Model save
Browse files- README.md +68 -0
- all_results.json +8 -0
- generation_config.json +6 -0
- train_results.json +8 -0
- trainer_state.json +751 -0
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: allenai/OLMoE-1B-7B-0125-Instruct
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: OLMoE-1B-7B-0125-Instruct-grpo-E6-D8000-L4096
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- grpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for OLMoE-1B-7B-0125-Instruct-grpo-E6-D8000-L4096
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [allenai/OLMoE-1B-7B-0125-Instruct](https://huggingface.co/allenai/OLMoE-1B-7B-0125-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="chenggong1995/OLMoE-1B-7B-0125-Instruct-grpo-E6-D8000-L4096", 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/gongc1995-city-university-of-hong-kong/huggingface/runs/z7mkjfwj)
|
| 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": 0.03362821891328426,
|
| 4 |
+
"train_runtime": 77437.8148,
|
| 5 |
+
"train_samples": 8000,
|
| 6 |
+
"train_samples_per_second": 0.62,
|
| 7 |
+
"train_steps_per_second": 0.003
|
| 8 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"eos_token_id": 50279,
|
| 4 |
+
"pad_token_id": 1,
|
| 5 |
+
"transformers_version": "4.49.0"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.03362821891328426,
|
| 4 |
+
"train_runtime": 77437.8148,
|
| 5 |
+
"train_samples": 8000,
|
| 6 |
+
"train_samples_per_second": 0.62,
|
| 7 |
+
"train_steps_per_second": 0.003
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,751 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.984,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 246,
|
| 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": 583.2621688842773,
|
| 14 |
+
"epoch": 0.12,
|
| 15 |
+
"grad_norm": 0.12920165061950684,
|
| 16 |
+
"kl": 0.0007047017415364583,
|
| 17 |
+
"learning_rate": 4e-07,
|
| 18 |
+
"loss": 0.0039,
|
| 19 |
+
"reward": 0.07673611293236414,
|
| 20 |
+
"reward_std": 0.12897611850251753,
|
| 21 |
+
"rewards/accuracy_reward": 0.07100694642091791,
|
| 22 |
+
"rewards/format_reward": 0.0057291668374091385,
|
| 23 |
+
"step": 5
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"clip_ratio": 0.0,
|
| 27 |
+
"completion_length": 576.849148050944,
|
| 28 |
+
"epoch": 0.24,
|
| 29 |
+
"grad_norm": 0.14590902626514435,
|
| 30 |
+
"kl": 0.0013418197631835938,
|
| 31 |
+
"learning_rate": 8e-07,
|
| 32 |
+
"loss": 0.0005,
|
| 33 |
+
"reward": 0.0871527800646921,
|
| 34 |
+
"reward_std": 0.13713855588187773,
|
| 35 |
+
"rewards/accuracy_reward": 0.081250002173086,
|
| 36 |
+
"rewards/format_reward": 0.005902777938172221,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"clip_ratio": 0.0,
|
| 41 |
+
"completion_length": 595.0527969360352,
|
| 42 |
+
"epoch": 0.36,
|
| 43 |
+
"grad_norm": 0.13242916762828827,
|
| 44 |
+
"kl": 0.0022638956705729168,
|
| 45 |
+
"learning_rate": 1.2e-06,
|
| 46 |
+
"loss": 0.0035,
|
| 47 |
+
"reward": 0.08645833590999245,
|
| 48 |
+
"reward_std": 0.12982427552342415,
|
| 49 |
+
"rewards/accuracy_reward": 0.07795139101023475,
|
| 50 |
+
"rewards/format_reward": 0.008506944651405017,
|
| 51 |
+
"step": 15
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"clip_ratio": 0.0,
|
| 55 |
+
"completion_length": 589.0404724121094,
|
| 56 |
+
"epoch": 0.48,
|
| 57 |
+
"grad_norm": 0.19261103868484497,
|
| 58 |
+
"kl": 0.006444295247395833,
|
| 59 |
+
"learning_rate": 1.6e-06,
|
| 60 |
+
"loss": 0.0083,
|
| 61 |
+
"reward": 0.12013889234513045,
|
| 62 |
+
"reward_std": 0.17057897535463173,
|
| 63 |
+
"rewards/accuracy_reward": 0.10694444783342381,
|
| 64 |
+
"rewards/format_reward": 0.013194444729015231,
|
| 65 |
+
"step": 20
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"clip_ratio": 0.0,
|
| 69 |
+
"completion_length": 600.2750193277994,
|
| 70 |
+
"epoch": 0.6,
|
| 71 |
+
"grad_norm": 0.18090881407260895,
|
| 72 |
+
"kl": 0.00933685302734375,
|
| 73 |
+
"learning_rate": 2e-06,
|
| 74 |
+
"loss": 0.011,
|
| 75 |
+
"reward": 0.12621528124436737,
|
| 76 |
+
"reward_std": 0.1751370935390393,
|
| 77 |
+
"rewards/accuracy_reward": 0.11458333623595536,
|
| 78 |
+
"rewards/format_reward": 0.011631944697971146,
|
| 79 |
+
"step": 25
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"clip_ratio": 0.0,
|
| 83 |
+
"completion_length": 577.3231079101563,
|
| 84 |
+
"epoch": 0.72,
|
| 85 |
+
"grad_norm": 0.14372943341732025,
|
| 86 |
+
"kl": 0.009203084309895833,
|
| 87 |
+
"learning_rate": 1.9974751105436262e-06,
|
| 88 |
+
"loss": 0.0063,
|
| 89 |
+
"reward": 0.13611111496575176,
|
| 90 |
+
"reward_std": 0.17103372573231657,
|
| 91 |
+
"rewards/accuracy_reward": 0.1302083367947489,
|
| 92 |
+
"rewards/format_reward": 0.005902777938172221,
|
| 93 |
+
"step": 30
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"clip_ratio": 0.0,
|
| 97 |
+
"completion_length": 564.1434224446615,
|
| 98 |
+
"epoch": 0.84,
|
| 99 |
+
"grad_norm": 0.15495018661022186,
|
| 100 |
+
"kl": 0.009105428059895834,
|
| 101 |
+
"learning_rate": 1.98991319230804e-06,
|
| 102 |
+
"loss": 0.0107,
|
| 103 |
+
"reward": 0.1550347256163756,
|
| 104 |
+
"reward_std": 0.19651179468880098,
|
| 105 |
+
"rewards/accuracy_reward": 0.14583333718279998,
|
| 106 |
+
"rewards/format_reward": 0.009201389101023475,
|
| 107 |
+
"step": 35
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"clip_ratio": 0.0,
|
| 111 |
+
"completion_length": 600.4140803019205,
|
| 112 |
+
"epoch": 0.96,
|
| 113 |
+
"grad_norm": 0.17615483701229095,
|
| 114 |
+
"kl": 0.012515767415364584,
|
| 115 |
+
"learning_rate": 1.9773524313084854e-06,
|
| 116 |
+
"loss": 0.0119,
|
| 117 |
+
"reward": 0.15885417150954406,
|
| 118 |
+
"reward_std": 0.2019161203255256,
|
| 119 |
+
"rewards/accuracy_reward": 0.14166667039195696,
|
| 120 |
+
"rewards/format_reward": 0.017187500388051072,
|
| 121 |
+
"step": 40
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"clip_ratio": 0.0,
|
| 125 |
+
"completion_length": 631.5385569254557,
|
| 126 |
+
"epoch": 1.096,
|
| 127 |
+
"grad_norm": 0.16588236391544342,
|
| 128 |
+
"kl": 0.014839680989583333,
|
| 129 |
+
"learning_rate": 1.959856256610988e-06,
|
| 130 |
+
"loss": 0.0096,
|
| 131 |
+
"reward": 0.18350694837669532,
|
| 132 |
+
"reward_std": 0.2348036120335261,
|
| 133 |
+
"rewards/accuracy_reward": 0.15277778275000553,
|
| 134 |
+
"rewards/format_reward": 0.030729167473812897,
|
| 135 |
+
"step": 45
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"clip_ratio": 0.0,
|
| 139 |
+
"completion_length": 593.8875162760417,
|
| 140 |
+
"epoch": 1.216,
|
| 141 |
+
"grad_norm": 0.18021586537361145,
|
| 142 |
+
"kl": 0.016481526692708335,
|
| 143 |
+
"learning_rate": 1.9375130200295876e-06,
|
| 144 |
+
"loss": 0.0189,
|
| 145 |
+
"reward": 0.19756944961845874,
|
| 146 |
+
"reward_std": 0.24355731457471846,
|
| 147 |
+
"rewards/accuracy_reward": 0.16006944881131252,
|
| 148 |
+
"rewards/format_reward": 0.03750000107102096,
|
| 149 |
+
"step": 50
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"clip_ratio": 0.0,
|
| 153 |
+
"completion_length": 587.4776229858398,
|
| 154 |
+
"epoch": 1.336,
|
| 155 |
+
"grad_norm": 0.1994011402130127,
|
| 156 |
+
"kl": 0.0164459228515625,
|
| 157 |
+
"learning_rate": 1.9104355499692162e-06,
|
| 158 |
+
"loss": 0.0176,
|
| 159 |
+
"reward": 0.2189236176510652,
|
| 160 |
+
"reward_std": 0.28772813665370145,
|
| 161 |
+
"rewards/accuracy_reward": 0.14583333767950535,
|
| 162 |
+
"rewards/format_reward": 0.07309028026647865,
|
| 163 |
+
"step": 55
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"clip_ratio": 0.0,
|
| 167 |
+
"completion_length": 575.6356079101563,
|
| 168 |
+
"epoch": 1.456,
|
| 169 |
+
"grad_norm": 0.23497304320335388,
|
| 170 |
+
"kl": 0.02615966796875,
|
| 171 |
+
"learning_rate": 1.8787605816671951e-06,
|
| 172 |
+
"loss": 0.0222,
|
| 173 |
+
"reward": 0.33281251018246016,
|
| 174 |
+
"reward_std": 0.3787623425324758,
|
| 175 |
+
"rewards/accuracy_reward": 0.13923611460874477,
|
| 176 |
+
"rewards/format_reward": 0.19357639501492183,
|
| 177 |
+
"step": 60
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"clip_ratio": 0.0,
|
| 181 |
+
"completion_length": 581.5133870442709,
|
| 182 |
+
"epoch": 1.576,
|
| 183 |
+
"grad_norm": 0.2583252191543579,
|
| 184 |
+
"kl": 0.039479573567708336,
|
| 185 |
+
"learning_rate": 1.8426480667105175e-06,
|
| 186 |
+
"loss": 0.0407,
|
| 187 |
+
"reward": 0.4626736263434092,
|
| 188 |
+
"reward_std": 0.45475957343975704,
|
| 189 |
+
"rewards/accuracy_reward": 0.13229167129223546,
|
| 190 |
+
"rewards/format_reward": 0.33038195346792537,
|
| 191 |
+
"step": 65
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"clip_ratio": 0.0,
|
| 195 |
+
"completion_length": 575.3401219685873,
|
| 196 |
+
"epoch": 1.696,
|
| 197 |
+
"grad_norm": 0.2477940022945404,
|
| 198 |
+
"kl": 0.04772135416666667,
|
| 199 |
+
"learning_rate": 1.8022803653156982e-06,
|
| 200 |
+
"loss": 0.037,
|
| 201 |
+
"reward": 0.6152777954936027,
|
| 202 |
+
"reward_std": 0.4820722574989001,
|
| 203 |
+
"rewards/accuracy_reward": 0.11440972487131755,
|
| 204 |
+
"rewards/format_reward": 0.5008680661519368,
|
| 205 |
+
"step": 70
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"clip_ratio": 0.0,
|
| 209 |
+
"completion_length": 592.2732818603515,
|
| 210 |
+
"epoch": 1.8159999999999998,
|
| 211 |
+
"grad_norm": 0.240928515791893,
|
| 212 |
+
"kl": 0.045609537760416666,
|
| 213 |
+
"learning_rate": 1.7578613254499968e-06,
|
| 214 |
+
"loss": 0.0367,
|
| 215 |
+
"reward": 0.6859375188748041,
|
| 216 |
+
"reward_std": 0.48161858022212983,
|
| 217 |
+
"rewards/accuracy_reward": 0.1130208361428231,
|
| 218 |
+
"rewards/format_reward": 0.5729166840513548,
|
| 219 |
+
"step": 75
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"clip_ratio": 0.0,
|
| 223 |
+
"completion_length": 570.3640808105469,
|
| 224 |
+
"epoch": 1.936,
|
| 225 |
+
"grad_norm": 0.30649441480636597,
|
| 226 |
+
"kl": 0.05423177083333333,
|
| 227 |
+
"learning_rate": 1.7096152534442513e-06,
|
| 228 |
+
"loss": 0.0334,
|
| 229 |
+
"reward": 0.7704861332972844,
|
| 230 |
+
"reward_std": 0.4483942608038584,
|
| 231 |
+
"rewards/accuracy_reward": 0.10572916980211934,
|
| 232 |
+
"rewards/format_reward": 0.6647569671273231,
|
| 233 |
+
"step": 80
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"clip_ratio": 0.0,
|
| 237 |
+
"completion_length": 560.8760559082032,
|
| 238 |
+
"epoch": 2.072,
|
| 239 |
+
"grad_norm": 0.25217875838279724,
|
| 240 |
+
"kl": 0.054768880208333336,
|
| 241 |
+
"learning_rate": 1.6577857812954991e-06,
|
| 242 |
+
"loss": 0.0375,
|
| 243 |
+
"reward": 0.8114583571751912,
|
| 244 |
+
"reward_std": 0.4467007691661517,
|
| 245 |
+
"rewards/accuracy_reward": 0.12256944736776253,
|
| 246 |
+
"rewards/format_reward": 0.6888889064391454,
|
| 247 |
+
"step": 85
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"clip_ratio": 0.0,
|
| 251 |
+
"completion_length": 565.7517557779948,
|
| 252 |
+
"epoch": 2.192,
|
| 253 |
+
"grad_norm": 0.27332058548927307,
|
| 254 |
+
"kl": 0.0546630859375,
|
| 255 |
+
"learning_rate": 1.6026346363792564e-06,
|
| 256 |
+
"loss": 0.0329,
|
| 257 |
+
"reward": 0.8633680770794551,
|
| 258 |
+
"reward_std": 0.47069497853517533,
|
| 259 |
+
"rewards/accuracy_reward": 0.15156250428408385,
|
| 260 |
+
"rewards/format_reward": 0.711805577079455,
|
| 261 |
+
"step": 90
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"clip_ratio": 0.0,
|
| 265 |
+
"completion_length": 564.6762344360352,
|
| 266 |
+
"epoch": 2.312,
|
| 267 |
+
"grad_norm": 0.22232523560523987,
|
| 268 |
+
"kl": 0.05868123372395833,
|
| 269 |
+
"learning_rate": 1.5444403197841344e-06,
|
| 270 |
+
"loss": 0.0211,
|
| 271 |
+
"reward": 0.8944444666306178,
|
| 272 |
+
"reward_std": 0.42260901977618537,
|
| 273 |
+
"rewards/accuracy_reward": 0.1461805594774584,
|
| 274 |
+
"rewards/format_reward": 0.7482639104127884,
|
| 275 |
+
"step": 95
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 2.432,
|
| 279 |
+
"grad_norm": 0.21494214236736298,
|
| 280 |
+
"learning_rate": 1.4834966999429178e-06,
|
| 281 |
+
"loss": 0.0271,
|
| 282 |
+
"step": 100
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 2.432,
|
| 286 |
+
"eval_clip_ratio": 0.0,
|
| 287 |
+
"eval_completion_length": 528.4134756234976,
|
| 288 |
+
"eval_kl": 0.060819185697115384,
|
| 289 |
+
"eval_loss": 0.013302656821906567,
|
| 290 |
+
"eval_reward": 0.97275644999284,
|
| 291 |
+
"eval_reward_std": 0.405555764069924,
|
| 292 |
+
"eval_rewards/accuracy_reward": 0.16025641531898424,
|
| 293 |
+
"eval_rewards/format_reward": 0.8125000275098361,
|
| 294 |
+
"eval_runtime": 164.4929,
|
| 295 |
+
"eval_samples_per_second": 0.602,
|
| 296 |
+
"eval_steps_per_second": 0.018,
|
| 297 |
+
"step": 100
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"clip_ratio": 0.0,
|
| 301 |
+
"completion_length": 562.4533172607422,
|
| 302 |
+
"epoch": 2.552,
|
| 303 |
+
"grad_norm": 0.24463680386543274,
|
| 304 |
+
"kl": 0.06415608723958334,
|
| 305 |
+
"learning_rate": 1.4201115286619464e-06,
|
| 306 |
+
"loss": 0.0246,
|
| 307 |
+
"reward": 0.9512153029441833,
|
| 308 |
+
"reward_std": 0.4153951602677504,
|
| 309 |
+
"rewards/accuracy_reward": 0.16093750478078922,
|
| 310 |
+
"rewards/format_reward": 0.7902777964870135,
|
| 311 |
+
"step": 105
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"clip_ratio": 0.0,
|
| 315 |
+
"completion_length": 558.5052256266276,
|
| 316 |
+
"epoch": 2.672,
|
| 317 |
+
"grad_norm": 0.22716720402240753,
|
| 318 |
+
"kl": 0.066162109375,
|
| 319 |
+
"learning_rate": 1.3546048870425354e-06,
|
| 320 |
+
"loss": 0.0242,
|
| 321 |
+
"reward": 0.9326389094193777,
|
| 322 |
+
"reward_std": 0.4124096731344859,
|
| 323 |
+
"rewards/accuracy_reward": 0.15625000453243654,
|
| 324 |
+
"rewards/format_reward": 0.7763889074325562,
|
| 325 |
+
"step": 110
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"clip_ratio": 0.0,
|
| 329 |
+
"completion_length": 547.4661656697591,
|
| 330 |
+
"epoch": 2.792,
|
| 331 |
+
"grad_norm": 0.2325204312801361,
|
| 332 |
+
"kl": 0.07105712890625,
|
| 333 |
+
"learning_rate": 1.2873075691421806e-06,
|
| 334 |
+
"loss": 0.0198,
|
| 335 |
+
"reward": 0.9222222457329432,
|
| 336 |
+
"reward_std": 0.4309779698650042,
|
| 337 |
+
"rewards/accuracy_reward": 0.16579861616094907,
|
| 338 |
+
"rewards/format_reward": 0.7564236293236415,
|
| 339 |
+
"step": 115
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"clip_ratio": 0.0,
|
| 343 |
+
"completion_length": 547.4175537109375,
|
| 344 |
+
"epoch": 2.912,
|
| 345 |
+
"grad_norm": 0.23773518204689026,
|
| 346 |
+
"kl": 0.06711832682291667,
|
| 347 |
+
"learning_rate": 1.218559411537699e-06,
|
| 348 |
+
"loss": 0.0211,
|
| 349 |
+
"reward": 0.894270858168602,
|
| 350 |
+
"reward_std": 0.41917893588542937,
|
| 351 |
+
"rewards/accuracy_reward": 0.15920139361793798,
|
| 352 |
+
"rewards/format_reward": 0.7350694666306178,
|
| 353 |
+
"step": 120
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"clip_ratio": 0.0,
|
| 357 |
+
"completion_length": 544.4187637329102,
|
| 358 |
+
"epoch": 3.048,
|
| 359 |
+
"grad_norm": 0.25982293486595154,
|
| 360 |
+
"kl": 0.07174072265625,
|
| 361 |
+
"learning_rate": 1.1487075772256517e-06,
|
| 362 |
+
"loss": 0.029,
|
| 363 |
+
"reward": 0.8729166895151138,
|
| 364 |
+
"reward_std": 0.4223095287879308,
|
| 365 |
+
"rewards/accuracy_reward": 0.16006944874922435,
|
| 366 |
+
"rewards/format_reward": 0.7128472457329432,
|
| 367 |
+
"step": 125
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"clip_ratio": 0.0,
|
| 371 |
+
"completion_length": 520.4824803670248,
|
| 372 |
+
"epoch": 3.168,
|
| 373 |
+
"grad_norm": 0.24558736383914948,
|
| 374 |
+
"kl": 0.08404541015625,
|
| 375 |
+
"learning_rate": 1.0781048025259646e-06,
|
| 376 |
+
"loss": 0.0261,
|
| 377 |
+
"reward": 0.8881944636503856,
|
| 378 |
+
"reward_std": 0.42243550966183346,
|
| 379 |
+
"rewards/accuracy_reward": 0.16284722611308097,
|
| 380 |
+
"rewards/format_reward": 0.725347238779068,
|
| 381 |
+
"step": 130
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"clip_ratio": 0.0,
|
| 385 |
+
"completion_length": 537.1248448689779,
|
| 386 |
+
"epoch": 3.288,
|
| 387 |
+
"grad_norm": 0.29106763005256653,
|
| 388 |
+
"kl": 0.07824300130208334,
|
| 389 |
+
"learning_rate": 1.0071076158414974e-06,
|
| 390 |
+
"loss": 0.0249,
|
| 391 |
+
"reward": 0.9265625198682149,
|
| 392 |
+
"reward_std": 0.41332067002852757,
|
| 393 |
+
"rewards/accuracy_reward": 0.17378472685813903,
|
| 394 |
+
"rewards/format_reward": 0.7527777969837188,
|
| 395 |
+
"step": 135
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"clip_ratio": 0.0,
|
| 399 |
+
"completion_length": 550.350016784668,
|
| 400 |
+
"epoch": 3.408,
|
| 401 |
+
"grad_norm": 0.2443486452102661,
|
| 402 |
+
"kl": 0.071923828125,
|
| 403 |
+
"learning_rate": 9.360745372684345e-07,
|
| 404 |
+
"loss": 0.0295,
|
| 405 |
+
"reward": 0.8960069666306177,
|
| 406 |
+
"reward_std": 0.4272393837571144,
|
| 407 |
+
"rewards/accuracy_reward": 0.17343750571211178,
|
| 408 |
+
"rewards/format_reward": 0.7225694636503855,
|
| 409 |
+
"step": 140
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"clip_ratio": 0.0,
|
| 413 |
+
"completion_length": 543.1215454101563,
|
| 414 |
+
"epoch": 3.528,
|
| 415 |
+
"grad_norm": 0.24748581647872925,
|
| 416 |
+
"kl": 0.07239176432291666,
|
| 417 |
+
"learning_rate": 8.653642681490607e-07,
|
| 418 |
+
"loss": 0.0247,
|
| 419 |
+
"reward": 0.9074653029441834,
|
| 420 |
+
"reward_std": 0.4173097605506579,
|
| 421 |
+
"rewards/accuracy_reward": 0.1642361162851254,
|
| 422 |
+
"rewards/format_reward": 0.7432291815678279,
|
| 423 |
+
"step": 145
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"clip_ratio": 0.0,
|
| 427 |
+
"completion_length": 539.6696355183919,
|
| 428 |
+
"epoch": 3.648,
|
| 429 |
+
"grad_norm": 0.2420652210712433,
|
| 430 |
+
"kl": 0.0806884765625,
|
| 431 |
+
"learning_rate": 7.953338797092901e-07,
|
| 432 |
+
"loss": 0.0247,
|
| 433 |
+
"reward": 0.9553819715976715,
|
| 434 |
+
"reward_std": 0.39353689054648083,
|
| 435 |
+
"rewards/accuracy_reward": 0.15902778361923992,
|
| 436 |
+
"rewards/format_reward": 0.7963541895151138,
|
| 437 |
+
"step": 150
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"clip_ratio": 0.0,
|
| 441 |
+
"completion_length": 542.0953282674153,
|
| 442 |
+
"epoch": 3.768,
|
| 443 |
+
"grad_norm": 0.22927226126194,
|
| 444 |
+
"kl": 0.08297119140625,
|
| 445 |
+
"learning_rate": 7.263370099279171e-07,
|
| 446 |
+
"loss": 0.0375,
|
| 447 |
+
"reward": 0.9421875288089117,
|
| 448 |
+
"reward_std": 0.3737917934854825,
|
| 449 |
+
"rewards/accuracy_reward": 0.14618056000520785,
|
| 450 |
+
"rewards/format_reward": 0.7960069606701533,
|
| 451 |
+
"step": 155
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"clip_ratio": 0.0,
|
| 455 |
+
"completion_length": 560.2560948689778,
|
| 456 |
+
"epoch": 3.888,
|
| 457 |
+
"grad_norm": 0.2442265897989273,
|
| 458 |
+
"kl": 0.08262532552083333,
|
| 459 |
+
"learning_rate": 6.587220777430095e-07,
|
| 460 |
+
"loss": 0.0438,
|
| 461 |
+
"reward": 0.9187500218550364,
|
| 462 |
+
"reward_std": 0.3985011622309685,
|
| 463 |
+
"rewards/accuracy_reward": 0.1730902827034394,
|
| 464 |
+
"rewards/format_reward": 0.7456597417593003,
|
| 465 |
+
"step": 160
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"clip_ratio": 0.0,
|
| 469 |
+
"completion_length": 549.7503621419271,
|
| 470 |
+
"epoch": 4.024,
|
| 471 |
+
"grad_norm": 0.36080020666122437,
|
| 472 |
+
"kl": 0.09295247395833334,
|
| 473 |
+
"learning_rate": 5.928305236133016e-07,
|
| 474 |
+
"loss": 0.0348,
|
| 475 |
+
"reward": 0.8918403009573619,
|
| 476 |
+
"reward_std": 0.42846539815266926,
|
| 477 |
+
"rewards/accuracy_reward": 0.16701389451821644,
|
| 478 |
+
"rewards/format_reward": 0.7248264094193776,
|
| 479 |
+
"step": 165
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"clip_ratio": 0.0,
|
| 483 |
+
"completion_length": 527.6090423583985,
|
| 484 |
+
"epoch": 4.144,
|
| 485 |
+
"grad_norm": 0.2888505756855011,
|
| 486 |
+
"kl": 0.0927978515625,
|
| 487 |
+
"learning_rate": 5.289950853193652e-07,
|
| 488 |
+
"loss": 0.0569,
|
| 489 |
+
"reward": 0.9263889143864313,
|
| 490 |
+
"reward_std": 0.4129257212082545,
|
| 491 |
+
"rewards/accuracy_reward": 0.17309028257926304,
|
| 492 |
+
"rewards/format_reward": 0.75329862733682,
|
| 493 |
+
"step": 170
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"clip_ratio": 0.0,
|
| 497 |
+
"completion_length": 521.2873443603515,
|
| 498 |
+
"epoch": 4.264,
|
| 499 |
+
"grad_norm": 0.2377632111310959,
|
| 500 |
+
"kl": 0.09527180989583334,
|
| 501 |
+
"learning_rate": 4.6753811771138365e-07,
|
| 502 |
+
"loss": 0.0401,
|
| 503 |
+
"reward": 0.9312500258286794,
|
| 504 |
+
"reward_std": 0.3798152153690656,
|
| 505 |
+
"rewards/accuracy_reward": 0.16128472660978635,
|
| 506 |
+
"rewards/format_reward": 0.7699652989705403,
|
| 507 |
+
"step": 175
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"clip_ratio": 0.0,
|
| 511 |
+
"completion_length": 545.0975886027019,
|
| 512 |
+
"epoch": 4.384,
|
| 513 |
+
"grad_norm": 0.2456796020269394,
|
| 514 |
+
"kl": 0.093505859375,
|
| 515 |
+
"learning_rate": 4.0876996488842475e-07,
|
| 516 |
+
"loss": 0.0539,
|
| 517 |
+
"reward": 0.9411458532015483,
|
| 518 |
+
"reward_std": 0.40113388895988467,
|
| 519 |
+
"rewards/accuracy_reward": 0.17621528282761573,
|
| 520 |
+
"rewards/format_reward": 0.7649305760860443,
|
| 521 |
+
"step": 180
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"clip_ratio": 0.0,
|
| 525 |
+
"completion_length": 552.6007125854492,
|
| 526 |
+
"epoch": 4.504,
|
| 527 |
+
"grad_norm": 0.24820686876773834,
|
| 528 |
+
"kl": 0.09252522786458334,
|
| 529 |
+
"learning_rate": 3.529873930293545e-07,
|
| 530 |
+
"loss": 0.0572,
|
| 531 |
+
"reward": 0.9163194666306178,
|
| 532 |
+
"reward_std": 0.38814649879932406,
|
| 533 |
+
"rewards/accuracy_reward": 0.15885417160267631,
|
| 534 |
+
"rewards/format_reward": 0.7574652969837189,
|
| 535 |
+
"step": 185
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"clip_ratio": 0.0,
|
| 539 |
+
"completion_length": 533.9510559082031,
|
| 540 |
+
"epoch": 4.624,
|
| 541 |
+
"grad_norm": 0.2605077922344208,
|
| 542 |
+
"kl": 0.09794921875,
|
| 543 |
+
"learning_rate": 3.0047209178924635e-07,
|
| 544 |
+
"loss": 0.0484,
|
| 545 |
+
"reward": 0.9178819686174393,
|
| 546 |
+
"reward_std": 0.4066275705893834,
|
| 547 |
+
"rewards/accuracy_reward": 0.16649306093653043,
|
| 548 |
+
"rewards/format_reward": 0.7513889104127884,
|
| 549 |
+
"step": 190
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"clip_ratio": 0.0,
|
| 553 |
+
"completion_length": 535.4948069254557,
|
| 554 |
+
"epoch": 4.744,
|
| 555 |
+
"grad_norm": 0.24027736485004425,
|
| 556 |
+
"kl": 0.10040690104166666,
|
| 557 |
+
"learning_rate": 2.514892518288988e-07,
|
| 558 |
+
"loss": 0.0508,
|
| 559 |
+
"reward": 0.9135416855414709,
|
| 560 |
+
"reward_std": 0.3980190739035606,
|
| 561 |
+
"rewards/accuracy_reward": 0.1682291696468989,
|
| 562 |
+
"rewards/format_reward": 0.7453125208616257,
|
| 563 |
+
"step": 195
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.864,
|
| 567 |
+
"grad_norm": 0.24247297644615173,
|
| 568 |
+
"learning_rate": 2.0628622566063058e-07,
|
| 569 |
+
"loss": 0.0556,
|
| 570 |
+
"step": 200
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.864,
|
| 574 |
+
"eval_clip_ratio": 0.0,
|
| 575 |
+
"eval_completion_length": 497.66507427509015,
|
| 576 |
+
"eval_kl": 0.10235126201923077,
|
| 577 |
+
"eval_loss": 0.09290527552366257,
|
| 578 |
+
"eval_reward": 0.8525641239606417,
|
| 579 |
+
"eval_reward_std": 0.4311282199162703,
|
| 580 |
+
"eval_rewards/accuracy_reward": 0.1314102616161108,
|
| 581 |
+
"eval_rewards/format_reward": 0.7211538690787095,
|
| 582 |
+
"eval_runtime": 164.8378,
|
| 583 |
+
"eval_samples_per_second": 0.601,
|
| 584 |
+
"eval_steps_per_second": 0.018,
|
| 585 |
+
"step": 200
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"clip_ratio": 0.0,
|
| 589 |
+
"completion_length": 527.5610371907552,
|
| 590 |
+
"epoch": 4.984,
|
| 591 |
+
"grad_norm": 0.24340181052684784,
|
| 592 |
+
"kl": 0.103265380859375,
|
| 593 |
+
"learning_rate": 1.6509127857277782e-07,
|
| 594 |
+
"loss": 0.0592,
|
| 595 |
+
"reward": 0.9018229390184085,
|
| 596 |
+
"reward_std": 0.38689753947158656,
|
| 597 |
+
"rewards/accuracy_reward": 0.16232639336958526,
|
| 598 |
+
"rewards/format_reward": 0.7394965469837189,
|
| 599 |
+
"step": 205
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"clip_ratio": 0.0,
|
| 603 |
+
"completion_length": 535.6948079427083,
|
| 604 |
+
"epoch": 5.12,
|
| 605 |
+
"grad_norm": 0.2896415889263153,
|
| 606 |
+
"kl": 0.10675455729166666,
|
| 607 |
+
"learning_rate": 1.2811243594045694e-07,
|
| 608 |
+
"loss": 0.059,
|
| 609 |
+
"reward": 0.8942708512147267,
|
| 610 |
+
"reward_std": 0.4014328221480052,
|
| 611 |
+
"rewards/accuracy_reward": 0.15416667043852308,
|
| 612 |
+
"rewards/format_reward": 0.7401041835546494,
|
| 613 |
+
"step": 210
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"clip_ratio": 0.0,
|
| 617 |
+
"completion_length": 525.1632136027018,
|
| 618 |
+
"epoch": 5.24,
|
| 619 |
+
"grad_norm": 0.2585560977458954,
|
| 620 |
+
"kl": 0.10707194010416667,
|
| 621 |
+
"learning_rate": 9.55364327434105e-08,
|
| 622 |
+
"loss": 0.0576,
|
| 623 |
+
"reward": 0.8967014074325561,
|
| 624 |
+
"reward_std": 0.3964859182635943,
|
| 625 |
+
"rewards/accuracy_reward": 0.1552083384245634,
|
| 626 |
+
"rewards/format_reward": 0.7414930721124013,
|
| 627 |
+
"step": 215
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"clip_ratio": 0.0,
|
| 631 |
+
"completion_length": 518.6449793497721,
|
| 632 |
+
"epoch": 5.36,
|
| 633 |
+
"grad_norm": 0.2638151943683624,
|
| 634 |
+
"kl": 0.110107421875,
|
| 635 |
+
"learning_rate": 6.75277705956443e-08,
|
| 636 |
+
"loss": 0.0621,
|
| 637 |
+
"reward": 0.910069465637207,
|
| 638 |
+
"reward_std": 0.38575134972731273,
|
| 639 |
+
"rewards/accuracy_reward": 0.1703125045945247,
|
| 640 |
+
"rewards/format_reward": 0.739756965637207,
|
| 641 |
+
"step": 220
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"clip_ratio": 0.0,
|
| 645 |
+
"completion_length": 519.6583480834961,
|
| 646 |
+
"epoch": 5.48,
|
| 647 |
+
"grad_norm": 0.2671918570995331,
|
| 648 |
+
"kl": 0.10970865885416667,
|
| 649 |
+
"learning_rate": 4.422788704864633e-08,
|
| 650 |
+
"loss": 0.0555,
|
| 651 |
+
"reward": 0.9017361313104629,
|
| 652 |
+
"reward_std": 0.39265564555923144,
|
| 653 |
+
"rewards/accuracy_reward": 0.16128472667187452,
|
| 654 |
+
"rewards/format_reward": 0.7404514094193776,
|
| 655 |
+
"step": 225
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"clip_ratio": 0.0,
|
| 659 |
+
"completion_length": 518.6614771525066,
|
| 660 |
+
"epoch": 5.6,
|
| 661 |
+
"grad_norm": 0.2516481876373291,
|
| 662 |
+
"kl": 0.10393880208333334,
|
| 663 |
+
"learning_rate": 2.575444136302185e-08,
|
| 664 |
+
"loss": 0.0585,
|
| 665 |
+
"reward": 0.9203125298023224,
|
| 666 |
+
"reward_std": 0.3854361062248548,
|
| 667 |
+
"rewards/accuracy_reward": 0.16527778220673403,
|
| 668 |
+
"rewards/format_reward": 0.7550347407658895,
|
| 669 |
+
"step": 230
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"clip_ratio": 0.0,
|
| 673 |
+
"completion_length": 523.0328257242838,
|
| 674 |
+
"epoch": 5.72,
|
| 675 |
+
"grad_norm": 0.24842773377895355,
|
| 676 |
+
"kl": 0.10460611979166666,
|
| 677 |
+
"learning_rate": 1.220072035523989e-08,
|
| 678 |
+
"loss": 0.0615,
|
| 679 |
+
"reward": 0.9104166885217031,
|
| 680 |
+
"reward_std": 0.4020949920018514,
|
| 681 |
+
"rewards/accuracy_reward": 0.16076389284183581,
|
| 682 |
+
"rewards/format_reward": 0.7496527930100759,
|
| 683 |
+
"step": 235
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"clip_ratio": 0.0,
|
| 687 |
+
"completion_length": 548.0658167521159,
|
| 688 |
+
"epoch": 5.84,
|
| 689 |
+
"grad_norm": 0.2960110008716583,
|
| 690 |
+
"kl": 0.10340983072916667,
|
| 691 |
+
"learning_rate": 3.6351673198347087e-09,
|
| 692 |
+
"loss": 0.0572,
|
| 693 |
+
"reward": 0.894097242752711,
|
| 694 |
+
"reward_std": 0.39487800349791846,
|
| 695 |
+
"rewards/accuracy_reward": 0.15312500384946665,
|
| 696 |
+
"rewards/format_reward": 0.740972242752711,
|
| 697 |
+
"step": 240
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"clip_ratio": 0.0,
|
| 701 |
+
"completion_length": 524.3946365356445,
|
| 702 |
+
"epoch": 5.96,
|
| 703 |
+
"grad_norm": 0.28276899456977844,
|
| 704 |
+
"kl": 0.1115234375,
|
| 705 |
+
"learning_rate": 1.0103640590064522e-10,
|
| 706 |
+
"loss": 0.0614,
|
| 707 |
+
"reward": 0.8968750178813935,
|
| 708 |
+
"reward_std": 0.40502374321222306,
|
| 709 |
+
"rewards/accuracy_reward": 0.15798611640930177,
|
| 710 |
+
"rewards/format_reward": 0.7388889054457347,
|
| 711 |
+
"step": 245
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"clip_ratio": 0.0,
|
| 715 |
+
"completion_length": 484.8463541666667,
|
| 716 |
+
"epoch": 5.984,
|
| 717 |
+
"kl": 0.11258951822916667,
|
| 718 |
+
"reward": 0.8923611293236414,
|
| 719 |
+
"reward_std": 0.40955925981203717,
|
| 720 |
+
"rewards/accuracy_reward": 0.1605902835726738,
|
| 721 |
+
"rewards/format_reward": 0.731770858168602,
|
| 722 |
+
"step": 246,
|
| 723 |
+
"total_flos": 0.0,
|
| 724 |
+
"train_loss": 0.03362821891328426,
|
| 725 |
+
"train_runtime": 77437.8148,
|
| 726 |
+
"train_samples_per_second": 0.62,
|
| 727 |
+
"train_steps_per_second": 0.003
|
| 728 |
+
}
|
| 729 |
+
],
|
| 730 |
+
"logging_steps": 5,
|
| 731 |
+
"max_steps": 246,
|
| 732 |
+
"num_input_tokens_seen": 0,
|
| 733 |
+
"num_train_epochs": 6,
|
| 734 |
+
"save_steps": 500,
|
| 735 |
+
"stateful_callbacks": {
|
| 736 |
+
"TrainerControl": {
|
| 737 |
+
"args": {
|
| 738 |
+
"should_epoch_stop": false,
|
| 739 |
+
"should_evaluate": false,
|
| 740 |
+
"should_log": false,
|
| 741 |
+
"should_save": true,
|
| 742 |
+
"should_training_stop": true
|
| 743 |
+
},
|
| 744 |
+
"attributes": {}
|
| 745 |
+
}
|
| 746 |
+
},
|
| 747 |
+
"total_flos": 0.0,
|
| 748 |
+
"train_batch_size": 12,
|
| 749 |
+
"trial_name": null,
|
| 750 |
+
"trial_params": null
|
| 751 |
+
}
|