
| 716 |
+
"grad_norm": 0.05569971352815628,
|
| 717 |
+
"kl": 0.007915496826171875,
|
| 718 |
+
"learning_rate": 6.126278954320294e-07,
|
| 719 |
+
"loss": 0.0188,
|
| 720 |
+
"reward": 0.6250000149011612,
|
| 721 |
+
"reward_std": 0.31560593098402023,
|
| 722 |
+
"rewards/accuracy_reward": 0.1145833395421505,
|
| 723 |
+
"rewards/format_reward": 0.5104166716337204,
|
| 724 |
+
"step": 51
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"completion_length": 3200.7083740234375,
|
| 728 |
+
"entropy": 0.4736328125,
|
| 729 |
+
"epoch": 0.05942857142857143,
|
| 730 |
+
"grad_norm": 0.10418283194303513,
|
| 731 |
+
"kl": 0.0034847259521484375,
|
| 732 |
+
"learning_rate": 5.97037808470444e-07,
|
| 733 |
+
"loss": 0.0464,
|
| 734 |
+
"reward": 0.7187500149011612,
|
| 735 |
+
"reward_std": 0.5199342221021652,
|
| 736 |
+
"rewards/accuracy_reward": 0.3020833432674408,
|
| 737 |
+
"rewards/format_reward": 0.416666679084301,
|
| 738 |
+
"step": 52
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"completion_length": 2897.354248046875,
|
| 742 |
+
"entropy": 0.46533203125,
|
| 743 |
+
"epoch": 0.060571428571428575,
|
| 744 |
+
"grad_norm": 0.0808432400226593,
|
| 745 |
+
"kl": 0.0022940635681152344,
|
| 746 |
+
"learning_rate": 5.813904131848564e-07,
|
| 747 |
+
"loss": 0.0321,
|
| 748 |
+
"reward": 0.8125000149011612,
|
| 749 |
+
"reward_std": 0.5500369966030121,
|
| 750 |
+
"rewards/accuracy_reward": 0.2812500037252903,
|
| 751 |
+
"rewards/format_reward": 0.5312500074505806,
|
| 752 |
+
"step": 53
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"completion_length": 3137.5834350585938,
|
| 756 |
+
"entropy": 0.48291015625,
|
| 757 |
+
"epoch": 0.061714285714285715,
|
| 758 |
+
"grad_norm": 0.14153237640857697,
|
| 759 |
+
"kl": 0.0033512115478515625,
|
| 760 |
+
"learning_rate": 5.657047735161255e-07,
|
| 761 |
+
"loss": 0.0765,
|
| 762 |
+
"reward": 0.8437500149011612,
|
| 763 |
+
"reward_std": 0.568057231605053,
|
| 764 |
+
"rewards/accuracy_reward": 0.354166679084301,
|
| 765 |
+
"rewards/format_reward": 0.4895833358168602,
|
| 766 |
+
"step": 54
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"completion_length": 3302.8021850585938,
|
| 770 |
+
"entropy": 0.5078125,
|
| 771 |
+
"epoch": 0.06285714285714286,
|
| 772 |
+
"grad_norm": 0.06538388878107071,
|
| 773 |
+
"kl": 0.004062652587890625,
|
| 774 |
+
"learning_rate": 5.5e-07,
|
| 775 |
+
"loss": 0.0373,
|
| 776 |
+
"reward": 0.6145833358168602,
|
| 777 |
+
"reward_std": 0.49769678711891174,
|
| 778 |
+
"rewards/accuracy_reward": 0.19791667070239782,
|
| 779 |
+
"rewards/format_reward": 0.4166666753590107,
|
| 780 |
+
"step": 55
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"completion_length": 3132.322998046875,
|
| 784 |
+
"entropy": 0.474609375,
|
| 785 |
+
"epoch": 0.064,
|
| 786 |
+
"grad_norm": 0.10536816716194153,
|
| 787 |
+
"kl": 0.00362396240234375,
|
| 788 |
+
"learning_rate": 5.342952264838747e-07,
|
| 789 |
+
"loss": 0.034,
|
| 790 |
+
"reward": 0.6041666669771075,
|
| 791 |
+
"reward_std": 0.31034940481185913,
|
| 792 |
+
"rewards/accuracy_reward": 0.1875,
|
| 793 |
+
"rewards/format_reward": 0.4166666669771075,
|
| 794 |
+
"step": 56
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"completion_length": 3647.697998046875,
|
| 798 |
+
"entropy": 0.39453125,
|
| 799 |
+
"epoch": 0.06514285714285714,
|
| 800 |
+
"grad_norm": 0.056989964097738266,
|
| 801 |
+
"kl": 0.002593994140625,
|
| 802 |
+
"learning_rate": 5.186095868151436e-07,
|
| 803 |
+
"loss": 0.0117,
|
| 804 |
+
"reward": 0.46875002048909664,
|
| 805 |
+
"reward_std": 0.5007961094379425,
|
| 806 |
+
"rewards/accuracy_reward": 0.15625000279396772,
|
| 807 |
+
"rewards/format_reward": 0.3125000176951289,
|
| 808 |
+
"step": 57
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"completion_length": 2546.010498046875,
|
| 812 |
+
"entropy": 0.3984375,
|
| 813 |
+
"epoch": 0.06628571428571428,
|
| 814 |
+
"grad_norm": 0.06985888630151749,
|
| 815 |
+
"kl": 0.005336761474609375,
|
| 816 |
+
"learning_rate": 5.02962191529556e-07,
|
| 817 |
+
"loss": 0.0615,
|
| 818 |
+
"reward": 0.9062500409781933,
|
| 819 |
+
"reward_std": 0.49653898924589157,
|
| 820 |
+
"rewards/accuracy_reward": 0.2708333395421505,
|
| 821 |
+
"rewards/format_reward": 0.6354166828095913,
|
| 822 |
+
"step": 58
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"completion_length": 3380.75,
|
| 826 |
+
"entropy": 0.4609375,
|
| 827 |
+
"epoch": 0.06742857142857143,
|
| 828 |
+
"grad_norm": 0.05882963165640831,
|
| 829 |
+
"kl": 0.004192352294921875,
|
| 830 |
+
"learning_rate": 4.873721045679706e-07,
|
| 831 |
+
"loss": 0.0238,
|
| 832 |
+
"reward": 0.4687500298023224,
|
| 833 |
+
"reward_std": 0.2912241667509079,
|
| 834 |
+
"rewards/accuracy_reward": 0.1770833432674408,
|
| 835 |
+
"rewards/format_reward": 0.2916666716337204,
|
| 836 |
+
"step": 59
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"completion_length": 3201.0208740234375,
|
| 840 |
+
"entropy": 0.45556640625,
|
| 841 |
+
"epoch": 0.06857142857142857,
|
| 842 |
+
"grad_norm": 0.04845750704407692,
|
| 843 |
+
"kl": 0.003467559814453125,
|
| 844 |
+
"learning_rate": 4.7185832004988133e-07,
|
| 845 |
+
"loss": 0.0125,
|
| 846 |
+
"reward": 0.5208333488553762,
|
| 847 |
+
"reward_std": 0.3204401135444641,
|
| 848 |
+
"rewards/accuracy_reward": 0.1354166679084301,
|
| 849 |
+
"rewards/format_reward": 0.3854166669771075,
|
| 850 |
+
"step": 60
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"completion_length": 3301.5626220703125,
|
| 854 |
+
"entropy": 0.44921875,
|
| 855 |
+
"epoch": 0.06971428571428571,
|
| 856 |
+
"grad_norm": 0.1134149506688118,
|
| 857 |
+
"kl": 0.00356292724609375,
|
| 858 |
+
"learning_rate": 4.5643973913200837e-07,
|
| 859 |
+
"loss": 0.077,
|
| 860 |
+
"reward": 0.6666666893288493,
|
| 861 |
+
"reward_std": 0.5132372379302979,
|
| 862 |
+
"rewards/accuracy_reward": 0.2291666716337204,
|
| 863 |
+
"rewards/format_reward": 0.4375000176951289,
|
| 864 |
+
"step": 61
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"completion_length": 2941.3958740234375,
|
| 868 |
+
"entropy": 0.437744140625,
|
| 869 |
+
"epoch": 0.07085714285714285,
|
| 870 |
+
"grad_norm": 0.11249862611293793,
|
| 871 |
+
"kl": 0.014247894287109375,
|
| 872 |
+
"learning_rate": 4.4113514698014953e-07,
|
| 873 |
+
"loss": 0.0926,
|
| 874 |
+
"reward": 0.833333358168602,
|
| 875 |
+
"reward_std": 0.6926020532846451,
|
| 876 |
+
"rewards/accuracy_reward": 0.28125000558793545,
|
| 877 |
+
"rewards/format_reward": 0.5520833507180214,
|
| 878 |
+
"step": 62
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"completion_length": 2724.9063110351562,
|
| 882 |
+
"entropy": 0.4765625,
|
| 883 |
+
"epoch": 0.072,
|
| 884 |
+
"grad_norm": 0.12434863299131393,
|
| 885 |
+
"kl": 0.00435638427734375,
|
| 886 |
+
"learning_rate": 4.2596318988235037e-07,
|
| 887 |
+
"loss": 0.0403,
|
| 888 |
+
"reward": 0.9791666865348816,
|
| 889 |
+
"reward_std": 0.46333979070186615,
|
| 890 |
+
"rewards/accuracy_reward": 0.3229166669771075,
|
| 891 |
+
"rewards/format_reward": 0.6562500074505806,
|
| 892 |
+
"step": 63
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"completion_length": 3465.1563720703125,
|
| 896 |
+
"entropy": 0.52001953125,
|
| 897 |
+
"epoch": 0.07314285714285715,
|
| 898 |
+
"grad_norm": 0.09894906729459763,
|
| 899 |
+
"kl": 0.006351470947265625,
|
| 900 |
+
"learning_rate": 4.1094235253127374e-07,
|
| 901 |
+
"loss": 0.079,
|
| 902 |
+
"reward": 0.5000000074505806,
|
| 903 |
+
"reward_std": 0.564013384282589,
|
| 904 |
+
"rewards/accuracy_reward": 0.17708333861082792,
|
| 905 |
+
"rewards/format_reward": 0.3229166753590107,
|
| 906 |
+
"step": 64
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"completion_length": 3121.6041870117188,
|
| 910 |
+
"entropy": 0.48046875,
|
| 911 |
+
"epoch": 0.07428571428571429,
|
| 912 |
+
"grad_norm": 0.053176186978816986,
|
| 913 |
+
"kl": 0.00548553466796875,
|
| 914 |
+
"learning_rate": 3.9609093550344907e-07,
|
| 915 |
+
"loss": 0.0199,
|
| 916 |
+
"reward": 0.5937500176951289,
|
| 917 |
+
"reward_std": 0.32823745906352997,
|
| 918 |
+
"rewards/accuracy_reward": 0.16666666697710752,
|
| 919 |
+
"rewards/format_reward": 0.4270833535119891,
|
| 920 |
+
"step": 65
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"completion_length": 2457.375,
|
| 924 |
+
"entropy": 0.4365234375,
|
| 925 |
+
"epoch": 0.07542857142857143,
|
| 926 |
+
"grad_norm": 0.040902264416217804,
|
| 927 |
+
"kl": 0.00630950927734375,
|
| 928 |
+
"learning_rate": 3.8142703296283953e-07,
|
| 929 |
+
"loss": -0.0028,
|
| 930 |
+
"reward": 0.9166666828095913,
|
| 931 |
+
"reward_std": 0.33521074056625366,
|
| 932 |
+
"rewards/accuracy_reward": 0.38541667722165585,
|
| 933 |
+
"rewards/format_reward": 0.5312500055879354,
|
| 934 |
+
"step": 66
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"completion_length": 3835.1354370117188,
|
| 938 |
+
"entropy": 0.4384765625,
|
| 939 |
+
"epoch": 0.07657142857142857,
|
| 940 |
+
"grad_norm": 0.033646635711193085,
|
| 941 |
+
"kl": 0.00453948974609375,
|
| 942 |
+
"learning_rate": 3.6696851061588994e-07,
|
| 943 |
+
"loss": 0.0167,
|
| 944 |
+
"reward": 0.1458333358168602,
|
| 945 |
+
"reward_std": 0.21662194281816483,
|
| 946 |
+
"rewards/accuracy_reward": 0.031250000931322575,
|
| 947 |
+
"rewards/format_reward": 0.1145833358168602,
|
| 948 |
+
"step": 67
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"completion_length": 2685.2188110351562,
|
| 952 |
+
"entropy": 0.5234375,
|
| 953 |
+
"epoch": 0.07771428571428571,
|
| 954 |
+
"grad_norm": 0.1085897833108902,
|
| 955 |
+
"kl": 0.0067596435546875,
|
| 956 |
+
"learning_rate": 3.5273298394491515e-07,
|
| 957 |
+
"loss": 0.0768,
|
| 958 |
+
"reward": 0.7083333656191826,
|
| 959 |
+
"reward_std": 0.4535161405801773,
|
| 960 |
+
"rewards/accuracy_reward": 0.1770833358168602,
|
| 961 |
+
"rewards/format_reward": 0.5312500074505806,
|
| 962 |
+
"step": 68
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"completion_length": 3197.5105590820312,
|
| 966 |
+
"entropy": 0.62109375,
|
| 967 |
+
"epoch": 0.07885714285714286,
|
| 968 |
+
"grad_norm": 0.05905470252037048,
|
| 969 |
+
"kl": 0.0102996826171875,
|
| 970 |
+
"learning_rate": 3.387377967463493e-07,
|
| 971 |
+
"loss": 0.0289,
|
| 972 |
+
"reward": 0.3854166828095913,
|
| 973 |
+
"reward_std": 0.3971848413348198,
|
| 974 |
+
"rewards/accuracy_reward": 0.0416666679084301,
|
| 975 |
+
"rewards/format_reward": 0.3437500074505806,
|
| 976 |
+
"step": 69
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"completion_length": 3256.760498046875,
|
| 980 |
+
"entropy": 0.4345703125,
|
| 981 |
+
"epoch": 0.08,
|
| 982 |
+
"grad_norm": 0.05583971366286278,
|
| 983 |
+
"kl": 0.004856109619140625,
|
| 984 |
+
"learning_rate": 3.250000000000001e-07,
|
| 985 |
+
"loss": 0.0337,
|
| 986 |
+
"reward": 0.5416666697710752,
|
| 987 |
+
"reward_std": 0.44431453198194504,
|
| 988 |
+
"rewards/accuracy_reward": 0.15625000279396772,
|
| 989 |
+
"rewards/format_reward": 0.3854166744276881,
|
| 990 |
+
"step": 70
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"completion_length": 3021.197967529297,
|
| 994 |
+
"entropy": 0.529296875,
|
| 995 |
+
"epoch": 0.08114285714285714,
|
| 996 |
+
"grad_norm": 0.06864861398935318,
|
| 997 |
+
"kl": 0.0126190185546875,
|
| 998 |
+
"learning_rate": 3.115363310950578e-07,
|
| 999 |
+
"loss": 0.016,
|
| 1000 |
+
"reward": 0.5520833386108279,
|
| 1001 |
+
"reward_std": 0.2611164078116417,
|
| 1002 |
+
"rewards/accuracy_reward": 0.1875,
|
| 1003 |
+
"rewards/format_reward": 0.3645833386108279,
|
| 1004 |
+
"step": 71
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"completion_length": 3671.9168090820312,
|
| 1008 |
+
"entropy": 0.62890625,
|
| 1009 |
+
"epoch": 0.08228571428571428,
|
| 1010 |
+
"grad_norm": 0.06177105754613876,
|
| 1011 |
+
"kl": 0.00665283203125,
|
| 1012 |
+
"learning_rate": 2.9836319343816397e-07,
|
| 1013 |
+
"loss": 0.0391,
|
| 1014 |
+
"reward": 0.3645833432674408,
|
| 1015 |
+
"reward_std": 0.5157303959131241,
|
| 1016 |
+
"rewards/accuracy_reward": 0.08333333674818277,
|
| 1017 |
+
"rewards/format_reward": 0.28125000558793545,
|
| 1018 |
+
"step": 72
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"completion_length": 3792.6563720703125,
|
| 1022 |
+
"entropy": 0.5341796875,
|
| 1023 |
+
"epoch": 0.08342857142857144,
|
| 1024 |
+
"grad_norm": 0.061636123806238174,
|
| 1025 |
+
"kl": 0.00244903564453125,
|
| 1026 |
+
"learning_rate": 2.854966364683872e-07,
|
| 1027 |
+
"loss": 0.0344,
|
| 1028 |
+
"reward": 0.2708333386108279,
|
| 1029 |
+
"reward_std": 0.4069225564599037,
|
| 1030 |
+
"rewards/accuracy_reward": 0.10416666977107525,
|
| 1031 |
+
"rewards/format_reward": 0.16666667070239782,
|
| 1032 |
+
"step": 73
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"completion_length": 3621.5313720703125,
|
| 1036 |
+
"entropy": 0.5068359375,
|
| 1037 |
+
"epoch": 0.08457142857142858,
|
| 1038 |
+
"grad_norm": 0.05920031666755676,
|
| 1039 |
+
"kl": 0.00554656982421875,
|
| 1040 |
+
"learning_rate": 2.729523361034538e-07,
|
| 1041 |
+
"loss": 0.0216,
|
| 1042 |
+
"reward": 0.4791666939854622,
|
| 1043 |
+
"reward_std": 0.4260884076356888,
|
| 1044 |
+
"rewards/accuracy_reward": 0.21875000186264515,
|
| 1045 |
+
"rewards/format_reward": 0.2604166753590107,
|
| 1046 |
+
"step": 74
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"completion_length": 3386.3751220703125,
|
| 1050 |
+
"entropy": 0.46826171875,
|
| 1051 |
+
"epoch": 0.08571428571428572,
|
| 1052 |
+
"grad_norm": 0.07354261726140976,
|
| 1053 |
+
"kl": 0.00623321533203125,
|
| 1054 |
+
"learning_rate": 2.6074557564105724e-07,
|
| 1055 |
+
"loss": 0.046,
|
| 1056 |
+
"reward": 0.5833333432674408,
|
| 1057 |
+
"reward_std": 0.47260352596640587,
|
| 1058 |
+
"rewards/accuracy_reward": 0.17708333488553762,
|
| 1059 |
+
"rewards/format_reward": 0.4062500149011612,
|
| 1060 |
+
"step": 75
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"completion_length": 3012.104248046875,
|
| 1064 |
+
"entropy": 0.4912109375,
|
| 1065 |
+
"epoch": 0.08685714285714285,
|
| 1066 |
+
"grad_norm": 0.08954478800296783,
|
| 1067 |
+
"kl": 0.00344085693359375,
|
| 1068 |
+
"learning_rate": 2.488912271385139e-07,
|
| 1069 |
+
"loss": 0.0614,
|
| 1070 |
+
"reward": 0.5833333432674408,
|
| 1071 |
+
"reward_std": 0.4225878678262234,
|
| 1072 |
+
"rewards/accuracy_reward": 0.08333333395421505,
|
| 1073 |
+
"rewards/format_reward": 0.5000000149011612,
|
| 1074 |
+
"step": 76
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"completion_length": 3350.6666870117188,
|
| 1078 |
+
"entropy": 0.52392578125,
|
| 1079 |
+
"epoch": 0.088,
|
| 1080 |
+
"grad_norm": 0.06333605200052261,
|
| 1081 |
+
"kl": 0.004314422607421875,
|
| 1082 |
+
"learning_rate": 2.374037332934512e-07,
|
| 1083 |
+
"loss": 0.0365,
|
| 1084 |
+
"reward": 0.5312500214204192,
|
| 1085 |
+
"reward_std": 0.3916928321123123,
|
| 1086 |
+
"rewards/accuracy_reward": 0.1041666716337204,
|
| 1087 |
+
"rewards/format_reward": 0.4270833497866988,
|
| 1088 |
+
"step": 77
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"completion_length": 3443.8438720703125,
|
| 1092 |
+
"entropy": 0.4697265625,
|
| 1093 |
+
"epoch": 0.08914285714285715,
|
| 1094 |
+
"grad_norm": 0.10582572221755981,
|
| 1095 |
+
"kl": 0.00374603271484375,
|
| 1096 |
+
"learning_rate": 2.2629708984760706e-07,
|
| 1097 |
+
"loss": 0.0468,
|
| 1098 |
+
"reward": 0.6354166865348816,
|
| 1099 |
+
"reward_std": 0.6427144631743431,
|
| 1100 |
+
"rewards/accuracy_reward": 0.2291666716337204,
|
| 1101 |
+
"rewards/format_reward": 0.4062500074505806,
|
| 1102 |
+
"step": 78
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"completion_length": 2630.135467529297,
|
| 1106 |
+
"entropy": 0.41455078125,
|
| 1107 |
+
"epoch": 0.09028571428571429,
|
| 1108 |
+
"grad_norm": 0.07537446171045303,
|
| 1109 |
+
"kl": 0.01548004150390625,
|
| 1110 |
+
"learning_rate": 2.1558482853517253e-07,
|
| 1111 |
+
"loss": 0.0285,
|
| 1112 |
+
"reward": 0.8750000298023224,
|
| 1113 |
+
"reward_std": 0.44084493070840836,
|
| 1114 |
+
"rewards/accuracy_reward": 0.27083334140479565,
|
| 1115 |
+
"rewards/format_reward": 0.604166679084301,
|
| 1116 |
+
"step": 79
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"completion_length": 3601.322998046875,
|
| 1120 |
+
"entropy": 0.57275390625,
|
| 1121 |
+
"epoch": 0.09142857142857143,
|
| 1122 |
+
"grad_norm": 0.050721827894449234,
|
| 1123 |
+
"kl": 0.0056438446044921875,
|
| 1124 |
+
"learning_rate": 2.0528000059645995e-07,
|
| 1125 |
+
"loss": 0.0131,
|
| 1126 |
+
"reward": 0.5104166772216558,
|
| 1127 |
+
"reward_std": 0.27086786925792694,
|
| 1128 |
+
"rewards/accuracy_reward": 0.1770833432674408,
|
| 1129 |
+
"rewards/format_reward": 0.33333333395421505,
|
| 1130 |
+
"step": 80
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"completion_length": 3389.9063110351562,
|
| 1134 |
+
"entropy": 0.716796875,
|
| 1135 |
+
"epoch": 0.09257142857142857,
|
| 1136 |
+
"grad_norm": 0.04999749734997749,
|
| 1137 |
+
"kl": 0.01094818115234375,
|
| 1138 |
+
"learning_rate": 1.9539516087697517e-07,
|
| 1139 |
+
"loss": 0.011,
|
| 1140 |
+
"reward": 0.39583334885537624,
|
| 1141 |
+
"reward_std": 0.30708885192871094,
|
| 1142 |
+
"rewards/accuracy_reward": 0.11458333674818277,
|
| 1143 |
+
"rewards/format_reward": 0.2812500027939677,
|
| 1144 |
+
"step": 81
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"completion_length": 3012.3959350585938,
|
| 1148 |
+
"entropy": 0.51806640625,
|
| 1149 |
+
"epoch": 0.09371428571428571,
|
| 1150 |
+
"grad_norm": 0.054486002773046494,
|
| 1151 |
+
"kl": 0.009601593017578125,
|
| 1152 |
+
"learning_rate": 1.8594235253127372e-07,
|
| 1153 |
+
"loss": 0.0318,
|
| 1154 |
+
"reward": 0.572916679084301,
|
| 1155 |
+
"reward_std": 0.30159565061330795,
|
| 1156 |
+
"rewards/accuracy_reward": 0.18750000558793545,
|
| 1157 |
+
"rewards/format_reward": 0.385416679084301,
|
| 1158 |
+
"step": 82
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"completion_length": 3209.9271850585938,
|
| 1162 |
+
"entropy": 0.5732421875,
|
| 1163 |
+
"epoch": 0.09485714285714286,
|
| 1164 |
+
"grad_norm": 0.04813052713871002,
|
| 1165 |
+
"kl": 0.00592041015625,
|
| 1166 |
+
"learning_rate": 1.7693309235023127e-07,
|
| 1167 |
+
"loss": 0.012,
|
| 1168 |
+
"reward": 0.41666667349636555,
|
| 1169 |
+
"reward_std": 0.3812928982079029,
|
| 1170 |
+
"rewards/accuracy_reward": 0.1354166679084301,
|
| 1171 |
+
"rewards/format_reward": 0.2812500009313226,
|
| 1172 |
+
"step": 83
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"completion_length": 3261.291748046875,
|
| 1176 |
+
"entropy": 0.521484375,
|
| 1177 |
+
"epoch": 0.096,
|
| 1178 |
+
"grad_norm": 0.07685024291276932,
|
| 1179 |
+
"kl": 0.0028629302978515625,
|
| 1180 |
+
"learning_rate": 1.6837835672960831e-07,
|
| 1181 |
+
"loss": 0.0347,
|
| 1182 |
+
"reward": 0.8125,
|
| 1183 |
+
"reward_std": 0.452178031206131,
|
| 1184 |
+
"rewards/accuracy_reward": 0.34375,
|
| 1185 |
+
"rewards/format_reward": 0.46875,
|
| 1186 |
+
"step": 84
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"completion_length": 3507.2916870117188,
|
| 1190 |
+
"entropy": 0.45458984375,
|
| 1191 |
+
"epoch": 0.09714285714285714,
|
| 1192 |
+
"grad_norm": 0.0731167271733284,
|
| 1193 |
+
"kl": 0.004367828369140625,
|
| 1194 |
+
"learning_rate": 1.6028856829700258e-07,
|
| 1195 |
+
"loss": 0.0261,
|
| 1196 |
+
"reward": 0.677083358168602,
|
| 1197 |
+
"reward_std": 0.6612033396959305,
|
| 1198 |
+
"rewards/accuracy_reward": 0.2500000102445483,
|
| 1199 |
+
"rewards/format_reward": 0.42708334885537624,
|
| 1200 |
+
"step": 85
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"completion_length": 3195.3646850585938,
|
| 1204 |
+
"entropy": 0.5634765625,
|
| 1205 |
+
"epoch": 0.09828571428571428,
|
| 1206 |
+
"grad_norm": 0.06479182839393616,
|
| 1207 |
+
"kl": 0.00772857666015625,
|
| 1208 |
+
"learning_rate": 1.5267358321348285e-07,
|
| 1209 |
+
"loss": 0.028,
|
| 1210 |
+
"reward": 0.5729166865348816,
|
| 1211 |
+
"reward_std": 0.24059326201677322,
|
| 1212 |
+
"rewards/accuracy_reward": 0.1354166716337204,
|
| 1213 |
+
"rewards/format_reward": 0.4375000074505806,
|
| 1214 |
+
"step": 86
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"completion_length": 3176.0833740234375,
|
| 1218 |
+
"entropy": 0.646484375,
|
| 1219 |
+
"epoch": 0.09942857142857142,
|
| 1220 |
+
"grad_norm": 0.10074935853481293,
|
| 1221 |
+
"kl": 0.0096588134765625,
|
| 1222 |
+
"learning_rate": 1.4554267916537495e-07,
|
| 1223 |
+
"loss": 0.0461,
|
| 1224 |
+
"reward": 0.7291666939854622,
|
| 1225 |
+
"reward_std": 0.45753754302859306,
|
| 1226 |
+
"rewards/accuracy_reward": 0.1979166716337204,
|
| 1227 |
+
"rewards/format_reward": 0.5312500149011612,
|
| 1228 |
+
"step": 87
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"completion_length": 3095.9584350585938,
|
| 1232 |
+
"entropy": 0.5673828125,
|
| 1233 |
+
"epoch": 0.10057142857142858,
|
| 1234 |
+
"grad_norm": 0.1463245153427124,
|
| 1235 |
+
"kl": 0.02208709716796875,
|
| 1236 |
+
"learning_rate": 1.3890454406082956e-07,
|
| 1237 |
+
"loss": 0.0721,
|
| 1238 |
+
"reward": 0.833333358168602,
|
| 1239 |
+
"reward_std": 0.7277083396911621,
|
| 1240 |
+
"rewards/accuracy_reward": 0.3437500149011612,
|
| 1241 |
+
"rewards/format_reward": 0.4895833507180214,
|
| 1242 |
+
"step": 88
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"completion_length": 3526.2605590820312,
|
| 1246 |
+
"entropy": 0.51171875,
|
| 1247 |
+
"epoch": 0.10171428571428572,
|
| 1248 |
+
"grad_norm": 0.058177653700113297,
|
| 1249 |
+
"kl": 0.00699615478515625,
|
| 1250 |
+
"learning_rate": 1.3276726544494571e-07,
|
| 1251 |
+
"loss": 0.024,
|
| 1252 |
+
"reward": 0.4375000111758709,
|
| 1253 |
+
"reward_std": 0.34694401547312737,
|
| 1254 |
+
"rewards/accuracy_reward": 0.14583334140479565,
|
| 1255 |
+
"rewards/format_reward": 0.29166666977107525,
|
| 1256 |
+
"step": 89
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"completion_length": 2816.729248046875,
|
| 1260 |
+
"entropy": 0.724609375,
|
| 1261 |
+
"epoch": 0.10285714285714286,
|
| 1262 |
+
"grad_norm": 0.0412394180893898,
|
| 1263 |
+
"kl": 0.01495361328125,
|
| 1264 |
+
"learning_rate": 1.2713832064634125e-07,
|
| 1265 |
+
"loss": 0.0118,
|
| 1266 |
+
"reward": 0.53125,
|
| 1267 |
+
"reward_std": 0.19076444953680038,
|
| 1268 |
+
"rewards/accuracy_reward": 0.041666666977107525,
|
| 1269 |
+
"rewards/format_reward": 0.4895833432674408,
|
| 1270 |
+
"step": 90
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"completion_length": 3363.0521240234375,
|
| 1274 |
+
"entropy": 0.548828125,
|
| 1275 |
+
"epoch": 0.104,
|
| 1276 |
+
"grad_norm": 0.09054948389530182,
|
| 1277 |
+
"kl": 0.0079193115234375,
|
| 1278 |
+
"learning_rate": 1.220245676671809e-07,
|
| 1279 |
+
"loss": 0.0385,
|
| 1280 |
+
"reward": 0.6458333358168602,
|
| 1281 |
+
"reward_std": 0.48818454146385193,
|
| 1282 |
+
"rewards/accuracy_reward": 0.2291666679084301,
|
| 1283 |
+
"rewards/format_reward": 0.4166666679084301,
|
| 1284 |
+
"step": 91
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"completion_length": 3039.9896850585938,
|
| 1288 |
+
"entropy": 0.52685546875,
|
| 1289 |
+
"epoch": 0.10514285714285715,
|
| 1290 |
+
"grad_norm": 0.3311564326286316,
|
| 1291 |
+
"kl": 0.111328125,
|
| 1292 |
+
"learning_rate": 1.1743223682775649e-07,
|
| 1293 |
+
"loss": 0.0405,
|
| 1294 |
+
"reward": 0.7187500298023224,
|
| 1295 |
+
"reward_std": 0.5042702481150627,
|
| 1296 |
+
"rewards/accuracy_reward": 0.17708333861082792,
|
| 1297 |
+
"rewards/format_reward": 0.541666679084301,
|
| 1298 |
+
"step": 92
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"completion_length": 3752.4166870117188,
|
| 1302 |
+
"entropy": 0.68115234375,
|
| 1303 |
+
"epoch": 0.10628571428571429,
|
| 1304 |
+
"grad_norm": 0.07829831540584564,
|
| 1305 |
+
"kl": 0.0079345703125,
|
| 1306 |
+
"learning_rate": 1.1336692317580158e-07,
|
| 1307 |
+
"loss": 0.0213,
|
| 1308 |
+
"reward": 0.07291666697710752,
|
| 1309 |
+
"reward_std": 0.10136350989341736,
|
| 1310 |
+
"rewards/accuracy_reward": 0.0,
|
| 1311 |
+
"rewards/format_reward": 0.07291666697710752,
|
| 1312 |
+
"step": 93
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"completion_length": 3379.666748046875,
|
| 1316 |
+
"entropy": 0.6376953125,
|
| 1317 |
+
"epoch": 0.10742857142857143,
|
| 1318 |
+
"grad_norm": 0.0632358267903328,
|
| 1319 |
+
"kl": 0.009815216064453125,
|
| 1320 |
+
"learning_rate": 1.0983357966978745e-07,
|
| 1321 |
+
"loss": 0.0332,
|
| 1322 |
+
"reward": 0.4687500186264515,
|
| 1323 |
+
"reward_std": 0.3699027970433235,
|
| 1324 |
+
"rewards/accuracy_reward": 0.1354166716337204,
|
| 1325 |
+
"rewards/format_reward": 0.3333333395421505,
|
| 1326 |
+
"step": 94
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"completion_length": 3759.7500610351562,
|
| 1330 |
+
"entropy": 0.50048828125,
|
| 1331 |
+
"epoch": 0.10857142857142857,
|
| 1332 |
+
"grad_norm": 0.06647934019565582,
|
| 1333 |
+
"kl": 0.0034885406494140625,
|
| 1334 |
+
"learning_rate": 1.068365111445064e-07,
|
| 1335 |
+
"loss": 0.0212,
|
| 1336 |
+
"reward": 0.4270833386108279,
|
| 1337 |
+
"reward_std": 0.5315546244382858,
|
| 1338 |
+
"rewards/accuracy_reward": 0.13541667442768812,
|
| 1339 |
+
"rewards/format_reward": 0.2916666744276881,
|
| 1340 |
+
"step": 95
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"completion_length": 3273.229248046875,
|
| 1344 |
+
"entropy": 0.5556640625,
|
| 1345 |
+
"epoch": 0.10971428571428571,
|
| 1346 |
+
"grad_norm": 0.08489354699850082,
|
| 1347 |
+
"kl": 0.008378982543945312,
|
| 1348 |
+
"learning_rate": 1.0437936906629334e-07,
|
| 1349 |
+
"loss": 0.0477,
|
| 1350 |
+
"reward": 0.6458333507180214,
|
| 1351 |
+
"reward_std": 0.48296865820884705,
|
| 1352 |
+
"rewards/accuracy_reward": 0.2604166716337204,
|
| 1353 |
+
"rewards/format_reward": 0.385416679084301,
|
| 1354 |
+
"step": 96
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"completion_length": 3500.1563110351562,
|
| 1358 |
+
"entropy": 0.546875,
|
| 1359 |
+
"epoch": 0.11085714285714286,
|
| 1360 |
+
"grad_norm": 0.03905348479747772,
|
| 1361 |
+
"kl": 0.0056133270263671875,
|
| 1362 |
+
"learning_rate": 1.0246514708427701e-07,
|
| 1363 |
+
"loss": 0.0213,
|
| 1364 |
+
"reward": 0.531250013038516,
|
| 1365 |
+
"reward_std": 0.2828656807541847,
|
| 1366 |
+
"rewards/accuracy_reward": 0.22916667722165585,
|
| 1367 |
+
"rewards/format_reward": 0.3020833386108279,
|
| 1368 |
+
"step": 97
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"completion_length": 3151.5625610351562,
|
| 1372 |
+
"entropy": 0.49755859375,
|
| 1373 |
+
"epoch": 0.112,
|
| 1374 |
+
"grad_norm": 0.06831816583871841,
|
| 1375 |
+
"kl": 0.00501251220703125,
|
| 1376 |
+
"learning_rate": 1.0109617738307911e-07,
|
| 1377 |
+
"loss": 0.0345,
|
| 1378 |
+
"reward": 0.7187500298023224,
|
| 1379 |
+
"reward_std": 0.5354187488555908,
|
| 1380 |
+
"rewards/accuracy_reward": 0.18750001024454832,
|
| 1381 |
+
"rewards/format_reward": 0.5312500149011612,
|
| 1382 |
+
"step": 98
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"completion_length": 3112.8646545410156,
|
| 1386 |
+
"entropy": 0.4921875,
|
| 1387 |
+
"epoch": 0.11314285714285714,
|
| 1388 |
+
"grad_norm": 0.05772259458899498,
|
| 1389 |
+
"kl": 0.0065765380859375,
|
| 1390 |
+
"learning_rate": 1.002741278414069e-07,
|
| 1391 |
+
"loss": 0.0306,
|
| 1392 |
+
"reward": 0.5312500055879354,
|
| 1393 |
+
"reward_std": 0.38146835565567017,
|
| 1394 |
+
"rewards/accuracy_reward": 0.1979166716337204,
|
| 1395 |
+
"rewards/format_reward": 0.33333334140479565,
|
| 1396 |
+
"step": 99
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"completion_length": 3131.6146240234375,
|
| 1400 |
+
"entropy": 0.49755859375,
|
| 1401 |
+
"epoch": 0.11428571428571428,
|
| 1402 |
+
"grad_norm": 0.10680815577507019,
|
| 1403 |
+
"kl": 0.00878143310546875,
|
| 1404 |
+
"learning_rate": 1e-07,
|
| 1405 |
+
"loss": 0.0572,
|
| 1406 |
+
"reward": 0.7395833358168602,
|
| 1407 |
+
"reward_std": 0.4930148720741272,
|
| 1408 |
+
"rewards/accuracy_reward": 0.2708333358168602,
|
| 1409 |
+
"rewards/format_reward": 0.4687500074505806,
|
| 1410 |
+
"step": 100
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.11428571428571428,
|
| 1414 |
+
"step": 100,
|
| 1415 |
+
"total_flos": 0.0,
|
| 1416 |
+
"train_loss": 0.03577750400174409,
|
| 1417 |
+
"train_runtime": 8397.3301,
|
| 1418 |
+
"train_samples_per_second": 1.143,
|
| 1419 |
+
"train_steps_per_second": 0.012
|
| 1420 |
+
}
|
| 1421 |
+
],
|
| 1422 |
+
"logging_steps": 1,
|
| 1423 |
+
"max_steps": 100,
|
| 1424 |
+
"num_input_tokens_seen": 0,
|
| 1425 |
+
"num_train_epochs": 1,
|
| 1426 |
+
"save_steps": 50,
|
| 1427 |
+
"stateful_callbacks": {
|
| 1428 |
+
"TrainerControl": {
|
| 1429 |
+
"args": {
|
| 1430 |
+
"should_epoch_stop": false,
|
| 1431 |
+
"should_evaluate": false,
|
| 1432 |
+
"should_log": false,
|
| 1433 |
+
"should_save": true,
|
| 1434 |
+
"should_training_stop": true
|
| 1435 |
+
},
|
| 1436 |
+
"attributes": {}
|
| 1437 |
+
}
|
| 1438 |
+
},
|
| 1439 |
+
"total_flos": 0.0,
|
| 1440 |
+
"train_batch_size": 6,
|
| 1441 |
+
"trial_name": null,
|
| 1442 |
+
"trial_params": null
|
| 1443 |
+
}
|