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 +743 -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: Qwen2.5-1.5B-CCRL-2
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- grpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for Qwen2.5-1.5B-CCRL-2
|
| 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="chansung/Qwen2.5-1.5B-CCRL-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/chansung18/huggingface/runs/xstyel4m)
|
| 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
|
| 38 |
+
- Transformers: 4.50.0
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.4.1
|
| 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": 0.008462187110126251,
|
| 4 |
+
"train_runtime": 7668.932,
|
| 5 |
+
"train_samples": 949,
|
| 6 |
+
"train_samples_per_second": 0.835,
|
| 7 |
+
"train_steps_per_second": 0.007
|
| 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": "4.50.0"
|
| 14 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.008462187110126251,
|
| 4 |
+
"train_runtime": 7668.932,
|
| 5 |
+
"train_samples": 949,
|
| 6 |
+
"train_samples_per_second": 0.835,
|
| 7 |
+
"train_steps_per_second": 0.007
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,743 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.8403361344537815,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 50,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"clip_ratio": 0.0,
|
| 14 |
+
"completion_length": 641.9296875,
|
| 15 |
+
"epoch": 0.01680672268907563,
|
| 16 |
+
"grad_norm": 0.1829434037208557,
|
| 17 |
+
"kl": 0.0,
|
| 18 |
+
"learning_rate": 5e-07,
|
| 19 |
+
"loss": -0.0027,
|
| 20 |
+
"num_tokens": 143935.0,
|
| 21 |
+
"reward": 0.015625,
|
| 22 |
+
"reward_std": 0.0289318785071373,
|
| 23 |
+
"rewards/format_reward": 0.015625,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"clip_ratio": 0.0,
|
| 28 |
+
"completion_length": 558.2109375,
|
| 29 |
+
"epoch": 0.03361344537815126,
|
| 30 |
+
"grad_norm": 0.3770347237586975,
|
| 31 |
+
"kl": 0.0,
|
| 32 |
+
"learning_rate": 1e-06,
|
| 33 |
+
"loss": 0.0243,
|
| 34 |
+
"num_tokens": 275234.0,
|
| 35 |
+
"reward": 0.03125,
|
| 36 |
+
"reward_std": 0.0883883461356163,
|
| 37 |
+
"rewards/format_reward": 0.03125,
|
| 38 |
+
"step": 2
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"clip_ratio": 0.0,
|
| 42 |
+
"completion_length": 577.3203125,
|
| 43 |
+
"epoch": 0.05042016806722689,
|
| 44 |
+
"grad_norm": 0.2114293873310089,
|
| 45 |
+
"kl": 0.00027871131896972656,
|
| 46 |
+
"learning_rate": 9.990365154573716e-07,
|
| 47 |
+
"loss": 0.0109,
|
| 48 |
+
"num_tokens": 411211.0,
|
| 49 |
+
"reward": 0.015625,
|
| 50 |
+
"reward_std": 0.04419417306780815,
|
| 51 |
+
"rewards/format_reward": 0.015625,
|
| 52 |
+
"step": 3
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"clip_ratio": 0.0,
|
| 56 |
+
"completion_length": 551.625,
|
| 57 |
+
"epoch": 0.06722689075630252,
|
| 58 |
+
"grad_norm": 0.32700982689857483,
|
| 59 |
+
"kl": 0.00025844573974609375,
|
| 60 |
+
"learning_rate": 9.961501876182148e-07,
|
| 61 |
+
"loss": 0.0038,
|
| 62 |
+
"num_tokens": 541427.0,
|
| 63 |
+
"reward": 0.0234375,
|
| 64 |
+
"reward_std": 0.06629125960171223,
|
| 65 |
+
"rewards/format_reward": 0.0234375,
|
| 66 |
+
"step": 4
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"clip_ratio": 0.0,
|
| 70 |
+
"completion_length": 577.4296875,
|
| 71 |
+
"epoch": 0.08403361344537816,
|
| 72 |
+
"grad_norm": 0.2744513154029846,
|
| 73 |
+
"kl": 0.0002903938293457031,
|
| 74 |
+
"learning_rate": 9.913533761814537e-07,
|
| 75 |
+
"loss": 0.0185,
|
| 76 |
+
"num_tokens": 674322.0,
|
| 77 |
+
"reward": 0.0390625,
|
| 78 |
+
"reward_std": 0.0765409953892231,
|
| 79 |
+
"rewards/format_reward": 0.0390625,
|
| 80 |
+
"step": 5
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"clip_ratio": 0.0,
|
| 84 |
+
"completion_length": 582.046875,
|
| 85 |
+
"epoch": 0.10084033613445378,
|
| 86 |
+
"grad_norm": 0.2994139492511749,
|
| 87 |
+
"kl": 0.0002753734588623047,
|
| 88 |
+
"learning_rate": 9.846666218300807e-07,
|
| 89 |
+
"loss": 0.011,
|
| 90 |
+
"num_tokens": 808448.0,
|
| 91 |
+
"reward": 0.03125,
|
| 92 |
+
"reward_std": 0.07312605157494545,
|
| 93 |
+
"rewards/format_reward": 0.03125,
|
| 94 |
+
"step": 6
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"clip_ratio": 0.0,
|
| 98 |
+
"completion_length": 567.9296875,
|
| 99 |
+
"epoch": 0.11764705882352941,
|
| 100 |
+
"grad_norm": 0.2810344696044922,
|
| 101 |
+
"kl": 0.0003495216369628906,
|
| 102 |
+
"learning_rate": 9.761185582727976e-07,
|
| 103 |
+
"loss": 0.0199,
|
| 104 |
+
"num_tokens": 936975.0,
|
| 105 |
+
"reward": 0.0234375,
|
| 106 |
+
"reward_std": 0.051028965041041374,
|
| 107 |
+
"rewards/format_reward": 0.0234375,
|
| 108 |
+
"step": 7
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"clip_ratio": 0.0,
|
| 112 |
+
"completion_length": 565.2578125,
|
| 113 |
+
"epoch": 0.13445378151260504,
|
| 114 |
+
"grad_norm": 0.3799228370189667,
|
| 115 |
+
"kl": 0.00041484832763671875,
|
| 116 |
+
"learning_rate": 9.657457896300791e-07,
|
| 117 |
+
"loss": 0.022,
|
| 118 |
+
"num_tokens": 1071784.0,
|
| 119 |
+
"reward": 0.046875,
|
| 120 |
+
"reward_std": 0.1173202246427536,
|
| 121 |
+
"rewards/format_reward": 0.046875,
|
| 122 |
+
"step": 8
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"clip_ratio": 0.0,
|
| 126 |
+
"completion_length": 534.75,
|
| 127 |
+
"epoch": 0.15126050420168066,
|
| 128 |
+
"grad_norm": 0.42587605118751526,
|
| 129 |
+
"kl": 0.0007534027099609375,
|
| 130 |
+
"learning_rate": 9.535927336897097e-07,
|
| 131 |
+
"loss": 0.0392,
|
| 132 |
+
"num_tokens": 1194536.0,
|
| 133 |
+
"reward": 0.0625,
|
| 134 |
+
"reward_std": 0.14283225685358047,
|
| 135 |
+
"rewards/format_reward": 0.0625,
|
| 136 |
+
"step": 9
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"clip_ratio": 0.0,
|
| 140 |
+
"completion_length": 621.3828125,
|
| 141 |
+
"epoch": 0.16806722689075632,
|
| 142 |
+
"grad_norm": 0.41017991304397583,
|
| 143 |
+
"kl": 0.0006995201110839844,
|
| 144 |
+
"learning_rate": 9.397114317029974e-07,
|
| 145 |
+
"loss": -0.0271,
|
| 146 |
+
"num_tokens": 1336961.0,
|
| 147 |
+
"reward": 0.09375,
|
| 148 |
+
"reward_std": 0.13781969621777534,
|
| 149 |
+
"rewards/format_reward": 0.09375,
|
| 150 |
+
"step": 10
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"clip_ratio": 0.0,
|
| 154 |
+
"completion_length": 537.6796875,
|
| 155 |
+
"epoch": 0.18487394957983194,
|
| 156 |
+
"grad_norm": 0.6829501986503601,
|
| 157 |
+
"kl": 0.0015192031860351562,
|
| 158 |
+
"learning_rate": 9.241613255361454e-07,
|
| 159 |
+
"loss": 0.0224,
|
| 160 |
+
"num_tokens": 1462080.0,
|
| 161 |
+
"reward": 0.1953125,
|
| 162 |
+
"reward_std": 0.2961437702178955,
|
| 163 |
+
"rewards/format_reward": 0.1953125,
|
| 164 |
+
"step": 11
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"clip_ratio": 0.0,
|
| 168 |
+
"completion_length": 590.59375,
|
| 169 |
+
"epoch": 0.20168067226890757,
|
| 170 |
+
"grad_norm": 0.484163761138916,
|
| 171 |
+
"kl": 0.0014421939849853516,
|
| 172 |
+
"learning_rate": 9.070090031310558e-07,
|
| 173 |
+
"loss": 0.0102,
|
| 174 |
+
"num_tokens": 1600932.0,
|
| 175 |
+
"reward": 0.109375,
|
| 176 |
+
"reward_std": 0.12703318148851395,
|
| 177 |
+
"rewards/format_reward": 0.109375,
|
| 178 |
+
"step": 12
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"clip_ratio": 0.0,
|
| 182 |
+
"completion_length": 565.2109375,
|
| 183 |
+
"epoch": 0.2184873949579832,
|
| 184 |
+
"grad_norm": 0.5853496789932251,
|
| 185 |
+
"kl": 0.002685546875,
|
| 186 |
+
"learning_rate": 8.883279133655398e-07,
|
| 187 |
+
"loss": 0.0463,
|
| 188 |
+
"num_tokens": 1733479.0,
|
| 189 |
+
"reward": 0.203125,
|
| 190 |
+
"reward_std": 0.2335668858140707,
|
| 191 |
+
"rewards/format_reward": 0.203125,
|
| 192 |
+
"step": 13
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"clip_ratio": 0.0,
|
| 196 |
+
"completion_length": 489.96875,
|
| 197 |
+
"epoch": 0.23529411764705882,
|
| 198 |
+
"grad_norm": 0.7053346037864685,
|
| 199 |
+
"kl": 0.00472259521484375,
|
| 200 |
+
"learning_rate": 8.681980515339463e-07,
|
| 201 |
+
"loss": -0.0509,
|
| 202 |
+
"num_tokens": 1851611.0,
|
| 203 |
+
"reward": 0.34375,
|
| 204 |
+
"reward_std": 0.2740466818213463,
|
| 205 |
+
"rewards/format_reward": 0.34375,
|
| 206 |
+
"step": 14
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"clip_ratio": 0.0,
|
| 210 |
+
"completion_length": 555.671875,
|
| 211 |
+
"epoch": 0.25210084033613445,
|
| 212 |
+
"grad_norm": 0.6062747240066528,
|
| 213 |
+
"kl": 0.0043487548828125,
|
| 214 |
+
"learning_rate": 8.46705616795031e-07,
|
| 215 |
+
"loss": 0.0142,
|
| 216 |
+
"num_tokens": 1982657.0,
|
| 217 |
+
"reward": 0.4140625,
|
| 218 |
+
"reward_std": 0.30509532056748867,
|
| 219 |
+
"rewards/format_reward": 0.4140625,
|
| 220 |
+
"step": 15
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"clip_ratio": 0.0,
|
| 224 |
+
"completion_length": 567.4453125,
|
| 225 |
+
"epoch": 0.2689075630252101,
|
| 226 |
+
"grad_norm": 0.5138593912124634,
|
| 227 |
+
"kl": 0.004817962646484375,
|
| 228 |
+
"learning_rate": 8.239426430539243e-07,
|
| 229 |
+
"loss": -0.0185,
|
| 230 |
+
"num_tokens": 2112490.0,
|
| 231 |
+
"reward": 0.328125,
|
| 232 |
+
"reward_std": 0.2001592293381691,
|
| 233 |
+
"rewards/format_reward": 0.328125,
|
| 234 |
+
"step": 16
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"clip_ratio": 0.0,
|
| 238 |
+
"completion_length": 497.5859375,
|
| 239 |
+
"epoch": 0.2857142857142857,
|
| 240 |
+
"grad_norm": 0.5945853590965271,
|
| 241 |
+
"kl": 0.00463104248046875,
|
| 242 |
+
"learning_rate": 8.00006604858821e-07,
|
| 243 |
+
"loss": -0.0218,
|
| 244 |
+
"num_tokens": 2233365.0,
|
| 245 |
+
"reward": 0.40625,
|
| 246 |
+
"reward_std": 0.25224902853369713,
|
| 247 |
+
"rewards/format_reward": 0.40625,
|
| 248 |
+
"step": 17
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"clip_ratio": 0.0,
|
| 252 |
+
"completion_length": 670.125,
|
| 253 |
+
"epoch": 0.3025210084033613,
|
| 254 |
+
"grad_norm": 0.27369990944862366,
|
| 255 |
+
"kl": 0.000995635986328125,
|
| 256 |
+
"learning_rate": 7.75e-07,
|
| 257 |
+
"loss": 0.0205,
|
| 258 |
+
"num_tokens": 2383277.0,
|
| 259 |
+
"reward": 0.09375,
|
| 260 |
+
"reward_std": 0.05444391071796417,
|
| 261 |
+
"rewards/format_reward": 0.09375,
|
| 262 |
+
"step": 18
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"clip_ratio": 0.0,
|
| 266 |
+
"completion_length": 602.7265625,
|
| 267 |
+
"epoch": 0.31932773109243695,
|
| 268 |
+
"grad_norm": 0.5327508449554443,
|
| 269 |
+
"kl": 0.006078243255615234,
|
| 270 |
+
"learning_rate": 7.490299105985506e-07,
|
| 271 |
+
"loss": -0.0041,
|
| 272 |
+
"num_tokens": 2521058.0,
|
| 273 |
+
"reward": 0.296875,
|
| 274 |
+
"reward_std": 0.15650184080004692,
|
| 275 |
+
"rewards/format_reward": 0.296875,
|
| 276 |
+
"step": 19
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"clip_ratio": 0.0,
|
| 280 |
+
"completion_length": 473.875,
|
| 281 |
+
"epoch": 0.33613445378151263,
|
| 282 |
+
"grad_norm": 0.5724947452545166,
|
| 283 |
+
"kl": 0.008075714111328125,
|
| 284 |
+
"learning_rate": 7.222075445642904e-07,
|
| 285 |
+
"loss": 0.0444,
|
| 286 |
+
"num_tokens": 2641082.0,
|
| 287 |
+
"reward": 0.4453125,
|
| 288 |
+
"reward_std": 0.2212003916501999,
|
| 289 |
+
"rewards/format_reward": 0.4453125,
|
| 290 |
+
"step": 20
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"clip_ratio": 0.0,
|
| 294 |
+
"completion_length": 660.5703125,
|
| 295 |
+
"epoch": 0.35294117647058826,
|
| 296 |
+
"grad_norm": 0.3499065935611725,
|
| 297 |
+
"kl": 0.0032248497009277344,
|
| 298 |
+
"learning_rate": 6.946477593864227e-07,
|
| 299 |
+
"loss": -0.0102,
|
| 300 |
+
"num_tokens": 2788211.0,
|
| 301 |
+
"reward": 0.203125,
|
| 302 |
+
"reward_std": 0.08337578736245632,
|
| 303 |
+
"rewards/format_reward": 0.203125,
|
| 304 |
+
"step": 21
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"clip_ratio": 0.0,
|
| 308 |
+
"completion_length": 566.859375,
|
| 309 |
+
"epoch": 0.3697478991596639,
|
| 310 |
+
"grad_norm": 0.2869742810726166,
|
| 311 |
+
"kl": 0.00702667236328125,
|
| 312 |
+
"learning_rate": 6.664685702961344e-07,
|
| 313 |
+
"loss": 0.0002,
|
| 314 |
+
"num_tokens": 2917449.0,
|
| 315 |
+
"reward": 0.4296875,
|
| 316 |
+
"reward_std": 0.022097086533904076,
|
| 317 |
+
"rewards/format_reward": 0.4296875,
|
| 318 |
+
"step": 22
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"clip_ratio": 0.0,
|
| 322 |
+
"completion_length": 603.03125,
|
| 323 |
+
"epoch": 0.3865546218487395,
|
| 324 |
+
"grad_norm": 0.22740638256072998,
|
| 325 |
+
"kl": 0.005024433135986328,
|
| 326 |
+
"learning_rate": 6.377906449072577e-07,
|
| 327 |
+
"loss": -0.0122,
|
| 328 |
+
"num_tokens": 3056589.0,
|
| 329 |
+
"reward": 0.28125,
|
| 330 |
+
"reward_std": 0.05444390885531902,
|
| 331 |
+
"rewards/format_reward": 0.28125,
|
| 332 |
+
"step": 23
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"clip_ratio": 0.0,
|
| 336 |
+
"completion_length": 525.0234375,
|
| 337 |
+
"epoch": 0.40336134453781514,
|
| 338 |
+
"grad_norm": 0.486358642578125,
|
| 339 |
+
"kl": 0.005306243896484375,
|
| 340 |
+
"learning_rate": 6.087367864990232e-07,
|
| 341 |
+
"loss": 0.0533,
|
| 342 |
+
"num_tokens": 3187176.0,
|
| 343 |
+
"reward": 0.3828125,
|
| 344 |
+
"reward_std": 0.1207351740449667,
|
| 345 |
+
"rewards/format_reward": 0.3828125,
|
| 346 |
+
"step": 24
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"clip_ratio": 0.0,
|
| 350 |
+
"completion_length": 540.1953125,
|
| 351 |
+
"epoch": 0.42016806722689076,
|
| 352 |
+
"grad_norm": 0.3638220429420471,
|
| 353 |
+
"kl": 0.007305145263671875,
|
| 354 |
+
"learning_rate": 5.794314081535644e-07,
|
| 355 |
+
"loss": -0.0038,
|
| 356 |
+
"num_tokens": 3314737.0,
|
| 357 |
+
"reward": 0.4140625,
|
| 358 |
+
"reward_std": 0.09522314183413982,
|
| 359 |
+
"rewards/format_reward": 0.4140625,
|
| 360 |
+
"step": 25
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"clip_ratio": 0.0,
|
| 364 |
+
"completion_length": 497.3125,
|
| 365 |
+
"epoch": 0.4369747899159664,
|
| 366 |
+
"grad_norm": 0.5379955768585205,
|
| 367 |
+
"kl": 0.014757633209228516,
|
| 368 |
+
"learning_rate": 5.5e-07,
|
| 369 |
+
"loss": -0.0513,
|
| 370 |
+
"num_tokens": 3436265.0,
|
| 371 |
+
"reward": 0.6015625,
|
| 372 |
+
"reward_std": 0.19728107377886772,
|
| 373 |
+
"rewards/format_reward": 0.6015625,
|
| 374 |
+
"step": 26
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"clip_ratio": 0.0,
|
| 378 |
+
"completion_length": 627.1875,
|
| 379 |
+
"epoch": 0.453781512605042,
|
| 380 |
+
"grad_norm": 0.26467689871788025,
|
| 381 |
+
"kl": 0.005634307861328125,
|
| 382 |
+
"learning_rate": 5.205685918464355e-07,
|
| 383 |
+
"loss": -0.0003,
|
| 384 |
+
"num_tokens": 3577265.0,
|
| 385 |
+
"reward": 0.3515625,
|
| 386 |
+
"reward_std": 0.05102896690368652,
|
| 387 |
+
"rewards/format_reward": 0.3515625,
|
| 388 |
+
"step": 27
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"clip_ratio": 0.0,
|
| 392 |
+
"completion_length": 543.90625,
|
| 393 |
+
"epoch": 0.47058823529411764,
|
| 394 |
+
"grad_norm": 0.3865010142326355,
|
| 395 |
+
"kl": 0.0108489990234375,
|
| 396 |
+
"learning_rate": 4.912632135009769e-07,
|
| 397 |
+
"loss": 0.0253,
|
| 398 |
+
"num_tokens": 3703797.0,
|
| 399 |
+
"reward": 0.5390625,
|
| 400 |
+
"reward_std": 0.05102896690368652,
|
| 401 |
+
"rewards/format_reward": 0.5390625,
|
| 402 |
+
"step": 28
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"clip_ratio": 0.0,
|
| 406 |
+
"completion_length": 668.0078125,
|
| 407 |
+
"epoch": 0.48739495798319327,
|
| 408 |
+
"grad_norm": 0.29972589015960693,
|
| 409 |
+
"kl": 0.0080108642578125,
|
| 410 |
+
"learning_rate": 4.6220935509274227e-07,
|
| 411 |
+
"loss": 0.0094,
|
| 412 |
+
"num_tokens": 3850094.0,
|
| 413 |
+
"reward": 0.3515625,
|
| 414 |
+
"reward_std": 0.051028965041041374,
|
| 415 |
+
"rewards/format_reward": 0.3515625,
|
| 416 |
+
"step": 29
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"clip_ratio": 0.0,
|
| 420 |
+
"completion_length": 607.1875,
|
| 421 |
+
"epoch": 0.5042016806722689,
|
| 422 |
+
"grad_norm": 0.32297083735466003,
|
| 423 |
+
"kl": 0.007590293884277344,
|
| 424 |
+
"learning_rate": 4.3353142970386557e-07,
|
| 425 |
+
"loss": 0.0134,
|
| 426 |
+
"num_tokens": 3991174.0,
|
| 427 |
+
"reward": 0.4140625,
|
| 428 |
+
"reward_std": 0.06629125960171223,
|
| 429 |
+
"rewards/format_reward": 0.4140625,
|
| 430 |
+
"step": 30
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"clip_ratio": 0.0,
|
| 434 |
+
"completion_length": 599.109375,
|
| 435 |
+
"epoch": 0.5210084033613446,
|
| 436 |
+
"grad_norm": 0.004667076747864485,
|
| 437 |
+
"kl": 0.005215167999267578,
|
| 438 |
+
"learning_rate": 4.053522406135774e-07,
|
| 439 |
+
"loss": 0.0001,
|
| 440 |
+
"num_tokens": 4129020.0,
|
| 441 |
+
"reward": 0.1875,
|
| 442 |
+
"reward_std": 0.0,
|
| 443 |
+
"rewards/format_reward": 0.1875,
|
| 444 |
+
"step": 31
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"clip_ratio": 0.0,
|
| 448 |
+
"completion_length": 553.5546875,
|
| 449 |
+
"epoch": 0.5378151260504201,
|
| 450 |
+
"grad_norm": 0.23914550244808197,
|
| 451 |
+
"kl": 0.006618499755859375,
|
| 452 |
+
"learning_rate": 3.777924554357096e-07,
|
| 453 |
+
"loss": 0.0173,
|
| 454 |
+
"num_tokens": 4263587.0,
|
| 455 |
+
"reward": 0.4296875,
|
| 456 |
+
"reward_std": 0.022097086533904076,
|
| 457 |
+
"rewards/format_reward": 0.4296875,
|
| 458 |
+
"step": 32
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"clip_ratio": 0.0,
|
| 462 |
+
"completion_length": 614.109375,
|
| 463 |
+
"epoch": 0.5546218487394958,
|
| 464 |
+
"grad_norm": 0.38585954904556274,
|
| 465 |
+
"kl": 0.00874185562133789,
|
| 466 |
+
"learning_rate": 3.5097008940144955e-07,
|
| 467 |
+
"loss": 0.0186,
|
| 468 |
+
"num_tokens": 4404113.0,
|
| 469 |
+
"reward": 0.359375,
|
| 470 |
+
"reward_std": 0.04419417306780815,
|
| 471 |
+
"rewards/format_reward": 0.359375,
|
| 472 |
+
"step": 33
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"clip_ratio": 0.0,
|
| 476 |
+
"completion_length": 503.859375,
|
| 477 |
+
"epoch": 0.5714285714285714,
|
| 478 |
+
"grad_norm": 0.2774820923805237,
|
| 479 |
+
"kl": 0.0081939697265625,
|
| 480 |
+
"learning_rate": 3.250000000000001e-07,
|
| 481 |
+
"loss": 0.009,
|
| 482 |
+
"num_tokens": 4530399.0,
|
| 483 |
+
"reward": 0.421875,
|
| 484 |
+
"reward_std": 0.04419417306780815,
|
| 485 |
+
"rewards/format_reward": 0.421875,
|
| 486 |
+
"step": 34
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"clip_ratio": 0.0,
|
| 490 |
+
"completion_length": 510.8046875,
|
| 491 |
+
"epoch": 0.5882352941176471,
|
| 492 |
+
"grad_norm": 0.3793410360813141,
|
| 493 |
+
"kl": 0.007872581481933594,
|
| 494 |
+
"learning_rate": 2.999933951411791e-07,
|
| 495 |
+
"loss": 0.0096,
|
| 496 |
+
"num_tokens": 4656958.0,
|
| 497 |
+
"reward": 0.359375,
|
| 498 |
+
"reward_std": 0.04419417306780815,
|
| 499 |
+
"rewards/format_reward": 0.359375,
|
| 500 |
+
"step": 35
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"clip_ratio": 0.0,
|
| 504 |
+
"completion_length": 554.125,
|
| 505 |
+
"epoch": 0.6050420168067226,
|
| 506 |
+
"grad_norm": 0.23132917284965515,
|
| 507 |
+
"kl": 0.0100250244140625,
|
| 508 |
+
"learning_rate": 2.760573569460757e-07,
|
| 509 |
+
"loss": -0.0074,
|
| 510 |
+
"num_tokens": 4788182.0,
|
| 511 |
+
"reward": 0.484375,
|
| 512 |
+
"reward_std": 0.04419417306780815,
|
| 513 |
+
"rewards/format_reward": 0.484375,
|
| 514 |
+
"step": 36
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"clip_ratio": 0.0,
|
| 518 |
+
"completion_length": 503.140625,
|
| 519 |
+
"epoch": 0.6218487394957983,
|
| 520 |
+
"grad_norm": 0.41905921697616577,
|
| 521 |
+
"kl": 0.0104522705078125,
|
| 522 |
+
"learning_rate": 2.53294383204969e-07,
|
| 523 |
+
"loss": 0.004,
|
| 524 |
+
"num_tokens": 4910232.0,
|
| 525 |
+
"reward": 0.40625,
|
| 526 |
+
"reward_std": 0.0883883461356163,
|
| 527 |
+
"rewards/format_reward": 0.40625,
|
| 528 |
+
"step": 37
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"clip_ratio": 0.0,
|
| 532 |
+
"completion_length": 488.4453125,
|
| 533 |
+
"epoch": 0.6386554621848739,
|
| 534 |
+
"grad_norm": 0.009937948547303677,
|
| 535 |
+
"kl": 0.0146942138671875,
|
| 536 |
+
"learning_rate": 2.3180194846605364e-07,
|
| 537 |
+
"loss": 0.0001,
|
| 538 |
+
"num_tokens": 5028785.0,
|
| 539 |
+
"reward": 0.5625,
|
| 540 |
+
"reward_std": 0.0,
|
| 541 |
+
"rewards/format_reward": 0.5625,
|
| 542 |
+
"step": 38
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"clip_ratio": 0.0,
|
| 546 |
+
"completion_length": 535.7578125,
|
| 547 |
+
"epoch": 0.6554621848739496,
|
| 548 |
+
"grad_norm": 0.40260764956474304,
|
| 549 |
+
"kl": 0.00785064697265625,
|
| 550 |
+
"learning_rate": 2.1167208663446022e-07,
|
| 551 |
+
"loss": -0.0086,
|
| 552 |
+
"num_tokens": 5156906.0,
|
| 553 |
+
"reward": 0.3515625,
|
| 554 |
+
"reward_std": 0.06629125960171223,
|
| 555 |
+
"rewards/format_reward": 0.3515625,
|
| 556 |
+
"step": 39
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"clip_ratio": 0.0,
|
| 560 |
+
"completion_length": 584.6796875,
|
| 561 |
+
"epoch": 0.6722689075630253,
|
| 562 |
+
"grad_norm": 0.3577350676059723,
|
| 563 |
+
"kl": 0.008663177490234375,
|
| 564 |
+
"learning_rate": 1.929909968689442e-07,
|
| 565 |
+
"loss": 0.0166,
|
| 566 |
+
"num_tokens": 5292793.0,
|
| 567 |
+
"reward": 0.3125,
|
| 568 |
+
"reward_std": 0.0731260534375906,
|
| 569 |
+
"rewards/format_reward": 0.3125,
|
| 570 |
+
"step": 40
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"clip_ratio": 0.0,
|
| 574 |
+
"completion_length": 494.53125,
|
| 575 |
+
"epoch": 0.6890756302521008,
|
| 576 |
+
"grad_norm": 0.5300049781799316,
|
| 577 |
+
"kl": 0.016998291015625,
|
| 578 |
+
"learning_rate": 1.758386744638546e-07,
|
| 579 |
+
"loss": -0.0005,
|
| 580 |
+
"num_tokens": 5413677.0,
|
| 581 |
+
"reward": 0.5859375,
|
| 582 |
+
"reward_std": 0.11048543266952038,
|
| 583 |
+
"rewards/format_reward": 0.5859375,
|
| 584 |
+
"step": 41
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"clip_ratio": 0.0,
|
| 588 |
+
"completion_length": 454.0859375,
|
| 589 |
+
"epoch": 0.7058823529411765,
|
| 590 |
+
"grad_norm": 0.6002023220062256,
|
| 591 |
+
"kl": 0.012908935546875,
|
| 592 |
+
"learning_rate": 1.6028856829700258e-07,
|
| 593 |
+
"loss": 0.0249,
|
| 594 |
+
"num_tokens": 5530672.0,
|
| 595 |
+
"reward": 0.578125,
|
| 596 |
+
"reward_std": 0.1173202209174633,
|
| 597 |
+
"rewards/format_reward": 0.578125,
|
| 598 |
+
"step": 42
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"clip_ratio": 0.0,
|
| 602 |
+
"completion_length": 536.7890625,
|
| 603 |
+
"epoch": 0.7226890756302521,
|
| 604 |
+
"grad_norm": 0.40132713317871094,
|
| 605 |
+
"kl": 0.01432037353515625,
|
| 606 |
+
"learning_rate": 1.464072663102903e-07,
|
| 607 |
+
"loss": 0.0289,
|
| 608 |
+
"num_tokens": 5657981.0,
|
| 609 |
+
"reward": 0.4765625,
|
| 610 |
+
"reward_std": 0.06629125960171223,
|
| 611 |
+
"rewards/format_reward": 0.4765625,
|
| 612 |
+
"step": 43
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"clip_ratio": 0.0,
|
| 616 |
+
"completion_length": 477.53125,
|
| 617 |
+
"epoch": 0.7394957983193278,
|
| 618 |
+
"grad_norm": 0.5269466042518616,
|
| 619 |
+
"kl": 0.011138916015625,
|
| 620 |
+
"learning_rate": 1.3425421036992097e-07,
|
| 621 |
+
"loss": 0.0338,
|
| 622 |
+
"num_tokens": 5782097.0,
|
| 623 |
+
"reward": 0.5859375,
|
| 624 |
+
"reward_std": 0.09522313810884953,
|
| 625 |
+
"rewards/format_reward": 0.5859375,
|
| 626 |
+
"step": 44
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"clip_ratio": 0.0,
|
| 630 |
+
"completion_length": 583.5546875,
|
| 631 |
+
"epoch": 0.7563025210084033,
|
| 632 |
+
"grad_norm": 0.30939722061157227,
|
| 633 |
+
"kl": 0.00948333740234375,
|
| 634 |
+
"learning_rate": 1.238814417272025e-07,
|
| 635 |
+
"loss": 0.0011,
|
| 636 |
+
"num_tokens": 5915848.0,
|
| 637 |
+
"reward": 0.484375,
|
| 638 |
+
"reward_std": 0.04419417306780815,
|
| 639 |
+
"rewards/format_reward": 0.484375,
|
| 640 |
+
"step": 45
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"clip_ratio": 0.0,
|
| 644 |
+
"completion_length": 493.0,
|
| 645 |
+
"epoch": 0.773109243697479,
|
| 646 |
+
"grad_norm": 0.36379459500312805,
|
| 647 |
+
"kl": 0.0174407958984375,
|
| 648 |
+
"learning_rate": 1.1533337816991931e-07,
|
| 649 |
+
"loss": 0.0097,
|
| 650 |
+
"num_tokens": 6035144.0,
|
| 651 |
+
"reward": 0.609375,
|
| 652 |
+
"reward_std": 0.04419417306780815,
|
| 653 |
+
"rewards/format_reward": 0.609375,
|
| 654 |
+
"step": 46
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"clip_ratio": 0.0,
|
| 658 |
+
"completion_length": 586.8125,
|
| 659 |
+
"epoch": 0.7899159663865546,
|
| 660 |
+
"grad_norm": 0.3710026741027832,
|
| 661 |
+
"kl": 0.00640106201171875,
|
| 662 |
+
"learning_rate": 1.086466238185463e-07,
|
| 663 |
+
"loss": 0.0173,
|
| 664 |
+
"num_tokens": 6172192.0,
|
| 665 |
+
"reward": 0.359375,
|
| 666 |
+
"reward_std": 0.04419417306780815,
|
| 667 |
+
"rewards/format_reward": 0.359375,
|
| 668 |
+
"step": 47
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"clip_ratio": 0.0,
|
| 672 |
+
"completion_length": 575.15625,
|
| 673 |
+
"epoch": 0.8067226890756303,
|
| 674 |
+
"grad_norm": 0.010671989060938358,
|
| 675 |
+
"kl": 0.00979757308959961,
|
| 676 |
+
"learning_rate": 1.0384981238178533e-07,
|
| 677 |
+
"loss": 0.0001,
|
| 678 |
+
"num_tokens": 6307060.0,
|
| 679 |
+
"reward": 0.4375,
|
| 680 |
+
"reward_std": 0.0,
|
| 681 |
+
"rewards/format_reward": 0.4375,
|
| 682 |
+
"step": 48
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"clip_ratio": 0.0,
|
| 686 |
+
"completion_length": 570.890625,
|
| 687 |
+
"epoch": 0.8235294117647058,
|
| 688 |
+
"grad_norm": 0.40951627492904663,
|
| 689 |
+
"kl": 0.011271476745605469,
|
| 690 |
+
"learning_rate": 1.0096348454262843e-07,
|
| 691 |
+
"loss": 0.0422,
|
| 692 |
+
"num_tokens": 6441918.0,
|
| 693 |
+
"reward": 0.421875,
|
| 694 |
+
"reward_std": 0.04419417306780815,
|
| 695 |
+
"rewards/format_reward": 0.421875,
|
| 696 |
+
"step": 49
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"clip_ratio": 0.0,
|
| 700 |
+
"completion_length": 576.8515625,
|
| 701 |
+
"epoch": 0.8403361344537815,
|
| 702 |
+
"grad_norm": 0.013534413650631905,
|
| 703 |
+
"kl": 0.0068531036376953125,
|
| 704 |
+
"learning_rate": 1e-07,
|
| 705 |
+
"loss": 0.0001,
|
| 706 |
+
"num_tokens": 6575251.0,
|
| 707 |
+
"reward": 0.3125,
|
| 708 |
+
"reward_std": 0.0,
|
| 709 |
+
"rewards/format_reward": 0.3125,
|
| 710 |
+
"step": 50
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.8403361344537815,
|
| 714 |
+
"step": 50,
|
| 715 |
+
"total_flos": 0.0,
|
| 716 |
+
"train_loss": 0.008462187110126251,
|
| 717 |
+
"train_runtime": 7668.932,
|
| 718 |
+
"train_samples_per_second": 0.835,
|
| 719 |
+
"train_steps_per_second": 0.007
|
| 720 |
+
}
|
| 721 |
+
],
|
| 722 |
+
"logging_steps": 1,
|
| 723 |
+
"max_steps": 50,
|
| 724 |
+
"num_input_tokens_seen": 0,
|
| 725 |
+
"num_train_epochs": 1,
|
| 726 |
+
"save_steps": 50,
|
| 727 |
+
"stateful_callbacks": {
|
| 728 |
+
"TrainerControl": {
|
| 729 |
+
"args": {
|
| 730 |
+
"should_epoch_stop": false,
|
| 731 |
+
"should_evaluate": false,
|
| 732 |
+
"should_log": false,
|
| 733 |
+
"should_save": true,
|
| 734 |
+
"should_training_stop": true
|
| 735 |
+
},
|
| 736 |
+
"attributes": {}
|
| 737 |
+
}
|
| 738 |
+
},
|
| 739 |
+
"total_flos": 0.0,
|
| 740 |
+
"train_batch_size": 8,
|
| 741 |
+
"trial_name": null,
|
| 742 |
+
"trial_params": null
|
| 743 |
+
}
|