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 +822 -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_generation_num_2
|
| 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_generation_num_2
|
| 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="EricLabile/DeepSeek-R1-Distill-Qwen-1.5B-GRPO_generation_num_2", 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/ericlabile/GRPO-Experiments/runs/gqctecwk)
|
| 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: 3.4.0
|
| 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": 0.01933252687404792,
|
| 4 |
+
"train_runtime": 2498.5955,
|
| 5 |
+
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 3.002,
|
| 7 |
+
"train_steps_per_second": 0.021
|
| 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": 0.01933252687404792,
|
| 4 |
+
"train_runtime": 2498.5955,
|
| 5 |
+
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 3.002,
|
| 7 |
+
"train_steps_per_second": 0.021
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,822 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9952153110047847,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 52,
|
| 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": 885.7360992431641,
|
| 14 |
+
"epoch": 0.019138755980861243,
|
| 15 |
+
"grad_norm": 0.05717110271611869,
|
| 16 |
+
"kl": 0.0,
|
| 17 |
+
"learning_rate": 1.6666666666666665e-07,
|
| 18 |
+
"loss": 0.0224,
|
| 19 |
+
"reward": 0.6623263955116272,
|
| 20 |
+
"reward_std": 0.20746709778904915,
|
| 21 |
+
"rewards/accuracy_reward": 0.2500000037252903,
|
| 22 |
+
"rewards/format_reward": 0.0,
|
| 23 |
+
"rewards/tag_count_reward": 0.4123263880610466,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"clip_ratio": 0.0,
|
| 28 |
+
"completion_length": 855.6076354980469,
|
| 29 |
+
"epoch": 0.03827751196172249,
|
| 30 |
+
"grad_norm": 0.06704605474432411,
|
| 31 |
+
"kl": 0.0,
|
| 32 |
+
"learning_rate": 3.333333333333333e-07,
|
| 33 |
+
"loss": 0.0199,
|
| 34 |
+
"reward": 0.6901041716337204,
|
| 35 |
+
"reward_std": 0.22956417128443718,
|
| 36 |
+
"rewards/accuracy_reward": 0.2465277761220932,
|
| 37 |
+
"rewards/format_reward": 0.0,
|
| 38 |
+
"rewards/tag_count_reward": 0.4435763880610466,
|
| 39 |
+
"step": 2
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"clip_ratio": 0.0,
|
| 43 |
+
"completion_length": 917.3541717529297,
|
| 44 |
+
"epoch": 0.05741626794258373,
|
| 45 |
+
"grad_norm": 0.05091560707629358,
|
| 46 |
+
"kl": 2.9027462005615234e-05,
|
| 47 |
+
"learning_rate": 5e-07,
|
| 48 |
+
"loss": 0.0137,
|
| 49 |
+
"reward": 0.5538194477558136,
|
| 50 |
+
"reward_std": 0.19887378066778183,
|
| 51 |
+
"rewards/accuracy_reward": 0.1770833358168602,
|
| 52 |
+
"rewards/format_reward": 0.0,
|
| 53 |
+
"rewards/tag_count_reward": 0.3767361119389534,
|
| 54 |
+
"step": 3
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"clip_ratio": 0.0,
|
| 58 |
+
"completion_length": 869.5937652587891,
|
| 59 |
+
"epoch": 0.07655502392344497,
|
| 60 |
+
"grad_norm": 0.06917728317111242,
|
| 61 |
+
"kl": 3.2007694244384766e-05,
|
| 62 |
+
"learning_rate": 6.666666666666666e-07,
|
| 63 |
+
"loss": 0.0258,
|
| 64 |
+
"reward": 0.634548619389534,
|
| 65 |
+
"reward_std": 0.2516612634062767,
|
| 66 |
+
"rewards/accuracy_reward": 0.2256944440305233,
|
| 67 |
+
"rewards/format_reward": 0.0,
|
| 68 |
+
"rewards/tag_count_reward": 0.4088541716337204,
|
| 69 |
+
"step": 4
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"clip_ratio": 0.0,
|
| 73 |
+
"completion_length": 860.3229217529297,
|
| 74 |
+
"epoch": 0.09569377990430622,
|
| 75 |
+
"grad_norm": 0.06655951788671848,
|
| 76 |
+
"kl": 2.7626752853393555e-05,
|
| 77 |
+
"learning_rate": 8.333333333333333e-07,
|
| 78 |
+
"loss": 0.0175,
|
| 79 |
+
"reward": 0.636284738779068,
|
| 80 |
+
"reward_std": 0.21483279019594193,
|
| 81 |
+
"rewards/accuracy_reward": 0.2222222238779068,
|
| 82 |
+
"rewards/format_reward": 0.0,
|
| 83 |
+
"rewards/tag_count_reward": 0.4140625074505806,
|
| 84 |
+
"step": 5
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"clip_ratio": 0.0,
|
| 88 |
+
"completion_length": 872.6909790039062,
|
| 89 |
+
"epoch": 0.11483253588516747,
|
| 90 |
+
"grad_norm": 0.06205063415295795,
|
| 91 |
+
"kl": 2.2977590560913086e-05,
|
| 92 |
+
"learning_rate": 1e-06,
|
| 93 |
+
"loss": 0.0139,
|
| 94 |
+
"reward": 0.6840277761220932,
|
| 95 |
+
"reward_std": 0.2086946927011013,
|
| 96 |
+
"rewards/accuracy_reward": 0.27083334140479565,
|
| 97 |
+
"rewards/format_reward": 0.0,
|
| 98 |
+
"rewards/tag_count_reward": 0.4131944477558136,
|
| 99 |
+
"step": 6
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"clip_ratio": 0.0,
|
| 103 |
+
"completion_length": 881.4861297607422,
|
| 104 |
+
"epoch": 0.1339712918660287,
|
| 105 |
+
"grad_norm": 0.06390474824070583,
|
| 106 |
+
"kl": 2.709031105041504e-05,
|
| 107 |
+
"learning_rate": 9.989509461357426e-07,
|
| 108 |
+
"loss": 0.0214,
|
| 109 |
+
"reward": 0.618923619389534,
|
| 110 |
+
"reward_std": 0.20255662873387337,
|
| 111 |
+
"rewards/accuracy_reward": 0.2256944477558136,
|
| 112 |
+
"rewards/format_reward": 0.0,
|
| 113 |
+
"rewards/tag_count_reward": 0.3932291641831398,
|
| 114 |
+
"step": 7
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"clip_ratio": 0.0,
|
| 118 |
+
"completion_length": 856.1423645019531,
|
| 119 |
+
"epoch": 0.15311004784688995,
|
| 120 |
+
"grad_norm": 0.06611386863145982,
|
| 121 |
+
"kl": 2.002716064453125e-05,
|
| 122 |
+
"learning_rate": 9.958086757163488e-07,
|
| 123 |
+
"loss": 0.0205,
|
| 124 |
+
"reward": 0.659722238779068,
|
| 125 |
+
"reward_std": 0.24306795373558998,
|
| 126 |
+
"rewards/accuracy_reward": 0.2291666716337204,
|
| 127 |
+
"rewards/format_reward": 0.0,
|
| 128 |
+
"rewards/tag_count_reward": 0.430555559694767,
|
| 129 |
+
"step": 8
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"clip_ratio": 0.0,
|
| 133 |
+
"completion_length": 875.7395935058594,
|
| 134 |
+
"epoch": 0.1722488038277512,
|
| 135 |
+
"grad_norm": 0.06272616959518927,
|
| 136 |
+
"kl": 1.7598271369934082e-05,
|
| 137 |
+
"learning_rate": 9.905878394570453e-07,
|
| 138 |
+
"loss": 0.0155,
|
| 139 |
+
"reward": 0.6875,
|
| 140 |
+
"reward_std": 0.22097085043787956,
|
| 141 |
+
"rewards/accuracy_reward": 0.2708333432674408,
|
| 142 |
+
"rewards/format_reward": 0.0,
|
| 143 |
+
"rewards/tag_count_reward": 0.4166666716337204,
|
| 144 |
+
"step": 9
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"clip_ratio": 0.0,
|
| 148 |
+
"completion_length": 885.6041717529297,
|
| 149 |
+
"epoch": 0.19138755980861244,
|
| 150 |
+
"grad_norm": 0.0566584544534715,
|
| 151 |
+
"kl": 1.3478100299835205e-05,
|
| 152 |
+
"learning_rate": 9.833127793065097e-07,
|
| 153 |
+
"loss": 0.0147,
|
| 154 |
+
"reward": 0.6840277910232544,
|
| 155 |
+
"reward_std": 0.21606039628386497,
|
| 156 |
+
"rewards/accuracy_reward": 0.2881944477558136,
|
| 157 |
+
"rewards/format_reward": 0.0,
|
| 158 |
+
"rewards/tag_count_reward": 0.3958333358168602,
|
| 159 |
+
"step": 10
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"clip_ratio": 0.0,
|
| 163 |
+
"completion_length": 849.3645782470703,
|
| 164 |
+
"epoch": 0.21052631578947367,
|
| 165 |
+
"grad_norm": 0.07330325365220204,
|
| 166 |
+
"kl": 2.1457672119140625e-05,
|
| 167 |
+
"learning_rate": 9.740174149534692e-07,
|
| 168 |
+
"loss": 0.0204,
|
| 169 |
+
"reward": 0.7526041716337204,
|
| 170 |
+
"reward_std": 0.2590269520878792,
|
| 171 |
+
"rewards/accuracy_reward": 0.3020833320915699,
|
| 172 |
+
"rewards/format_reward": 0.0,
|
| 173 |
+
"rewards/tag_count_reward": 0.4505208432674408,
|
| 174 |
+
"step": 11
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"clip_ratio": 0.0,
|
| 178 |
+
"completion_length": 876.4757080078125,
|
| 179 |
+
"epoch": 0.22966507177033493,
|
| 180 |
+
"grad_norm": 0.05630416735365695,
|
| 181 |
+
"kl": 1.9222497940063477e-05,
|
| 182 |
+
"learning_rate": 9.627450856774539e-07,
|
| 183 |
+
"loss": 0.0098,
|
| 184 |
+
"reward": 0.661458358168602,
|
| 185 |
+
"reward_std": 0.18905285000801086,
|
| 186 |
+
"rewards/accuracy_reward": 0.2291666716337204,
|
| 187 |
+
"rewards/format_reward": 0.0,
|
| 188 |
+
"rewards/tag_count_reward": 0.4322916716337204,
|
| 189 |
+
"step": 12
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"clip_ratio": 0.0,
|
| 193 |
+
"completion_length": 863.5590362548828,
|
| 194 |
+
"epoch": 0.24880382775119617,
|
| 195 |
+
"grad_norm": 0.060825286332779054,
|
| 196 |
+
"kl": 1.8551945686340332e-05,
|
| 197 |
+
"learning_rate": 9.495483482810687e-07,
|
| 198 |
+
"loss": 0.0179,
|
| 199 |
+
"reward": 0.6163194477558136,
|
| 200 |
+
"reward_std": 0.22097087278962135,
|
| 201 |
+
"rewards/accuracy_reward": 0.2048611119389534,
|
| 202 |
+
"rewards/format_reward": 0.0,
|
| 203 |
+
"rewards/tag_count_reward": 0.4114583358168602,
|
| 204 |
+
"step": 13
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"clip_ratio": 0.0,
|
| 208 |
+
"completion_length": 859.7778015136719,
|
| 209 |
+
"epoch": 0.2679425837320574,
|
| 210 |
+
"grad_norm": 0.06028261705382758,
|
| 211 |
+
"kl": 4.965066909790039e-05,
|
| 212 |
+
"learning_rate": 9.344887320459198e-07,
|
| 213 |
+
"loss": 0.0167,
|
| 214 |
+
"reward": 0.6354166641831398,
|
| 215 |
+
"reward_std": 0.2086947001516819,
|
| 216 |
+
"rewards/accuracy_reward": 0.21527778171002865,
|
| 217 |
+
"rewards/format_reward": 0.0,
|
| 218 |
+
"rewards/tag_count_reward": 0.4201388955116272,
|
| 219 |
+
"step": 14
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"clip_ratio": 0.0,
|
| 223 |
+
"completion_length": 889.1319427490234,
|
| 224 |
+
"epoch": 0.28708133971291866,
|
| 225 |
+
"grad_norm": 0.06077195852297118,
|
| 226 |
+
"kl": 7.545948028564453e-05,
|
| 227 |
+
"learning_rate": 9.176364518546988e-07,
|
| 228 |
+
"loss": 0.0159,
|
| 229 |
+
"reward": 0.6475694477558136,
|
| 230 |
+
"reward_std": 0.19641854241490364,
|
| 231 |
+
"rewards/accuracy_reward": 0.2465277723968029,
|
| 232 |
+
"rewards/format_reward": 0.0,
|
| 233 |
+
"rewards/tag_count_reward": 0.4010416716337204,
|
| 234 |
+
"step": 15
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"clip_ratio": 0.0,
|
| 238 |
+
"completion_length": 875.8854217529297,
|
| 239 |
+
"epoch": 0.3062200956937799,
|
| 240 |
+
"grad_norm": 0.06912072388582151,
|
| 241 |
+
"kl": 9.357929229736328e-05,
|
| 242 |
+
"learning_rate": 8.990700808169889e-07,
|
| 243 |
+
"loss": 0.0254,
|
| 244 |
+
"reward": 0.709201380610466,
|
| 245 |
+
"reward_std": 0.21483278647065163,
|
| 246 |
+
"rewards/accuracy_reward": 0.2708333320915699,
|
| 247 |
+
"rewards/format_reward": 0.0,
|
| 248 |
+
"rewards/tag_count_reward": 0.438368059694767,
|
| 249 |
+
"step": 16
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"clip_ratio": 0.0,
|
| 253 |
+
"completion_length": 865.875,
|
| 254 |
+
"epoch": 0.3253588516746411,
|
| 255 |
+
"grad_norm": 0.06267659137438868,
|
| 256 |
+
"kl": 0.00013184547424316406,
|
| 257 |
+
"learning_rate": 8.788761839251558e-07,
|
| 258 |
+
"loss": 0.0206,
|
| 259 |
+
"reward": 0.7309027910232544,
|
| 260 |
+
"reward_std": 0.262709803879261,
|
| 261 |
+
"rewards/accuracy_reward": 0.3090277835726738,
|
| 262 |
+
"rewards/format_reward": 0.0,
|
| 263 |
+
"rewards/tag_count_reward": 0.4218750074505806,
|
| 264 |
+
"step": 17
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"clip_ratio": 0.0,
|
| 268 |
+
"completion_length": 864.7777862548828,
|
| 269 |
+
"epoch": 0.3444976076555024,
|
| 270 |
+
"grad_norm": 0.06921820728455384,
|
| 271 |
+
"kl": 0.0001842975616455078,
|
| 272 |
+
"learning_rate": 8.571489144483944e-07,
|
| 273 |
+
"loss": 0.023,
|
| 274 |
+
"reward": 0.6979166716337204,
|
| 275 |
+
"reward_std": 0.23079178109765053,
|
| 276 |
+
"rewards/accuracy_reward": 0.2777777798473835,
|
| 277 |
+
"rewards/format_reward": 0.0,
|
| 278 |
+
"rewards/tag_count_reward": 0.4201388880610466,
|
| 279 |
+
"step": 18
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"clip_ratio": 0.0,
|
| 283 |
+
"completion_length": 864.5208435058594,
|
| 284 |
+
"epoch": 0.36363636363636365,
|
| 285 |
+
"grad_norm": 0.06828931603666485,
|
| 286 |
+
"kl": 0.0002467632293701172,
|
| 287 |
+
"learning_rate": 8.339895749467237e-07,
|
| 288 |
+
"loss": 0.021,
|
| 289 |
+
"reward": 0.7291666865348816,
|
| 290 |
+
"reward_std": 0.2357022613286972,
|
| 291 |
+
"rewards/accuracy_reward": 0.2986111082136631,
|
| 292 |
+
"rewards/format_reward": 0.0,
|
| 293 |
+
"rewards/tag_count_reward": 0.430555559694767,
|
| 294 |
+
"step": 19
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"clip_ratio": 0.0,
|
| 298 |
+
"completion_length": 857.4965362548828,
|
| 299 |
+
"epoch": 0.3827751196172249,
|
| 300 |
+
"grad_norm": 0.057114973470776605,
|
| 301 |
+
"kl": 0.00024008750915527344,
|
| 302 |
+
"learning_rate": 8.095061449516902e-07,
|
| 303 |
+
"loss": 0.0101,
|
| 304 |
+
"reward": 0.7317708283662796,
|
| 305 |
+
"reward_std": 0.23692986369132996,
|
| 306 |
+
"rewards/accuracy_reward": 0.3090277835726738,
|
| 307 |
+
"rewards/format_reward": 0.0,
|
| 308 |
+
"rewards/tag_count_reward": 0.4227430671453476,
|
| 309 |
+
"step": 20
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"clip_ratio": 0.0,
|
| 313 |
+
"completion_length": 855.9201354980469,
|
| 314 |
+
"epoch": 0.4019138755980861,
|
| 315 |
+
"grad_norm": 0.05706380401976011,
|
| 316 |
+
"kl": 0.0002942085266113281,
|
| 317 |
+
"learning_rate": 7.838127775159451e-07,
|
| 318 |
+
"loss": 0.0221,
|
| 319 |
+
"reward": 0.7118055671453476,
|
| 320 |
+
"reward_std": 0.25534411519765854,
|
| 321 |
+
"rewards/accuracy_reward": 0.2847222313284874,
|
| 322 |
+
"rewards/format_reward": 0.0,
|
| 323 |
+
"rewards/tag_count_reward": 0.4270833358168602,
|
| 324 |
+
"step": 21
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"clip_ratio": 0.0,
|
| 328 |
+
"completion_length": 860.5173645019531,
|
| 329 |
+
"epoch": 0.42105263157894735,
|
| 330 |
+
"grad_norm": 0.05882812256894408,
|
| 331 |
+
"kl": 0.0003452301025390625,
|
| 332 |
+
"learning_rate": 7.570292669790184e-07,
|
| 333 |
+
"loss": 0.0031,
|
| 334 |
+
"reward": 0.6935763955116272,
|
| 335 |
+
"reward_std": 0.20501185953617096,
|
| 336 |
+
"rewards/accuracy_reward": 0.2708333395421505,
|
| 337 |
+
"rewards/format_reward": 0.0,
|
| 338 |
+
"rewards/tag_count_reward": 0.422743059694767,
|
| 339 |
+
"step": 22
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"clip_ratio": 0.0,
|
| 343 |
+
"completion_length": 859.2639007568359,
|
| 344 |
+
"epoch": 0.44019138755980863,
|
| 345 |
+
"grad_norm": 0.06173500079510746,
|
| 346 |
+
"kl": 0.000537872314453125,
|
| 347 |
+
"learning_rate": 7.292804904308086e-07,
|
| 348 |
+
"loss": 0.0154,
|
| 349 |
+
"reward": 0.7161458283662796,
|
| 350 |
+
"reward_std": 0.21483277902007103,
|
| 351 |
+
"rewards/accuracy_reward": 0.2951388917863369,
|
| 352 |
+
"rewards/format_reward": 0.0,
|
| 353 |
+
"rewards/tag_count_reward": 0.4210069552063942,
|
| 354 |
+
"step": 23
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"clip_ratio": 0.0,
|
| 358 |
+
"completion_length": 858.9097290039062,
|
| 359 |
+
"epoch": 0.45933014354066987,
|
| 360 |
+
"grad_norm": 0.06893952606573064,
|
| 361 |
+
"kl": 0.0005884170532226562,
|
| 362 |
+
"learning_rate": 7.006958254769437e-07,
|
| 363 |
+
"loss": 0.0264,
|
| 364 |
+
"reward": 0.701388880610466,
|
| 365 |
+
"reward_std": 0.2602545768022537,
|
| 366 |
+
"rewards/accuracy_reward": 0.2881944440305233,
|
| 367 |
+
"rewards/format_reward": 0.0,
|
| 368 |
+
"rewards/tag_count_reward": 0.4131944477558136,
|
| 369 |
+
"step": 24
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"clip_ratio": 0.0,
|
| 373 |
+
"completion_length": 860.0416717529297,
|
| 374 |
+
"epoch": 0.4784688995215311,
|
| 375 |
+
"grad_norm": 0.0687100634503083,
|
| 376 |
+
"kl": 0.0006256103515625,
|
| 377 |
+
"learning_rate": 6.714085470206609e-07,
|
| 378 |
+
"loss": 0.0255,
|
| 379 |
+
"reward": 0.7230902910232544,
|
| 380 |
+
"reward_std": 0.24184033274650574,
|
| 381 |
+
"rewards/accuracy_reward": 0.2881944552063942,
|
| 382 |
+
"rewards/format_reward": 0.0,
|
| 383 |
+
"rewards/tag_count_reward": 0.4348958283662796,
|
| 384 |
+
"step": 25
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"clip_ratio": 0.0,
|
| 388 |
+
"completion_length": 864.3715209960938,
|
| 389 |
+
"epoch": 0.49760765550239233,
|
| 390 |
+
"grad_norm": 0.061459175011899785,
|
| 391 |
+
"kl": 0.0009174346923828125,
|
| 392 |
+
"learning_rate": 6.415552058736853e-07,
|
| 393 |
+
"loss": 0.0191,
|
| 394 |
+
"reward": 0.7335069477558136,
|
| 395 |
+
"reward_std": 0.2786688134074211,
|
| 396 |
+
"rewards/accuracy_reward": 0.3229166641831398,
|
| 397 |
+
"rewards/format_reward": 0.0,
|
| 398 |
+
"rewards/tag_count_reward": 0.4105902835726738,
|
| 399 |
+
"step": 26
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"clip_ratio": 0.0,
|
| 403 |
+
"completion_length": 896.4722290039062,
|
| 404 |
+
"epoch": 0.5167464114832536,
|
| 405 |
+
"grad_norm": 0.06334722551499229,
|
| 406 |
+
"kl": 0.000942230224609375,
|
| 407 |
+
"learning_rate": 6.11274992093311e-07,
|
| 408 |
+
"loss": 0.0245,
|
| 409 |
+
"reward": 0.6371527761220932,
|
| 410 |
+
"reward_std": 0.24306796491146088,
|
| 411 |
+
"rewards/accuracy_reward": 0.2430555559694767,
|
| 412 |
+
"rewards/format_reward": 0.0,
|
| 413 |
+
"rewards/tag_count_reward": 0.3940972313284874,
|
| 414 |
+
"step": 27
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"clip_ratio": 0.0,
|
| 418 |
+
"completion_length": 853.0451354980469,
|
| 419 |
+
"epoch": 0.5358851674641149,
|
| 420 |
+
"grad_norm": 0.05677441196904184,
|
| 421 |
+
"kl": 0.0012645721435546875,
|
| 422 |
+
"learning_rate": 5.80709086014102e-07,
|
| 423 |
+
"loss": 0.0077,
|
| 424 |
+
"reward": 0.740451380610466,
|
| 425 |
+
"reward_std": 0.16818338260054588,
|
| 426 |
+
"rewards/accuracy_reward": 0.3229166641831398,
|
| 427 |
+
"rewards/format_reward": 0.0,
|
| 428 |
+
"rewards/tag_count_reward": 0.4175347238779068,
|
| 429 |
+
"step": 28
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"clip_ratio": 0.0,
|
| 433 |
+
"completion_length": 859.9132080078125,
|
| 434 |
+
"epoch": 0.5550239234449761,
|
| 435 |
+
"grad_norm": 0.08121565074078244,
|
| 436 |
+
"kl": 0.001453399658203125,
|
| 437 |
+
"learning_rate": 5.5e-07,
|
| 438 |
+
"loss": 0.0325,
|
| 439 |
+
"reward": 0.743923619389534,
|
| 440 |
+
"reward_std": 0.3105868399143219,
|
| 441 |
+
"rewards/accuracy_reward": 0.3090277835726738,
|
| 442 |
+
"rewards/format_reward": 0.0,
|
| 443 |
+
"rewards/tag_count_reward": 0.4348958358168602,
|
| 444 |
+
"step": 29
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"clip_ratio": 0.0,
|
| 448 |
+
"completion_length": 843.4513854980469,
|
| 449 |
+
"epoch": 0.5741626794258373,
|
| 450 |
+
"grad_norm": 0.06778365437453367,
|
| 451 |
+
"kl": 0.001445770263671875,
|
| 452 |
+
"learning_rate": 5.192909139858981e-07,
|
| 453 |
+
"loss": 0.028,
|
| 454 |
+
"reward": 0.7717013955116272,
|
| 455 |
+
"reward_std": 0.27375834435224533,
|
| 456 |
+
"rewards/accuracy_reward": 0.3368055522441864,
|
| 457 |
+
"rewards/format_reward": 0.0,
|
| 458 |
+
"rewards/tag_count_reward": 0.4348958432674408,
|
| 459 |
+
"step": 30
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"clip_ratio": 0.0,
|
| 463 |
+
"completion_length": 887.6354370117188,
|
| 464 |
+
"epoch": 0.5933014354066986,
|
| 465 |
+
"grad_norm": 0.0590498329750237,
|
| 466 |
+
"kl": 0.0012683868408203125,
|
| 467 |
+
"learning_rate": 4.887250079066891e-07,
|
| 468 |
+
"loss": 0.0255,
|
| 469 |
+
"reward": 0.7057291865348816,
|
| 470 |
+
"reward_std": 0.21237755194306374,
|
| 471 |
+
"rewards/accuracy_reward": 0.2916666716337204,
|
| 472 |
+
"rewards/format_reward": 0.0,
|
| 473 |
+
"rewards/tag_count_reward": 0.4140625074505806,
|
| 474 |
+
"step": 31
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"clip_ratio": 0.0,
|
| 478 |
+
"completion_length": 832.625,
|
| 479 |
+
"epoch": 0.6124401913875598,
|
| 480 |
+
"grad_norm": 0.07056325724098401,
|
| 481 |
+
"kl": 0.0017757415771484375,
|
| 482 |
+
"learning_rate": 4.584447941263149e-07,
|
| 483 |
+
"loss": 0.0178,
|
| 484 |
+
"reward": 0.7578125149011612,
|
| 485 |
+
"reward_std": 0.26393740251660347,
|
| 486 |
+
"rewards/accuracy_reward": 0.3159722238779068,
|
| 487 |
+
"rewards/format_reward": 0.0,
|
| 488 |
+
"rewards/tag_count_reward": 0.4418402761220932,
|
| 489 |
+
"step": 32
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"clip_ratio": 0.0,
|
| 493 |
+
"completion_length": 856.9514007568359,
|
| 494 |
+
"epoch": 0.631578947368421,
|
| 495 |
+
"grad_norm": 0.06693116072370171,
|
| 496 |
+
"kl": 0.001903533935546875,
|
| 497 |
+
"learning_rate": 4.285914529793391e-07,
|
| 498 |
+
"loss": 0.0283,
|
| 499 |
+
"reward": 0.747395858168602,
|
| 500 |
+
"reward_std": 0.23447464406490326,
|
| 501 |
+
"rewards/accuracy_reward": 0.3229166716337204,
|
| 502 |
+
"rewards/format_reward": 0.0,
|
| 503 |
+
"rewards/tag_count_reward": 0.4244791716337204,
|
| 504 |
+
"step": 33
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"clip_ratio": 0.0,
|
| 508 |
+
"completion_length": 837.0069427490234,
|
| 509 |
+
"epoch": 0.6507177033492823,
|
| 510 |
+
"grad_norm": 0.06430130826482054,
|
| 511 |
+
"kl": 0.0016841888427734375,
|
| 512 |
+
"learning_rate": 3.993041745230562e-07,
|
| 513 |
+
"loss": 0.0254,
|
| 514 |
+
"reward": 0.8125000149011612,
|
| 515 |
+
"reward_std": 0.22833656147122383,
|
| 516 |
+
"rewards/accuracy_reward": 0.3888888955116272,
|
| 517 |
+
"rewards/format_reward": 0.0,
|
| 518 |
+
"rewards/tag_count_reward": 0.423611119389534,
|
| 519 |
+
"step": 34
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"clip_ratio": 0.0,
|
| 523 |
+
"completion_length": 833.3854217529297,
|
| 524 |
+
"epoch": 0.6698564593301436,
|
| 525 |
+
"grad_norm": 0.06623663931865324,
|
| 526 |
+
"kl": 0.0020465850830078125,
|
| 527 |
+
"learning_rate": 3.707195095691913e-07,
|
| 528 |
+
"loss": 0.0112,
|
| 529 |
+
"reward": 0.8081597238779068,
|
| 530 |
+
"reward_std": 0.254116490483284,
|
| 531 |
+
"rewards/accuracy_reward": 0.3576388955116272,
|
| 532 |
+
"rewards/format_reward": 0.0,
|
| 533 |
+
"rewards/tag_count_reward": 0.4505208358168602,
|
| 534 |
+
"step": 35
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"clip_ratio": 0.0,
|
| 538 |
+
"completion_length": 888.3264007568359,
|
| 539 |
+
"epoch": 0.6889952153110048,
|
| 540 |
+
"grad_norm": 0.05215266241295597,
|
| 541 |
+
"kl": 0.001735687255859375,
|
| 542 |
+
"learning_rate": 3.4297073302098155e-07,
|
| 543 |
+
"loss": 0.012,
|
| 544 |
+
"reward": 0.7022569477558136,
|
| 545 |
+
"reward_std": 0.22956418246030807,
|
| 546 |
+
"rewards/accuracy_reward": 0.2986111156642437,
|
| 547 |
+
"rewards/format_reward": 0.0,
|
| 548 |
+
"rewards/tag_count_reward": 0.4036458432674408,
|
| 549 |
+
"step": 36
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"clip_ratio": 0.0,
|
| 553 |
+
"completion_length": 855.5243225097656,
|
| 554 |
+
"epoch": 0.7081339712918661,
|
| 555 |
+
"grad_norm": 0.06846168763574559,
|
| 556 |
+
"kl": 0.0021762847900390625,
|
| 557 |
+
"learning_rate": 3.16187222484055e-07,
|
| 558 |
+
"loss": 0.0195,
|
| 559 |
+
"reward": 0.7786458283662796,
|
| 560 |
+
"reward_std": 0.29831067472696304,
|
| 561 |
+
"rewards/accuracy_reward": 0.3402777761220932,
|
| 562 |
+
"rewards/format_reward": 0.0,
|
| 563 |
+
"rewards/tag_count_reward": 0.4383680522441864,
|
| 564 |
+
"step": 37
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"clip_ratio": 0.0,
|
| 568 |
+
"completion_length": 867.7257080078125,
|
| 569 |
+
"epoch": 0.7272727272727273,
|
| 570 |
+
"grad_norm": 0.06807879530009997,
|
| 571 |
+
"kl": 0.0018215179443359375,
|
| 572 |
+
"learning_rate": 2.904938550483098e-07,
|
| 573 |
+
"loss": 0.0285,
|
| 574 |
+
"reward": 0.6675347238779068,
|
| 575 |
+
"reward_std": 0.21728801727294922,
|
| 576 |
+
"rewards/accuracy_reward": 0.2638888917863369,
|
| 577 |
+
"rewards/format_reward": 0.0,
|
| 578 |
+
"rewards/tag_count_reward": 0.4036458358168602,
|
| 579 |
+
"step": 38
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"clip_ratio": 0.0,
|
| 583 |
+
"completion_length": 877.5381927490234,
|
| 584 |
+
"epoch": 0.7464114832535885,
|
| 585 |
+
"grad_norm": 0.06659774200582641,
|
| 586 |
+
"kl": 0.001922607421875,
|
| 587 |
+
"learning_rate": 2.6601042505327635e-07,
|
| 588 |
+
"loss": 0.0224,
|
| 589 |
+
"reward": 0.732638880610466,
|
| 590 |
+
"reward_std": 0.2946278154850006,
|
| 591 |
+
"rewards/accuracy_reward": 0.3055555559694767,
|
| 592 |
+
"rewards/format_reward": 0.0,
|
| 593 |
+
"rewards/tag_count_reward": 0.4270833358168602,
|
| 594 |
+
"step": 39
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"clip_ratio": 0.0,
|
| 598 |
+
"completion_length": 848.8159790039062,
|
| 599 |
+
"epoch": 0.7655502392344498,
|
| 600 |
+
"grad_norm": 0.06715335856665118,
|
| 601 |
+
"kl": 0.0018100738525390625,
|
| 602 |
+
"learning_rate": 2.4285108555160575e-07,
|
| 603 |
+
"loss": 0.0216,
|
| 604 |
+
"reward": 0.8125000149011612,
|
| 605 |
+
"reward_std": 0.2357022687792778,
|
| 606 |
+
"rewards/accuracy_reward": 0.3854166716337204,
|
| 607 |
+
"rewards/format_reward": 0.0,
|
| 608 |
+
"rewards/tag_count_reward": 0.4270833358168602,
|
| 609 |
+
"step": 40
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"clip_ratio": 0.0,
|
| 613 |
+
"completion_length": 862.9201507568359,
|
| 614 |
+
"epoch": 0.784688995215311,
|
| 615 |
+
"grad_norm": 0.06259931940680961,
|
| 616 |
+
"kl": 0.00200653076171875,
|
| 617 |
+
"learning_rate": 2.2112381607484416e-07,
|
| 618 |
+
"loss": 0.0198,
|
| 619 |
+
"reward": 0.714409738779068,
|
| 620 |
+
"reward_std": 0.24429557099938393,
|
| 621 |
+
"rewards/accuracy_reward": 0.2916666716337204,
|
| 622 |
+
"rewards/format_reward": 0.0,
|
| 623 |
+
"rewards/tag_count_reward": 0.422743059694767,
|
| 624 |
+
"step": 41
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"clip_ratio": 0.0,
|
| 628 |
+
"completion_length": 863.4201354980469,
|
| 629 |
+
"epoch": 0.8038277511961722,
|
| 630 |
+
"grad_norm": 0.07414881526494982,
|
| 631 |
+
"kl": 0.002193450927734375,
|
| 632 |
+
"learning_rate": 2.0092991918301106e-07,
|
| 633 |
+
"loss": 0.022,
|
| 634 |
+
"reward": 0.7604166865348816,
|
| 635 |
+
"reward_std": 0.22833655402064323,
|
| 636 |
+
"rewards/accuracy_reward": 0.329861119389534,
|
| 637 |
+
"rewards/format_reward": 0.0,
|
| 638 |
+
"rewards/tag_count_reward": 0.4305555522441864,
|
| 639 |
+
"step": 42
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"clip_ratio": 0.0,
|
| 643 |
+
"completion_length": 843.4826507568359,
|
| 644 |
+
"epoch": 0.8229665071770335,
|
| 645 |
+
"grad_norm": 0.06426826229363446,
|
| 646 |
+
"kl": 0.0018749237060546875,
|
| 647 |
+
"learning_rate": 1.8236354814530112e-07,
|
| 648 |
+
"loss": 0.0145,
|
| 649 |
+
"reward": 0.796875,
|
| 650 |
+
"reward_std": 0.2111499235033989,
|
| 651 |
+
"rewards/accuracy_reward": 0.357638880610466,
|
| 652 |
+
"rewards/format_reward": 0.0,
|
| 653 |
+
"rewards/tag_count_reward": 0.4392361119389534,
|
| 654 |
+
"step": 43
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"clip_ratio": 0.0,
|
| 658 |
+
"completion_length": 878.6527709960938,
|
| 659 |
+
"epoch": 0.8421052631578947,
|
| 660 |
+
"grad_norm": 0.058633524755595696,
|
| 661 |
+
"kl": 0.0018100738525390625,
|
| 662 |
+
"learning_rate": 1.6551126795408015e-07,
|
| 663 |
+
"loss": 0.0143,
|
| 664 |
+
"reward": 0.6909722238779068,
|
| 665 |
+
"reward_std": 0.18659762665629387,
|
| 666 |
+
"rewards/accuracy_reward": 0.2951388880610466,
|
| 667 |
+
"rewards/format_reward": 0.0,
|
| 668 |
+
"rewards/tag_count_reward": 0.3958333358168602,
|
| 669 |
+
"step": 44
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"clip_ratio": 0.0,
|
| 673 |
+
"completion_length": 840.09375,
|
| 674 |
+
"epoch": 0.861244019138756,
|
| 675 |
+
"grad_norm": 0.07730488608092184,
|
| 676 |
+
"kl": 0.0020885467529296875,
|
| 677 |
+
"learning_rate": 1.5045165171893116e-07,
|
| 678 |
+
"loss": 0.0289,
|
| 679 |
+
"reward": 0.8420138955116272,
|
| 680 |
+
"reward_std": 0.2676202580332756,
|
| 681 |
+
"rewards/accuracy_reward": 0.3888888955116272,
|
| 682 |
+
"rewards/format_reward": 0.0,
|
| 683 |
+
"rewards/tag_count_reward": 0.453125,
|
| 684 |
+
"step": 45
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"clip_ratio": 0.0,
|
| 688 |
+
"completion_length": 847.3611297607422,
|
| 689 |
+
"epoch": 0.8803827751196173,
|
| 690 |
+
"grad_norm": 0.06479288468708434,
|
| 691 |
+
"kl": 0.002285003662109375,
|
| 692 |
+
"learning_rate": 1.3725491432254623e-07,
|
| 693 |
+
"loss": 0.0195,
|
| 694 |
+
"reward": 0.7777777910232544,
|
| 695 |
+
"reward_std": 0.27989643067121506,
|
| 696 |
+
"rewards/accuracy_reward": 0.3541666716337204,
|
| 697 |
+
"rewards/format_reward": 0.0,
|
| 698 |
+
"rewards/tag_count_reward": 0.4236111119389534,
|
| 699 |
+
"step": 46
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"clip_ratio": 0.0,
|
| 703 |
+
"completion_length": 851.6944427490234,
|
| 704 |
+
"epoch": 0.8995215311004785,
|
| 705 |
+
"grad_norm": 0.05948308775990511,
|
| 706 |
+
"kl": 0.0021038055419921875,
|
| 707 |
+
"learning_rate": 1.259825850465308e-07,
|
| 708 |
+
"loss": 0.0056,
|
| 709 |
+
"reward": 0.7213541716337204,
|
| 710 |
+
"reward_std": 0.19764616712927818,
|
| 711 |
+
"rewards/accuracy_reward": 0.3159722276031971,
|
| 712 |
+
"rewards/format_reward": 0.0,
|
| 713 |
+
"rewards/tag_count_reward": 0.405381940305233,
|
| 714 |
+
"step": 47
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"clip_ratio": 0.0,
|
| 718 |
+
"completion_length": 831.1041717529297,
|
| 719 |
+
"epoch": 0.9186602870813397,
|
| 720 |
+
"grad_norm": 0.06976945326641983,
|
| 721 |
+
"kl": 0.0020198822021484375,
|
| 722 |
+
"learning_rate": 1.166872206934904e-07,
|
| 723 |
+
"loss": 0.0315,
|
| 724 |
+
"reward": 0.7795138955116272,
|
| 725 |
+
"reward_std": 0.25043364614248276,
|
| 726 |
+
"rewards/accuracy_reward": 0.3402777835726738,
|
| 727 |
+
"rewards/format_reward": 0.0,
|
| 728 |
+
"rewards/tag_count_reward": 0.4392361119389534,
|
| 729 |
+
"step": 48
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"clip_ratio": 0.0,
|
| 733 |
+
"completion_length": 871.7847137451172,
|
| 734 |
+
"epoch": 0.937799043062201,
|
| 735 |
+
"grad_norm": 0.06380948814810407,
|
| 736 |
+
"kl": 0.002155303955078125,
|
| 737 |
+
"learning_rate": 1.0941216054295468e-07,
|
| 738 |
+
"loss": 0.0197,
|
| 739 |
+
"reward": 0.7526041716337204,
|
| 740 |
+
"reward_std": 0.23447464033961296,
|
| 741 |
+
"rewards/accuracy_reward": 0.3298611119389534,
|
| 742 |
+
"rewards/format_reward": 0.0,
|
| 743 |
+
"rewards/tag_count_reward": 0.4227430671453476,
|
| 744 |
+
"step": 49
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"clip_ratio": 0.0,
|
| 748 |
+
"completion_length": 847.6632080078125,
|
| 749 |
+
"epoch": 0.9569377990430622,
|
| 750 |
+
"grad_norm": 0.07507185787901022,
|
| 751 |
+
"kl": 0.0022182464599609375,
|
| 752 |
+
"learning_rate": 1.0419132428365116e-07,
|
| 753 |
+
"loss": 0.0201,
|
| 754 |
+
"reward": 0.7786458283662796,
|
| 755 |
+
"reward_std": 0.2590269520878792,
|
| 756 |
+
"rewards/accuracy_reward": 0.329861119389534,
|
| 757 |
+
"rewards/format_reward": 0.0,
|
| 758 |
+
"rewards/tag_count_reward": 0.4487847238779068,
|
| 759 |
+
"step": 50
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"clip_ratio": 0.0,
|
| 763 |
+
"completion_length": 848.7222290039062,
|
| 764 |
+
"epoch": 0.9760765550239234,
|
| 765 |
+
"grad_norm": 0.05934554686543501,
|
| 766 |
+
"kl": 0.0019321441650390625,
|
| 767 |
+
"learning_rate": 1.0104905386425732e-07,
|
| 768 |
+
"loss": 0.0147,
|
| 769 |
+
"reward": 0.7803819626569748,
|
| 770 |
+
"reward_std": 0.22219848446547985,
|
| 771 |
+
"rewards/accuracy_reward": 0.3402777835726738,
|
| 772 |
+
"rewards/format_reward": 0.0,
|
| 773 |
+
"rewards/tag_count_reward": 0.4401041641831398,
|
| 774 |
+
"step": 51
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"clip_ratio": 0.0,
|
| 778 |
+
"completion_length": 920.5416870117188,
|
| 779 |
+
"epoch": 0.9952153110047847,
|
| 780 |
+
"grad_norm": 0.05965196947908554,
|
| 781 |
+
"kl": 0.002071380615234375,
|
| 782 |
+
"learning_rate": 1e-07,
|
| 783 |
+
"loss": 0.0123,
|
| 784 |
+
"reward": 0.7968750149011612,
|
| 785 |
+
"reward_std": 0.2651650495827198,
|
| 786 |
+
"rewards/accuracy_reward": 0.3715277761220932,
|
| 787 |
+
"rewards/format_reward": 0.0,
|
| 788 |
+
"rewards/tag_count_reward": 0.4253472313284874,
|
| 789 |
+
"step": 52
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.9952153110047847,
|
| 793 |
+
"step": 52,
|
| 794 |
+
"total_flos": 0.0,
|
| 795 |
+
"train_loss": 0.01933252687404792,
|
| 796 |
+
"train_runtime": 2498.5955,
|
| 797 |
+
"train_samples_per_second": 3.002,
|
| 798 |
+
"train_steps_per_second": 0.021
|
| 799 |
+
}
|
| 800 |
+
],
|
| 801 |
+
"logging_steps": 1,
|
| 802 |
+
"max_steps": 52,
|
| 803 |
+
"num_input_tokens_seen": 0,
|
| 804 |
+
"num_train_epochs": 1,
|
| 805 |
+
"save_steps": 10,
|
| 806 |
+
"stateful_callbacks": {
|
| 807 |
+
"TrainerControl": {
|
| 808 |
+
"args": {
|
| 809 |
+
"should_epoch_stop": false,
|
| 810 |
+
"should_evaluate": false,
|
| 811 |
+
"should_log": false,
|
| 812 |
+
"should_save": true,
|
| 813 |
+
"should_training_stop": true
|
| 814 |
+
},
|
| 815 |
+
"attributes": {}
|
| 816 |
+
}
|
| 817 |
+
},
|
| 818 |
+
"total_flos": 0.0,
|
| 819 |
+
"train_batch_size": 24,
|
| 820 |
+
"trial_name": null,
|
| 821 |
+
"trial_params": null
|
| 822 |
+
}
|