Model save
Browse files- README.md +68 -0
- all_results.json +8 -0
- generation_config.json +14 -0
- train_results.json +8 -0
- trainer_state.json +796 -0
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: Qwen/Qwen2.5-1.5B-Instruct
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: Qwen-2.5-1.5B-Simple-RL
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- grpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for Qwen-2.5-1.5B-Simple-RL
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [Qwen/Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct).
|
| 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="saemin21/Qwen-2.5-1.5B-Simple-RL", 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/saemin21-postech/huggingface/runs/kc3dy4cu)
|
| 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: 0.15.0.dev0
|
| 38 |
+
- Transformers: 4.49.0.dev0
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.2.0
|
| 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": 0.20938398721150356,
|
| 4 |
+
"train_runtime": 4954.7093,
|
| 5 |
+
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 1.514,
|
| 7 |
+
"train_steps_per_second": 0.012
|
| 8 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
151645,
|
| 6 |
+
151643
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 151643,
|
| 9 |
+
"repetition_penalty": 1.1,
|
| 10 |
+
"temperature": 0.7,
|
| 11 |
+
"top_k": 20,
|
| 12 |
+
"top_p": 0.8,
|
| 13 |
+
"transformers_version": "4.49.0.dev0"
|
| 14 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.20938398721150356,
|
| 4 |
+
"train_runtime": 4954.7093,
|
| 5 |
+
"train_samples": 7500,
|
| 6 |
+
"train_samples_per_second": 1.514,
|
| 7 |
+
"train_steps_per_second": 0.012
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,796 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9893390191897654,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 58,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"completion_length": 365.3112087249756,
|
| 13 |
+
"epoch": 0.017057569296375266,
|
| 14 |
+
"grad_norm": 1.04221985968125,
|
| 15 |
+
"kl": 0.0,
|
| 16 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 17 |
+
"loss": 0.0,
|
| 18 |
+
"reward": 0.6432291865348816,
|
| 19 |
+
"reward_std": 0.4505743272602558,
|
| 20 |
+
"rewards/accuracy_reward": 0.11718750232830644,
|
| 21 |
+
"rewards/format_reward": 0.5260416772216558,
|
| 22 |
+
"step": 1
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"completion_length": 367.3958435058594,
|
| 26 |
+
"epoch": 0.03411513859275053,
|
| 27 |
+
"grad_norm": 1.1434252212815785,
|
| 28 |
+
"kl": 0.0,
|
| 29 |
+
"learning_rate": 6.666666666666667e-06,
|
| 30 |
+
"loss": 0.0,
|
| 31 |
+
"reward": 0.6783854328095913,
|
| 32 |
+
"reward_std": 0.46897680312395096,
|
| 33 |
+
"rewards/accuracy_reward": 0.12630208570044488,
|
| 34 |
+
"rewards/format_reward": 0.5520833525806665,
|
| 35 |
+
"step": 2
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"completion_length": 339.8724060058594,
|
| 39 |
+
"epoch": 0.0511727078891258,
|
| 40 |
+
"grad_norm": 0.8635963812006352,
|
| 41 |
+
"kl": 0.0010623931884765625,
|
| 42 |
+
"learning_rate": 1e-05,
|
| 43 |
+
"loss": 0.0,
|
| 44 |
+
"reward": 0.6835937723517418,
|
| 45 |
+
"reward_std": 0.43934670090675354,
|
| 46 |
+
"rewards/accuracy_reward": 0.09895833639893681,
|
| 47 |
+
"rewards/format_reward": 0.5846354383975267,
|
| 48 |
+
"step": 3
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"completion_length": 149.22526454925537,
|
| 52 |
+
"epoch": 0.06823027718550106,
|
| 53 |
+
"grad_norm": 307.2682075780713,
|
| 54 |
+
"kl": 2.980987548828125,
|
| 55 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 56 |
+
"loss": 0.1193,
|
| 57 |
+
"reward": 1.010416690260172,
|
| 58 |
+
"reward_std": 0.2095302422530949,
|
| 59 |
+
"rewards/accuracy_reward": 0.06640625139698386,
|
| 60 |
+
"rewards/format_reward": 0.9440104439854622,
|
| 61 |
+
"step": 4
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"completion_length": 112.50521183013916,
|
| 65 |
+
"epoch": 0.08528784648187633,
|
| 66 |
+
"grad_norm": 507.14990688734866,
|
| 67 |
+
"kl": 9.401611328125,
|
| 68 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 69 |
+
"loss": 0.3775,
|
| 70 |
+
"reward": 1.007812526077032,
|
| 71 |
+
"reward_std": 0.15946260537020862,
|
| 72 |
+
"rewards/accuracy_reward": 0.04427083441987634,
|
| 73 |
+
"rewards/format_reward": 0.9635416828095913,
|
| 74 |
+
"step": 5
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"completion_length": 100.34375333786011,
|
| 78 |
+
"epoch": 0.1023454157782516,
|
| 79 |
+
"grad_norm": 5.424616910263027,
|
| 80 |
+
"kl": 0.33245849609375,
|
| 81 |
+
"learning_rate": 2e-05,
|
| 82 |
+
"loss": 0.0133,
|
| 83 |
+
"reward": 1.0052083618938923,
|
| 84 |
+
"reward_std": 0.15602844418026507,
|
| 85 |
+
"rewards/accuracy_reward": 0.05338541930541396,
|
| 86 |
+
"rewards/format_reward": 0.9518229402601719,
|
| 87 |
+
"step": 6
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"completion_length": 133.2343783378601,
|
| 91 |
+
"epoch": 0.11940298507462686,
|
| 92 |
+
"grad_norm": 1.7843437966022808,
|
| 93 |
+
"kl": 0.23712158203125,
|
| 94 |
+
"learning_rate": 1.9981755542233175e-05,
|
| 95 |
+
"loss": 0.0095,
|
| 96 |
+
"reward": 0.9570312686264515,
|
| 97 |
+
"reward_std": 0.2223757691681385,
|
| 98 |
+
"rewards/accuracy_reward": 0.04166666802484542,
|
| 99 |
+
"rewards/format_reward": 0.9153646044433117,
|
| 100 |
+
"step": 7
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"completion_length": 114.69271230697632,
|
| 104 |
+
"epoch": 0.13646055437100213,
|
| 105 |
+
"grad_norm": 0.623961957547949,
|
| 106 |
+
"kl": 0.1710205078125,
|
| 107 |
+
"learning_rate": 1.992708874098054e-05,
|
| 108 |
+
"loss": 0.0068,
|
| 109 |
+
"reward": 0.9895833544433117,
|
| 110 |
+
"reward_std": 0.22396012931130826,
|
| 111 |
+
"rewards/accuracy_reward": 0.054687500931322575,
|
| 112 |
+
"rewards/format_reward": 0.934895858168602,
|
| 113 |
+
"step": 8
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"completion_length": 104.41406488418579,
|
| 117 |
+
"epoch": 0.1535181236673774,
|
| 118 |
+
"grad_norm": 1.5259640643959018,
|
| 119 |
+
"kl": 0.224365234375,
|
| 120 |
+
"learning_rate": 1.983619906947144e-05,
|
| 121 |
+
"loss": 0.009,
|
| 122 |
+
"reward": 0.9765625260770321,
|
| 123 |
+
"reward_std": 0.2754378484096378,
|
| 124 |
+
"rewards/accuracy_reward": 0.0664062516298145,
|
| 125 |
+
"rewards/format_reward": 0.9101562611758709,
|
| 126 |
+
"step": 9
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"completion_length": 231.64844512939453,
|
| 130 |
+
"epoch": 0.17057569296375266,
|
| 131 |
+
"grad_norm": 40.31352492325995,
|
| 132 |
+
"kl": 0.45977783203125,
|
| 133 |
+
"learning_rate": 1.9709418174260523e-05,
|
| 134 |
+
"loss": 0.0184,
|
| 135 |
+
"reward": 0.9257812723517418,
|
| 136 |
+
"reward_std": 0.39183398708701134,
|
| 137 |
+
"rewards/accuracy_reward": 0.0976562516298145,
|
| 138 |
+
"rewards/format_reward": 0.8281250186264515,
|
| 139 |
+
"step": 10
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"completion_length": 613.7070560455322,
|
| 143 |
+
"epoch": 0.18763326226012794,
|
| 144 |
+
"grad_norm": 840.2414358802456,
|
| 145 |
+
"kl": 3.21197509765625,
|
| 146 |
+
"learning_rate": 1.954720866508546e-05,
|
| 147 |
+
"loss": 0.1284,
|
| 148 |
+
"reward": 0.5429687686264515,
|
| 149 |
+
"reward_std": 0.4636515509337187,
|
| 150 |
+
"rewards/accuracy_reward": 0.076822918956168,
|
| 151 |
+
"rewards/format_reward": 0.46614584140479565,
|
| 152 |
+
"step": 11
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"completion_length": 762.7265853881836,
|
| 156 |
+
"epoch": 0.2046908315565032,
|
| 157 |
+
"grad_norm": 10.528180991568039,
|
| 158 |
+
"kl": 0.7625732421875,
|
| 159 |
+
"learning_rate": 1.9350162426854152e-05,
|
| 160 |
+
"loss": 0.0305,
|
| 161 |
+
"reward": 0.38411459513008595,
|
| 162 |
+
"reward_std": 0.3724043210968375,
|
| 163 |
+
"rewards/accuracy_reward": 0.09505208616610616,
|
| 164 |
+
"rewards/format_reward": 0.2890625102445483,
|
| 165 |
+
"step": 12
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"completion_length": 382.18230056762695,
|
| 169 |
+
"epoch": 0.22174840085287847,
|
| 170 |
+
"grad_norm": 2.761605849015544,
|
| 171 |
+
"kl": 0.384521484375,
|
| 172 |
+
"learning_rate": 1.91189984599209e-05,
|
| 173 |
+
"loss": 0.0154,
|
| 174 |
+
"reward": 0.6484375223517418,
|
| 175 |
+
"reward_std": 0.5331083796918392,
|
| 176 |
+
"rewards/accuracy_reward": 0.06901041930541396,
|
| 177 |
+
"rewards/format_reward": 0.5794271007180214,
|
| 178 |
+
"step": 13
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"completion_length": 198.7330780029297,
|
| 182 |
+
"epoch": 0.23880597014925373,
|
| 183 |
+
"grad_norm": 0.6078465083236722,
|
| 184 |
+
"kl": 0.23638916015625,
|
| 185 |
+
"learning_rate": 1.8854560256532098e-05,
|
| 186 |
+
"loss": 0.0095,
|
| 187 |
+
"reward": 0.912760429084301,
|
| 188 |
+
"reward_std": 0.4155522510409355,
|
| 189 |
+
"rewards/accuracy_reward": 0.10937500302679837,
|
| 190 |
+
"rewards/format_reward": 0.8033854328095913,
|
| 191 |
+
"step": 14
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"completion_length": 136.53516006469727,
|
| 195 |
+
"epoch": 0.255863539445629,
|
| 196 |
+
"grad_norm": 0.8617440561534008,
|
| 197 |
+
"kl": 0.21514892578125,
|
| 198 |
+
"learning_rate": 1.8557812723014476e-05,
|
| 199 |
+
"loss": 0.0086,
|
| 200 |
+
"reward": 0.9492187723517418,
|
| 201 |
+
"reward_std": 0.38575689122080803,
|
| 202 |
+
"rewards/accuracy_reward": 0.09895833593327552,
|
| 203 |
+
"rewards/format_reward": 0.8502604328095913,
|
| 204 |
+
"step": 15
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"completion_length": 98.51042032241821,
|
| 208 |
+
"epoch": 0.27292110874200426,
|
| 209 |
+
"grad_norm": 0.6595763666425144,
|
| 210 |
+
"kl": 0.2208251953125,
|
| 211 |
+
"learning_rate": 1.8229838658936566e-05,
|
| 212 |
+
"loss": 0.0088,
|
| 213 |
+
"reward": 0.9726562686264515,
|
| 214 |
+
"reward_std": 0.3207697505131364,
|
| 215 |
+
"rewards/accuracy_reward": 0.08854167000390589,
|
| 216 |
+
"rewards/format_reward": 0.8841146044433117,
|
| 217 |
+
"step": 16
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"completion_length": 76.04166841506958,
|
| 221 |
+
"epoch": 0.2899786780383795,
|
| 222 |
+
"grad_norm": 3.048254714245794,
|
| 223 |
+
"kl": 0.44549560546875,
|
| 224 |
+
"learning_rate": 1.7871834806090502e-05,
|
| 225 |
+
"loss": 0.0178,
|
| 226 |
+
"reward": 1.0468750298023224,
|
| 227 |
+
"reward_std": 0.15973232360556722,
|
| 228 |
+
"rewards/accuracy_reward": 0.07552083500195295,
|
| 229 |
+
"rewards/format_reward": 0.9713541865348816,
|
| 230 |
+
"step": 17
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"completion_length": 81.40234661102295,
|
| 234 |
+
"epoch": 0.3070362473347548,
|
| 235 |
+
"grad_norm": 0.4832905802799025,
|
| 236 |
+
"kl": 0.24029541015625,
|
| 237 |
+
"learning_rate": 1.7485107481711014e-05,
|
| 238 |
+
"loss": 0.0096,
|
| 239 |
+
"reward": 1.0742187798023224,
|
| 240 |
+
"reward_std": 0.20187148824334145,
|
| 241 |
+
"rewards/accuracy_reward": 0.10026042012032121,
|
| 242 |
+
"rewards/format_reward": 0.9739583507180214,
|
| 243 |
+
"step": 18
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"completion_length": 126.81640911102295,
|
| 247 |
+
"epoch": 0.32409381663113007,
|
| 248 |
+
"grad_norm": 0.43834266665765215,
|
| 249 |
+
"kl": 0.18353271484375,
|
| 250 |
+
"learning_rate": 1.7071067811865477e-05,
|
| 251 |
+
"loss": 0.0073,
|
| 252 |
+
"reward": 1.0924479514360428,
|
| 253 |
+
"reward_std": 0.2407330577261746,
|
| 254 |
+
"rewards/accuracy_reward": 0.13151042140088975,
|
| 255 |
+
"rewards/format_reward": 0.9609375186264515,
|
| 256 |
+
"step": 19
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"completion_length": 174.13021278381348,
|
| 260 |
+
"epoch": 0.3411513859275053,
|
| 261 |
+
"grad_norm": 0.3691172026913083,
|
| 262 |
+
"kl": 0.170654296875,
|
| 263 |
+
"learning_rate": 1.6631226582407954e-05,
|
| 264 |
+
"loss": 0.0068,
|
| 265 |
+
"reward": 1.0807292088866234,
|
| 266 |
+
"reward_std": 0.17400484485551715,
|
| 267 |
+
"rewards/accuracy_reward": 0.10546875465661287,
|
| 268 |
+
"rewards/format_reward": 0.9752604365348816,
|
| 269 |
+
"step": 20
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"completion_length": 226.26042366027832,
|
| 273 |
+
"epoch": 0.3582089552238806,
|
| 274 |
+
"grad_norm": 0.35348063732954105,
|
| 275 |
+
"kl": 0.162994384765625,
|
| 276 |
+
"learning_rate": 1.6167188726285433e-05,
|
| 277 |
+
"loss": 0.0065,
|
| 278 |
+
"reward": 1.0768229514360428,
|
| 279 |
+
"reward_std": 0.26323840813711286,
|
| 280 |
+
"rewards/accuracy_reward": 0.12239583872724324,
|
| 281 |
+
"rewards/format_reward": 0.9544271044433117,
|
| 282 |
+
"step": 21
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"completion_length": 226.471360206604,
|
| 286 |
+
"epoch": 0.3752665245202559,
|
| 287 |
+
"grad_norm": 0.2548959416361629,
|
| 288 |
+
"kl": 0.137939453125,
|
| 289 |
+
"learning_rate": 1.568064746731156e-05,
|
| 290 |
+
"loss": 0.0055,
|
| 291 |
+
"reward": 1.0937500447034836,
|
| 292 |
+
"reward_std": 0.19759728573262691,
|
| 293 |
+
"rewards/accuracy_reward": 0.11588542012032121,
|
| 294 |
+
"rewards/format_reward": 0.9778646007180214,
|
| 295 |
+
"step": 22
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"completion_length": 222.7395896911621,
|
| 299 |
+
"epoch": 0.39232409381663114,
|
| 300 |
+
"grad_norm": 0.29372855664077896,
|
| 301 |
+
"kl": 0.154876708984375,
|
| 302 |
+
"learning_rate": 1.5173378141776569e-05,
|
| 303 |
+
"loss": 0.0062,
|
| 304 |
+
"reward": 1.1015625409781933,
|
| 305 |
+
"reward_std": 0.21707096393220127,
|
| 306 |
+
"rewards/accuracy_reward": 0.12890625349245965,
|
| 307 |
+
"rewards/format_reward": 0.9726562686264515,
|
| 308 |
+
"step": 23
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"completion_length": 202.705735206604,
|
| 312 |
+
"epoch": 0.4093816631130064,
|
| 313 |
+
"grad_norm": 0.2969744538892161,
|
| 314 |
+
"kl": 0.155120849609375,
|
| 315 |
+
"learning_rate": 1.4647231720437687e-05,
|
| 316 |
+
"loss": 0.0062,
|
| 317 |
+
"reward": 1.1184896230697632,
|
| 318 |
+
"reward_std": 0.20930432621389627,
|
| 319 |
+
"rewards/accuracy_reward": 0.1328125053551048,
|
| 320 |
+
"rewards/format_reward": 0.9856770932674408,
|
| 321 |
+
"step": 24
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"completion_length": 208.4622449874878,
|
| 325 |
+
"epoch": 0.42643923240938164,
|
| 326 |
+
"grad_norm": 0.2551264579415379,
|
| 327 |
+
"kl": 0.145782470703125,
|
| 328 |
+
"learning_rate": 1.410412805452757e-05,
|
| 329 |
+
"loss": 0.0058,
|
| 330 |
+
"reward": 1.1341146230697632,
|
| 331 |
+
"reward_std": 0.20376197341829538,
|
| 332 |
+
"rewards/accuracy_reward": 0.1458333374466747,
|
| 333 |
+
"rewards/format_reward": 0.9882812611758709,
|
| 334 |
+
"step": 25
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"completion_length": 194.07292079925537,
|
| 338 |
+
"epoch": 0.44349680170575695,
|
| 339 |
+
"grad_norm": 0.29985133201161923,
|
| 340 |
+
"kl": 0.147552490234375,
|
| 341 |
+
"learning_rate": 1.3546048870425356e-05,
|
| 342 |
+
"loss": 0.0059,
|
| 343 |
+
"reward": 1.1875000298023224,
|
| 344 |
+
"reward_std": 0.22688957839272916,
|
| 345 |
+
"rewards/accuracy_reward": 0.19531250500585884,
|
| 346 |
+
"rewards/format_reward": 0.9921875074505806,
|
| 347 |
+
"step": 26
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"completion_length": 199.3333396911621,
|
| 351 |
+
"epoch": 0.4605543710021322,
|
| 352 |
+
"grad_norm": 0.2944306259052974,
|
| 353 |
+
"kl": 0.144775390625,
|
| 354 |
+
"learning_rate": 1.297503053855203e-05,
|
| 355 |
+
"loss": 0.0058,
|
| 356 |
+
"reward": 1.1458333656191826,
|
| 357 |
+
"reward_std": 0.19133792025968432,
|
| 358 |
+
"rewards/accuracy_reward": 0.15494792093522847,
|
| 359 |
+
"rewards/format_reward": 0.9908854253590107,
|
| 360 |
+
"step": 27
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"completion_length": 172.79427576065063,
|
| 364 |
+
"epoch": 0.47761194029850745,
|
| 365 |
+
"grad_norm": 0.3362991815387882,
|
| 366 |
+
"kl": 0.148773193359375,
|
| 367 |
+
"learning_rate": 1.2393156642875579e-05,
|
| 368 |
+
"loss": 0.006,
|
| 369 |
+
"reward": 1.178385466337204,
|
| 370 |
+
"reward_std": 0.22494715498760343,
|
| 371 |
+
"rewards/accuracy_reward": 0.18489583837799728,
|
| 372 |
+
"rewards/format_reward": 0.9934895895421505,
|
| 373 |
+
"step": 28
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"completion_length": 176.18490028381348,
|
| 377 |
+
"epoch": 0.4946695095948827,
|
| 378 |
+
"grad_norm": 0.3044368690957198,
|
| 379 |
+
"kl": 0.146514892578125,
|
| 380 |
+
"learning_rate": 1.180255037813906e-05,
|
| 381 |
+
"loss": 0.0059,
|
| 382 |
+
"reward": 1.1614583730697632,
|
| 383 |
+
"reward_std": 0.19968353700824082,
|
| 384 |
+
"rewards/accuracy_reward": 0.16796875244472176,
|
| 385 |
+
"rewards/format_reward": 0.9934895895421505,
|
| 386 |
+
"step": 29
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"completion_length": 169.21615076065063,
|
| 390 |
+
"epoch": 0.511727078891258,
|
| 391 |
+
"grad_norm": 0.3050078730361449,
|
| 392 |
+
"kl": 0.1395263671875,
|
| 393 |
+
"learning_rate": 1.1205366802553231e-05,
|
| 394 |
+
"loss": 0.0056,
|
| 395 |
+
"reward": 1.1601563021540642,
|
| 396 |
+
"reward_std": 0.19885429926216602,
|
| 397 |
+
"rewards/accuracy_reward": 0.167968753259629,
|
| 398 |
+
"rewards/format_reward": 0.9921875074505806,
|
| 399 |
+
"step": 30
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"completion_length": 191.99479484558105,
|
| 403 |
+
"epoch": 0.5287846481876333,
|
| 404 |
+
"grad_norm": 0.28939261315708026,
|
| 405 |
+
"kl": 0.145721435546875,
|
| 406 |
+
"learning_rate": 1.0603784974222862e-05,
|
| 407 |
+
"loss": 0.0058,
|
| 408 |
+
"reward": 1.1354167088866234,
|
| 409 |
+
"reward_std": 0.2171460180543363,
|
| 410 |
+
"rewards/accuracy_reward": 0.15104167070239782,
|
| 411 |
+
"rewards/format_reward": 0.9843750074505806,
|
| 412 |
+
"step": 31
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"completion_length": 233.89974689483643,
|
| 416 |
+
"epoch": 0.5458422174840085,
|
| 417 |
+
"grad_norm": 0.28487537328396517,
|
| 418 |
+
"kl": 0.13800048828125,
|
| 419 |
+
"learning_rate": 1e-05,
|
| 420 |
+
"loss": 0.0055,
|
| 421 |
+
"reward": 1.123697966337204,
|
| 422 |
+
"reward_std": 0.2813015836291015,
|
| 423 |
+
"rewards/accuracy_reward": 0.15885417093522847,
|
| 424 |
+
"rewards/format_reward": 0.9648437723517418,
|
| 425 |
+
"step": 32
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"completion_length": 253.55078887939453,
|
| 429 |
+
"epoch": 0.5628997867803838,
|
| 430 |
+
"grad_norm": 0.2737431545513293,
|
| 431 |
+
"kl": 0.137664794921875,
|
| 432 |
+
"learning_rate": 9.39621502577714e-06,
|
| 433 |
+
"loss": 0.0055,
|
| 434 |
+
"reward": 1.128906287252903,
|
| 435 |
+
"reward_std": 0.19965026015415788,
|
| 436 |
+
"rewards/accuracy_reward": 0.1497395880287513,
|
| 437 |
+
"rewards/format_reward": 0.9791666828095913,
|
| 438 |
+
"step": 33
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"completion_length": 239.842453956604,
|
| 442 |
+
"epoch": 0.579957356076759,
|
| 443 |
+
"grad_norm": 0.6254452308530076,
|
| 444 |
+
"kl": 0.14483642578125,
|
| 445 |
+
"learning_rate": 8.79463319744677e-06,
|
| 446 |
+
"loss": 0.0058,
|
| 447 |
+
"reward": 1.1757812798023224,
|
| 448 |
+
"reward_std": 0.26264199148863554,
|
| 449 |
+
"rewards/accuracy_reward": 0.195312503259629,
|
| 450 |
+
"rewards/format_reward": 0.9804687574505806,
|
| 451 |
+
"step": 34
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"completion_length": 241.17187976837158,
|
| 455 |
+
"epoch": 0.5970149253731343,
|
| 456 |
+
"grad_norm": 0.3224088206959575,
|
| 457 |
+
"kl": 0.14569091796875,
|
| 458 |
+
"learning_rate": 8.197449621860944e-06,
|
| 459 |
+
"loss": 0.0058,
|
| 460 |
+
"reward": 1.1757812947034836,
|
| 461 |
+
"reward_std": 0.27850970113649964,
|
| 462 |
+
"rewards/accuracy_reward": 0.199218753259629,
|
| 463 |
+
"rewards/format_reward": 0.9765625186264515,
|
| 464 |
+
"step": 35
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"completion_length": 231.39062976837158,
|
| 468 |
+
"epoch": 0.6140724946695096,
|
| 469 |
+
"grad_norm": 0.2936768596993624,
|
| 470 |
+
"kl": 0.142852783203125,
|
| 471 |
+
"learning_rate": 7.606843357124426e-06,
|
| 472 |
+
"loss": 0.0057,
|
| 473 |
+
"reward": 1.1510417088866234,
|
| 474 |
+
"reward_std": 0.26834863936528563,
|
| 475 |
+
"rewards/accuracy_reward": 0.17578125349245965,
|
| 476 |
+
"rewards/format_reward": 0.9752604328095913,
|
| 477 |
+
"step": 36
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"completion_length": 200.52344417572021,
|
| 481 |
+
"epoch": 0.6311300639658849,
|
| 482 |
+
"grad_norm": 0.3398881371255209,
|
| 483 |
+
"kl": 0.14337158203125,
|
| 484 |
+
"learning_rate": 7.024969461447973e-06,
|
| 485 |
+
"loss": 0.0057,
|
| 486 |
+
"reward": 1.140625037252903,
|
| 487 |
+
"reward_std": 0.30551271699368954,
|
| 488 |
+
"rewards/accuracy_reward": 0.20312500465661287,
|
| 489 |
+
"rewards/format_reward": 0.9375000223517418,
|
| 490 |
+
"step": 37
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"completion_length": 199.080735206604,
|
| 494 |
+
"epoch": 0.6481876332622601,
|
| 495 |
+
"grad_norm": 0.3950051985318622,
|
| 496 |
+
"kl": 0.15716552734375,
|
| 497 |
+
"learning_rate": 6.453951129574644e-06,
|
| 498 |
+
"loss": 0.0063,
|
| 499 |
+
"reward": 1.0898437723517418,
|
| 500 |
+
"reward_std": 0.3697906183078885,
|
| 501 |
+
"rewards/accuracy_reward": 0.1953125053551048,
|
| 502 |
+
"rewards/format_reward": 0.8945312686264515,
|
| 503 |
+
"step": 38
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"completion_length": 176.2083396911621,
|
| 507 |
+
"epoch": 0.6652452025586354,
|
| 508 |
+
"grad_norm": 0.37189719351321165,
|
| 509 |
+
"kl": 0.1624755859375,
|
| 510 |
+
"learning_rate": 5.895871945472434e-06,
|
| 511 |
+
"loss": 0.0065,
|
| 512 |
+
"reward": 1.1380208805203438,
|
| 513 |
+
"reward_std": 0.34118577465415,
|
| 514 |
+
"rewards/accuracy_reward": 0.20703125558793545,
|
| 515 |
+
"rewards/format_reward": 0.9309896044433117,
|
| 516 |
+
"step": 39
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"completion_length": 171.10677528381348,
|
| 520 |
+
"epoch": 0.6823027718550106,
|
| 521 |
+
"grad_norm": 0.3867912490480322,
|
| 522 |
+
"kl": 0.15789794921875,
|
| 523 |
+
"learning_rate": 5.352768279562315e-06,
|
| 524 |
+
"loss": 0.0063,
|
| 525 |
+
"reward": 1.1210937947034836,
|
| 526 |
+
"reward_std": 0.28729582112282515,
|
| 527 |
+
"rewards/accuracy_reward": 0.1640625053551048,
|
| 528 |
+
"rewards/format_reward": 0.9570312686264515,
|
| 529 |
+
"step": 40
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"completion_length": 159.16016101837158,
|
| 533 |
+
"epoch": 0.6993603411513859,
|
| 534 |
+
"grad_norm": 0.3463972010603223,
|
| 535 |
+
"kl": 0.162689208984375,
|
| 536 |
+
"learning_rate": 4.826621858223431e-06,
|
| 537 |
+
"loss": 0.0065,
|
| 538 |
+
"reward": 1.1770833656191826,
|
| 539 |
+
"reward_std": 0.2748530600219965,
|
| 540 |
+
"rewards/accuracy_reward": 0.22135417209938169,
|
| 541 |
+
"rewards/format_reward": 0.9557291865348816,
|
| 542 |
+
"step": 41
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"completion_length": 155.342453956604,
|
| 546 |
+
"epoch": 0.7164179104477612,
|
| 547 |
+
"grad_norm": 0.31066796112570005,
|
| 548 |
+
"kl": 0.16387939453125,
|
| 549 |
+
"learning_rate": 4.319352532688444e-06,
|
| 550 |
+
"loss": 0.0066,
|
| 551 |
+
"reward": 1.1692708805203438,
|
| 552 |
+
"reward_std": 0.24965191585943103,
|
| 553 |
+
"rewards/accuracy_reward": 0.19010417256504297,
|
| 554 |
+
"rewards/format_reward": 0.9791666828095913,
|
| 555 |
+
"step": 42
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"completion_length": 147.61849355697632,
|
| 559 |
+
"epoch": 0.7334754797441365,
|
| 560 |
+
"grad_norm": 0.3883785701764709,
|
| 561 |
+
"kl": 0.15966796875,
|
| 562 |
+
"learning_rate": 3.832811273714569e-06,
|
| 563 |
+
"loss": 0.0064,
|
| 564 |
+
"reward": 1.194010466337204,
|
| 565 |
+
"reward_std": 0.26059679966419935,
|
| 566 |
+
"rewards/accuracy_reward": 0.20442709000781178,
|
| 567 |
+
"rewards/format_reward": 0.9895833432674408,
|
| 568 |
+
"step": 43
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"completion_length": 160.03385829925537,
|
| 572 |
+
"epoch": 0.7505330490405118,
|
| 573 |
+
"grad_norm": 5788.508271348362,
|
| 574 |
+
"kl": 278.14984130859375,
|
| 575 |
+
"learning_rate": 3.3687734175920505e-06,
|
| 576 |
+
"loss": 11.1208,
|
| 577 |
+
"reward": 1.1757812947034836,
|
| 578 |
+
"reward_std": 0.23907144693657756,
|
| 579 |
+
"rewards/accuracy_reward": 0.18880208989139646,
|
| 580 |
+
"rewards/format_reward": 0.986979179084301,
|
| 581 |
+
"step": 44
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"completion_length": 149.94141149520874,
|
| 585 |
+
"epoch": 0.767590618336887,
|
| 586 |
+
"grad_norm": 0.36297405142580474,
|
| 587 |
+
"kl": 0.164794921875,
|
| 588 |
+
"learning_rate": 2.9289321881345257e-06,
|
| 589 |
+
"loss": 0.0066,
|
| 590 |
+
"reward": 1.199218787252903,
|
| 591 |
+
"reward_std": 0.2734901886433363,
|
| 592 |
+
"rewards/accuracy_reward": 0.21744792349636555,
|
| 593 |
+
"rewards/format_reward": 0.9817708469927311,
|
| 594 |
+
"step": 45
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"completion_length": 157.68620204925537,
|
| 598 |
+
"epoch": 0.7846481876332623,
|
| 599 |
+
"grad_norm": 0.3108491951827798,
|
| 600 |
+
"kl": 0.16436767578125,
|
| 601 |
+
"learning_rate": 2.514892518288988e-06,
|
| 602 |
+
"loss": 0.0066,
|
| 603 |
+
"reward": 1.1744792014360428,
|
| 604 |
+
"reward_std": 0.21565480902791023,
|
| 605 |
+
"rewards/accuracy_reward": 0.19010417233221233,
|
| 606 |
+
"rewards/format_reward": 0.9843750149011612,
|
| 607 |
+
"step": 46
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"completion_length": 168.7890682220459,
|
| 611 |
+
"epoch": 0.8017057569296375,
|
| 612 |
+
"grad_norm": 1.04029324716777,
|
| 613 |
+
"kl": 0.16632080078125,
|
| 614 |
+
"learning_rate": 2.1281651939094996e-06,
|
| 615 |
+
"loss": 0.0067,
|
| 616 |
+
"reward": 1.1197917014360428,
|
| 617 |
+
"reward_std": 0.20990665443241596,
|
| 618 |
+
"rewards/accuracy_reward": 0.13281250500585884,
|
| 619 |
+
"rewards/format_reward": 0.9869791753590107,
|
| 620 |
+
"step": 47
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"completion_length": 167.15625381469727,
|
| 624 |
+
"epoch": 0.8187633262260128,
|
| 625 |
+
"grad_norm": 0.32966812871466905,
|
| 626 |
+
"kl": 0.15936279296875,
|
| 627 |
+
"learning_rate": 1.7701613410634367e-06,
|
| 628 |
+
"loss": 0.0064,
|
| 629 |
+
"reward": 1.2018229588866234,
|
| 630 |
+
"reward_std": 0.2573004774749279,
|
| 631 |
+
"rewards/accuracy_reward": 0.22005208767950535,
|
| 632 |
+
"rewards/format_reward": 0.9817708469927311,
|
| 633 |
+
"step": 48
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"completion_length": 171.07682704925537,
|
| 637 |
+
"epoch": 0.835820895522388,
|
| 638 |
+
"grad_norm": 0.32208043175249607,
|
| 639 |
+
"kl": 0.151123046875,
|
| 640 |
+
"learning_rate": 1.4421872769855262e-06,
|
| 641 |
+
"loss": 0.006,
|
| 642 |
+
"reward": 1.203125037252903,
|
| 643 |
+
"reward_std": 0.26521802693605423,
|
| 644 |
+
"rewards/accuracy_reward": 0.22135417209938169,
|
| 645 |
+
"rewards/format_reward": 0.9817708469927311,
|
| 646 |
+
"step": 49
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"completion_length": 166.2864637374878,
|
| 650 |
+
"epoch": 0.8528784648187633,
|
| 651 |
+
"grad_norm": 0.34577456114094934,
|
| 652 |
+
"kl": 0.16162109375,
|
| 653 |
+
"learning_rate": 1.1454397434679022e-06,
|
| 654 |
+
"loss": 0.0065,
|
| 655 |
+
"reward": 1.2174479588866234,
|
| 656 |
+
"reward_std": 0.2847161674872041,
|
| 657 |
+
"rewards/accuracy_reward": 0.23307292023673654,
|
| 658 |
+
"rewards/format_reward": 0.9843750111758709,
|
| 659 |
+
"step": 50
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"completion_length": 176.26693153381348,
|
| 663 |
+
"epoch": 0.8699360341151386,
|
| 664 |
+
"grad_norm": 0.32710109344570104,
|
| 665 |
+
"kl": 0.151611328125,
|
| 666 |
+
"learning_rate": 8.810015400790994e-07,
|
| 667 |
+
"loss": 0.0061,
|
| 668 |
+
"reward": 1.1367187835276127,
|
| 669 |
+
"reward_std": 0.2029442568309605,
|
| 670 |
+
"rewards/accuracy_reward": 0.14713542093522847,
|
| 671 |
+
"rewards/format_reward": 0.9895833395421505,
|
| 672 |
+
"step": 51
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"completion_length": 181.84505653381348,
|
| 676 |
+
"epoch": 0.8869936034115139,
|
| 677 |
+
"grad_norm": 0.3181109150342723,
|
| 678 |
+
"kl": 0.177215576171875,
|
| 679 |
+
"learning_rate": 6.498375731458529e-07,
|
| 680 |
+
"loss": 0.0071,
|
| 681 |
+
"reward": 1.1692708618938923,
|
| 682 |
+
"reward_std": 0.21676216088235378,
|
| 683 |
+
"rewards/accuracy_reward": 0.18229167233221233,
|
| 684 |
+
"rewards/format_reward": 0.9869791753590107,
|
| 685 |
+
"step": 52
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"completion_length": 171.81901359558105,
|
| 689 |
+
"epoch": 0.9040511727078892,
|
| 690 |
+
"grad_norm": 0.3007348356231504,
|
| 691 |
+
"kl": 0.15460205078125,
|
| 692 |
+
"learning_rate": 4.5279133491454406e-07,
|
| 693 |
+
"loss": 0.0062,
|
| 694 |
+
"reward": 1.169270858168602,
|
| 695 |
+
"reward_std": 0.26831901678815484,
|
| 696 |
+
"rewards/accuracy_reward": 0.18880208872724324,
|
| 697 |
+
"rewards/format_reward": 0.9804687686264515,
|
| 698 |
+
"step": 53
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"completion_length": 165.6862015724182,
|
| 702 |
+
"epoch": 0.9211087420042644,
|
| 703 |
+
"grad_norm": 0.3465232635146208,
|
| 704 |
+
"kl": 0.160491943359375,
|
| 705 |
+
"learning_rate": 2.905818257394799e-07,
|
| 706 |
+
"loss": 0.0064,
|
| 707 |
+
"reward": 1.2343750223517418,
|
| 708 |
+
"reward_std": 0.2902528368867934,
|
| 709 |
+
"rewards/accuracy_reward": 0.2513020889600739,
|
| 710 |
+
"rewards/format_reward": 0.983072929084301,
|
| 711 |
+
"step": 54
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"completion_length": 185.47396278381348,
|
| 715 |
+
"epoch": 0.9381663113006397,
|
| 716 |
+
"grad_norm": 0.339986245997186,
|
| 717 |
+
"kl": 0.151763916015625,
|
| 718 |
+
"learning_rate": 1.6380093052856482e-07,
|
| 719 |
+
"loss": 0.0061,
|
| 720 |
+
"reward": 1.1757812909781933,
|
| 721 |
+
"reward_std": 0.2616687403060496,
|
| 722 |
+
"rewards/accuracy_reward": 0.1953125058207661,
|
| 723 |
+
"rewards/format_reward": 0.9804687611758709,
|
| 724 |
+
"step": 55
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"completion_length": 168.7773494720459,
|
| 728 |
+
"epoch": 0.9552238805970149,
|
| 729 |
+
"grad_norm": 0.33393014556768225,
|
| 730 |
+
"kl": 0.152862548828125,
|
| 731 |
+
"learning_rate": 7.291125901946027e-08,
|
| 732 |
+
"loss": 0.0061,
|
| 733 |
+
"reward": 1.2343750447034836,
|
| 734 |
+
"reward_std": 0.2642471818253398,
|
| 735 |
+
"rewards/accuracy_reward": 0.24479167396202683,
|
| 736 |
+
"rewards/format_reward": 0.9895833432674408,
|
| 737 |
+
"step": 56
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"completion_length": 175.51172637939453,
|
| 741 |
+
"epoch": 0.9722814498933902,
|
| 742 |
+
"grad_norm": 0.30356714080443054,
|
| 743 |
+
"kl": 0.1541748046875,
|
| 744 |
+
"learning_rate": 1.824445776682504e-08,
|
| 745 |
+
"loss": 0.0062,
|
| 746 |
+
"reward": 1.2239583656191826,
|
| 747 |
+
"reward_std": 0.2539575123228133,
|
| 748 |
+
"rewards/accuracy_reward": 0.24739584140479565,
|
| 749 |
+
"rewards/format_reward": 0.9765625149011612,
|
| 750 |
+
"step": 57
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"completion_length": 187.67057609558105,
|
| 754 |
+
"epoch": 0.9893390191897654,
|
| 755 |
+
"grad_norm": 0.3138605458712969,
|
| 756 |
+
"kl": 0.152740478515625,
|
| 757 |
+
"learning_rate": 0.0,
|
| 758 |
+
"loss": 0.0061,
|
| 759 |
+
"reward": 1.208333358168602,
|
| 760 |
+
"reward_std": 0.24711204366758466,
|
| 761 |
+
"rewards/accuracy_reward": 0.22786458861082792,
|
| 762 |
+
"rewards/format_reward": 0.9804687649011612,
|
| 763 |
+
"step": 58
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.9893390191897654,
|
| 767 |
+
"step": 58,
|
| 768 |
+
"total_flos": 0.0,
|
| 769 |
+
"train_loss": 0.20938398721150356,
|
| 770 |
+
"train_runtime": 4954.7093,
|
| 771 |
+
"train_samples_per_second": 1.514,
|
| 772 |
+
"train_steps_per_second": 0.012
|
| 773 |
+
}
|
| 774 |
+
],
|
| 775 |
+
"logging_steps": 1,
|
| 776 |
+
"max_steps": 58,
|
| 777 |
+
"num_input_tokens_seen": 0,
|
| 778 |
+
"num_train_epochs": 1,
|
| 779 |
+
"save_steps": 100,
|
| 780 |
+
"stateful_callbacks": {
|
| 781 |
+
"TrainerControl": {
|
| 782 |
+
"args": {
|
| 783 |
+
"should_epoch_stop": false,
|
| 784 |
+
"should_evaluate": false,
|
| 785 |
+
"should_log": false,
|
| 786 |
+
"should_save": true,
|
| 787 |
+
"should_training_stop": true
|
| 788 |
+
},
|
| 789 |
+
"attributes": {}
|
| 790 |
+
}
|
| 791 |
+
},
|
| 792 |
+
"total_flos": 0.0,
|
| 793 |
+
"train_batch_size": 16,
|
| 794 |
+
"trial_name": null,
|
| 795 |
+
"trial_params": null
|
| 796 |
+
}
|