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 +1197 -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: DeepSeek-R1-Distill-Qwen-1.5B-GRPO
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- grpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for DeepSeek-R1-Distill-Qwen-1.5B-GRPO
|
| 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="blacksunfm/DeepSeek-R1-Distill-Qwen-1.5B-GRPO", 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/blacksunfm23-nwpu/huggingface/runs/hvn5cd05)
|
| 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: 0.16.0.dev0
|
| 38 |
+
- Transformers: 4.49.0
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.3.2
|
| 41 |
+
- Tokenizers: 0.21.0
|
| 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": 0.13210956121238504,
|
| 4 |
+
"train_runtime": 6757.7515,
|
| 5 |
+
"train_samples": 311,
|
| 6 |
+
"train_samples_per_second": 0.046,
|
| 7 |
+
"train_steps_per_second": 0.011
|
| 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": "4.49.0"
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.13210956121238504,
|
| 4 |
+
"train_runtime": 6757.7515,
|
| 5 |
+
"train_samples": 311,
|
| 6 |
+
"train_samples_per_second": 0.046,
|
| 7 |
+
"train_steps_per_second": 0.011
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1197 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9903536977491961,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 77,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"clip_ratio": 0.0,
|
| 13 |
+
"completion_length": 1084.34375,
|
| 14 |
+
"epoch": 0.012861736334405145,
|
| 15 |
+
"grad_norm": 0.41294848918914795,
|
| 16 |
+
"kl": 0.0,
|
| 17 |
+
"learning_rate": 1.25e-07,
|
| 18 |
+
"loss": 0.053,
|
| 19 |
+
"reward": 1.54296875,
|
| 20 |
+
"reward_std": 0.11134403571486473,
|
| 21 |
+
"rewards/accuracy_reward": 1.0,
|
| 22 |
+
"rewards/format_reward": 0.0,
|
| 23 |
+
"rewards/tag_count_reward": 0.54296875,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"clip_ratio": 0.0,
|
| 28 |
+
"completion_length": 912.421875,
|
| 29 |
+
"epoch": 0.02572347266881029,
|
| 30 |
+
"grad_norm": 0.5866098403930664,
|
| 31 |
+
"kl": 0.0,
|
| 32 |
+
"learning_rate": 2.5e-07,
|
| 33 |
+
"loss": 0.1483,
|
| 34 |
+
"reward": 1.5625,
|
| 35 |
+
"reward_std": 0.15145022049546242,
|
| 36 |
+
"rewards/accuracy_reward": 1.0,
|
| 37 |
+
"rewards/format_reward": 0.0,
|
| 38 |
+
"rewards/tag_count_reward": 0.5625,
|
| 39 |
+
"step": 2
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"clip_ratio": 0.0,
|
| 43 |
+
"completion_length": 1136.859375,
|
| 44 |
+
"epoch": 0.03858520900321544,
|
| 45 |
+
"grad_norm": 0.5072430372238159,
|
| 46 |
+
"kl": 0.00023818016052246094,
|
| 47 |
+
"learning_rate": 3.75e-07,
|
| 48 |
+
"loss": 0.117,
|
| 49 |
+
"reward": 1.57421875,
|
| 50 |
+
"reward_std": 0.1861907932907343,
|
| 51 |
+
"rewards/accuracy_reward": 1.0,
|
| 52 |
+
"rewards/format_reward": 0.0,
|
| 53 |
+
"rewards/tag_count_reward": 0.57421875,
|
| 54 |
+
"step": 3
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"clip_ratio": 0.0,
|
| 58 |
+
"completion_length": 1155.015625,
|
| 59 |
+
"epoch": 0.05144694533762058,
|
| 60 |
+
"grad_norm": 0.4072016179561615,
|
| 61 |
+
"kl": 0.00027489662170410156,
|
| 62 |
+
"learning_rate": 5e-07,
|
| 63 |
+
"loss": 0.0995,
|
| 64 |
+
"reward": 1.5390625,
|
| 65 |
+
"reward_std": 0.10738485306501389,
|
| 66 |
+
"rewards/accuracy_reward": 1.0,
|
| 67 |
+
"rewards/format_reward": 0.0,
|
| 68 |
+
"rewards/tag_count_reward": 0.5390625,
|
| 69 |
+
"step": 4
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"clip_ratio": 0.0,
|
| 73 |
+
"completion_length": 966.703125,
|
| 74 |
+
"epoch": 0.06430868167202572,
|
| 75 |
+
"grad_norm": 0.5740306377410889,
|
| 76 |
+
"kl": 0.0003867149353027344,
|
| 77 |
+
"learning_rate": 6.249999999999999e-07,
|
| 78 |
+
"loss": 0.1656,
|
| 79 |
+
"reward": 1.57421875,
|
| 80 |
+
"reward_std": 0.11065823957324028,
|
| 81 |
+
"rewards/accuracy_reward": 1.0,
|
| 82 |
+
"rewards/format_reward": 0.0,
|
| 83 |
+
"rewards/tag_count_reward": 0.57421875,
|
| 84 |
+
"step": 5
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"clip_ratio": 0.0,
|
| 88 |
+
"completion_length": 1206.765625,
|
| 89 |
+
"epoch": 0.07717041800643087,
|
| 90 |
+
"grad_norm": 0.4439733028411865,
|
| 91 |
+
"kl": 0.00043845176696777344,
|
| 92 |
+
"learning_rate": 7.5e-07,
|
| 93 |
+
"loss": 0.0946,
|
| 94 |
+
"reward": 1.515625,
|
| 95 |
+
"reward_std": 0.08086910098791122,
|
| 96 |
+
"rewards/accuracy_reward": 1.0,
|
| 97 |
+
"rewards/format_reward": 0.0,
|
| 98 |
+
"rewards/tag_count_reward": 0.515625,
|
| 99 |
+
"step": 6
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"clip_ratio": 0.0,
|
| 103 |
+
"completion_length": 1010.46875,
|
| 104 |
+
"epoch": 0.09003215434083602,
|
| 105 |
+
"grad_norm": 0.2982307970523834,
|
| 106 |
+
"kl": 0.00031566619873046875,
|
| 107 |
+
"learning_rate": 8.75e-07,
|
| 108 |
+
"loss": 0.0943,
|
| 109 |
+
"reward": 1.53125,
|
| 110 |
+
"reward_std": 0.09774631634354591,
|
| 111 |
+
"rewards/accuracy_reward": 1.0,
|
| 112 |
+
"rewards/format_reward": 0.0,
|
| 113 |
+
"rewards/tag_count_reward": 0.53125,
|
| 114 |
+
"step": 7
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"clip_ratio": 0.0,
|
| 118 |
+
"completion_length": 1046.9375,
|
| 119 |
+
"epoch": 0.10289389067524116,
|
| 120 |
+
"grad_norm": 0.4670819938182831,
|
| 121 |
+
"kl": 0.00023698806762695312,
|
| 122 |
+
"learning_rate": 1e-06,
|
| 123 |
+
"loss": 0.1583,
|
| 124 |
+
"reward": 1.5390625,
|
| 125 |
+
"reward_std": 0.13101092725992203,
|
| 126 |
+
"rewards/accuracy_reward": 1.0,
|
| 127 |
+
"rewards/format_reward": 0.0,
|
| 128 |
+
"rewards/tag_count_reward": 0.5390625,
|
| 129 |
+
"step": 8
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"clip_ratio": 0.0,
|
| 133 |
+
"completion_length": 966.28125,
|
| 134 |
+
"epoch": 0.1157556270096463,
|
| 135 |
+
"grad_norm": 0.40219369530677795,
|
| 136 |
+
"kl": 0.0002582073211669922,
|
| 137 |
+
"learning_rate": 9.995336531410273e-07,
|
| 138 |
+
"loss": 0.1739,
|
| 139 |
+
"reward": 1.546875,
|
| 140 |
+
"reward_std": 0.1388406604528427,
|
| 141 |
+
"rewards/accuracy_reward": 1.0,
|
| 142 |
+
"rewards/format_reward": 0.0,
|
| 143 |
+
"rewards/tag_count_reward": 0.546875,
|
| 144 |
+
"step": 9
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"clip_ratio": 0.0,
|
| 148 |
+
"completion_length": 1078.390625,
|
| 149 |
+
"epoch": 0.12861736334405144,
|
| 150 |
+
"grad_norm": 0.291391521692276,
|
| 151 |
+
"kl": 0.00041747093200683594,
|
| 152 |
+
"learning_rate": 9.98135579139189e-07,
|
| 153 |
+
"loss": -0.0025,
|
| 154 |
+
"reward": 1.53125,
|
| 155 |
+
"reward_std": 0.04564354941248894,
|
| 156 |
+
"rewards/accuracy_reward": 1.0,
|
| 157 |
+
"rewards/format_reward": 0.0,
|
| 158 |
+
"rewards/tag_count_reward": 0.53125,
|
| 159 |
+
"step": 10
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"clip_ratio": 0.0,
|
| 163 |
+
"completion_length": 1055.640625,
|
| 164 |
+
"epoch": 0.1414790996784566,
|
| 165 |
+
"grad_norm": 0.34915173053741455,
|
| 166 |
+
"kl": 0.00037789344787597656,
|
| 167 |
+
"learning_rate": 9.958086757163488e-07,
|
| 168 |
+
"loss": 0.0703,
|
| 169 |
+
"reward": 1.56640625,
|
| 170 |
+
"reward_std": 0.10219397768378258,
|
| 171 |
+
"rewards/accuracy_reward": 1.0,
|
| 172 |
+
"rewards/format_reward": 0.0,
|
| 173 |
+
"rewards/tag_count_reward": 0.56640625,
|
| 174 |
+
"step": 11
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"clip_ratio": 0.0,
|
| 178 |
+
"completion_length": 1095.640625,
|
| 179 |
+
"epoch": 0.15434083601286175,
|
| 180 |
+
"grad_norm": 0.550199031829834,
|
| 181 |
+
"kl": 0.0002205371856689453,
|
| 182 |
+
"learning_rate": 9.92557765735184e-07,
|
| 183 |
+
"loss": 0.2093,
|
| 184 |
+
"reward": 1.58203125,
|
| 185 |
+
"reward_std": 0.19091684743762016,
|
| 186 |
+
"rewards/accuracy_reward": 1.0,
|
| 187 |
+
"rewards/format_reward": 0.0,
|
| 188 |
+
"rewards/tag_count_reward": 0.58203125,
|
| 189 |
+
"step": 12
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"clip_ratio": 0.0,
|
| 193 |
+
"completion_length": 1095.921875,
|
| 194 |
+
"epoch": 0.16720257234726688,
|
| 195 |
+
"grad_norm": 0.4369826316833496,
|
| 196 |
+
"kl": 0.00036525726318359375,
|
| 197 |
+
"learning_rate": 9.883895872030657e-07,
|
| 198 |
+
"loss": 0.1704,
|
| 199 |
+
"reward": 1.52734375,
|
| 200 |
+
"reward_std": 0.09039457142353058,
|
| 201 |
+
"rewards/accuracy_reward": 1.0,
|
| 202 |
+
"rewards/format_reward": 0.0,
|
| 203 |
+
"rewards/tag_count_reward": 0.52734375,
|
| 204 |
+
"step": 13
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"clip_ratio": 0.0,
|
| 208 |
+
"completion_length": 837.65625,
|
| 209 |
+
"epoch": 0.18006430868167203,
|
| 210 |
+
"grad_norm": 0.25077515840530396,
|
| 211 |
+
"kl": 0.0001766681671142578,
|
| 212 |
+
"learning_rate": 9.833127793065097e-07,
|
| 213 |
+
"loss": 0.0155,
|
| 214 |
+
"reward": 1.50390625,
|
| 215 |
+
"reward_std": 0.015625,
|
| 216 |
+
"rewards/accuracy_reward": 1.0,
|
| 217 |
+
"rewards/format_reward": 0.0,
|
| 218 |
+
"rewards/tag_count_reward": 0.50390625,
|
| 219 |
+
"step": 14
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"clip_ratio": 0.0,
|
| 223 |
+
"completion_length": 1183.109375,
|
| 224 |
+
"epoch": 0.19292604501607716,
|
| 225 |
+
"grad_norm": 0.42581164836883545,
|
| 226 |
+
"kl": 0.00038433074951171875,
|
| 227 |
+
"learning_rate": 9.773378645051436e-07,
|
| 228 |
+
"loss": 0.1642,
|
| 229 |
+
"reward": 1.5,
|
| 230 |
+
"reward_std": 0.06994619965553284,
|
| 231 |
+
"rewards/accuracy_reward": 1.0,
|
| 232 |
+
"rewards/format_reward": 0.0,
|
| 233 |
+
"rewards/tag_count_reward": 0.5,
|
| 234 |
+
"step": 15
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"clip_ratio": 0.0,
|
| 238 |
+
"completion_length": 1042.71875,
|
| 239 |
+
"epoch": 0.2057877813504823,
|
| 240 |
+
"grad_norm": 0.371883749961853,
|
| 241 |
+
"kl": 0.00041294097900390625,
|
| 242 |
+
"learning_rate": 9.704772267223019e-07,
|
| 243 |
+
"loss": 0.0703,
|
| 244 |
+
"reward": 1.54296875,
|
| 245 |
+
"reward_std": 0.08989579975605011,
|
| 246 |
+
"rewards/accuracy_reward": 1.0,
|
| 247 |
+
"rewards/format_reward": 0.0,
|
| 248 |
+
"rewards/tag_count_reward": 0.54296875,
|
| 249 |
+
"step": 16
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"clip_ratio": 0.0,
|
| 253 |
+
"completion_length": 978.828125,
|
| 254 |
+
"epoch": 0.21864951768488747,
|
| 255 |
+
"grad_norm": 0.4985962212085724,
|
| 256 |
+
"kl": 0.00021839141845703125,
|
| 257 |
+
"learning_rate": 9.627450856774539e-07,
|
| 258 |
+
"loss": 0.079,
|
| 259 |
+
"reward": 1.5703125,
|
| 260 |
+
"reward_std": 0.11574668437242508,
|
| 261 |
+
"rewards/accuracy_reward": 1.0,
|
| 262 |
+
"rewards/format_reward": 0.0,
|
| 263 |
+
"rewards/tag_count_reward": 0.5703125,
|
| 264 |
+
"step": 17
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"clip_ratio": 0.0,
|
| 268 |
+
"completion_length": 850.4375,
|
| 269 |
+
"epoch": 0.2315112540192926,
|
| 270 |
+
"grad_norm": 0.49231991171836853,
|
| 271 |
+
"kl": 0.0002932548522949219,
|
| 272 |
+
"learning_rate": 9.541574674136632e-07,
|
| 273 |
+
"loss": 0.0227,
|
| 274 |
+
"reward": 1.53125,
|
| 275 |
+
"reward_std": 0.0816391110420227,
|
| 276 |
+
"rewards/accuracy_reward": 1.0,
|
| 277 |
+
"rewards/format_reward": 0.0,
|
| 278 |
+
"rewards/tag_count_reward": 0.53125,
|
| 279 |
+
"step": 18
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"clip_ratio": 0.0,
|
| 283 |
+
"completion_length": 1099.109375,
|
| 284 |
+
"epoch": 0.24437299035369775,
|
| 285 |
+
"grad_norm": 0.5843875408172607,
|
| 286 |
+
"kl": 0.0002193450927734375,
|
| 287 |
+
"learning_rate": 9.447321710811674e-07,
|
| 288 |
+
"loss": 0.1417,
|
| 289 |
+
"reward": 1.5390625,
|
| 290 |
+
"reward_std": 0.1500675454735756,
|
| 291 |
+
"rewards/accuracy_reward": 1.0,
|
| 292 |
+
"rewards/format_reward": 0.0,
|
| 293 |
+
"rewards/tag_count_reward": 0.5390625,
|
| 294 |
+
"step": 19
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"clip_ratio": 0.0,
|
| 298 |
+
"completion_length": 1058.03125,
|
| 299 |
+
"epoch": 0.2572347266881029,
|
| 300 |
+
"grad_norm": 0.6623404026031494,
|
| 301 |
+
"kl": 0.0004153251647949219,
|
| 302 |
+
"learning_rate": 9.344887320459198e-07,
|
| 303 |
+
"loss": 0.1164,
|
| 304 |
+
"reward": 1.55078125,
|
| 305 |
+
"reward_std": 0.10405365750193596,
|
| 306 |
+
"rewards/accuracy_reward": 1.0,
|
| 307 |
+
"rewards/format_reward": 0.0,
|
| 308 |
+
"rewards/tag_count_reward": 0.55078125,
|
| 309 |
+
"step": 20
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"clip_ratio": 0.0,
|
| 313 |
+
"completion_length": 820.734375,
|
| 314 |
+
"epoch": 0.27009646302250806,
|
| 315 |
+
"grad_norm": 0.6680777072906494,
|
| 316 |
+
"kl": 0.0003333091735839844,
|
| 317 |
+
"learning_rate": 9.234483813995613e-07,
|
| 318 |
+
"loss": 0.2424,
|
| 319 |
+
"reward": 1.55078125,
|
| 320 |
+
"reward_std": 0.174929304048419,
|
| 321 |
+
"rewards/accuracy_reward": 1.0,
|
| 322 |
+
"rewards/format_reward": 0.0,
|
| 323 |
+
"rewards/tag_count_reward": 0.55078125,
|
| 324 |
+
"step": 21
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"clip_ratio": 0.0,
|
| 328 |
+
"completion_length": 1022.640625,
|
| 329 |
+
"epoch": 0.2829581993569132,
|
| 330 |
+
"grad_norm": 0.6333020329475403,
|
| 331 |
+
"kl": 0.0004470348358154297,
|
| 332 |
+
"learning_rate": 9.116340019547401e-07,
|
| 333 |
+
"loss": 0.0632,
|
| 334 |
+
"reward": 1.51171875,
|
| 335 |
+
"reward_std": 0.08957062661647797,
|
| 336 |
+
"rewards/accuracy_reward": 1.0,
|
| 337 |
+
"rewards/format_reward": 0.0,
|
| 338 |
+
"rewards/tag_count_reward": 0.51171875,
|
| 339 |
+
"step": 22
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"clip_ratio": 0.0,
|
| 343 |
+
"completion_length": 1067.015625,
|
| 344 |
+
"epoch": 0.2958199356913183,
|
| 345 |
+
"grad_norm": 0.5886364579200745,
|
| 346 |
+
"kl": 0.0003254413604736328,
|
| 347 |
+
"learning_rate": 8.990700808169889e-07,
|
| 348 |
+
"loss": 0.1254,
|
| 349 |
+
"reward": 1.5703125,
|
| 350 |
+
"reward_std": 0.20152484998106956,
|
| 351 |
+
"rewards/accuracy_reward": 1.0,
|
| 352 |
+
"rewards/format_reward": 0.0,
|
| 353 |
+
"rewards/tag_count_reward": 0.5703125,
|
| 354 |
+
"step": 23
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"clip_ratio": 0.0,
|
| 358 |
+
"completion_length": 1248.078125,
|
| 359 |
+
"epoch": 0.3086816720257235,
|
| 360 |
+
"grad_norm": 0.6907972693443298,
|
| 361 |
+
"kl": 0.0006000995635986328,
|
| 362 |
+
"learning_rate": 8.857826586314586e-07,
|
| 363 |
+
"loss": 0.0326,
|
| 364 |
+
"reward": 1.54296875,
|
| 365 |
+
"reward_std": 0.13257849216461182,
|
| 366 |
+
"rewards/accuracy_reward": 1.0,
|
| 367 |
+
"rewards/format_reward": 0.0,
|
| 368 |
+
"rewards/tag_count_reward": 0.54296875,
|
| 369 |
+
"step": 24
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"clip_ratio": 0.0,
|
| 373 |
+
"completion_length": 1049.546875,
|
| 374 |
+
"epoch": 0.3215434083601286,
|
| 375 |
+
"grad_norm": 0.45920121669769287,
|
| 376 |
+
"kl": 0.0003612041473388672,
|
| 377 |
+
"learning_rate": 8.717992756097047e-07,
|
| 378 |
+
"loss": 0.0834,
|
| 379 |
+
"reward": 1.5546875,
|
| 380 |
+
"reward_std": 0.1335771605372429,
|
| 381 |
+
"rewards/accuracy_reward": 1.0,
|
| 382 |
+
"rewards/format_reward": 0.0,
|
| 383 |
+
"rewards/tag_count_reward": 0.5546875,
|
| 384 |
+
"step": 25
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"clip_ratio": 0.0,
|
| 388 |
+
"completion_length": 912.46875,
|
| 389 |
+
"epoch": 0.33440514469453375,
|
| 390 |
+
"grad_norm": 0.336818665266037,
|
| 391 |
+
"kl": 0.00023627281188964844,
|
| 392 |
+
"learning_rate": 8.571489144483944e-07,
|
| 393 |
+
"loss": 0.04,
|
| 394 |
+
"reward": 1.51953125,
|
| 395 |
+
"reward_std": 0.06524410098791122,
|
| 396 |
+
"rewards/accuracy_reward": 1.0,
|
| 397 |
+
"rewards/format_reward": 0.0,
|
| 398 |
+
"rewards/tag_count_reward": 0.51953125,
|
| 399 |
+
"step": 26
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"clip_ratio": 0.0,
|
| 403 |
+
"completion_length": 1066.484375,
|
| 404 |
+
"epoch": 0.34726688102893893,
|
| 405 |
+
"grad_norm": 0.39180704951286316,
|
| 406 |
+
"kl": 0.00023651123046875,
|
| 407 |
+
"learning_rate": 8.418619402582402e-07,
|
| 408 |
+
"loss": 0.1659,
|
| 409 |
+
"reward": 1.546875,
|
| 410 |
+
"reward_std": 0.12253788113594055,
|
| 411 |
+
"rewards/accuracy_reward": 1.0,
|
| 412 |
+
"rewards/format_reward": 0.0,
|
| 413 |
+
"rewards/tag_count_reward": 0.546875,
|
| 414 |
+
"step": 27
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"clip_ratio": 0.0,
|
| 418 |
+
"completion_length": 1128.390625,
|
| 419 |
+
"epoch": 0.36012861736334406,
|
| 420 |
+
"grad_norm": 0.7383729815483093,
|
| 421 |
+
"kl": 0.0007467269897460938,
|
| 422 |
+
"learning_rate": 8.259700376276723e-07,
|
| 423 |
+
"loss": 0.15,
|
| 424 |
+
"reward": 1.578125,
|
| 425 |
+
"reward_std": 0.14458869770169258,
|
| 426 |
+
"rewards/accuracy_reward": 1.0,
|
| 427 |
+
"rewards/format_reward": 0.0,
|
| 428 |
+
"rewards/tag_count_reward": 0.578125,
|
| 429 |
+
"step": 28
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"clip_ratio": 0.0,
|
| 433 |
+
"completion_length": 1031.703125,
|
| 434 |
+
"epoch": 0.3729903536977492,
|
| 435 |
+
"grad_norm": 0.730670154094696,
|
| 436 |
+
"kl": 0.0005695819854736328,
|
| 437 |
+
"learning_rate": 8.095061449516902e-07,
|
| 438 |
+
"loss": -0.0424,
|
| 439 |
+
"reward": 1.55859375,
|
| 440 |
+
"reward_std": 0.11572204157710075,
|
| 441 |
+
"rewards/accuracy_reward": 1.0,
|
| 442 |
+
"rewards/format_reward": 0.0,
|
| 443 |
+
"rewards/tag_count_reward": 0.55859375,
|
| 444 |
+
"step": 29
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"clip_ratio": 0.0,
|
| 448 |
+
"completion_length": 1088.359375,
|
| 449 |
+
"epoch": 0.3858520900321543,
|
| 450 |
+
"grad_norm": 0.4401358366012573,
|
| 451 |
+
"kl": 0.0004668235778808594,
|
| 452 |
+
"learning_rate": 7.92504386162009e-07,
|
| 453 |
+
"loss": 0.0101,
|
| 454 |
+
"reward": 1.58984375,
|
| 455 |
+
"reward_std": 0.10618764162063599,
|
| 456 |
+
"rewards/accuracy_reward": 1.0,
|
| 457 |
+
"rewards/format_reward": 0.0,
|
| 458 |
+
"rewards/tag_count_reward": 0.58984375,
|
| 459 |
+
"step": 30
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"clip_ratio": 0.0,
|
| 463 |
+
"completion_length": 970.046875,
|
| 464 |
+
"epoch": 0.3987138263665595,
|
| 465 |
+
"grad_norm": 0.57388836145401,
|
| 466 |
+
"kl": 0.00038433074951171875,
|
| 467 |
+
"learning_rate": 7.75e-07,
|
| 468 |
+
"loss": 0.0789,
|
| 469 |
+
"reward": 1.64453125,
|
| 470 |
+
"reward_std": 0.18705939501523972,
|
| 471 |
+
"rewards/accuracy_reward": 1.0,
|
| 472 |
+
"rewards/format_reward": 0.0,
|
| 473 |
+
"rewards/tag_count_reward": 0.64453125,
|
| 474 |
+
"step": 31
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"clip_ratio": 0.0,
|
| 478 |
+
"completion_length": 852.6875,
|
| 479 |
+
"epoch": 0.4115755627009646,
|
| 480 |
+
"grad_norm": 0.5620797872543335,
|
| 481 |
+
"kl": 0.0002961158752441406,
|
| 482 |
+
"learning_rate": 7.570292669790184e-07,
|
| 483 |
+
"loss": 0.0866,
|
| 484 |
+
"reward": 1.6171875,
|
| 485 |
+
"reward_std": 0.21375522762537003,
|
| 486 |
+
"rewards/accuracy_reward": 1.0,
|
| 487 |
+
"rewards/format_reward": 0.0,
|
| 488 |
+
"rewards/tag_count_reward": 0.6171875,
|
| 489 |
+
"step": 32
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"clip_ratio": 0.0,
|
| 493 |
+
"completion_length": 1189.84375,
|
| 494 |
+
"epoch": 0.42443729903536975,
|
| 495 |
+
"grad_norm": 0.4780595302581787,
|
| 496 |
+
"kl": 0.00036525726318359375,
|
| 497 |
+
"learning_rate": 7.38629434187499e-07,
|
| 498 |
+
"loss": -0.0424,
|
| 499 |
+
"reward": 1.5546875,
|
| 500 |
+
"reward_std": 0.10984911397099495,
|
| 501 |
+
"rewards/accuracy_reward": 1.0,
|
| 502 |
+
"rewards/format_reward": 0.0,
|
| 503 |
+
"rewards/tag_count_reward": 0.5546875,
|
| 504 |
+
"step": 33
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"clip_ratio": 0.0,
|
| 508 |
+
"completion_length": 1082.984375,
|
| 509 |
+
"epoch": 0.43729903536977494,
|
| 510 |
+
"grad_norm": 0.6845836043357849,
|
| 511 |
+
"kl": 0.0010368824005126953,
|
| 512 |
+
"learning_rate": 7.198386380886764e-07,
|
| 513 |
+
"loss": 0.0489,
|
| 514 |
+
"reward": 1.57421875,
|
| 515 |
+
"reward_std": 0.15323764458298683,
|
| 516 |
+
"rewards/accuracy_reward": 1.0,
|
| 517 |
+
"rewards/format_reward": 0.0,
|
| 518 |
+
"rewards/tag_count_reward": 0.57421875,
|
| 519 |
+
"step": 34
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"clip_ratio": 0.0,
|
| 523 |
+
"completion_length": 1034.34375,
|
| 524 |
+
"epoch": 0.45016077170418006,
|
| 525 |
+
"grad_norm": 0.46563801169395447,
|
| 526 |
+
"kl": 0.0003986358642578125,
|
| 527 |
+
"learning_rate": 7.006958254769437e-07,
|
| 528 |
+
"loss": 0.166,
|
| 529 |
+
"reward": 1.57421875,
|
| 530 |
+
"reward_std": 0.13553934916853905,
|
| 531 |
+
"rewards/accuracy_reward": 1.0,
|
| 532 |
+
"rewards/format_reward": 0.0,
|
| 533 |
+
"rewards/tag_count_reward": 0.57421875,
|
| 534 |
+
"step": 35
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"clip_ratio": 0.0,
|
| 538 |
+
"completion_length": 918.84375,
|
| 539 |
+
"epoch": 0.4630225080385852,
|
| 540 |
+
"grad_norm": 0.5318493247032166,
|
| 541 |
+
"kl": 0.0004329681396484375,
|
| 542 |
+
"learning_rate": 6.812406727546712e-07,
|
| 543 |
+
"loss": 0.0854,
|
| 544 |
+
"reward": 1.5625,
|
| 545 |
+
"reward_std": 0.1329439952969551,
|
| 546 |
+
"rewards/accuracy_reward": 1.0,
|
| 547 |
+
"rewards/format_reward": 0.0,
|
| 548 |
+
"rewards/tag_count_reward": 0.5625,
|
| 549 |
+
"step": 36
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"clip_ratio": 0.0,
|
| 553 |
+
"completion_length": 984.84375,
|
| 554 |
+
"epoch": 0.4758842443729904,
|
| 555 |
+
"grad_norm": 0.5245525240898132,
|
| 556 |
+
"kl": 0.00041675567626953125,
|
| 557 |
+
"learning_rate": 6.615135036968049e-07,
|
| 558 |
+
"loss": 0.1983,
|
| 559 |
+
"reward": 1.640625,
|
| 560 |
+
"reward_std": 0.18561063334345818,
|
| 561 |
+
"rewards/accuracy_reward": 1.0,
|
| 562 |
+
"rewards/format_reward": 0.015625,
|
| 563 |
+
"rewards/tag_count_reward": 0.625,
|
| 564 |
+
"step": 37
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"clip_ratio": 0.0,
|
| 568 |
+
"completion_length": 948.4375,
|
| 569 |
+
"epoch": 0.4887459807073955,
|
| 570 |
+
"grad_norm": 0.617463231086731,
|
| 571 |
+
"kl": 0.0012621879577636719,
|
| 572 |
+
"learning_rate": 6.415552058736853e-07,
|
| 573 |
+
"loss": 0.1642,
|
| 574 |
+
"reward": 1.65234375,
|
| 575 |
+
"reward_std": 0.205689437687397,
|
| 576 |
+
"rewards/accuracy_reward": 1.0,
|
| 577 |
+
"rewards/format_reward": 0.0,
|
| 578 |
+
"rewards/tag_count_reward": 0.65234375,
|
| 579 |
+
"step": 38
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"clip_ratio": 0.0,
|
| 583 |
+
"completion_length": 824.5,
|
| 584 |
+
"epoch": 0.5016077170418006,
|
| 585 |
+
"grad_norm": 0.6972614526748657,
|
| 586 |
+
"kl": 0.000629425048828125,
|
| 587 |
+
"learning_rate": 6.21407145905313e-07,
|
| 588 |
+
"loss": 0.1105,
|
| 589 |
+
"reward": 1.6484375,
|
| 590 |
+
"reward_std": 0.2235238291323185,
|
| 591 |
+
"rewards/accuracy_reward": 1.0,
|
| 592 |
+
"rewards/format_reward": 0.0,
|
| 593 |
+
"rewards/tag_count_reward": 0.6484375,
|
| 594 |
+
"step": 39
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"clip_ratio": 0.0,
|
| 598 |
+
"completion_length": 1114.5,
|
| 599 |
+
"epoch": 0.5144694533762058,
|
| 600 |
+
"grad_norm": 0.5985956788063049,
|
| 601 |
+
"kl": 0.0006375312805175781,
|
| 602 |
+
"learning_rate": 6.011110837227137e-07,
|
| 603 |
+
"loss": 0.1203,
|
| 604 |
+
"reward": 1.6015625,
|
| 605 |
+
"reward_std": 0.17044181004166603,
|
| 606 |
+
"rewards/accuracy_reward": 1.0,
|
| 607 |
+
"rewards/format_reward": 0.0,
|
| 608 |
+
"rewards/tag_count_reward": 0.6015625,
|
| 609 |
+
"step": 40
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"clip_ratio": 0.0,
|
| 613 |
+
"completion_length": 884.921875,
|
| 614 |
+
"epoch": 0.5273311897106109,
|
| 615 |
+
"grad_norm": 0.5041890740394592,
|
| 616 |
+
"kl": 0.000698089599609375,
|
| 617 |
+
"learning_rate": 5.80709086014102e-07,
|
| 618 |
+
"loss": 0.1753,
|
| 619 |
+
"reward": 1.609375,
|
| 620 |
+
"reward_std": 0.19224052503705025,
|
| 621 |
+
"rewards/accuracy_reward": 1.0,
|
| 622 |
+
"rewards/format_reward": 0.0,
|
| 623 |
+
"rewards/tag_count_reward": 0.609375,
|
| 624 |
+
"step": 41
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"clip_ratio": 0.0,
|
| 628 |
+
"completion_length": 1000.265625,
|
| 629 |
+
"epoch": 0.5401929260450161,
|
| 630 |
+
"grad_norm": 0.5194684267044067,
|
| 631 |
+
"kl": 0.000476837158203125,
|
| 632 |
+
"learning_rate": 5.602434390352476e-07,
|
| 633 |
+
"loss": 0.1213,
|
| 634 |
+
"reward": 1.6015625,
|
| 635 |
+
"reward_std": 0.19315097481012344,
|
| 636 |
+
"rewards/accuracy_reward": 1.0,
|
| 637 |
+
"rewards/format_reward": 0.0,
|
| 638 |
+
"rewards/tag_count_reward": 0.6015625,
|
| 639 |
+
"step": 42
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"clip_ratio": 0.0,
|
| 643 |
+
"completion_length": 973.03125,
|
| 644 |
+
"epoch": 0.5530546623794212,
|
| 645 |
+
"grad_norm": 0.6358352303504944,
|
| 646 |
+
"kl": 0.0004420280456542969,
|
| 647 |
+
"learning_rate": 5.397565609647524e-07,
|
| 648 |
+
"loss": 0.1533,
|
| 649 |
+
"reward": 1.671875,
|
| 650 |
+
"reward_std": 0.20554664731025696,
|
| 651 |
+
"rewards/accuracy_reward": 1.0,
|
| 652 |
+
"rewards/format_reward": 0.0,
|
| 653 |
+
"rewards/tag_count_reward": 0.671875,
|
| 654 |
+
"step": 43
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"clip_ratio": 0.0,
|
| 658 |
+
"completion_length": 1015.921875,
|
| 659 |
+
"epoch": 0.5659163987138264,
|
| 660 |
+
"grad_norm": 0.5505901575088501,
|
| 661 |
+
"kl": 0.0008034706115722656,
|
| 662 |
+
"learning_rate": 5.192909139858981e-07,
|
| 663 |
+
"loss": 0.1629,
|
| 664 |
+
"reward": 1.59375,
|
| 665 |
+
"reward_std": 0.16546358913183212,
|
| 666 |
+
"rewards/accuracy_reward": 1.0,
|
| 667 |
+
"rewards/format_reward": 0.0,
|
| 668 |
+
"rewards/tag_count_reward": 0.59375,
|
| 669 |
+
"step": 44
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"clip_ratio": 0.0,
|
| 673 |
+
"completion_length": 726.28125,
|
| 674 |
+
"epoch": 0.5787781350482315,
|
| 675 |
+
"grad_norm": 0.6235621571540833,
|
| 676 |
+
"kl": 0.0012774467468261719,
|
| 677 |
+
"learning_rate": 4.988889162772862e-07,
|
| 678 |
+
"loss": 0.2198,
|
| 679 |
+
"reward": 1.640625,
|
| 680 |
+
"reward_std": 0.19715734012424946,
|
| 681 |
+
"rewards/accuracy_reward": 1.0,
|
| 682 |
+
"rewards/format_reward": 0.0,
|
| 683 |
+
"rewards/tag_count_reward": 0.640625,
|
| 684 |
+
"step": 45
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"clip_ratio": 0.0,
|
| 688 |
+
"completion_length": 1159.125,
|
| 689 |
+
"epoch": 0.5916398713826366,
|
| 690 |
+
"grad_norm": 0.5912997722625732,
|
| 691 |
+
"kl": 0.0010008811950683594,
|
| 692 |
+
"learning_rate": 4.785928540946868e-07,
|
| 693 |
+
"loss": 0.179,
|
| 694 |
+
"reward": 1.56640625,
|
| 695 |
+
"reward_std": 0.12528811395168304,
|
| 696 |
+
"rewards/accuracy_reward": 1.0,
|
| 697 |
+
"rewards/format_reward": 0.0,
|
| 698 |
+
"rewards/tag_count_reward": 0.56640625,
|
| 699 |
+
"step": 46
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"clip_ratio": 0.0,
|
| 703 |
+
"completion_length": 1131.15625,
|
| 704 |
+
"epoch": 0.6045016077170418,
|
| 705 |
+
"grad_norm": 0.521943986415863,
|
| 706 |
+
"kl": 0.0008096694946289062,
|
| 707 |
+
"learning_rate": 4.584447941263149e-07,
|
| 708 |
+
"loss": 0.0605,
|
| 709 |
+
"reward": 1.58984375,
|
| 710 |
+
"reward_std": 0.12631270289421082,
|
| 711 |
+
"rewards/accuracy_reward": 1.0,
|
| 712 |
+
"rewards/format_reward": 0.0,
|
| 713 |
+
"rewards/tag_count_reward": 0.58984375,
|
| 714 |
+
"step": 47
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"clip_ratio": 0.0,
|
| 718 |
+
"completion_length": 1118.828125,
|
| 719 |
+
"epoch": 0.617363344051447,
|
| 720 |
+
"grad_norm": 0.4915071725845337,
|
| 721 |
+
"kl": 0.0010766983032226562,
|
| 722 |
+
"learning_rate": 4.384864963031951e-07,
|
| 723 |
+
"loss": 0.2512,
|
| 724 |
+
"reward": 1.6484375,
|
| 725 |
+
"reward_std": 0.2036767303943634,
|
| 726 |
+
"rewards/accuracy_reward": 1.0,
|
| 727 |
+
"rewards/format_reward": 0.0,
|
| 728 |
+
"rewards/tag_count_reward": 0.6484375,
|
| 729 |
+
"step": 48
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"clip_ratio": 0.0,
|
| 733 |
+
"completion_length": 1062.84375,
|
| 734 |
+
"epoch": 0.6302250803858521,
|
| 735 |
+
"grad_norm": 0.529114305973053,
|
| 736 |
+
"kl": 0.0007066726684570312,
|
| 737 |
+
"learning_rate": 4.187593272453288e-07,
|
| 738 |
+
"loss": 0.1496,
|
| 739 |
+
"reward": 1.60546875,
|
| 740 |
+
"reward_std": 0.15380773693323135,
|
| 741 |
+
"rewards/accuracy_reward": 1.0,
|
| 742 |
+
"rewards/format_reward": 0.0,
|
| 743 |
+
"rewards/tag_count_reward": 0.60546875,
|
| 744 |
+
"step": 49
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"clip_ratio": 0.0,
|
| 748 |
+
"completion_length": 694.140625,
|
| 749 |
+
"epoch": 0.6430868167202572,
|
| 750 |
+
"grad_norm": 0.6105093359947205,
|
| 751 |
+
"kl": 0.0016994476318359375,
|
| 752 |
+
"learning_rate": 3.993041745230562e-07,
|
| 753 |
+
"loss": 0.3051,
|
| 754 |
+
"reward": 1.203125,
|
| 755 |
+
"reward_std": 0.2077712006866932,
|
| 756 |
+
"rewards/accuracy_reward": 0.5,
|
| 757 |
+
"rewards/format_reward": 0.0,
|
| 758 |
+
"rewards/tag_count_reward": 0.703125,
|
| 759 |
+
"step": 50
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"clip_ratio": 0.0,
|
| 763 |
+
"completion_length": 982.140625,
|
| 764 |
+
"epoch": 0.6559485530546624,
|
| 765 |
+
"grad_norm": 0.540160596370697,
|
| 766 |
+
"kl": 0.0007996559143066406,
|
| 767 |
+
"learning_rate": 3.8016136191132354e-07,
|
| 768 |
+
"loss": 0.0934,
|
| 769 |
+
"reward": 1.609375,
|
| 770 |
+
"reward_std": 0.1588139720261097,
|
| 771 |
+
"rewards/accuracy_reward": 1.0,
|
| 772 |
+
"rewards/format_reward": 0.0,
|
| 773 |
+
"rewards/tag_count_reward": 0.609375,
|
| 774 |
+
"step": 51
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"clip_ratio": 0.0,
|
| 778 |
+
"completion_length": 937.734375,
|
| 779 |
+
"epoch": 0.6688102893890675,
|
| 780 |
+
"grad_norm": 0.6522411108016968,
|
| 781 |
+
"kl": 0.0015106201171875,
|
| 782 |
+
"learning_rate": 3.6137056581250137e-07,
|
| 783 |
+
"loss": 0.1704,
|
| 784 |
+
"reward": 1.6171875,
|
| 785 |
+
"reward_std": 0.1773756481707096,
|
| 786 |
+
"rewards/accuracy_reward": 1.0,
|
| 787 |
+
"rewards/format_reward": 0.0,
|
| 788 |
+
"rewards/tag_count_reward": 0.6171875,
|
| 789 |
+
"step": 52
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"clip_ratio": 0.0,
|
| 793 |
+
"completion_length": 831.9375,
|
| 794 |
+
"epoch": 0.6816720257234726,
|
| 795 |
+
"grad_norm": 0.6507072448730469,
|
| 796 |
+
"kl": 0.0011072158813476562,
|
| 797 |
+
"learning_rate": 3.4297073302098155e-07,
|
| 798 |
+
"loss": 0.1508,
|
| 799 |
+
"reward": 1.65625,
|
| 800 |
+
"reward_std": 0.1978173814713955,
|
| 801 |
+
"rewards/accuracy_reward": 1.0,
|
| 802 |
+
"rewards/format_reward": 0.0,
|
| 803 |
+
"rewards/tag_count_reward": 0.65625,
|
| 804 |
+
"step": 53
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"clip_ratio": 0.0,
|
| 808 |
+
"completion_length": 1121.0625,
|
| 809 |
+
"epoch": 0.6945337620578779,
|
| 810 |
+
"grad_norm": 0.5701295137405396,
|
| 811 |
+
"kl": 0.000888824462890625,
|
| 812 |
+
"learning_rate": 3.250000000000001e-07,
|
| 813 |
+
"loss": 0.204,
|
| 814 |
+
"reward": 1.65234375,
|
| 815 |
+
"reward_std": 0.20623698085546494,
|
| 816 |
+
"rewards/accuracy_reward": 1.0,
|
| 817 |
+
"rewards/format_reward": 0.0,
|
| 818 |
+
"rewards/tag_count_reward": 0.65234375,
|
| 819 |
+
"step": 54
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"clip_ratio": 0.0,
|
| 823 |
+
"completion_length": 880.375,
|
| 824 |
+
"epoch": 0.707395498392283,
|
| 825 |
+
"grad_norm": 0.7120354175567627,
|
| 826 |
+
"kl": 0.0023937225341796875,
|
| 827 |
+
"learning_rate": 3.0749561383799107e-07,
|
| 828 |
+
"loss": 0.1468,
|
| 829 |
+
"reward": 1.6640625,
|
| 830 |
+
"reward_std": 0.2058405578136444,
|
| 831 |
+
"rewards/accuracy_reward": 1.0,
|
| 832 |
+
"rewards/format_reward": 0.0,
|
| 833 |
+
"rewards/tag_count_reward": 0.6640625,
|
| 834 |
+
"step": 55
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"clip_ratio": 0.0,
|
| 838 |
+
"completion_length": 1107.71875,
|
| 839 |
+
"epoch": 0.7202572347266881,
|
| 840 |
+
"grad_norm": 0.49433115124702454,
|
| 841 |
+
"kl": 0.0011415481567382812,
|
| 842 |
+
"learning_rate": 2.904938550483098e-07,
|
| 843 |
+
"loss": 0.2208,
|
| 844 |
+
"reward": 1.62890625,
|
| 845 |
+
"reward_std": 0.18008438870310783,
|
| 846 |
+
"rewards/accuracy_reward": 1.0,
|
| 847 |
+
"rewards/format_reward": 0.0,
|
| 848 |
+
"rewards/tag_count_reward": 0.62890625,
|
| 849 |
+
"step": 56
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"clip_ratio": 0.0,
|
| 853 |
+
"completion_length": 1063.40625,
|
| 854 |
+
"epoch": 0.7331189710610932,
|
| 855 |
+
"grad_norm": 0.5648707151412964,
|
| 856 |
+
"kl": 0.0015516281127929688,
|
| 857 |
+
"learning_rate": 2.7402996237232757e-07,
|
| 858 |
+
"loss": 0.1323,
|
| 859 |
+
"reward": 1.58203125,
|
| 860 |
+
"reward_std": 0.1835322491824627,
|
| 861 |
+
"rewards/accuracy_reward": 1.0,
|
| 862 |
+
"rewards/format_reward": 0.0,
|
| 863 |
+
"rewards/tag_count_reward": 0.58203125,
|
| 864 |
+
"step": 57
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"clip_ratio": 0.0,
|
| 868 |
+
"completion_length": 842.5625,
|
| 869 |
+
"epoch": 0.7459807073954984,
|
| 870 |
+
"grad_norm": 0.6252850890159607,
|
| 871 |
+
"kl": 0.0014705657958984375,
|
| 872 |
+
"learning_rate": 2.5813805974175984e-07,
|
| 873 |
+
"loss": 0.2248,
|
| 874 |
+
"reward": 1.625,
|
| 875 |
+
"reward_std": 0.18506385013461113,
|
| 876 |
+
"rewards/accuracy_reward": 1.0,
|
| 877 |
+
"rewards/format_reward": 0.0,
|
| 878 |
+
"rewards/tag_count_reward": 0.625,
|
| 879 |
+
"step": 58
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"clip_ratio": 0.0,
|
| 883 |
+
"completion_length": 879.40625,
|
| 884 |
+
"epoch": 0.7588424437299035,
|
| 885 |
+
"grad_norm": 0.628139317035675,
|
| 886 |
+
"kl": 0.0014867782592773438,
|
| 887 |
+
"learning_rate": 2.4285108555160575e-07,
|
| 888 |
+
"loss": 0.0506,
|
| 889 |
+
"reward": 1.60546875,
|
| 890 |
+
"reward_std": 0.18587078154087067,
|
| 891 |
+
"rewards/accuracy_reward": 1.0,
|
| 892 |
+
"rewards/format_reward": 0.0,
|
| 893 |
+
"rewards/tag_count_reward": 0.60546875,
|
| 894 |
+
"step": 59
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"clip_ratio": 0.0,
|
| 898 |
+
"completion_length": 1132.859375,
|
| 899 |
+
"epoch": 0.7717041800643086,
|
| 900 |
+
"grad_norm": 0.4942401647567749,
|
| 901 |
+
"kl": 0.00098419189453125,
|
| 902 |
+
"learning_rate": 2.2820072439029523e-07,
|
| 903 |
+
"loss": 0.1978,
|
| 904 |
+
"reward": 1.671875,
|
| 905 |
+
"reward_std": 0.2286946102976799,
|
| 906 |
+
"rewards/accuracy_reward": 1.0,
|
| 907 |
+
"rewards/format_reward": 0.0,
|
| 908 |
+
"rewards/tag_count_reward": 0.671875,
|
| 909 |
+
"step": 60
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"clip_ratio": 0.0,
|
| 913 |
+
"completion_length": 1108.09375,
|
| 914 |
+
"epoch": 0.7845659163987139,
|
| 915 |
+
"grad_norm": 0.6833640336990356,
|
| 916 |
+
"kl": 0.0009565353393554688,
|
| 917 |
+
"learning_rate": 2.1421734136854153e-07,
|
| 918 |
+
"loss": 0.1842,
|
| 919 |
+
"reward": 1.625,
|
| 920 |
+
"reward_std": 0.20673153176903725,
|
| 921 |
+
"rewards/accuracy_reward": 1.0,
|
| 922 |
+
"rewards/format_reward": 0.0,
|
| 923 |
+
"rewards/tag_count_reward": 0.625,
|
| 924 |
+
"step": 61
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"clip_ratio": 0.0,
|
| 928 |
+
"completion_length": 1164.8125,
|
| 929 |
+
"epoch": 0.797427652733119,
|
| 930 |
+
"grad_norm": 0.550803005695343,
|
| 931 |
+
"kl": 0.0009765625,
|
| 932 |
+
"learning_rate": 2.0092991918301106e-07,
|
| 933 |
+
"loss": 0.0963,
|
| 934 |
+
"reward": 1.578125,
|
| 935 |
+
"reward_std": 0.18822026252746582,
|
| 936 |
+
"rewards/accuracy_reward": 1.0,
|
| 937 |
+
"rewards/format_reward": 0.0,
|
| 938 |
+
"rewards/tag_count_reward": 0.578125,
|
| 939 |
+
"step": 62
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"clip_ratio": 0.0,
|
| 943 |
+
"completion_length": 1222.84375,
|
| 944 |
+
"epoch": 0.8102893890675241,
|
| 945 |
+
"grad_norm": 0.47955459356307983,
|
| 946 |
+
"kl": 0.0008296966552734375,
|
| 947 |
+
"learning_rate": 1.883659980452598e-07,
|
| 948 |
+
"loss": 0.092,
|
| 949 |
+
"reward": 1.640625,
|
| 950 |
+
"reward_std": 0.1938280612230301,
|
| 951 |
+
"rewards/accuracy_reward": 1.0,
|
| 952 |
+
"rewards/format_reward": 0.0,
|
| 953 |
+
"rewards/tag_count_reward": 0.640625,
|
| 954 |
+
"step": 63
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"clip_ratio": 0.0,
|
| 958 |
+
"completion_length": 1009.125,
|
| 959 |
+
"epoch": 0.8231511254019293,
|
| 960 |
+
"grad_norm": 0.6472100615501404,
|
| 961 |
+
"kl": 0.0012683868408203125,
|
| 962 |
+
"learning_rate": 1.765516186004387e-07,
|
| 963 |
+
"loss": 0.1924,
|
| 964 |
+
"reward": 1.671875,
|
| 965 |
+
"reward_std": 0.20133696496486664,
|
| 966 |
+
"rewards/accuracy_reward": 1.0,
|
| 967 |
+
"rewards/format_reward": 0.0,
|
| 968 |
+
"rewards/tag_count_reward": 0.671875,
|
| 969 |
+
"step": 64
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"clip_ratio": 0.0,
|
| 973 |
+
"completion_length": 997.796875,
|
| 974 |
+
"epoch": 0.8360128617363344,
|
| 975 |
+
"grad_norm": 0.6434372067451477,
|
| 976 |
+
"kl": 0.00380706787109375,
|
| 977 |
+
"learning_rate": 1.6551126795408015e-07,
|
| 978 |
+
"loss": 0.2009,
|
| 979 |
+
"reward": 1.66796875,
|
| 980 |
+
"reward_std": 0.15900417417287827,
|
| 981 |
+
"rewards/accuracy_reward": 1.0,
|
| 982 |
+
"rewards/format_reward": 0.0,
|
| 983 |
+
"rewards/tag_count_reward": 0.66796875,
|
| 984 |
+
"step": 65
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"clip_ratio": 0.0,
|
| 988 |
+
"completion_length": 942.125,
|
| 989 |
+
"epoch": 0.8488745980707395,
|
| 990 |
+
"grad_norm": 0.5874162912368774,
|
| 991 |
+
"kl": 0.0014162063598632812,
|
| 992 |
+
"learning_rate": 1.552678289188326e-07,
|
| 993 |
+
"loss": 0.2217,
|
| 994 |
+
"reward": 1.6484375,
|
| 995 |
+
"reward_std": 0.18080222606658936,
|
| 996 |
+
"rewards/accuracy_reward": 1.0,
|
| 997 |
+
"rewards/format_reward": 0.0,
|
| 998 |
+
"rewards/tag_count_reward": 0.6484375,
|
| 999 |
+
"step": 66
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"clip_ratio": 0.0,
|
| 1003 |
+
"completion_length": 847.015625,
|
| 1004 |
+
"epoch": 0.8617363344051447,
|
| 1005 |
+
"grad_norm": 0.6690056920051575,
|
| 1006 |
+
"kl": 0.0014781951904296875,
|
| 1007 |
+
"learning_rate": 1.4584253258633681e-07,
|
| 1008 |
+
"loss": 0.1697,
|
| 1009 |
+
"reward": 1.6484375,
|
| 1010 |
+
"reward_std": 0.20908700302243233,
|
| 1011 |
+
"rewards/accuracy_reward": 1.0,
|
| 1012 |
+
"rewards/format_reward": 0.0,
|
| 1013 |
+
"rewards/tag_count_reward": 0.6484375,
|
| 1014 |
+
"step": 67
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"clip_ratio": 0.0,
|
| 1018 |
+
"completion_length": 967.515625,
|
| 1019 |
+
"epoch": 0.8745980707395499,
|
| 1020 |
+
"grad_norm": 0.6969915628433228,
|
| 1021 |
+
"kl": 0.0023241043090820312,
|
| 1022 |
+
"learning_rate": 1.3725491432254623e-07,
|
| 1023 |
+
"loss": 0.0535,
|
| 1024 |
+
"reward": 1.64453125,
|
| 1025 |
+
"reward_std": 0.12684179469943047,
|
| 1026 |
+
"rewards/accuracy_reward": 1.0,
|
| 1027 |
+
"rewards/format_reward": 0.0,
|
| 1028 |
+
"rewards/tag_count_reward": 0.64453125,
|
| 1029 |
+
"step": 68
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"clip_ratio": 0.0,
|
| 1033 |
+
"completion_length": 1140.578125,
|
| 1034 |
+
"epoch": 0.887459807073955,
|
| 1035 |
+
"grad_norm": 0.5747631192207336,
|
| 1036 |
+
"kl": 0.0011663436889648438,
|
| 1037 |
+
"learning_rate": 1.2952277327769804e-07,
|
| 1038 |
+
"loss": 0.1446,
|
| 1039 |
+
"reward": 1.6171875,
|
| 1040 |
+
"reward_std": 0.1967878546565771,
|
| 1041 |
+
"rewards/accuracy_reward": 1.0,
|
| 1042 |
+
"rewards/format_reward": 0.0,
|
| 1043 |
+
"rewards/tag_count_reward": 0.6171875,
|
| 1044 |
+
"step": 69
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"clip_ratio": 0.0,
|
| 1048 |
+
"completion_length": 1219.296875,
|
| 1049 |
+
"epoch": 0.9003215434083601,
|
| 1050 |
+
"grad_norm": 0.5590356588363647,
|
| 1051 |
+
"kl": 0.0020923614501953125,
|
| 1052 |
+
"learning_rate": 1.2266213549485637e-07,
|
| 1053 |
+
"loss": 0.1291,
|
| 1054 |
+
"reward": 1.58984375,
|
| 1055 |
+
"reward_std": 0.16909249685704708,
|
| 1056 |
+
"rewards/accuracy_reward": 1.0,
|
| 1057 |
+
"rewards/format_reward": 0.0,
|
| 1058 |
+
"rewards/tag_count_reward": 0.58984375,
|
| 1059 |
+
"step": 70
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"clip_ratio": 0.0,
|
| 1063 |
+
"completion_length": 871.53125,
|
| 1064 |
+
"epoch": 0.9131832797427653,
|
| 1065 |
+
"grad_norm": 0.6128804087638855,
|
| 1066 |
+
"kl": 0.002285003662109375,
|
| 1067 |
+
"learning_rate": 1.166872206934904e-07,
|
| 1068 |
+
"loss": 0.2013,
|
| 1069 |
+
"reward": 1.66015625,
|
| 1070 |
+
"reward_std": 0.18546020612120628,
|
| 1071 |
+
"rewards/accuracy_reward": 1.0,
|
| 1072 |
+
"rewards/format_reward": 0.0,
|
| 1073 |
+
"rewards/tag_count_reward": 0.66015625,
|
| 1074 |
+
"step": 71
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"clip_ratio": 0.0,
|
| 1078 |
+
"completion_length": 804.84375,
|
| 1079 |
+
"epoch": 0.9260450160771704,
|
| 1080 |
+
"grad_norm": 0.6195582151412964,
|
| 1081 |
+
"kl": 0.00205230712890625,
|
| 1082 |
+
"learning_rate": 1.1161041279693445e-07,
|
| 1083 |
+
"loss": 0.3434,
|
| 1084 |
+
"reward": 1.71875,
|
| 1085 |
+
"reward_std": 0.2264367826282978,
|
| 1086 |
+
"rewards/accuracy_reward": 1.0,
|
| 1087 |
+
"rewards/format_reward": 0.0,
|
| 1088 |
+
"rewards/tag_count_reward": 0.71875,
|
| 1089 |
+
"step": 72
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"clip_ratio": 0.0,
|
| 1093 |
+
"completion_length": 1082.171875,
|
| 1094 |
+
"epoch": 0.9389067524115756,
|
| 1095 |
+
"grad_norm": 0.6192907691001892,
|
| 1096 |
+
"kl": 0.0029478073120117188,
|
| 1097 |
+
"learning_rate": 1.074422342648161e-07,
|
| 1098 |
+
"loss": 0.2337,
|
| 1099 |
+
"reward": 1.625,
|
| 1100 |
+
"reward_std": 0.2141621820628643,
|
| 1101 |
+
"rewards/accuracy_reward": 1.0,
|
| 1102 |
+
"rewards/format_reward": 0.0,
|
| 1103 |
+
"rewards/tag_count_reward": 0.625,
|
| 1104 |
+
"step": 73
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"clip_ratio": 0.0,
|
| 1108 |
+
"completion_length": 614.75,
|
| 1109 |
+
"epoch": 0.9517684887459807,
|
| 1110 |
+
"grad_norm": 0.75156170129776,
|
| 1111 |
+
"kl": 0.0022068023681640625,
|
| 1112 |
+
"learning_rate": 1.0419132428365116e-07,
|
| 1113 |
+
"loss": 0.1886,
|
| 1114 |
+
"reward": 1.796875,
|
| 1115 |
+
"reward_std": 0.18980605155229568,
|
| 1116 |
+
"rewards/accuracy_reward": 1.0,
|
| 1117 |
+
"rewards/format_reward": 0.0,
|
| 1118 |
+
"rewards/tag_count_reward": 0.796875,
|
| 1119 |
+
"step": 74
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"clip_ratio": 0.0,
|
| 1123 |
+
"completion_length": 1100.421875,
|
| 1124 |
+
"epoch": 0.9646302250803859,
|
| 1125 |
+
"grad_norm": 0.4935654401779175,
|
| 1126 |
+
"kl": 0.0010290145874023438,
|
| 1127 |
+
"learning_rate": 1.0186442086081092e-07,
|
| 1128 |
+
"loss": 0.0693,
|
| 1129 |
+
"reward": 1.59375,
|
| 1130 |
+
"reward_std": 0.19873745739459991,
|
| 1131 |
+
"rewards/accuracy_reward": 1.0,
|
| 1132 |
+
"rewards/format_reward": 0.0,
|
| 1133 |
+
"rewards/tag_count_reward": 0.59375,
|
| 1134 |
+
"step": 75
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"clip_ratio": 0.0,
|
| 1138 |
+
"completion_length": 1082.59375,
|
| 1139 |
+
"epoch": 0.977491961414791,
|
| 1140 |
+
"grad_norm": 0.4804929196834564,
|
| 1141 |
+
"kl": 0.0011501312255859375,
|
| 1142 |
+
"learning_rate": 1.004663468589726e-07,
|
| 1143 |
+
"loss": 0.1592,
|
| 1144 |
+
"reward": 1.64453125,
|
| 1145 |
+
"reward_std": 0.19965127855539322,
|
| 1146 |
+
"rewards/accuracy_reward": 1.0,
|
| 1147 |
+
"rewards/format_reward": 0.0,
|
| 1148 |
+
"rewards/tag_count_reward": 0.64453125,
|
| 1149 |
+
"step": 76
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"clip_ratio": 0.0,
|
| 1153 |
+
"completion_length": 972.4643249511719,
|
| 1154 |
+
"epoch": 0.9903536977491961,
|
| 1155 |
+
"grad_norm": 0.5502123236656189,
|
| 1156 |
+
"kl": 0.0015010833740234375,
|
| 1157 |
+
"learning_rate": 1e-07,
|
| 1158 |
+
"loss": 0.0737,
|
| 1159 |
+
"reward": 1.6640625,
|
| 1160 |
+
"reward_std": 0.212241068482399,
|
| 1161 |
+
"rewards/accuracy_reward": 1.0,
|
| 1162 |
+
"rewards/format_reward": 0.0,
|
| 1163 |
+
"rewards/tag_count_reward": 0.6640625,
|
| 1164 |
+
"step": 77
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.9903536977491961,
|
| 1168 |
+
"step": 77,
|
| 1169 |
+
"total_flos": 0.0,
|
| 1170 |
+
"train_loss": 0.13210956121238504,
|
| 1171 |
+
"train_runtime": 6757.7515,
|
| 1172 |
+
"train_samples_per_second": 0.046,
|
| 1173 |
+
"train_steps_per_second": 0.011
|
| 1174 |
+
}
|
| 1175 |
+
],
|
| 1176 |
+
"logging_steps": 1,
|
| 1177 |
+
"max_steps": 77,
|
| 1178 |
+
"num_input_tokens_seen": 0,
|
| 1179 |
+
"num_train_epochs": 1,
|
| 1180 |
+
"save_steps": 500,
|
| 1181 |
+
"stateful_callbacks": {
|
| 1182 |
+
"TrainerControl": {
|
| 1183 |
+
"args": {
|
| 1184 |
+
"should_epoch_stop": false,
|
| 1185 |
+
"should_evaluate": false,
|
| 1186 |
+
"should_log": false,
|
| 1187 |
+
"should_save": true,
|
| 1188 |
+
"should_training_stop": true
|
| 1189 |
+
},
|
| 1190 |
+
"attributes": {}
|
| 1191 |
+
}
|
| 1192 |
+
},
|
| 1193 |
+
"total_flos": 0.0,
|
| 1194 |
+
"train_batch_size": 16,
|
| 1195 |
+
"trial_name": null,
|
| 1196 |
+
"trial_params": null
|
| 1197 |
+
}
|