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 +2787 -0
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: Qwen/Qwen2.5-3B
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: Qwen2.5-3B-Open-R1-GRPO-math-selected-cosine-v2
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- grpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for Qwen2.5-3B-Open-R1-GRPO-math-selected-cosine-v2
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [Qwen/Qwen2.5-3B](https://huggingface.co/Qwen/Qwen2.5-3B).
|
| 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="Lansechen/Qwen2.5-3B-Open-R1-GRPO-math-selected-cosine-v2", 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/chenran1995-the-chinese-university-of-hong-kong/huggingface/runs/3knvoabh)
|
| 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+cu121
|
| 40 |
+
- Datasets: 3.5.0
|
| 41 |
+
- Tokenizers: 0.21.1
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite GRPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@article{zhihong2024deepseekmath,
|
| 49 |
+
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
|
| 50 |
+
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
|
| 51 |
+
year = 2024,
|
| 52 |
+
eprint = {arXiv:2402.03300},
|
| 53 |
+
}
|
| 54 |
+
|
| 55 |
+
```
|
| 56 |
+
|
| 57 |
+
Cite TRL as:
|
| 58 |
+
|
| 59 |
+
```bibtex
|
| 60 |
+
@misc{vonwerra2022trl,
|
| 61 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 62 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
|
| 63 |
+
year = 2020,
|
| 64 |
+
journal = {GitHub repository},
|
| 65 |
+
publisher = {GitHub},
|
| 66 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 67 |
+
}
|
| 68 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": -0.009652293185873921,
|
| 4 |
+
"train_runtime": 33755.1654,
|
| 5 |
+
"train_samples": 11040,
|
| 6 |
+
"train_samples_per_second": 0.654,
|
| 7 |
+
"train_steps_per_second": 0.006
|
| 8 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"eos_token_id": 151643,
|
| 4 |
+
"max_new_tokens": 2048,
|
| 5 |
+
"transformers_version": "4.50.0"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": -0.009652293185873921,
|
| 4 |
+
"train_runtime": 33755.1654,
|
| 5 |
+
"train_samples": 11040,
|
| 6 |
+
"train_samples_per_second": 0.654,
|
| 7 |
+
"train_steps_per_second": 0.006
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2787 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.9936628643852978,
|
| 6 |
+
"eval_steps": 100,
|
| 7 |
+
"global_step": 196,
|
| 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": 544.7745819091797,
|
| 15 |
+
"epoch": 0.010139416983523447,
|
| 16 |
+
"grad_norm": 0.2070285975933075,
|
| 17 |
+
"learning_rate": 5e-08,
|
| 18 |
+
"loss": -0.1577,
|
| 19 |
+
"num_tokens": 636118.0,
|
| 20 |
+
"reward": -0.35065557062625885,
|
| 21 |
+
"reward_std": 0.656378798186779,
|
| 22 |
+
"rewards/cosine_scaled_reward": -0.35400376841425896,
|
| 23 |
+
"rewards/format_reward": 0.003348214435391128,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"clip_ratio": 0.0,
|
| 28 |
+
"completion_length": 549.2812843322754,
|
| 29 |
+
"epoch": 0.020278833967046894,
|
| 30 |
+
"grad_norm": 0.2276979684829712,
|
| 31 |
+
"learning_rate": 1e-07,
|
| 32 |
+
"loss": -0.1565,
|
| 33 |
+
"num_tokens": 1263330.0,
|
| 34 |
+
"reward": -0.3044062051922083,
|
| 35 |
+
"reward_std": 0.6904742792248726,
|
| 36 |
+
"rewards/cosine_scaled_reward": -0.31221868470311165,
|
| 37 |
+
"rewards/format_reward": 0.007812500349245965,
|
| 38 |
+
"step": 2
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"clip_ratio": 0.0,
|
| 42 |
+
"completion_length": 588.2891006469727,
|
| 43 |
+
"epoch": 0.030418250950570342,
|
| 44 |
+
"grad_norm": 0.19815148413181305,
|
| 45 |
+
"learning_rate": 1.5e-07,
|
| 46 |
+
"loss": -0.1123,
|
| 47 |
+
"num_tokens": 1930557.0,
|
| 48 |
+
"reward": -0.2540861386805773,
|
| 49 |
+
"reward_std": 0.6807773411273956,
|
| 50 |
+
"rewards/cosine_scaled_reward": -0.2618986228480935,
|
| 51 |
+
"rewards/format_reward": 0.007812500232830644,
|
| 52 |
+
"step": 3
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"clip_ratio": 0.0,
|
| 56 |
+
"completion_length": 552.0669898986816,
|
| 57 |
+
"epoch": 0.04055766793409379,
|
| 58 |
+
"grad_norm": 0.20235052704811096,
|
| 59 |
+
"learning_rate": 2e-07,
|
| 60 |
+
"loss": -0.1652,
|
| 61 |
+
"num_tokens": 2562161.0,
|
| 62 |
+
"reward": -0.3022355027496815,
|
| 63 |
+
"reward_std": 0.6755559742450714,
|
| 64 |
+
"rewards/cosine_scaled_reward": -0.3122801296412945,
|
| 65 |
+
"rewards/format_reward": 0.010044643189758062,
|
| 66 |
+
"step": 4
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"clip_ratio": 0.0,
|
| 70 |
+
"completion_length": 571.6172142028809,
|
| 71 |
+
"epoch": 0.050697084917617236,
|
| 72 |
+
"grad_norm": 0.19000709056854248,
|
| 73 |
+
"learning_rate": 2.5e-07,
|
| 74 |
+
"loss": -0.1267,
|
| 75 |
+
"num_tokens": 3201114.0,
|
| 76 |
+
"reward": -0.2917697671800852,
|
| 77 |
+
"reward_std": 0.6964326798915863,
|
| 78 |
+
"rewards/cosine_scaled_reward": -0.2984661813825369,
|
| 79 |
+
"rewards/format_reward": 0.006696428870782256,
|
| 80 |
+
"step": 5
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"clip_ratio": 0.0,
|
| 84 |
+
"completion_length": 561.6384201049805,
|
| 85 |
+
"epoch": 0.060836501901140684,
|
| 86 |
+
"grad_norm": 0.21064706146717072,
|
| 87 |
+
"learning_rate": 3e-07,
|
| 88 |
+
"loss": -0.1771,
|
| 89 |
+
"num_tokens": 3833534.0,
|
| 90 |
+
"reward": -0.28072800021618605,
|
| 91 |
+
"reward_std": 0.6883805021643639,
|
| 92 |
+
"rewards/cosine_scaled_reward": -0.2840762068517506,
|
| 93 |
+
"rewards/format_reward": 0.003348214435391128,
|
| 94 |
+
"step": 6
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"clip_ratio": 0.0,
|
| 98 |
+
"completion_length": 542.1149749755859,
|
| 99 |
+
"epoch": 0.07097591888466413,
|
| 100 |
+
"grad_norm": 0.22860734164714813,
|
| 101 |
+
"learning_rate": 3.5e-07,
|
| 102 |
+
"loss": -0.131,
|
| 103 |
+
"num_tokens": 4443693.0,
|
| 104 |
+
"reward": -0.23171719023957849,
|
| 105 |
+
"reward_std": 0.7431022375822067,
|
| 106 |
+
"rewards/cosine_scaled_reward": -0.24287789640948176,
|
| 107 |
+
"rewards/format_reward": 0.011160714668221772,
|
| 108 |
+
"step": 7
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"clip_ratio": 0.0,
|
| 112 |
+
"completion_length": 563.7890892028809,
|
| 113 |
+
"epoch": 0.08111533586818757,
|
| 114 |
+
"grad_norm": 0.19847826659679413,
|
| 115 |
+
"learning_rate": 4e-07,
|
| 116 |
+
"loss": -0.142,
|
| 117 |
+
"num_tokens": 5076400.0,
|
| 118 |
+
"reward": -0.3053848221898079,
|
| 119 |
+
"reward_std": 0.6786770969629288,
|
| 120 |
+
"rewards/cosine_scaled_reward": -0.3109651692211628,
|
| 121 |
+
"rewards/format_reward": 0.005580357392318547,
|
| 122 |
+
"step": 8
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"clip_ratio": 0.0,
|
| 126 |
+
"completion_length": 585.7589569091797,
|
| 127 |
+
"epoch": 0.09125475285171103,
|
| 128 |
+
"grad_norm": 0.1838308870792389,
|
| 129 |
+
"learning_rate": 4.5e-07,
|
| 130 |
+
"loss": -0.1486,
|
| 131 |
+
"num_tokens": 5724568.0,
|
| 132 |
+
"reward": -0.26803811825811863,
|
| 133 |
+
"reward_std": 0.6804082244634628,
|
| 134 |
+
"rewards/cosine_scaled_reward": -0.26915417425334454,
|
| 135 |
+
"rewards/format_reward": 0.0011160714784637094,
|
| 136 |
+
"step": 9
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"clip_ratio": 0.0,
|
| 140 |
+
"completion_length": 516.0636291503906,
|
| 141 |
+
"epoch": 0.10139416983523447,
|
| 142 |
+
"grad_norm": 0.6098268032073975,
|
| 143 |
+
"learning_rate": 5e-07,
|
| 144 |
+
"loss": -0.1271,
|
| 145 |
+
"num_tokens": 6313689.0,
|
| 146 |
+
"reward": -0.2813316825777292,
|
| 147 |
+
"reward_std": 0.7330739945173264,
|
| 148 |
+
"rewards/cosine_scaled_reward": -0.2880280986428261,
|
| 149 |
+
"rewards/format_reward": 0.006696428870782256,
|
| 150 |
+
"step": 10
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"clip_ratio": 0.0,
|
| 154 |
+
"completion_length": 568.8984718322754,
|
| 155 |
+
"epoch": 0.11153358681875793,
|
| 156 |
+
"grad_norm": 0.5140683054924011,
|
| 157 |
+
"learning_rate": 5.5e-07,
|
| 158 |
+
"loss": -0.1441,
|
| 159 |
+
"num_tokens": 6960958.0,
|
| 160 |
+
"reward": -0.2723624687641859,
|
| 161 |
+
"reward_std": 0.6729818060994148,
|
| 162 |
+
"rewards/cosine_scaled_reward": -0.275710666552186,
|
| 163 |
+
"rewards/format_reward": 0.003348214435391128,
|
| 164 |
+
"step": 11
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"clip_ratio": 0.0,
|
| 168 |
+
"completion_length": 584.2366409301758,
|
| 169 |
+
"epoch": 0.12167300380228137,
|
| 170 |
+
"grad_norm": 0.20556923747062683,
|
| 171 |
+
"learning_rate": 6e-07,
|
| 172 |
+
"loss": -0.1173,
|
| 173 |
+
"num_tokens": 7615746.0,
|
| 174 |
+
"reward": -0.21572664007544518,
|
| 175 |
+
"reward_std": 0.7041841298341751,
|
| 176 |
+
"rewards/cosine_scaled_reward": -0.2201909152790904,
|
| 177 |
+
"rewards/format_reward": 0.004464285913854837,
|
| 178 |
+
"step": 12
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"clip_ratio": 0.0,
|
| 182 |
+
"completion_length": 598.1797065734863,
|
| 183 |
+
"epoch": 0.13181242078580482,
|
| 184 |
+
"grad_norm": 0.19161486625671387,
|
| 185 |
+
"learning_rate": 6.5e-07,
|
| 186 |
+
"loss": -0.1216,
|
| 187 |
+
"num_tokens": 8279899.0,
|
| 188 |
+
"reward": -0.23263627663254738,
|
| 189 |
+
"reward_std": 0.7194378077983856,
|
| 190 |
+
"rewards/cosine_scaled_reward": -0.23933268897235394,
|
| 191 |
+
"rewards/format_reward": 0.006696428870782256,
|
| 192 |
+
"step": 13
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"clip_ratio": 0.0,
|
| 196 |
+
"completion_length": 539.4096183776855,
|
| 197 |
+
"epoch": 0.14195183776932827,
|
| 198 |
+
"grad_norm": 0.20079070329666138,
|
| 199 |
+
"learning_rate": 7e-07,
|
| 200 |
+
"loss": -0.1091,
|
| 201 |
+
"num_tokens": 8895386.0,
|
| 202 |
+
"reward": -0.21676091896370053,
|
| 203 |
+
"reward_std": 0.7302467301487923,
|
| 204 |
+
"rewards/cosine_scaled_reward": -0.22209045942872763,
|
| 205 |
+
"rewards/format_reward": 0.0011160714784637094,
|
| 206 |
+
"step": 14
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"clip_ratio": 0.0,
|
| 210 |
+
"completion_length": 581.7645378112793,
|
| 211 |
+
"epoch": 0.1520912547528517,
|
| 212 |
+
"grad_norm": 0.21947209537029266,
|
| 213 |
+
"learning_rate": 7.5e-07,
|
| 214 |
+
"loss": -0.097,
|
| 215 |
+
"num_tokens": 9545087.0,
|
| 216 |
+
"reward": -0.10511145042255521,
|
| 217 |
+
"reward_std": 0.7483287081122398,
|
| 218 |
+
"rewards/cosine_scaled_reward": -0.10957573121413589,
|
| 219 |
+
"rewards/format_reward": 0.004464285797439516,
|
| 220 |
+
"step": 15
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"clip_ratio": 0.0,
|
| 224 |
+
"completion_length": 587.3013725280762,
|
| 225 |
+
"epoch": 0.16223067173637515,
|
| 226 |
+
"grad_norm": 0.20107129216194153,
|
| 227 |
+
"learning_rate": 8e-07,
|
| 228 |
+
"loss": -0.0894,
|
| 229 |
+
"num_tokens": 10198237.0,
|
| 230 |
+
"reward": -0.14020115474704653,
|
| 231 |
+
"reward_std": 0.7795401588082314,
|
| 232 |
+
"rewards/cosine_scaled_reward": -0.14689757686574012,
|
| 233 |
+
"rewards/format_reward": 0.006696428870782256,
|
| 234 |
+
"step": 16
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"clip_ratio": 0.0,
|
| 238 |
+
"completion_length": 616.6361923217773,
|
| 239 |
+
"epoch": 0.17237008871989862,
|
| 240 |
+
"grad_norm": 0.17595961689949036,
|
| 241 |
+
"learning_rate": 8.499999999999999e-07,
|
| 242 |
+
"loss": -0.1258,
|
| 243 |
+
"num_tokens": 10885255.0,
|
| 244 |
+
"reward": -0.16032503126189113,
|
| 245 |
+
"reward_std": 0.7151542007923126,
|
| 246 |
+
"rewards/cosine_scaled_reward": -0.16478930541779846,
|
| 247 |
+
"rewards/format_reward": 0.004464285913854837,
|
| 248 |
+
"step": 17
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"clip_ratio": 0.0,
|
| 252 |
+
"completion_length": 614.3348388671875,
|
| 253 |
+
"epoch": 0.18250950570342206,
|
| 254 |
+
"grad_norm": 0.17326559126377106,
|
| 255 |
+
"learning_rate": 9e-07,
|
| 256 |
+
"loss": -0.1223,
|
| 257 |
+
"num_tokens": 11559043.0,
|
| 258 |
+
"reward": -0.05317553365603089,
|
| 259 |
+
"reward_std": 0.8043533340096474,
|
| 260 |
+
"rewards/cosine_scaled_reward": -0.055407675448805094,
|
| 261 |
+
"rewards/format_reward": 0.0022321429569274187,
|
| 262 |
+
"step": 18
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"clip_ratio": 0.0,
|
| 266 |
+
"completion_length": 576.6774749755859,
|
| 267 |
+
"epoch": 0.1926489226869455,
|
| 268 |
+
"grad_norm": 0.19331927597522736,
|
| 269 |
+
"learning_rate": 9.499999999999999e-07,
|
| 270 |
+
"loss": -0.0965,
|
| 271 |
+
"num_tokens": 12207730.0,
|
| 272 |
+
"reward": -0.01400054944679141,
|
| 273 |
+
"reward_std": 0.8140848129987717,
|
| 274 |
+
"rewards/cosine_scaled_reward": -0.017348763067275286,
|
| 275 |
+
"rewards/format_reward": 0.003348214435391128,
|
| 276 |
+
"step": 19
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"clip_ratio": 0.0,
|
| 280 |
+
"completion_length": 659.9911117553711,
|
| 281 |
+
"epoch": 0.20278833967046894,
|
| 282 |
+
"grad_norm": 0.16300565004348755,
|
| 283 |
+
"learning_rate": 1e-06,
|
| 284 |
+
"loss": -0.0491,
|
| 285 |
+
"num_tokens": 12919506.0,
|
| 286 |
+
"reward": 0.12035302398726344,
|
| 287 |
+
"reward_std": 0.7825085148215294,
|
| 288 |
+
"rewards/cosine_scaled_reward": 0.11588872922584414,
|
| 289 |
+
"rewards/format_reward": 0.004464285797439516,
|
| 290 |
+
"step": 20
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"clip_ratio": 0.0,
|
| 294 |
+
"completion_length": 641.3013687133789,
|
| 295 |
+
"epoch": 0.21292775665399238,
|
| 296 |
+
"grad_norm": 0.24253493547439575,
|
| 297 |
+
"learning_rate": 9.999203468625015e-07,
|
| 298 |
+
"loss": -0.0523,
|
| 299 |
+
"num_tokens": 13615384.0,
|
| 300 |
+
"reward": 0.19580535404384136,
|
| 301 |
+
"reward_std": 0.8355192169547081,
|
| 302 |
+
"rewards/cosine_scaled_reward": 0.19357320480048656,
|
| 303 |
+
"rewards/format_reward": 0.0022321429569274187,
|
| 304 |
+
"step": 21
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"clip_ratio": 0.0,
|
| 308 |
+
"completion_length": 694.412971496582,
|
| 309 |
+
"epoch": 0.22306717363751585,
|
| 310 |
+
"grad_norm": 0.22512537240982056,
|
| 311 |
+
"learning_rate": 9.99681412828496e-07,
|
| 312 |
+
"loss": -0.0494,
|
| 313 |
+
"num_tokens": 14370738.0,
|
| 314 |
+
"reward": 0.2461150388699025,
|
| 315 |
+
"reward_std": 0.7426658719778061,
|
| 316 |
+
"rewards/cosine_scaled_reward": 0.24388288520276546,
|
| 317 |
+
"rewards/format_reward": 0.0022321429569274187,
|
| 318 |
+
"step": 22
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"clip_ratio": 0.0,
|
| 322 |
+
"completion_length": 675.4542617797852,
|
| 323 |
+
"epoch": 0.2332065906210393,
|
| 324 |
+
"grad_norm": 0.16492308676242828,
|
| 325 |
+
"learning_rate": 9.992832740253644e-07,
|
| 326 |
+
"loss": -0.0398,
|
| 327 |
+
"num_tokens": 15114193.0,
|
| 328 |
+
"reward": 0.2721172561869025,
|
| 329 |
+
"reward_std": 0.8025888651609421,
|
| 330 |
+
"rewards/cosine_scaled_reward": 0.2710011739982292,
|
| 331 |
+
"rewards/format_reward": 0.0011160714784637094,
|
| 332 |
+
"step": 23
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"clip_ratio": 0.0,
|
| 336 |
+
"completion_length": 719.147346496582,
|
| 337 |
+
"epoch": 0.24334600760456274,
|
| 338 |
+
"grad_norm": 0.1536169797182083,
|
| 339 |
+
"learning_rate": 9.987260573051267e-07,
|
| 340 |
+
"loss": -0.0321,
|
| 341 |
+
"num_tokens": 15888093.0,
|
| 342 |
+
"reward": 0.30394730158150196,
|
| 343 |
+
"reward_std": 0.7235733941197395,
|
| 344 |
+
"rewards/cosine_scaled_reward": 0.30394728668034077,
|
| 345 |
+
"rewards/format_reward": 0.0,
|
| 346 |
+
"step": 24
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"clip_ratio": 0.0,
|
| 350 |
+
"completion_length": 665.2589645385742,
|
| 351 |
+
"epoch": 0.2534854245880862,
|
| 352 |
+
"grad_norm": 0.15167948603630066,
|
| 353 |
+
"learning_rate": 9.98009940204023e-07,
|
| 354 |
+
"loss": -0.0196,
|
| 355 |
+
"num_tokens": 16615493.0,
|
| 356 |
+
"reward": 0.4243004210293293,
|
| 357 |
+
"reward_std": 0.784424401819706,
|
| 358 |
+
"rewards/cosine_scaled_reward": 0.42318433709442616,
|
| 359 |
+
"rewards/format_reward": 0.0011160714784637094,
|
| 360 |
+
"step": 25
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"clip_ratio": 0.0,
|
| 364 |
+
"completion_length": 680.0814971923828,
|
| 365 |
+
"epoch": 0.26362484157160965,
|
| 366 |
+
"grad_norm": 0.1787332445383072,
|
| 367 |
+
"learning_rate": 9.971351508859486e-07,
|
| 368 |
+
"loss": -0.0554,
|
| 369 |
+
"num_tokens": 17343678.0,
|
| 370 |
+
"reward": 0.5126287564635277,
|
| 371 |
+
"reward_std": 0.7500302940607071,
|
| 372 |
+
"rewards/cosine_scaled_reward": 0.5126287154853344,
|
| 373 |
+
"rewards/format_reward": 0.0,
|
| 374 |
+
"step": 26
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"clip_ratio": 0.0,
|
| 378 |
+
"completion_length": 691.2957992553711,
|
| 379 |
+
"epoch": 0.2737642585551331,
|
| 380 |
+
"grad_norm": 0.15422694385051727,
|
| 381 |
+
"learning_rate": 9.961019680697591e-07,
|
| 382 |
+
"loss": 0.0019,
|
| 383 |
+
"num_tokens": 18091255.0,
|
| 384 |
+
"reward": 0.478223854675889,
|
| 385 |
+
"reward_std": 0.7884277924895287,
|
| 386 |
+
"rewards/cosine_scaled_reward": 0.4771077586337924,
|
| 387 |
+
"rewards/format_reward": 0.0011160714784637094,
|
| 388 |
+
"step": 27
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"clip_ratio": 0.0,
|
| 392 |
+
"completion_length": 724.2433319091797,
|
| 393 |
+
"epoch": 0.28390367553865653,
|
| 394 |
+
"grad_norm": 0.14453695714473724,
|
| 395 |
+
"learning_rate": 9.949107209404663e-07,
|
| 396 |
+
"loss": 0.0015,
|
| 397 |
+
"num_tokens": 18873817.0,
|
| 398 |
+
"reward": 0.6080499514937401,
|
| 399 |
+
"reward_std": 0.7710720151662827,
|
| 400 |
+
"rewards/cosine_scaled_reward": 0.6080499142408371,
|
| 401 |
+
"rewards/format_reward": 0.0,
|
| 402 |
+
"step": 28
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"clip_ratio": 0.0,
|
| 406 |
+
"completion_length": 719.5837326049805,
|
| 407 |
+
"epoch": 0.29404309252217997,
|
| 408 |
+
"grad_norm": 0.14530107378959656,
|
| 409 |
+
"learning_rate": 9.935617890443554e-07,
|
| 410 |
+
"loss": 0.0026,
|
| 411 |
+
"num_tokens": 19659300.0,
|
| 412 |
+
"reward": 0.6268112398684025,
|
| 413 |
+
"reward_std": 0.6641118600964546,
|
| 414 |
+
"rewards/cosine_scaled_reward": 0.6268112063407898,
|
| 415 |
+
"rewards/format_reward": 0.0,
|
| 416 |
+
"step": 29
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"clip_ratio": 0.0,
|
| 420 |
+
"completion_length": 793.0312805175781,
|
| 421 |
+
"epoch": 0.3041825095057034,
|
| 422 |
+
"grad_norm": 0.1358717679977417,
|
| 423 |
+
"learning_rate": 9.92055602168058e-07,
|
| 424 |
+
"loss": -0.0059,
|
| 425 |
+
"num_tokens": 20515344.0,
|
| 426 |
+
"reward": 0.6840809024870396,
|
| 427 |
+
"reward_std": 0.6523778513073921,
|
| 428 |
+
"rewards/cosine_scaled_reward": 0.6840808726847172,
|
| 429 |
+
"rewards/format_reward": 0.0,
|
| 430 |
+
"step": 30
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"clip_ratio": 0.0,
|
| 434 |
+
"completion_length": 766.2176818847656,
|
| 435 |
+
"epoch": 0.31432192648922685,
|
| 436 |
+
"grad_norm": 0.13609299063682556,
|
| 437 |
+
"learning_rate": 9.90392640201615e-07,
|
| 438 |
+
"loss": 0.0048,
|
| 439 |
+
"num_tokens": 21329483.0,
|
| 440 |
+
"reward": 0.7440241314470768,
|
| 441 |
+
"reward_std": 0.6845037117600441,
|
| 442 |
+
"rewards/cosine_scaled_reward": 0.7440241128206253,
|
| 443 |
+
"rewards/format_reward": 0.0,
|
| 444 |
+
"step": 31
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"clip_ratio": 0.0,
|
| 448 |
+
"completion_length": 760.3225784301758,
|
| 449 |
+
"epoch": 0.3244613434727503,
|
| 450 |
+
"grad_norm": 0.1422865241765976,
|
| 451 |
+
"learning_rate": 9.885734329855797e-07,
|
| 452 |
+
"loss": 0.0076,
|
| 453 |
+
"num_tokens": 22130476.0,
|
| 454 |
+
"reward": 0.8551607951521873,
|
| 455 |
+
"reward_std": 0.6532174497842789,
|
| 456 |
+
"rewards/cosine_scaled_reward": 0.8551607728004456,
|
| 457 |
+
"rewards/format_reward": 0.0,
|
| 458 |
+
"step": 32
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"clip_ratio": 0.0,
|
| 462 |
+
"completion_length": 788.7120895385742,
|
| 463 |
+
"epoch": 0.33460076045627374,
|
| 464 |
+
"grad_norm": 0.13245096802711487,
|
| 465 |
+
"learning_rate": 9.865985601422017e-07,
|
| 466 |
+
"loss": -0.0077,
|
| 467 |
+
"num_tokens": 22971218.0,
|
| 468 |
+
"reward": 0.7257559858262539,
|
| 469 |
+
"reward_std": 0.6099549047648907,
|
| 470 |
+
"rewards/cosine_scaled_reward": 0.7302779853343964,
|
| 471 |
+
"rewards/format_reward": 0.0,
|
| 472 |
+
"step": 33
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"clip_ratio": 0.0,
|
| 476 |
+
"completion_length": 802.2344055175781,
|
| 477 |
+
"epoch": 0.34474017743979724,
|
| 478 |
+
"grad_norm": 0.13606438040733337,
|
| 479 |
+
"learning_rate": 9.844686508907537e-07,
|
| 480 |
+
"loss": -0.0022,
|
| 481 |
+
"num_tokens": 23817508.0,
|
| 482 |
+
"reward": 0.9126424044370651,
|
| 483 |
+
"reward_std": 0.6410668790340424,
|
| 484 |
+
"rewards/cosine_scaled_reward": 0.9126423671841621,
|
| 485 |
+
"rewards/format_reward": 0.0,
|
| 486 |
+
"step": 34
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"clip_ratio": 0.0,
|
| 490 |
+
"completion_length": 779.1975784301758,
|
| 491 |
+
"epoch": 0.3548795944233207,
|
| 492 |
+
"grad_norm": 0.13803985714912415,
|
| 493 |
+
"learning_rate": 9.821843838470534e-07,
|
| 494 |
+
"loss": -0.0014,
|
| 495 |
+
"num_tokens": 24658397.0,
|
| 496 |
+
"reward": 0.951663188636303,
|
| 497 |
+
"reward_std": 0.6128395721316338,
|
| 498 |
+
"rewards/cosine_scaled_reward": 0.9516631364822388,
|
| 499 |
+
"rewards/format_reward": 0.0,
|
| 500 |
+
"step": 35
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"clip_ratio": 0.0,
|
| 504 |
+
"completion_length": 790.4141006469727,
|
| 505 |
+
"epoch": 0.3650190114068441,
|
| 506 |
+
"grad_norm": 0.15671393275260925,
|
| 507 |
+
"learning_rate": 9.797464868072486e-07,
|
| 508 |
+
"loss": 0.0435,
|
| 509 |
+
"num_tokens": 25494720.0,
|
| 510 |
+
"reward": 0.952189639210701,
|
| 511 |
+
"reward_std": 0.6134774163365364,
|
| 512 |
+
"rewards/cosine_scaled_reward": 0.9521896243095398,
|
| 513 |
+
"rewards/format_reward": 0.0,
|
| 514 |
+
"step": 36
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"clip_ratio": 0.0,
|
| 518 |
+
"completion_length": 780.6808395385742,
|
| 519 |
+
"epoch": 0.37515842839036756,
|
| 520 |
+
"grad_norm": 0.13327902555465698,
|
| 521 |
+
"learning_rate": 9.771557365159319e-07,
|
| 522 |
+
"loss": -0.0051,
|
| 523 |
+
"num_tokens": 26319738.0,
|
| 524 |
+
"reward": 0.9673999920487404,
|
| 525 |
+
"reward_std": 0.5673549436032772,
|
| 526 |
+
"rewards/cosine_scaled_reward": 0.9673999547958374,
|
| 527 |
+
"rewards/format_reward": 0.0,
|
| 528 |
+
"step": 37
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"clip_ratio": 0.0,
|
| 532 |
+
"completion_length": 797.1105194091797,
|
| 533 |
+
"epoch": 0.385297845373891,
|
| 534 |
+
"grad_norm": 0.14108316600322723,
|
| 535 |
+
"learning_rate": 9.744129584186597e-07,
|
| 536 |
+
"loss": -0.0108,
|
| 537 |
+
"num_tokens": 27160557.0,
|
| 538 |
+
"reward": 0.9532012119889259,
|
| 539 |
+
"reward_std": 0.5637913830578327,
|
| 540 |
+
"rewards/cosine_scaled_reward": 0.9532012045383453,
|
| 541 |
+
"rewards/format_reward": 0.0,
|
| 542 |
+
"step": 38
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"clip_ratio": 0.0,
|
| 546 |
+
"completion_length": 857.9230270385742,
|
| 547 |
+
"epoch": 0.39543726235741444,
|
| 548 |
+
"grad_norm": 0.13035330176353455,
|
| 549 |
+
"learning_rate": 9.71519026398956e-07,
|
| 550 |
+
"loss": 0.0268,
|
| 551 |
+
"num_tokens": 28067040.0,
|
| 552 |
+
"reward": 0.9902723953127861,
|
| 553 |
+
"reward_std": 0.5246665701270103,
|
| 554 |
+
"rewards/cosine_scaled_reward": 0.9902723580598831,
|
| 555 |
+
"rewards/format_reward": 0.0,
|
| 556 |
+
"step": 39
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"clip_ratio": 0.0,
|
| 560 |
+
"completion_length": 793.2377471923828,
|
| 561 |
+
"epoch": 0.4055766793409379,
|
| 562 |
+
"grad_norm": 0.13015364110469818,
|
| 563 |
+
"learning_rate": 9.68474862499881e-07,
|
| 564 |
+
"loss": 0.0268,
|
| 565 |
+
"num_tokens": 28897045.0,
|
| 566 |
+
"reward": 1.063771203160286,
|
| 567 |
+
"reward_std": 0.49807993322610855,
|
| 568 |
+
"rewards/cosine_scaled_reward": 1.0637711435556412,
|
| 569 |
+
"rewards/format_reward": 0.0,
|
| 570 |
+
"step": 40
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"clip_ratio": 0.0,
|
| 574 |
+
"completion_length": 776.3783798217773,
|
| 575 |
+
"epoch": 0.4157160963244613,
|
| 576 |
+
"grad_norm": 0.12312573939561844,
|
| 577 |
+
"learning_rate": 9.652814366302568e-07,
|
| 578 |
+
"loss": 0.0136,
|
| 579 |
+
"num_tokens": 29718656.0,
|
| 580 |
+
"reward": 1.1480179950594902,
|
| 581 |
+
"reward_std": 0.5026725828647614,
|
| 582 |
+
"rewards/cosine_scaled_reward": 1.1480179578065872,
|
| 583 |
+
"rewards/format_reward": 0.0,
|
| 584 |
+
"step": 41
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"clip_ratio": 0.0,
|
| 588 |
+
"completion_length": 782.9832916259766,
|
| 589 |
+
"epoch": 0.42585551330798477,
|
| 590 |
+
"grad_norm": 0.12757915258407593,
|
| 591 |
+
"learning_rate": 9.619397662556433e-07,
|
| 592 |
+
"loss": 0.0089,
|
| 593 |
+
"num_tokens": 30549177.0,
|
| 594 |
+
"reward": 1.153884269297123,
|
| 595 |
+
"reward_std": 0.43890631198883057,
|
| 596 |
+
"rewards/cosine_scaled_reward": 1.1538842469453812,
|
| 597 |
+
"rewards/format_reward": 0.0,
|
| 598 |
+
"step": 42
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"clip_ratio": 0.0,
|
| 602 |
+
"completion_length": 855.8906631469727,
|
| 603 |
+
"epoch": 0.43599493029150826,
|
| 604 |
+
"grad_norm": 0.11601021140813828,
|
| 605 |
+
"learning_rate": 9.5845091607416e-07,
|
| 606 |
+
"loss": -0.0027,
|
| 607 |
+
"num_tokens": 31434759.0,
|
| 608 |
+
"reward": 1.1496833115816116,
|
| 609 |
+
"reward_std": 0.44448646157979965,
|
| 610 |
+
"rewards/cosine_scaled_reward": 1.1496832445263863,
|
| 611 |
+
"rewards/format_reward": 0.0,
|
| 612 |
+
"step": 43
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"clip_ratio": 0.0,
|
| 616 |
+
"completion_length": 862.2332916259766,
|
| 617 |
+
"epoch": 0.4461343472750317,
|
| 618 |
+
"grad_norm": 0.11746807396411896,
|
| 619 |
+
"learning_rate": 9.548159976772592e-07,
|
| 620 |
+
"loss": 0.0197,
|
| 621 |
+
"num_tokens": 32337368.0,
|
| 622 |
+
"reward": 1.1307310312986374,
|
| 623 |
+
"reward_std": 0.4776446260511875,
|
| 624 |
+
"rewards/cosine_scaled_reward": 1.130730964243412,
|
| 625 |
+
"rewards/format_reward": 0.0,
|
| 626 |
+
"step": 44
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"clip_ratio": 0.0,
|
| 630 |
+
"completion_length": 900.1975860595703,
|
| 631 |
+
"epoch": 0.45627376425855515,
|
| 632 |
+
"grad_norm": 0.1248885914683342,
|
| 633 |
+
"learning_rate": 9.510361691955606e-07,
|
| 634 |
+
"loss": 0.0165,
|
| 635 |
+
"num_tokens": 33266393.0,
|
| 636 |
+
"reward": 1.060933604836464,
|
| 637 |
+
"reward_std": 0.46989936381578445,
|
| 638 |
+
"rewards/cosine_scaled_reward": 1.060933567583561,
|
| 639 |
+
"rewards/format_reward": 0.0,
|
| 640 |
+
"step": 45
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"clip_ratio": 0.0,
|
| 644 |
+
"completion_length": 815.2812881469727,
|
| 645 |
+
"epoch": 0.4664131812420786,
|
| 646 |
+
"grad_norm": 0.9569998979568481,
|
| 647 |
+
"learning_rate": 9.471126349298556e-07,
|
| 648 |
+
"loss": 0.0354,
|
| 649 |
+
"num_tokens": 34127101.0,
|
| 650 |
+
"reward": 1.151917465031147,
|
| 651 |
+
"reward_std": 0.5036191828548908,
|
| 652 |
+
"rewards/cosine_scaled_reward": 1.1519174352288246,
|
| 653 |
+
"rewards/format_reward": 0.0,
|
| 654 |
+
"step": 46
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"clip_ratio": 0.0,
|
| 658 |
+
"completion_length": 833.6239242553711,
|
| 659 |
+
"epoch": 0.47655259822560203,
|
| 660 |
+
"grad_norm": 0.15996558964252472,
|
| 661 |
+
"learning_rate": 9.430466449674013e-07,
|
| 662 |
+
"loss": -0.0196,
|
| 663 |
+
"num_tokens": 35022380.0,
|
| 664 |
+
"reward": 1.0830270573496819,
|
| 665 |
+
"reward_std": 0.4072240814566612,
|
| 666 |
+
"rewards/cosine_scaled_reward": 1.0830269828438759,
|
| 667 |
+
"rewards/format_reward": 0.0,
|
| 668 |
+
"step": 47
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"clip_ratio": 0.0,
|
| 672 |
+
"completion_length": 892.449821472168,
|
| 673 |
+
"epoch": 0.4866920152091255,
|
| 674 |
+
"grad_norm": 0.11413749307394028,
|
| 675 |
+
"learning_rate": 9.388394947836278e-07,
|
| 676 |
+
"loss": -0.0125,
|
| 677 |
+
"num_tokens": 35948751.0,
|
| 678 |
+
"reward": 1.0179572254419327,
|
| 679 |
+
"reward_std": 0.45433366298675537,
|
| 680 |
+
"rewards/cosine_scaled_reward": 1.0179571583867073,
|
| 681 |
+
"rewards/format_reward": 0.0,
|
| 682 |
+
"step": 48
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"clip_ratio": 0.0,
|
| 686 |
+
"completion_length": 912.2645492553711,
|
| 687 |
+
"epoch": 0.4968314321926489,
|
| 688 |
+
"grad_norm": 0.11429373919963837,
|
| 689 |
+
"learning_rate": 9.344925248293835e-07,
|
| 690 |
+
"loss": -0.001,
|
| 691 |
+
"num_tokens": 36887148.0,
|
| 692 |
+
"reward": 1.1400989294052124,
|
| 693 |
+
"reward_std": 0.4401072785258293,
|
| 694 |
+
"rewards/cosine_scaled_reward": 1.1400988548994064,
|
| 695 |
+
"rewards/format_reward": 0.0,
|
| 696 |
+
"step": 49
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"clip_ratio": 0.0,
|
| 700 |
+
"completion_length": 866.2511596679688,
|
| 701 |
+
"epoch": 0.5069708491761724,
|
| 702 |
+
"grad_norm": 0.11980035156011581,
|
| 703 |
+
"learning_rate": 9.300071201038501e-07,
|
| 704 |
+
"loss": 0.0215,
|
| 705 |
+
"num_tokens": 37789485.0,
|
| 706 |
+
"reward": 1.13294617831707,
|
| 707 |
+
"reward_std": 0.4442785568535328,
|
| 708 |
+
"rewards/cosine_scaled_reward": 1.1329461187124252,
|
| 709 |
+
"rewards/format_reward": 0.0,
|
| 710 |
+
"step": 50
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"clip_ratio": 0.0,
|
| 714 |
+
"completion_length": 799.482177734375,
|
| 715 |
+
"epoch": 0.5171102661596958,
|
| 716 |
+
"grad_norm": 0.122144915163517,
|
| 717 |
+
"learning_rate": 9.253847097132655e-07,
|
| 718 |
+
"loss": 0.0044,
|
| 719 |
+
"num_tokens": 38636645.0,
|
| 720 |
+
"reward": 1.2101461067795753,
|
| 721 |
+
"reward_std": 0.41771573945879936,
|
| 722 |
+
"rewards/cosine_scaled_reward": 1.210146002471447,
|
| 723 |
+
"rewards/format_reward": 0.0,
|
| 724 |
+
"step": 51
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"clip_ratio": 0.0,
|
| 728 |
+
"completion_length": 847.4330749511719,
|
| 729 |
+
"epoch": 0.5272496831432193,
|
| 730 |
+
"grad_norm": 0.1199701800942421,
|
| 731 |
+
"learning_rate": 9.206267664155906e-07,
|
| 732 |
+
"loss": 0.0082,
|
| 733 |
+
"num_tokens": 39516321.0,
|
| 734 |
+
"reward": 1.1085221618413925,
|
| 735 |
+
"reward_std": 0.4500999916344881,
|
| 736 |
+
"rewards/cosine_scaled_reward": 1.1085221245884895,
|
| 737 |
+
"rewards/format_reward": 0.0,
|
| 738 |
+
"step": 52
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"clip_ratio": 0.0,
|
| 742 |
+
"completion_length": 873.8102951049805,
|
| 743 |
+
"epoch": 0.5373891001267427,
|
| 744 |
+
"grad_norm": 0.14303860068321228,
|
| 745 |
+
"learning_rate": 9.157348061512726e-07,
|
| 746 |
+
"loss": -0.0033,
|
| 747 |
+
"num_tokens": 40435335.0,
|
| 748 |
+
"reward": 1.1145585402846336,
|
| 749 |
+
"reward_std": 0.47080911323428154,
|
| 750 |
+
"rewards/cosine_scaled_reward": 1.1145584508776665,
|
| 751 |
+
"rewards/format_reward": 0.0,
|
| 752 |
+
"step": 53
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"clip_ratio": 0.0,
|
| 756 |
+
"completion_length": 961.9029312133789,
|
| 757 |
+
"epoch": 0.5475285171102662,
|
| 758 |
+
"grad_norm": 0.11308898031711578,
|
| 759 |
+
"learning_rate": 9.107103875602458e-07,
|
| 760 |
+
"loss": -0.0123,
|
| 761 |
+
"num_tokens": 41415752.0,
|
| 762 |
+
"reward": 1.0546733513474464,
|
| 763 |
+
"reward_std": 0.41840433329343796,
|
| 764 |
+
"rewards/cosine_scaled_reward": 1.054673284292221,
|
| 765 |
+
"rewards/format_reward": 0.0,
|
| 766 |
+
"step": 54
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"clip_ratio": 0.0,
|
| 770 |
+
"completion_length": 892.6484680175781,
|
| 771 |
+
"epoch": 0.5576679340937896,
|
| 772 |
+
"grad_norm": 0.12083946913480759,
|
| 773 |
+
"learning_rate": 9.055551114853295e-07,
|
| 774 |
+
"loss": 0.0043,
|
| 775 |
+
"num_tokens": 42345957.0,
|
| 776 |
+
"reward": 1.0607689917087555,
|
| 777 |
+
"reward_std": 0.4757756106555462,
|
| 778 |
+
"rewards/cosine_scaled_reward": 1.0607689544558525,
|
| 779 |
+
"rewards/format_reward": 0.0,
|
| 780 |
+
"step": 55
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"clip_ratio": 0.0,
|
| 784 |
+
"completion_length": 823.7578506469727,
|
| 785 |
+
"epoch": 0.5678073510773131,
|
| 786 |
+
"grad_norm": 0.21660108864307404,
|
| 787 |
+
"learning_rate": 9.002706204621802e-07,
|
| 788 |
+
"loss": 0.0311,
|
| 789 |
+
"num_tokens": 43207724.0,
|
| 790 |
+
"reward": 1.1668334901332855,
|
| 791 |
+
"reward_std": 0.482263408601284,
|
| 792 |
+
"rewards/cosine_scaled_reward": 1.1668334528803825,
|
| 793 |
+
"rewards/format_reward": 0.0,
|
| 794 |
+
"step": 56
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"clip_ratio": 0.0,
|
| 798 |
+
"completion_length": 844.7210159301758,
|
| 799 |
+
"epoch": 0.5779467680608364,
|
| 800 |
+
"grad_norm": 0.4439857602119446,
|
| 801 |
+
"learning_rate": 8.948585981959578e-07,
|
| 802 |
+
"loss": 0.0112,
|
| 803 |
+
"num_tokens": 44096178.0,
|
| 804 |
+
"reward": 1.1872114390134811,
|
| 805 |
+
"reward_std": 0.44185368344187737,
|
| 806 |
+
"rewards/cosine_scaled_reward": 1.1872113794088364,
|
| 807 |
+
"rewards/format_reward": 0.0,
|
| 808 |
+
"step": 57
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"clip_ratio": 0.0,
|
| 812 |
+
"completion_length": 832.4553909301758,
|
| 813 |
+
"epoch": 0.5880861850443599,
|
| 814 |
+
"grad_norm": 0.12814383208751678,
|
| 815 |
+
"learning_rate": 8.893207690248775e-07,
|
| 816 |
+
"loss": -0.0022,
|
| 817 |
+
"num_tokens": 44963122.0,
|
| 818 |
+
"reward": 1.1495825573801994,
|
| 819 |
+
"reward_std": 0.43896420300006866,
|
| 820 |
+
"rewards/cosine_scaled_reward": 1.149582527577877,
|
| 821 |
+
"rewards/format_reward": 0.0,
|
| 822 |
+
"step": 58
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"clip_ratio": 0.0,
|
| 826 |
+
"completion_length": 819.3661117553711,
|
| 827 |
+
"epoch": 0.5982256020278834,
|
| 828 |
+
"grad_norm": 0.12473784387111664,
|
| 829 |
+
"learning_rate": 8.836588973708128e-07,
|
| 830 |
+
"loss": -0.0152,
|
| 831 |
+
"num_tokens": 45838866.0,
|
| 832 |
+
"reward": 1.101597860455513,
|
| 833 |
+
"reward_std": 0.44026929885149,
|
| 834 |
+
"rewards/cosine_scaled_reward": 1.1015978381037712,
|
| 835 |
+
"rewards/format_reward": 0.0,
|
| 836 |
+
"step": 59
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"clip_ratio": 0.0,
|
| 840 |
+
"completion_length": 883.5223617553711,
|
| 841 |
+
"epoch": 0.6083650190114068,
|
| 842 |
+
"grad_norm": 0.12360174208879471,
|
| 843 |
+
"learning_rate": 8.778747871771291e-07,
|
| 844 |
+
"loss": 0.0119,
|
| 845 |
+
"num_tokens": 46767750.0,
|
| 846 |
+
"reward": 1.1120553016662598,
|
| 847 |
+
"reward_std": 0.43890899419784546,
|
| 848 |
+
"rewards/cosine_scaled_reward": 1.1120552495121956,
|
| 849 |
+
"rewards/format_reward": 0.0,
|
| 850 |
+
"step": 60
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"clip_ratio": 0.0,
|
| 854 |
+
"completion_length": 830.4174499511719,
|
| 855 |
+
"epoch": 0.6185044359949303,
|
| 856 |
+
"grad_norm": 0.12516207993030548,
|
| 857 |
+
"learning_rate": 8.719702813339247e-07,
|
| 858 |
+
"loss": 0.007,
|
| 859 |
+
"num_tokens": 47634148.0,
|
| 860 |
+
"reward": 1.1350341737270355,
|
| 861 |
+
"reward_std": 0.3932863809168339,
|
| 862 |
+
"rewards/cosine_scaled_reward": 1.1350341215729713,
|
| 863 |
+
"rewards/format_reward": 0.0,
|
| 864 |
+
"step": 61
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"clip_ratio": 0.0,
|
| 868 |
+
"completion_length": 860.8895492553711,
|
| 869 |
+
"epoch": 0.6286438529784537,
|
| 870 |
+
"grad_norm": 0.11765479296445847,
|
| 871 |
+
"learning_rate": 8.659472610908627e-07,
|
| 872 |
+
"loss": -0.0127,
|
| 873 |
+
"num_tokens": 48534897.0,
|
| 874 |
+
"reward": 1.0959327295422554,
|
| 875 |
+
"reward_std": 0.43786268681287766,
|
| 876 |
+
"rewards/cosine_scaled_reward": 1.09593266248703,
|
| 877 |
+
"rewards/format_reward": 0.0,
|
| 878 |
+
"step": 62
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"clip_ratio": 0.0,
|
| 882 |
+
"completion_length": 789.957633972168,
|
| 883 |
+
"epoch": 0.6387832699619772,
|
| 884 |
+
"grad_norm": 0.12538108229637146,
|
| 885 |
+
"learning_rate": 8.598076454577814e-07,
|
| 886 |
+
"loss": 0.0043,
|
| 887 |
+
"num_tokens": 49366995.0,
|
| 888 |
+
"reward": 1.2128012925386429,
|
| 889 |
+
"reward_std": 0.5132176093757153,
|
| 890 |
+
"rewards/cosine_scaled_reward": 1.2128012254834175,
|
| 891 |
+
"rewards/format_reward": 0.0,
|
| 892 |
+
"step": 63
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"clip_ratio": 0.0,
|
| 896 |
+
"completion_length": 814.1540603637695,
|
| 897 |
+
"epoch": 0.6489226869455006,
|
| 898 |
+
"grad_norm": 0.12521201372146606,
|
| 899 |
+
"learning_rate": 8.535533905932737e-07,
|
| 900 |
+
"loss": 0.031,
|
| 901 |
+
"num_tokens": 50244605.0,
|
| 902 |
+
"reward": 1.2290623039007187,
|
| 903 |
+
"reward_std": 0.35417212545871735,
|
| 904 |
+
"rewards/cosine_scaled_reward": 1.2290622666478157,
|
| 905 |
+
"rewards/format_reward": 0.0,
|
| 906 |
+
"step": 64
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"clip_ratio": 0.0,
|
| 910 |
+
"completion_length": 826.7444610595703,
|
| 911 |
+
"epoch": 0.6590621039290241,
|
| 912 |
+
"grad_norm": 0.1192706748843193,
|
| 913 |
+
"learning_rate": 8.471864891814304e-07,
|
| 914 |
+
"loss": 0.0187,
|
| 915 |
+
"num_tokens": 51104472.0,
|
| 916 |
+
"reward": 1.2224428355693817,
|
| 917 |
+
"reward_std": 0.4238445721566677,
|
| 918 |
+
"rewards/cosine_scaled_reward": 1.2224428057670593,
|
| 919 |
+
"rewards/format_reward": 0.0,
|
| 920 |
+
"step": 65
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"clip_ratio": 0.0,
|
| 924 |
+
"completion_length": 890.8181228637695,
|
| 925 |
+
"epoch": 0.6692015209125475,
|
| 926 |
+
"grad_norm": 0.11663277447223663,
|
| 927 |
+
"learning_rate": 8.407089697969456e-07,
|
| 928 |
+
"loss": -0.0075,
|
| 929 |
+
"num_tokens": 52027509.0,
|
| 930 |
+
"reward": 1.1307330876588821,
|
| 931 |
+
"reward_std": 0.43055112659931183,
|
| 932 |
+
"rewards/cosine_scaled_reward": 1.130733035504818,
|
| 933 |
+
"rewards/format_reward": 0.0,
|
| 934 |
+
"step": 66
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"clip_ratio": 0.0,
|
| 938 |
+
"completion_length": 896.5524978637695,
|
| 939 |
+
"epoch": 0.679340937896071,
|
| 940 |
+
"grad_norm": 0.11125318706035614,
|
| 941 |
+
"learning_rate": 8.341228962587881e-07,
|
| 942 |
+
"loss": -0.0128,
|
| 943 |
+
"num_tokens": 52966948.0,
|
| 944 |
+
"reward": 1.1304329186677933,
|
| 945 |
+
"reward_std": 0.38643651083111763,
|
| 946 |
+
"rewards/cosine_scaled_reward": 1.1304328739643097,
|
| 947 |
+
"rewards/format_reward": 0.0,
|
| 948 |
+
"step": 67
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"clip_ratio": 0.0,
|
| 952 |
+
"completion_length": 805.1919937133789,
|
| 953 |
+
"epoch": 0.6894803548795945,
|
| 954 |
+
"grad_norm": 0.1318301111459732,
|
| 955 |
+
"learning_rate": 8.274303669726426e-07,
|
| 956 |
+
"loss": 0.0267,
|
| 957 |
+
"num_tokens": 53809720.0,
|
| 958 |
+
"reward": 1.2848119288682938,
|
| 959 |
+
"reward_std": 0.4024609997868538,
|
| 960 |
+
"rewards/cosine_scaled_reward": 1.2848118543624878,
|
| 961 |
+
"rewards/format_reward": 0.0,
|
| 962 |
+
"step": 68
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"clip_ratio": 0.0,
|
| 966 |
+
"completion_length": 897.3783950805664,
|
| 967 |
+
"epoch": 0.6996197718631179,
|
| 968 |
+
"grad_norm": 0.11534889042377472,
|
| 969 |
+
"learning_rate": 8.206335142623304e-07,
|
| 970 |
+
"loss": -0.0213,
|
| 971 |
+
"num_tokens": 54743531.0,
|
| 972 |
+
"reward": 1.0870068296790123,
|
| 973 |
+
"reward_std": 0.396763913333416,
|
| 974 |
+
"rewards/cosine_scaled_reward": 1.0870067551732063,
|
| 975 |
+
"rewards/format_reward": 0.0,
|
| 976 |
+
"step": 69
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"clip_ratio": 0.0,
|
| 980 |
+
"completion_length": 849.7935562133789,
|
| 981 |
+
"epoch": 0.7097591888466414,
|
| 982 |
+
"grad_norm": 0.11737702041864395,
|
| 983 |
+
"learning_rate": 8.137345036904259e-07,
|
| 984 |
+
"loss": 0.0049,
|
| 985 |
+
"num_tokens": 55624322.0,
|
| 986 |
+
"reward": 1.1217299401760101,
|
| 987 |
+
"reward_std": 0.4600865840911865,
|
| 988 |
+
"rewards/cosine_scaled_reward": 1.1217298731207848,
|
| 989 |
+
"rewards/format_reward": 0.0,
|
| 990 |
+
"step": 70
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"clip_ratio": 0.0,
|
| 994 |
+
"completion_length": 843.4699020385742,
|
| 995 |
+
"epoch": 0.7198986058301647,
|
| 996 |
+
"grad_norm": 0.12105900794267654,
|
| 997 |
+
"learning_rate": 8.067355333682797e-07,
|
| 998 |
+
"loss": 0.0058,
|
| 999 |
+
"num_tokens": 56518839.0,
|
| 1000 |
+
"reward": 1.128207340836525,
|
| 1001 |
+
"reward_std": 0.39387310668826103,
|
| 1002 |
+
"rewards/cosine_scaled_reward": 1.1282072588801384,
|
| 1003 |
+
"rewards/format_reward": 0.0,
|
| 1004 |
+
"step": 71
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"clip_ratio": 0.0,
|
| 1008 |
+
"completion_length": 854.9007110595703,
|
| 1009 |
+
"epoch": 0.7300380228136882,
|
| 1010 |
+
"grad_norm": 0.11754778772592545,
|
| 1011 |
+
"learning_rate": 7.996388332556734e-07,
|
| 1012 |
+
"loss": 0.0184,
|
| 1013 |
+
"num_tokens": 57413550.0,
|
| 1014 |
+
"reward": 1.1404274106025696,
|
| 1015 |
+
"reward_std": 0.45089153945446014,
|
| 1016 |
+
"rewards/cosine_scaled_reward": 1.1404273584485054,
|
| 1017 |
+
"rewards/format_reward": 0.0,
|
| 1018 |
+
"step": 72
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"clip_ratio": 0.0,
|
| 1022 |
+
"completion_length": 900.8973541259766,
|
| 1023 |
+
"epoch": 0.7401774397972116,
|
| 1024 |
+
"grad_norm": 0.11590257287025452,
|
| 1025 |
+
"learning_rate": 7.924466644503264e-07,
|
| 1026 |
+
"loss": -0.0059,
|
| 1027 |
+
"num_tokens": 58353250.0,
|
| 1028 |
+
"reward": 1.1319697871804237,
|
| 1029 |
+
"reward_std": 0.4219101257622242,
|
| 1030 |
+
"rewards/cosine_scaled_reward": 1.131969727575779,
|
| 1031 |
+
"rewards/format_reward": 0.0,
|
| 1032 |
+
"step": 73
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"clip_ratio": 0.0,
|
| 1036 |
+
"completion_length": 806.4665603637695,
|
| 1037 |
+
"epoch": 0.7503168567807351,
|
| 1038 |
+
"grad_norm": 0.12108156830072403,
|
| 1039 |
+
"learning_rate": 7.85161318467482e-07,
|
| 1040 |
+
"loss": -0.0001,
|
| 1041 |
+
"num_tokens": 59211228.0,
|
| 1042 |
+
"reward": 1.1962338984012604,
|
| 1043 |
+
"reward_std": 0.45812737941741943,
|
| 1044 |
+
"rewards/cosine_scaled_reward": 1.196233868598938,
|
| 1045 |
+
"rewards/format_reward": 0.0,
|
| 1046 |
+
"step": 74
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"clip_ratio": 0.0,
|
| 1050 |
+
"completion_length": 812.6629791259766,
|
| 1051 |
+
"epoch": 0.7604562737642585,
|
| 1052 |
+
"grad_norm": 0.12231780588626862,
|
| 1053 |
+
"learning_rate": 7.777851165098011e-07,
|
| 1054 |
+
"loss": 0.0223,
|
| 1055 |
+
"num_tokens": 60070054.0,
|
| 1056 |
+
"reward": 1.2705483883619308,
|
| 1057 |
+
"reward_std": 0.4315441697835922,
|
| 1058 |
+
"rewards/cosine_scaled_reward": 1.2705483138561249,
|
| 1059 |
+
"rewards/format_reward": 0.0,
|
| 1060 |
+
"step": 75
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"clip_ratio": 0.0,
|
| 1064 |
+
"completion_length": 792.5323944091797,
|
| 1065 |
+
"epoch": 0.770595690747782,
|
| 1066 |
+
"grad_norm": 0.13156548142433167,
|
| 1067 |
+
"learning_rate": 7.703204087277988e-07,
|
| 1068 |
+
"loss": 0.0054,
|
| 1069 |
+
"num_tokens": 60907403.0,
|
| 1070 |
+
"reward": 1.0341035649180412,
|
| 1071 |
+
"reward_std": 0.4903283417224884,
|
| 1072 |
+
"rewards/cosine_scaled_reward": 1.0341035276651382,
|
| 1073 |
+
"rewards/format_reward": 0.0,
|
| 1074 |
+
"step": 76
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"clip_ratio": 0.0,
|
| 1078 |
+
"completion_length": 843.8315200805664,
|
| 1079 |
+
"epoch": 0.7807351077313055,
|
| 1080 |
+
"grad_norm": 0.12101699411869049,
|
| 1081 |
+
"learning_rate": 7.627695734710564e-07,
|
| 1082 |
+
"loss": 0.0109,
|
| 1083 |
+
"num_tokens": 61785700.0,
|
| 1084 |
+
"reward": 1.226922646164894,
|
| 1085 |
+
"reward_std": 0.38846197351813316,
|
| 1086 |
+
"rewards/cosine_scaled_reward": 1.2269226014614105,
|
| 1087 |
+
"rewards/format_reward": 0.0,
|
| 1088 |
+
"step": 77
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"clip_ratio": 0.0,
|
| 1092 |
+
"completion_length": 859.2433395385742,
|
| 1093 |
+
"epoch": 0.7908745247148289,
|
| 1094 |
+
"grad_norm": 0.1146969199180603,
|
| 1095 |
+
"learning_rate": 7.551350165304499e-07,
|
| 1096 |
+
"loss": 0.0219,
|
| 1097 |
+
"num_tokens": 62673358.0,
|
| 1098 |
+
"reward": 1.2758738100528717,
|
| 1099 |
+
"reward_std": 0.4126880206167698,
|
| 1100 |
+
"rewards/cosine_scaled_reward": 1.275873750448227,
|
| 1101 |
+
"rewards/format_reward": 0.0,
|
| 1102 |
+
"step": 78
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"clip_ratio": 0.0,
|
| 1106 |
+
"completion_length": 847.4732513427734,
|
| 1107 |
+
"epoch": 0.8010139416983524,
|
| 1108 |
+
"grad_norm": 0.11725428700447083,
|
| 1109 |
+
"learning_rate": 7.474191703716338e-07,
|
| 1110 |
+
"loss": -0.0134,
|
| 1111 |
+
"num_tokens": 63563254.0,
|
| 1112 |
+
"reward": 1.1727145612239838,
|
| 1113 |
+
"reward_std": 0.3684752322733402,
|
| 1114 |
+
"rewards/cosine_scaled_reward": 1.1727144792675972,
|
| 1115 |
+
"rewards/format_reward": 0.0,
|
| 1116 |
+
"step": 79
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"clip_ratio": 0.0,
|
| 1120 |
+
"completion_length": 831.0089721679688,
|
| 1121 |
+
"epoch": 0.8111533586818758,
|
| 1122 |
+
"grad_norm": 0.1215280145406723,
|
| 1123 |
+
"learning_rate": 7.396244933600284e-07,
|
| 1124 |
+
"loss": 0.0269,
|
| 1125 |
+
"num_tokens": 64432126.0,
|
| 1126 |
+
"reward": 1.1777353882789612,
|
| 1127 |
+
"reward_std": 0.420895554125309,
|
| 1128 |
+
"rewards/cosine_scaled_reward": 1.1777353659272194,
|
| 1129 |
+
"rewards/format_reward": 0.0,
|
| 1130 |
+
"step": 80
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"clip_ratio": 0.0,
|
| 1134 |
+
"completion_length": 848.4598617553711,
|
| 1135 |
+
"epoch": 0.8212927756653993,
|
| 1136 |
+
"grad_norm": 0.13223014771938324,
|
| 1137 |
+
"learning_rate": 7.317534689775527e-07,
|
| 1138 |
+
"loss": 0.0072,
|
| 1139 |
+
"num_tokens": 65312050.0,
|
| 1140 |
+
"reward": 1.1698129922151566,
|
| 1141 |
+
"reward_std": 0.4710509404540062,
|
| 1142 |
+
"rewards/cosine_scaled_reward": 1.1698129400610924,
|
| 1143 |
+
"rewards/format_reward": 0.0,
|
| 1144 |
+
"step": 81
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"clip_ratio": 0.0,
|
| 1148 |
+
"completion_length": 811.4230346679688,
|
| 1149 |
+
"epoch": 0.8314321926489227,
|
| 1150 |
+
"grad_norm": 0.12015990167856216,
|
| 1151 |
+
"learning_rate": 7.238086050313562e-07,
|
| 1152 |
+
"loss": -0.0264,
|
| 1153 |
+
"num_tokens": 66174341.0,
|
| 1154 |
+
"reward": 1.124290645122528,
|
| 1155 |
+
"reward_std": 0.40254683420062065,
|
| 1156 |
+
"rewards/cosine_scaled_reward": 1.1242906004190445,
|
| 1157 |
+
"rewards/format_reward": 0.0,
|
| 1158 |
+
"step": 82
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"clip_ratio": 0.0,
|
| 1162 |
+
"completion_length": 819.8382110595703,
|
| 1163 |
+
"epoch": 0.8415716096324461,
|
| 1164 |
+
"grad_norm": 0.14578643441200256,
|
| 1165 |
+
"learning_rate": 7.157924328548002e-07,
|
| 1166 |
+
"loss": 0.026,
|
| 1167 |
+
"num_tokens": 67042508.0,
|
| 1168 |
+
"reward": 1.2035124003887177,
|
| 1169 |
+
"reward_std": 0.4501505568623543,
|
| 1170 |
+
"rewards/cosine_scaled_reward": 1.2035123333334923,
|
| 1171 |
+
"rewards/format_reward": 0.0,
|
| 1172 |
+
"step": 83
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"clip_ratio": 0.0,
|
| 1176 |
+
"completion_length": 832.864990234375,
|
| 1177 |
+
"epoch": 0.8517110266159695,
|
| 1178 |
+
"grad_norm": 0.11915823817253113,
|
| 1179 |
+
"learning_rate": 7.077075065009433e-07,
|
| 1180 |
+
"loss": 0.0419,
|
| 1181 |
+
"num_tokens": 67922067.0,
|
| 1182 |
+
"reward": 1.2239351123571396,
|
| 1183 |
+
"reward_std": 0.4415438733994961,
|
| 1184 |
+
"rewards/cosine_scaled_reward": 1.2239350602030754,
|
| 1185 |
+
"rewards/format_reward": 0.0,
|
| 1186 |
+
"step": 84
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"clip_ratio": 0.0,
|
| 1190 |
+
"completion_length": 794.1897659301758,
|
| 1191 |
+
"epoch": 0.861850443599493,
|
| 1192 |
+
"grad_norm": 0.12470311671495438,
|
| 1193 |
+
"learning_rate": 6.995564019287869e-07,
|
| 1194 |
+
"loss": 0.0085,
|
| 1195 |
+
"num_tokens": 68763477.0,
|
| 1196 |
+
"reward": 1.1386212185025215,
|
| 1197 |
+
"reward_std": 0.4416712336242199,
|
| 1198 |
+
"rewards/cosine_scaled_reward": 1.1386211588978767,
|
| 1199 |
+
"rewards/format_reward": 0.0,
|
| 1200 |
+
"step": 85
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"clip_ratio": 0.0,
|
| 1204 |
+
"completion_length": 865.2344207763672,
|
| 1205 |
+
"epoch": 0.8719898605830165,
|
| 1206 |
+
"grad_norm": 0.1167558878660202,
|
| 1207 |
+
"learning_rate": 6.913417161825449e-07,
|
| 1208 |
+
"loss": -0.0029,
|
| 1209 |
+
"num_tokens": 69673639.0,
|
| 1210 |
+
"reward": 1.1339842602610588,
|
| 1211 |
+
"reward_std": 0.44360488280653954,
|
| 1212 |
+
"rewards/cosine_scaled_reward": 1.1339841783046722,
|
| 1213 |
+
"rewards/format_reward": 0.0,
|
| 1214 |
+
"step": 86
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"clip_ratio": 0.0,
|
| 1218 |
+
"completion_length": 807.192008972168,
|
| 1219 |
+
"epoch": 0.8821292775665399,
|
| 1220 |
+
"grad_norm": 0.22383332252502441,
|
| 1221 |
+
"learning_rate": 6.830660665641897e-07,
|
| 1222 |
+
"loss": 0.0167,
|
| 1223 |
+
"num_tokens": 70526603.0,
|
| 1224 |
+
"reward": 1.1783367395401,
|
| 1225 |
+
"reward_std": 0.44941645860671997,
|
| 1226 |
+
"rewards/cosine_scaled_reward": 1.1783367022871971,
|
| 1227 |
+
"rewards/format_reward": 0.0,
|
| 1228 |
+
"step": 87
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"clip_ratio": 0.0,
|
| 1232 |
+
"completion_length": 842.9832916259766,
|
| 1233 |
+
"epoch": 0.8922686945500634,
|
| 1234 |
+
"grad_norm": 0.12214665859937668,
|
| 1235 |
+
"learning_rate": 6.747320897995492e-07,
|
| 1236 |
+
"loss": 0.0028,
|
| 1237 |
+
"num_tokens": 71410908.0,
|
| 1238 |
+
"reward": 1.1979937851428986,
|
| 1239 |
+
"reward_std": 0.41686391085386276,
|
| 1240 |
+
"rewards/cosine_scaled_reward": 1.1979937255382538,
|
| 1241 |
+
"rewards/format_reward": 0.0,
|
| 1242 |
+
"step": 88
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"clip_ratio": 0.0,
|
| 1246 |
+
"completion_length": 752.7991409301758,
|
| 1247 |
+
"epoch": 0.9024081115335868,
|
| 1248 |
+
"grad_norm": 0.12918852269649506,
|
| 1249 |
+
"learning_rate": 6.66342441198212e-07,
|
| 1250 |
+
"loss": -0.0199,
|
| 1251 |
+
"num_tokens": 72209320.0,
|
| 1252 |
+
"reward": 1.1556047648191452,
|
| 1253 |
+
"reward_std": 0.39559851959347725,
|
| 1254 |
+
"rewards/cosine_scaled_reward": 1.1556047424674034,
|
| 1255 |
+
"rewards/format_reward": 0.0,
|
| 1256 |
+
"step": 89
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"clip_ratio": 0.0,
|
| 1260 |
+
"completion_length": 788.5815048217773,
|
| 1261 |
+
"epoch": 0.9125475285171103,
|
| 1262 |
+
"grad_norm": 0.2962648272514343,
|
| 1263 |
+
"learning_rate": 6.578997938075125e-07,
|
| 1264 |
+
"loss": 0.0258,
|
| 1265 |
+
"num_tokens": 73040361.0,
|
| 1266 |
+
"reward": 1.1919841021299362,
|
| 1267 |
+
"reward_std": 0.40032988227903843,
|
| 1268 |
+
"rewards/cosine_scaled_reward": 1.1919840648770332,
|
| 1269 |
+
"rewards/format_reward": 0.0,
|
| 1270 |
+
"step": 90
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"clip_ratio": 0.0,
|
| 1274 |
+
"completion_length": 787.3214645385742,
|
| 1275 |
+
"epoch": 0.9226869455006337,
|
| 1276 |
+
"grad_norm": 0.15694542229175568,
|
| 1277 |
+
"learning_rate": 6.494068375608646e-07,
|
| 1278 |
+
"loss": -0.0147,
|
| 1279 |
+
"num_tokens": 73880425.0,
|
| 1280 |
+
"reward": 1.1829868927598,
|
| 1281 |
+
"reward_std": 0.3718248065561056,
|
| 1282 |
+
"rewards/cosine_scaled_reward": 1.1829868257045746,
|
| 1283 |
+
"rewards/format_reward": 0.0,
|
| 1284 |
+
"step": 91
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"clip_ratio": 0.0,
|
| 1288 |
+
"completion_length": 821.3850708007812,
|
| 1289 |
+
"epoch": 0.9328263624841572,
|
| 1290 |
+
"grad_norm": 0.11674854159355164,
|
| 1291 |
+
"learning_rate": 6.408662784207149e-07,
|
| 1292 |
+
"loss": 0.0095,
|
| 1293 |
+
"num_tokens": 74755762.0,
|
| 1294 |
+
"reward": 1.2243514209985733,
|
| 1295 |
+
"reward_std": 0.4170701913535595,
|
| 1296 |
+
"rewards/cosine_scaled_reward": 1.2243513762950897,
|
| 1297 |
+
"rewards/format_reward": 0.0,
|
| 1298 |
+
"step": 92
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"clip_ratio": 0.0,
|
| 1302 |
+
"completion_length": 761.3248138427734,
|
| 1303 |
+
"epoch": 0.9429657794676806,
|
| 1304 |
+
"grad_norm": 0.12954537570476532,
|
| 1305 |
+
"learning_rate": 6.322808375163895e-07,
|
| 1306 |
+
"loss": 0.0386,
|
| 1307 |
+
"num_tokens": 75564789.0,
|
| 1308 |
+
"reward": 1.1891243010759354,
|
| 1309 |
+
"reward_std": 0.45389125496149063,
|
| 1310 |
+
"rewards/cosine_scaled_reward": 1.1891242563724518,
|
| 1311 |
+
"rewards/format_reward": 0.0,
|
| 1312 |
+
"step": 93
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"clip_ratio": 0.0,
|
| 1316 |
+
"completion_length": 793.9643249511719,
|
| 1317 |
+
"epoch": 0.9531051964512041,
|
| 1318 |
+
"grad_norm": 0.12744048237800598,
|
| 1319 |
+
"learning_rate": 6.236532502771077e-07,
|
| 1320 |
+
"loss": 0.0271,
|
| 1321 |
+
"num_tokens": 76409781.0,
|
| 1322 |
+
"reward": 1.1992642730474472,
|
| 1323 |
+
"reward_std": 0.4242793843150139,
|
| 1324 |
+
"rewards/cosine_scaled_reward": 1.1992641612887383,
|
| 1325 |
+
"rewards/format_reward": 0.0,
|
| 1326 |
+
"step": 94
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"clip_ratio": 0.0,
|
| 1330 |
+
"completion_length": 789.4219207763672,
|
| 1331 |
+
"epoch": 0.9632446134347274,
|
| 1332 |
+
"grad_norm": 0.12654094398021698,
|
| 1333 |
+
"learning_rate": 6.149862655604403e-07,
|
| 1334 |
+
"loss": 0.0339,
|
| 1335 |
+
"num_tokens": 77242791.0,
|
| 1336 |
+
"reward": 1.2284981906414032,
|
| 1337 |
+
"reward_std": 0.42066042125225067,
|
| 1338 |
+
"rewards/cosine_scaled_reward": 1.2284981459379196,
|
| 1339 |
+
"rewards/format_reward": 0.0,
|
| 1340 |
+
"step": 95
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"clip_ratio": 0.0,
|
| 1344 |
+
"completion_length": 795.716552734375,
|
| 1345 |
+
"epoch": 0.973384030418251,
|
| 1346 |
+
"grad_norm": 0.12548381090164185,
|
| 1347 |
+
"learning_rate": 6.062826447764883e-07,
|
| 1348 |
+
"loss": 0.008,
|
| 1349 |
+
"num_tokens": 78078497.0,
|
| 1350 |
+
"reward": 1.2442717999219894,
|
| 1351 |
+
"reward_std": 0.35396804101765156,
|
| 1352 |
+
"rewards/cosine_scaled_reward": 1.2442717254161835,
|
| 1353 |
+
"rewards/format_reward": 0.0,
|
| 1354 |
+
"step": 96
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"clip_ratio": 0.0,
|
| 1358 |
+
"completion_length": 793.0692291259766,
|
| 1359 |
+
"epoch": 0.9835234474017744,
|
| 1360 |
+
"grad_norm": 0.12744414806365967,
|
| 1361 |
+
"learning_rate": 5.975451610080642e-07,
|
| 1362 |
+
"loss": -0.0194,
|
| 1363 |
+
"num_tokens": 78916895.0,
|
| 1364 |
+
"reward": 1.1445267572999,
|
| 1365 |
+
"reward_std": 0.37857919186353683,
|
| 1366 |
+
"rewards/cosine_scaled_reward": 1.1445267051458359,
|
| 1367 |
+
"rewards/format_reward": 0.0,
|
| 1368 |
+
"step": 97
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"clip_ratio": 0.0,
|
| 1372 |
+
"completion_length": 782.1171875,
|
| 1373 |
+
"epoch": 0.9936628643852978,
|
| 1374 |
+
"grad_norm": 0.12226420640945435,
|
| 1375 |
+
"learning_rate": 5.887765981271517e-07,
|
| 1376 |
+
"loss": 0.0066,
|
| 1377 |
+
"num_tokens": 79784779.0,
|
| 1378 |
+
"reward": 1.168272152543068,
|
| 1379 |
+
"reward_std": 0.39856788516044617,
|
| 1380 |
+
"rewards/cosine_scaled_reward": 1.1682721078395844,
|
| 1381 |
+
"rewards/format_reward": 0.0,
|
| 1382 |
+
"step": 98
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"clip_ratio": 0.0,
|
| 1386 |
+
"completion_length": 783.4386367797852,
|
| 1387 |
+
"epoch": 1.0101394169835234,
|
| 1388 |
+
"grad_norm": 0.13228166103363037,
|
| 1389 |
+
"learning_rate": 5.7997974990793e-07,
|
| 1390 |
+
"loss": 0.0145,
|
| 1391 |
+
"num_tokens": 80615732.0,
|
| 1392 |
+
"reward": 1.2336678951978683,
|
| 1393 |
+
"reward_std": 0.38141428492963314,
|
| 1394 |
+
"rewards/cosine_scaled_reward": 1.233667865395546,
|
| 1395 |
+
"rewards/format_reward": 0.0,
|
| 1396 |
+
"step": 99
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"clip_ratio": 0.0,
|
| 1400 |
+
"completion_length": 806.2355346679688,
|
| 1401 |
+
"epoch": 1.020278833967047,
|
| 1402 |
+
"grad_norm": 0.12670038640499115,
|
| 1403 |
+
"learning_rate": 5.711574191366427e-07,
|
| 1404 |
+
"loss": 0.0028,
|
| 1405 |
+
"num_tokens": 81465495.0,
|
| 1406 |
+
"reward": 1.1623064577579498,
|
| 1407 |
+
"reward_std": 0.3932801876217127,
|
| 1408 |
+
"rewards/cosine_scaled_reward": 1.1623064130544662,
|
| 1409 |
+
"rewards/format_reward": 0.0,
|
| 1410 |
+
"step": 100
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"clip_ratio": 0.0,
|
| 1414 |
+
"completion_length": 790.1205749511719,
|
| 1415 |
+
"epoch": 1.0304182509505704,
|
| 1416 |
+
"grad_norm": 0.12086564302444458,
|
| 1417 |
+
"learning_rate": 5.623124167185929e-07,
|
| 1418 |
+
"loss": 0.0063,
|
| 1419 |
+
"num_tokens": 82295971.0,
|
| 1420 |
+
"reward": 1.2574269622564316,
|
| 1421 |
+
"reward_std": 0.3898525554686785,
|
| 1422 |
+
"rewards/cosine_scaled_reward": 1.257426917552948,
|
| 1423 |
+
"rewards/format_reward": 0.0,
|
| 1424 |
+
"step": 101
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"clip_ratio": 0.0,
|
| 1428 |
+
"completion_length": 758.897346496582,
|
| 1429 |
+
"epoch": 1.0405576679340938,
|
| 1430 |
+
"grad_norm": 0.14097033441066742,
|
| 1431 |
+
"learning_rate": 5.534475607825565e-07,
|
| 1432 |
+
"loss": -0.0126,
|
| 1433 |
+
"num_tokens": 83103231.0,
|
| 1434 |
+
"reward": 1.2127035036683083,
|
| 1435 |
+
"reward_std": 0.3951477538794279,
|
| 1436 |
+
"rewards/cosine_scaled_reward": 1.2127034589648247,
|
| 1437 |
+
"rewards/format_reward": 0.0,
|
| 1438 |
+
"step": 102
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"clip_ratio": 0.0,
|
| 1442 |
+
"completion_length": 823.2377624511719,
|
| 1443 |
+
"epoch": 1.0506970849176172,
|
| 1444 |
+
"grad_norm": 0.11900278925895691,
|
| 1445 |
+
"learning_rate": 5.445656757828879e-07,
|
| 1446 |
+
"loss": -0.0341,
|
| 1447 |
+
"num_tokens": 83972156.0,
|
| 1448 |
+
"reward": 1.0786512047052383,
|
| 1449 |
+
"reward_std": 0.386748855933547,
|
| 1450 |
+
"rewards/cosine_scaled_reward": 1.078651174902916,
|
| 1451 |
+
"rewards/format_reward": 0.0,
|
| 1452 |
+
"step": 103
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"clip_ratio": 0.0,
|
| 1456 |
+
"completion_length": 778.1082992553711,
|
| 1457 |
+
"epoch": 1.0608365019011408,
|
| 1458 |
+
"grad_norm": 0.12485373765230179,
|
| 1459 |
+
"learning_rate": 5.356695915996161e-07,
|
| 1460 |
+
"loss": -0.0152,
|
| 1461 |
+
"num_tokens": 84803701.0,
|
| 1462 |
+
"reward": 1.1336441561579704,
|
| 1463 |
+
"reward_std": 0.43268290534615517,
|
| 1464 |
+
"rewards/cosine_scaled_reward": 1.1336440965533257,
|
| 1465 |
+
"rewards/format_reward": 0.0,
|
| 1466 |
+
"step": 104
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"clip_ratio": 0.0,
|
| 1470 |
+
"completion_length": 747.7790451049805,
|
| 1471 |
+
"epoch": 1.0709759188846641,
|
| 1472 |
+
"grad_norm": 0.14275644719600677,
|
| 1473 |
+
"learning_rate": 5.267621426368075e-07,
|
| 1474 |
+
"loss": 0.0402,
|
| 1475 |
+
"num_tokens": 85592927.0,
|
| 1476 |
+
"reward": 1.2611572295427322,
|
| 1477 |
+
"reward_std": 0.4418743886053562,
|
| 1478 |
+
"rewards/cosine_scaled_reward": 1.2611571997404099,
|
| 1479 |
+
"rewards/format_reward": 0.0,
|
| 1480 |
+
"step": 105
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"clip_ratio": 0.0,
|
| 1484 |
+
"completion_length": 686.8861770629883,
|
| 1485 |
+
"epoch": 1.0811153358681875,
|
| 1486 |
+
"grad_norm": 0.2392619252204895,
|
| 1487 |
+
"learning_rate": 5.178461669194903e-07,
|
| 1488 |
+
"loss": 0.0233,
|
| 1489 |
+
"num_tokens": 86335585.0,
|
| 1490 |
+
"reward": 1.2902618646621704,
|
| 1491 |
+
"reward_std": 0.4318929500877857,
|
| 1492 |
+
"rewards/cosine_scaled_reward": 1.2902617901563644,
|
| 1493 |
+
"rewards/format_reward": 0.0,
|
| 1494 |
+
"step": 106
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"clip_ratio": 0.0,
|
| 1498 |
+
"completion_length": 765.6752548217773,
|
| 1499 |
+
"epoch": 1.091254752851711,
|
| 1500 |
+
"grad_norm": 0.12888778746128082,
|
| 1501 |
+
"learning_rate": 5.08924505189423e-07,
|
| 1502 |
+
"loss": -0.0053,
|
| 1503 |
+
"num_tokens": 87147590.0,
|
| 1504 |
+
"reward": 1.2406404912471771,
|
| 1505 |
+
"reward_std": 0.3950204625725746,
|
| 1506 |
+
"rewards/cosine_scaled_reward": 1.2406404465436935,
|
| 1507 |
+
"rewards/format_reward": 0.0,
|
| 1508 |
+
"step": 107
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"clip_ratio": 0.0,
|
| 1512 |
+
"completion_length": 774.6596298217773,
|
| 1513 |
+
"epoch": 1.1013941698352345,
|
| 1514 |
+
"grad_norm": 0.16063496470451355,
|
| 1515 |
+
"learning_rate": 5e-07,
|
| 1516 |
+
"loss": 0.0407,
|
| 1517 |
+
"num_tokens": 87965805.0,
|
| 1518 |
+
"reward": 1.2785077691078186,
|
| 1519 |
+
"reward_std": 0.39564304798841476,
|
| 1520 |
+
"rewards/cosine_scaled_reward": 1.2785076946020126,
|
| 1521 |
+
"rewards/format_reward": 0.0,
|
| 1522 |
+
"step": 108
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"clip_ratio": 0.0,
|
| 1526 |
+
"completion_length": 800.3393249511719,
|
| 1527 |
+
"epoch": 1.111533586818758,
|
| 1528 |
+
"grad_norm": 0.23322370648384094,
|
| 1529 |
+
"learning_rate": 4.91075494810577e-07,
|
| 1530 |
+
"loss": 0.0181,
|
| 1531 |
+
"num_tokens": 88800901.0,
|
| 1532 |
+
"reward": 1.2374378964304924,
|
| 1533 |
+
"reward_std": 0.39184754714369774,
|
| 1534 |
+
"rewards/cosine_scaled_reward": 1.2374378442764282,
|
| 1535 |
+
"rewards/format_reward": 0.0,
|
| 1536 |
+
"step": 109
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"clip_ratio": 0.0,
|
| 1540 |
+
"completion_length": 802.3839721679688,
|
| 1541 |
+
"epoch": 1.1216730038022813,
|
| 1542 |
+
"grad_norm": 0.12637439370155334,
|
| 1543 |
+
"learning_rate": 4.821538330805098e-07,
|
| 1544 |
+
"loss": -0.0142,
|
| 1545 |
+
"num_tokens": 89646357.0,
|
| 1546 |
+
"reward": 1.2594422698020935,
|
| 1547 |
+
"reward_std": 0.3456545788794756,
|
| 1548 |
+
"rewards/cosine_scaled_reward": 1.2594422399997711,
|
| 1549 |
+
"rewards/format_reward": 0.0,
|
| 1550 |
+
"step": 110
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"clip_ratio": 0.0,
|
| 1554 |
+
"completion_length": 804.2109680175781,
|
| 1555 |
+
"epoch": 1.131812420785805,
|
| 1556 |
+
"grad_norm": 0.4338585138320923,
|
| 1557 |
+
"learning_rate": 4.732378573631924e-07,
|
| 1558 |
+
"loss": -0.0124,
|
| 1559 |
+
"num_tokens": 90489658.0,
|
| 1560 |
+
"reward": 1.204567864537239,
|
| 1561 |
+
"reward_std": 0.44952966272830963,
|
| 1562 |
+
"rewards/cosine_scaled_reward": 1.2045678049325943,
|
| 1563 |
+
"rewards/format_reward": 0.0,
|
| 1564 |
+
"step": 111
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"clip_ratio": 0.0,
|
| 1568 |
+
"completion_length": 814.7243728637695,
|
| 1569 |
+
"epoch": 1.1419518377693283,
|
| 1570 |
+
"grad_norm": 0.12191635370254517,
|
| 1571 |
+
"learning_rate": 4.643304084003838e-07,
|
| 1572 |
+
"loss": 0.0138,
|
| 1573 |
+
"num_tokens": 91347099.0,
|
| 1574 |
+
"reward": 1.1750361323356628,
|
| 1575 |
+
"reward_std": 0.4308737814426422,
|
| 1576 |
+
"rewards/cosine_scaled_reward": 1.1750360652804375,
|
| 1577 |
+
"rewards/format_reward": 0.0,
|
| 1578 |
+
"step": 112
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"clip_ratio": 0.0,
|
| 1582 |
+
"completion_length": 767.3794937133789,
|
| 1583 |
+
"epoch": 1.1520912547528517,
|
| 1584 |
+
"grad_norm": 0.13367751240730286,
|
| 1585 |
+
"learning_rate": 4.55434324217112e-07,
|
| 1586 |
+
"loss": -0.026,
|
| 1587 |
+
"num_tokens": 92163911.0,
|
| 1588 |
+
"reward": 1.0875676944851875,
|
| 1589 |
+
"reward_std": 0.4285357743501663,
|
| 1590 |
+
"rewards/cosine_scaled_reward": 1.0875676572322845,
|
| 1591 |
+
"rewards/format_reward": 0.0,
|
| 1592 |
+
"step": 113
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"clip_ratio": 0.0,
|
| 1596 |
+
"completion_length": 759.1138687133789,
|
| 1597 |
+
"epoch": 1.162230671736375,
|
| 1598 |
+
"grad_norm": 0.6044357419013977,
|
| 1599 |
+
"learning_rate": 4.4655243921744367e-07,
|
| 1600 |
+
"loss": -0.0034,
|
| 1601 |
+
"num_tokens": 92969037.0,
|
| 1602 |
+
"reward": 1.1716800779104233,
|
| 1603 |
+
"reward_std": 0.40124501287937164,
|
| 1604 |
+
"rewards/cosine_scaled_reward": 1.1716800406575203,
|
| 1605 |
+
"rewards/format_reward": 0.0,
|
| 1606 |
+
"step": 114
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"clip_ratio": 0.0,
|
| 1610 |
+
"completion_length": 742.3248138427734,
|
| 1611 |
+
"epoch": 1.1723700887198987,
|
| 1612 |
+
"grad_norm": 0.12603311240673065,
|
| 1613 |
+
"learning_rate": 4.37687583281407e-07,
|
| 1614 |
+
"loss": -0.0106,
|
| 1615 |
+
"num_tokens": 93761376.0,
|
| 1616 |
+
"reward": 1.2006227523088455,
|
| 1617 |
+
"reward_std": 0.3855649037286639,
|
| 1618 |
+
"rewards/cosine_scaled_reward": 1.2006226852536201,
|
| 1619 |
+
"rewards/format_reward": 0.0,
|
| 1620 |
+
"step": 115
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"clip_ratio": 0.0,
|
| 1624 |
+
"completion_length": 750.4765930175781,
|
| 1625 |
+
"epoch": 1.182509505703422,
|
| 1626 |
+
"grad_norm": 0.13189645111560822,
|
| 1627 |
+
"learning_rate": 4.2884258086335745e-07,
|
| 1628 |
+
"loss": 0.0114,
|
| 1629 |
+
"num_tokens": 94573043.0,
|
| 1630 |
+
"reward": 1.2511205673217773,
|
| 1631 |
+
"reward_std": 0.408273559063673,
|
| 1632 |
+
"rewards/cosine_scaled_reward": 1.2511205524206161,
|
| 1633 |
+
"rewards/format_reward": 0.0,
|
| 1634 |
+
"step": 116
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"clip_ratio": 0.0,
|
| 1638 |
+
"completion_length": 748.3750305175781,
|
| 1639 |
+
"epoch": 1.1926489226869454,
|
| 1640 |
+
"grad_norm": 0.13386943936347961,
|
| 1641 |
+
"learning_rate": 4.2002025009206987e-07,
|
| 1642 |
+
"loss": 0.0081,
|
| 1643 |
+
"num_tokens": 95375539.0,
|
| 1644 |
+
"reward": 1.1935091465711594,
|
| 1645 |
+
"reward_std": 0.39292260631918907,
|
| 1646 |
+
"rewards/cosine_scaled_reward": 1.1935090869665146,
|
| 1647 |
+
"rewards/format_reward": 0.0,
|
| 1648 |
+
"step": 117
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"clip_ratio": 0.0,
|
| 1652 |
+
"completion_length": 752.021240234375,
|
| 1653 |
+
"epoch": 1.202788339670469,
|
| 1654 |
+
"grad_norm": 0.13410834968090057,
|
| 1655 |
+
"learning_rate": 4.1122340187284845e-07,
|
| 1656 |
+
"loss": -0.0055,
|
| 1657 |
+
"num_tokens": 96171638.0,
|
| 1658 |
+
"reward": 1.215317651629448,
|
| 1659 |
+
"reward_std": 0.4204988572746515,
|
| 1660 |
+
"rewards/cosine_scaled_reward": 1.2153176292777061,
|
| 1661 |
+
"rewards/format_reward": 0.0,
|
| 1662 |
+
"step": 118
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"clip_ratio": 0.0,
|
| 1666 |
+
"completion_length": 741.1294937133789,
|
| 1667 |
+
"epoch": 1.2129277566539924,
|
| 1668 |
+
"grad_norm": 0.14352215826511383,
|
| 1669 |
+
"learning_rate": 4.0245483899193586e-07,
|
| 1670 |
+
"loss": -0.0021,
|
| 1671 |
+
"num_tokens": 96963066.0,
|
| 1672 |
+
"reward": 1.171108528971672,
|
| 1673 |
+
"reward_std": 0.390108622610569,
|
| 1674 |
+
"rewards/cosine_scaled_reward": 1.171108491718769,
|
| 1675 |
+
"rewards/format_reward": 0.0,
|
| 1676 |
+
"step": 119
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"clip_ratio": 0.0,
|
| 1680 |
+
"completion_length": 731.9676666259766,
|
| 1681 |
+
"epoch": 1.2230671736375158,
|
| 1682 |
+
"grad_norm": 0.3331538140773773,
|
| 1683 |
+
"learning_rate": 3.937173552235116e-07,
|
| 1684 |
+
"loss": -0.0142,
|
| 1685 |
+
"num_tokens": 97751717.0,
|
| 1686 |
+
"reward": 1.1743304282426834,
|
| 1687 |
+
"reward_std": 0.3502536006271839,
|
| 1688 |
+
"rewards/cosine_scaled_reward": 1.174330398440361,
|
| 1689 |
+
"rewards/format_reward": 0.0,
|
| 1690 |
+
"step": 120
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"clip_ratio": 0.0,
|
| 1694 |
+
"completion_length": 722.037971496582,
|
| 1695 |
+
"epoch": 1.2332065906210392,
|
| 1696 |
+
"grad_norm": 0.5637516975402832,
|
| 1697 |
+
"learning_rate": 3.850137344395598e-07,
|
| 1698 |
+
"loss": 0.0232,
|
| 1699 |
+
"num_tokens": 98521879.0,
|
| 1700 |
+
"reward": 1.261700451374054,
|
| 1701 |
+
"reward_std": 0.34792254120111465,
|
| 1702 |
+
"rewards/cosine_scaled_reward": 1.2617004364728928,
|
| 1703 |
+
"rewards/format_reward": 0.0,
|
| 1704 |
+
"step": 121
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"clip_ratio": 0.0,
|
| 1708 |
+
"completion_length": 810.3984832763672,
|
| 1709 |
+
"epoch": 1.2433460076045628,
|
| 1710 |
+
"grad_norm": 0.12412155419588089,
|
| 1711 |
+
"learning_rate": 3.763467497228922e-07,
|
| 1712 |
+
"loss": -0.0175,
|
| 1713 |
+
"num_tokens": 99386028.0,
|
| 1714 |
+
"reward": 1.1121523007750511,
|
| 1715 |
+
"reward_std": 0.4185631051659584,
|
| 1716 |
+
"rewards/cosine_scaled_reward": 1.11215228587389,
|
| 1717 |
+
"rewards/format_reward": 0.0,
|
| 1718 |
+
"step": 122
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"clip_ratio": 0.0,
|
| 1722 |
+
"completion_length": 762.1272659301758,
|
| 1723 |
+
"epoch": 1.2534854245880862,
|
| 1724 |
+
"grad_norm": 0.5518940091133118,
|
| 1725 |
+
"learning_rate": 3.677191624836106e-07,
|
| 1726 |
+
"loss": -0.0156,
|
| 1727 |
+
"num_tokens": 100190606.0,
|
| 1728 |
+
"reward": 1.2098966389894485,
|
| 1729 |
+
"reward_std": 0.42287516221404076,
|
| 1730 |
+
"rewards/cosine_scaled_reward": 1.2098965868353844,
|
| 1731 |
+
"rewards/format_reward": 0.0,
|
| 1732 |
+
"step": 123
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"clip_ratio": 0.0,
|
| 1736 |
+
"completion_length": 797.1741409301758,
|
| 1737 |
+
"epoch": 1.2636248415716096,
|
| 1738 |
+
"grad_norm": 0.13090519607067108,
|
| 1739 |
+
"learning_rate": 3.591337215792851e-07,
|
| 1740 |
+
"loss": -0.0049,
|
| 1741 |
+
"num_tokens": 101033042.0,
|
| 1742 |
+
"reward": 1.2026162892580032,
|
| 1743 |
+
"reward_std": 0.38727976381778717,
|
| 1744 |
+
"rewards/cosine_scaled_reward": 1.2026162147521973,
|
| 1745 |
+
"rewards/format_reward": 0.0,
|
| 1746 |
+
"step": 124
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"clip_ratio": 0.0,
|
| 1750 |
+
"completion_length": 755.0368728637695,
|
| 1751 |
+
"epoch": 1.2737642585551332,
|
| 1752 |
+
"grad_norm": 0.13797762989997864,
|
| 1753 |
+
"learning_rate": 3.505931624391355e-07,
|
| 1754 |
+
"loss": 0.0129,
|
| 1755 |
+
"num_tokens": 101827819.0,
|
| 1756 |
+
"reward": 1.1926257461309433,
|
| 1757 |
+
"reward_std": 0.44687049090862274,
|
| 1758 |
+
"rewards/cosine_scaled_reward": 1.1926256865262985,
|
| 1759 |
+
"rewards/format_reward": 0.0,
|
| 1760 |
+
"step": 125
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"clip_ratio": 0.0,
|
| 1764 |
+
"completion_length": 741.3024978637695,
|
| 1765 |
+
"epoch": 1.2839036755386566,
|
| 1766 |
+
"grad_norm": 0.18938252329826355,
|
| 1767 |
+
"learning_rate": 3.421002061924876e-07,
|
| 1768 |
+
"loss": 0.0201,
|
| 1769 |
+
"num_tokens": 102618890.0,
|
| 1770 |
+
"reward": 1.263178527355194,
|
| 1771 |
+
"reward_std": 0.3987303748726845,
|
| 1772 |
+
"rewards/cosine_scaled_reward": 1.2631784602999687,
|
| 1773 |
+
"rewards/format_reward": 0.0,
|
| 1774 |
+
"step": 126
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"clip_ratio": 0.0,
|
| 1778 |
+
"completion_length": 747.1696701049805,
|
| 1779 |
+
"epoch": 1.29404309252218,
|
| 1780 |
+
"grad_norm": 0.13063876330852509,
|
| 1781 |
+
"learning_rate": 3.3365755880178807e-07,
|
| 1782 |
+
"loss": 0.0312,
|
| 1783 |
+
"num_tokens": 103410058.0,
|
| 1784 |
+
"reward": 1.2827118784189224,
|
| 1785 |
+
"reward_std": 0.3673780020326376,
|
| 1786 |
+
"rewards/cosine_scaled_reward": 1.2827118039131165,
|
| 1787 |
+
"rewards/format_reward": 0.0,
|
| 1788 |
+
"step": 127
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"clip_ratio": 0.0,
|
| 1792 |
+
"completion_length": 796.0893173217773,
|
| 1793 |
+
"epoch": 1.3041825095057034,
|
| 1794 |
+
"grad_norm": 0.12410367280244827,
|
| 1795 |
+
"learning_rate": 3.2526791020045087e-07,
|
| 1796 |
+
"loss": 0.0153,
|
| 1797 |
+
"num_tokens": 104255954.0,
|
| 1798 |
+
"reward": 1.13737141340971,
|
| 1799 |
+
"reward_std": 0.43270402774214745,
|
| 1800 |
+
"rewards/cosine_scaled_reward": 1.1373713836073875,
|
| 1801 |
+
"rewards/format_reward": 0.0,
|
| 1802 |
+
"step": 128
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"clip_ratio": 0.0,
|
| 1806 |
+
"completion_length": 709.1953353881836,
|
| 1807 |
+
"epoch": 1.3143219264892267,
|
| 1808 |
+
"grad_norm": 0.13791009783744812,
|
| 1809 |
+
"learning_rate": 3.169339334358104e-07,
|
| 1810 |
+
"loss": 0.0012,
|
| 1811 |
+
"num_tokens": 105024737.0,
|
| 1812 |
+
"reward": 1.2293454632163048,
|
| 1813 |
+
"reward_std": 0.36772353760898113,
|
| 1814 |
+
"rewards/cosine_scaled_reward": 1.2293454110622406,
|
| 1815 |
+
"rewards/format_reward": 0.0,
|
| 1816 |
+
"step": 129
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"clip_ratio": 0.0,
|
| 1820 |
+
"completion_length": 754.825927734375,
|
| 1821 |
+
"epoch": 1.3244613434727504,
|
| 1822 |
+
"grad_norm": 0.13509666919708252,
|
| 1823 |
+
"learning_rate": 3.086582838174551e-07,
|
| 1824 |
+
"loss": -0.0033,
|
| 1825 |
+
"num_tokens": 105833549.0,
|
| 1826 |
+
"reward": 1.1948751211166382,
|
| 1827 |
+
"reward_std": 0.37703990191221237,
|
| 1828 |
+
"rewards/cosine_scaled_reward": 1.1948750466108322,
|
| 1829 |
+
"rewards/format_reward": 0.0,
|
| 1830 |
+
"step": 130
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"clip_ratio": 0.0,
|
| 1834 |
+
"completion_length": 752.9855194091797,
|
| 1835 |
+
"epoch": 1.3346007604562737,
|
| 1836 |
+
"grad_norm": 0.13409507274627686,
|
| 1837 |
+
"learning_rate": 3.004435980712129e-07,
|
| 1838 |
+
"loss": -0.0299,
|
| 1839 |
+
"num_tokens": 106631688.0,
|
| 1840 |
+
"reward": 1.2489657402038574,
|
| 1841 |
+
"reward_std": 0.3884260356426239,
|
| 1842 |
+
"rewards/cosine_scaled_reward": 1.248965710401535,
|
| 1843 |
+
"rewards/format_reward": 0.0,
|
| 1844 |
+
"step": 131
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"clip_ratio": 0.0,
|
| 1848 |
+
"completion_length": 728.0178909301758,
|
| 1849 |
+
"epoch": 1.3447401774397973,
|
| 1850 |
+
"grad_norm": 0.14096473157405853,
|
| 1851 |
+
"learning_rate": 2.922924934990568e-07,
|
| 1852 |
+
"loss": 0.0421,
|
| 1853 |
+
"num_tokens": 107409144.0,
|
| 1854 |
+
"reward": 1.2679341062903404,
|
| 1855 |
+
"reward_std": 0.4113157168030739,
|
| 1856 |
+
"rewards/cosine_scaled_reward": 1.2679340541362762,
|
| 1857 |
+
"rewards/format_reward": 0.0,
|
| 1858 |
+
"step": 132
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"clip_ratio": 0.0,
|
| 1862 |
+
"completion_length": 708.2924423217773,
|
| 1863 |
+
"epoch": 1.3548795944233207,
|
| 1864 |
+
"grad_norm": 0.13527466356754303,
|
| 1865 |
+
"learning_rate": 2.8420756714519954e-07,
|
| 1866 |
+
"loss": 0.0099,
|
| 1867 |
+
"num_tokens": 108188078.0,
|
| 1868 |
+
"reward": 1.2067532986402512,
|
| 1869 |
+
"reward_std": 0.4394964314997196,
|
| 1870 |
+
"rewards/cosine_scaled_reward": 1.2067532688379288,
|
| 1871 |
+
"rewards/format_reward": 0.0,
|
| 1872 |
+
"step": 133
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"clip_ratio": 0.0,
|
| 1876 |
+
"completion_length": 745.264533996582,
|
| 1877 |
+
"epoch": 1.3650190114068441,
|
| 1878 |
+
"grad_norm": 0.12941937148571014,
|
| 1879 |
+
"learning_rate": 2.7619139496864376e-07,
|
| 1880 |
+
"loss": -0.0117,
|
| 1881 |
+
"num_tokens": 108991819.0,
|
| 1882 |
+
"reward": 1.1342971697449684,
|
| 1883 |
+
"reward_std": 0.35589409433305264,
|
| 1884 |
+
"rewards/cosine_scaled_reward": 1.1342971250414848,
|
| 1885 |
+
"rewards/format_reward": 0.0,
|
| 1886 |
+
"step": 134
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"clip_ratio": 0.0,
|
| 1890 |
+
"completion_length": 690.6328506469727,
|
| 1891 |
+
"epoch": 1.3751584283903675,
|
| 1892 |
+
"grad_norm": 0.1342969536781311,
|
| 1893 |
+
"learning_rate": 2.6824653102244727e-07,
|
| 1894 |
+
"loss": 0.0057,
|
| 1895 |
+
"num_tokens": 109744698.0,
|
| 1896 |
+
"reward": 1.20839624106884,
|
| 1897 |
+
"reward_std": 0.44357743114233017,
|
| 1898 |
+
"rewards/cosine_scaled_reward": 1.208396166563034,
|
| 1899 |
+
"rewards/format_reward": 0.0,
|
| 1900 |
+
"step": 135
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"clip_ratio": 0.0,
|
| 1904 |
+
"completion_length": 739.3605270385742,
|
| 1905 |
+
"epoch": 1.385297845373891,
|
| 1906 |
+
"grad_norm": 0.22325177490711212,
|
| 1907 |
+
"learning_rate": 2.603755066399718e-07,
|
| 1908 |
+
"loss": 0.0066,
|
| 1909 |
+
"num_tokens": 110537245.0,
|
| 1910 |
+
"reward": 1.1991283893585205,
|
| 1911 |
+
"reward_std": 0.4422716274857521,
|
| 1912 |
+
"rewards/cosine_scaled_reward": 1.1991283744573593,
|
| 1913 |
+
"rewards/format_reward": 0.0,
|
| 1914 |
+
"step": 136
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"clip_ratio": 0.0,
|
| 1918 |
+
"completion_length": 771.5045013427734,
|
| 1919 |
+
"epoch": 1.3954372623574145,
|
| 1920 |
+
"grad_norm": 0.131967693567276,
|
| 1921 |
+
"learning_rate": 2.5258082962836614e-07,
|
| 1922 |
+
"loss": -0.0609,
|
| 1923 |
+
"num_tokens": 111363073.0,
|
| 1924 |
+
"reward": 1.0822726637125015,
|
| 1925 |
+
"reward_std": 0.31478939671069384,
|
| 1926 |
+
"rewards/cosine_scaled_reward": 1.082272619009018,
|
| 1927 |
+
"rewards/format_reward": 0.0,
|
| 1928 |
+
"step": 137
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"clip_ratio": 0.0,
|
| 1932 |
+
"completion_length": 745.6908798217773,
|
| 1933 |
+
"epoch": 1.4055766793409379,
|
| 1934 |
+
"grad_norm": 0.13431350886821747,
|
| 1935 |
+
"learning_rate": 2.4486498346955023e-07,
|
| 1936 |
+
"loss": 0.0393,
|
| 1937 |
+
"num_tokens": 112154044.0,
|
| 1938 |
+
"reward": 1.2083490192890167,
|
| 1939 |
+
"reward_std": 0.4234139323234558,
|
| 1940 |
+
"rewards/cosine_scaled_reward": 1.2083489745855331,
|
| 1941 |
+
"rewards/format_reward": 0.0,
|
| 1942 |
+
"step": 138
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"clip_ratio": 0.0,
|
| 1946 |
+
"completion_length": 730.8303909301758,
|
| 1947 |
+
"epoch": 1.4157160963244613,
|
| 1948 |
+
"grad_norm": 0.37255144119262695,
|
| 1949 |
+
"learning_rate": 2.372304265289436e-07,
|
| 1950 |
+
"loss": 0.0129,
|
| 1951 |
+
"num_tokens": 112933164.0,
|
| 1952 |
+
"reward": 1.14224873483181,
|
| 1953 |
+
"reward_std": 0.47245585173368454,
|
| 1954 |
+
"rewards/cosine_scaled_reward": 1.142248660326004,
|
| 1955 |
+
"rewards/format_reward": 0.0,
|
| 1956 |
+
"step": 139
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"clip_ratio": 0.0,
|
| 1960 |
+
"completion_length": 723.6975708007812,
|
| 1961 |
+
"epoch": 1.4258555133079849,
|
| 1962 |
+
"grad_norm": 0.13437899947166443,
|
| 1963 |
+
"learning_rate": 2.2967959127220137e-07,
|
| 1964 |
+
"loss": 0.0149,
|
| 1965 |
+
"num_tokens": 113703557.0,
|
| 1966 |
+
"reward": 1.2409879192709923,
|
| 1967 |
+
"reward_std": 0.39572571590542793,
|
| 1968 |
+
"rewards/cosine_scaled_reward": 1.240987867116928,
|
| 1969 |
+
"rewards/format_reward": 0.0,
|
| 1970 |
+
"step": 140
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"clip_ratio": 0.0,
|
| 1974 |
+
"completion_length": 715.0223617553711,
|
| 1975 |
+
"epoch": 1.4359949302915083,
|
| 1976 |
+
"grad_norm": 0.5516566634178162,
|
| 1977 |
+
"learning_rate": 2.2221488349019902e-07,
|
| 1978 |
+
"loss": -0.0049,
|
| 1979 |
+
"num_tokens": 114492577.0,
|
| 1980 |
+
"reward": 1.1441184803843498,
|
| 1981 |
+
"reward_std": 0.4257240351289511,
|
| 1982 |
+
"rewards/cosine_scaled_reward": 1.1441184356808662,
|
| 1983 |
+
"rewards/format_reward": 0.0,
|
| 1984 |
+
"step": 141
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"clip_ratio": 0.0,
|
| 1988 |
+
"completion_length": 683.0033798217773,
|
| 1989 |
+
"epoch": 1.4461343472750317,
|
| 1990 |
+
"grad_norm": 0.135258749127388,
|
| 1991 |
+
"learning_rate": 2.1483868153251788e-07,
|
| 1992 |
+
"loss": 0.0098,
|
| 1993 |
+
"num_tokens": 115236092.0,
|
| 1994 |
+
"reward": 1.2539073526859283,
|
| 1995 |
+
"reward_std": 0.42142050713300705,
|
| 1996 |
+
"rewards/cosine_scaled_reward": 1.2539073079824448,
|
| 1997 |
+
"rewards/format_reward": 0.0,
|
| 1998 |
+
"step": 142
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"clip_ratio": 0.0,
|
| 2002 |
+
"completion_length": 783.0201263427734,
|
| 2003 |
+
"epoch": 1.456273764258555,
|
| 2004 |
+
"grad_norm": 0.17022480070590973,
|
| 2005 |
+
"learning_rate": 2.0755333554967346e-07,
|
| 2006 |
+
"loss": -0.0005,
|
| 2007 |
+
"num_tokens": 116074878.0,
|
| 2008 |
+
"reward": 1.0983483716845512,
|
| 2009 |
+
"reward_std": 0.42650508880615234,
|
| 2010 |
+
"rewards/cosine_scaled_reward": 1.0983483269810677,
|
| 2011 |
+
"rewards/format_reward": 0.0,
|
| 2012 |
+
"step": 143
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"clip_ratio": 0.0,
|
| 2016 |
+
"completion_length": 774.2790603637695,
|
| 2017 |
+
"epoch": 1.4664131812420786,
|
| 2018 |
+
"grad_norm": 0.1283206343650818,
|
| 2019 |
+
"learning_rate": 2.0036116674432652e-07,
|
| 2020 |
+
"loss": -0.0156,
|
| 2021 |
+
"num_tokens": 116905080.0,
|
| 2022 |
+
"reward": 1.1725050881505013,
|
| 2023 |
+
"reward_std": 0.35209081321954727,
|
| 2024 |
+
"rewards/cosine_scaled_reward": 1.172505035996437,
|
| 2025 |
+
"rewards/format_reward": 0.0,
|
| 2026 |
+
"step": 144
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"clip_ratio": 0.0,
|
| 2030 |
+
"completion_length": 658.7600746154785,
|
| 2031 |
+
"epoch": 1.476552598225602,
|
| 2032 |
+
"grad_norm": 0.14014765620231628,
|
| 2033 |
+
"learning_rate": 1.9326446663172035e-07,
|
| 2034 |
+
"loss": 0.0474,
|
| 2035 |
+
"num_tokens": 117626809.0,
|
| 2036 |
+
"reward": 1.341113954782486,
|
| 2037 |
+
"reward_std": 0.4142299462109804,
|
| 2038 |
+
"rewards/cosine_scaled_reward": 1.3411139100790024,
|
| 2039 |
+
"rewards/format_reward": 0.0,
|
| 2040 |
+
"step": 145
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"clip_ratio": 0.0,
|
| 2044 |
+
"completion_length": 679.3192367553711,
|
| 2045 |
+
"epoch": 1.4866920152091254,
|
| 2046 |
+
"grad_norm": 0.15331801772117615,
|
| 2047 |
+
"learning_rate": 1.8626549630957395e-07,
|
| 2048 |
+
"loss": 0.0136,
|
| 2049 |
+
"num_tokens": 118363623.0,
|
| 2050 |
+
"reward": 1.19945827126503,
|
| 2051 |
+
"reward_std": 0.4244145043194294,
|
| 2052 |
+
"rewards/cosine_scaled_reward": 1.1994582116603851,
|
| 2053 |
+
"rewards/format_reward": 0.0,
|
| 2054 |
+
"step": 146
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"clip_ratio": 0.0,
|
| 2058 |
+
"completion_length": 752.2890930175781,
|
| 2059 |
+
"epoch": 1.496831432192649,
|
| 2060 |
+
"grad_norm": 0.13101238012313843,
|
| 2061 |
+
"learning_rate": 1.7936648573766954e-07,
|
| 2062 |
+
"loss": -0.0056,
|
| 2063 |
+
"num_tokens": 119174386.0,
|
| 2064 |
+
"reward": 1.2080383449792862,
|
| 2065 |
+
"reward_std": 0.4013860188424587,
|
| 2066 |
+
"rewards/cosine_scaled_reward": 1.2080383002758026,
|
| 2067 |
+
"rewards/format_reward": 0.0,
|
| 2068 |
+
"step": 147
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"clip_ratio": 0.0,
|
| 2072 |
+
"completion_length": 762.8594131469727,
|
| 2073 |
+
"epoch": 1.5069708491761724,
|
| 2074 |
+
"grad_norm": 0.12902143597602844,
|
| 2075 |
+
"learning_rate": 1.725696330273575e-07,
|
| 2076 |
+
"loss": -0.0383,
|
| 2077 |
+
"num_tokens": 119988252.0,
|
| 2078 |
+
"reward": 1.169223740696907,
|
| 2079 |
+
"reward_std": 0.39561279490590096,
|
| 2080 |
+
"rewards/cosine_scaled_reward": 1.169223703444004,
|
| 2081 |
+
"rewards/format_reward": 0.0,
|
| 2082 |
+
"step": 148
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"clip_ratio": 0.0,
|
| 2086 |
+
"completion_length": 700.4531478881836,
|
| 2087 |
+
"epoch": 1.5171102661596958,
|
| 2088 |
+
"grad_norm": 0.1390618234872818,
|
| 2089 |
+
"learning_rate": 1.65877103741212e-07,
|
| 2090 |
+
"loss": 0.0519,
|
| 2091 |
+
"num_tokens": 120740586.0,
|
| 2092 |
+
"reward": 1.37008798122406,
|
| 2093 |
+
"reward_std": 0.3735492154955864,
|
| 2094 |
+
"rewards/cosine_scaled_reward": 1.3700879365205765,
|
| 2095 |
+
"rewards/format_reward": 0.0,
|
| 2096 |
+
"step": 149
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"clip_ratio": 0.0,
|
| 2100 |
+
"completion_length": 706.7366485595703,
|
| 2101 |
+
"epoch": 1.5272496831432192,
|
| 2102 |
+
"grad_norm": 0.13693000376224518,
|
| 2103 |
+
"learning_rate": 1.592910302030544e-07,
|
| 2104 |
+
"loss": 0.0186,
|
| 2105 |
+
"num_tokens": 121493126.0,
|
| 2106 |
+
"reward": 1.1771222800016403,
|
| 2107 |
+
"reward_std": 0.4693867526948452,
|
| 2108 |
+
"rewards/cosine_scaled_reward": 1.1771221980452538,
|
| 2109 |
+
"rewards/format_reward": 0.0,
|
| 2110 |
+
"step": 150
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"clip_ratio": 0.0,
|
| 2114 |
+
"completion_length": 725.279052734375,
|
| 2115 |
+
"epoch": 1.5373891001267426,
|
| 2116 |
+
"grad_norm": 0.16313283145427704,
|
| 2117 |
+
"learning_rate": 1.5281351081856976e-07,
|
| 2118 |
+
"loss": 0.0023,
|
| 2119 |
+
"num_tokens": 122274232.0,
|
| 2120 |
+
"reward": 1.203133448958397,
|
| 2121 |
+
"reward_std": 0.41216023825109005,
|
| 2122 |
+
"rewards/cosine_scaled_reward": 1.2031333893537521,
|
| 2123 |
+
"rewards/format_reward": 0.0,
|
| 2124 |
+
"step": 151
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"clip_ratio": 0.0,
|
| 2128 |
+
"completion_length": 703.8917770385742,
|
| 2129 |
+
"epoch": 1.5475285171102662,
|
| 2130 |
+
"grad_norm": 0.16185523569583893,
|
| 2131 |
+
"learning_rate": 1.4644660940672627e-07,
|
| 2132 |
+
"loss": -0.0126,
|
| 2133 |
+
"num_tokens": 123030511.0,
|
| 2134 |
+
"reward": 1.2789395153522491,
|
| 2135 |
+
"reward_std": 0.382132101804018,
|
| 2136 |
+
"rewards/cosine_scaled_reward": 1.2789394706487656,
|
| 2137 |
+
"rewards/format_reward": 0.0,
|
| 2138 |
+
"step": 152
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"clip_ratio": 0.0,
|
| 2142 |
+
"completion_length": 690.9888610839844,
|
| 2143 |
+
"epoch": 1.5576679340937896,
|
| 2144 |
+
"grad_norm": 0.6896323561668396,
|
| 2145 |
+
"learning_rate": 1.4019235454221856e-07,
|
| 2146 |
+
"loss": -0.0251,
|
| 2147 |
+
"num_tokens": 123783709.0,
|
| 2148 |
+
"reward": 1.230302333831787,
|
| 2149 |
+
"reward_std": 0.3682190030813217,
|
| 2150 |
+
"rewards/cosine_scaled_reward": 1.2303023040294647,
|
| 2151 |
+
"rewards/format_reward": 0.0,
|
| 2152 |
+
"step": 153
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"clip_ratio": 0.0,
|
| 2156 |
+
"completion_length": 702.8527069091797,
|
| 2157 |
+
"epoch": 1.5678073510773132,
|
| 2158 |
+
"grad_norm": 0.14129234850406647,
|
| 2159 |
+
"learning_rate": 1.3405273890913737e-07,
|
| 2160 |
+
"loss": 0.0005,
|
| 2161 |
+
"num_tokens": 124535481.0,
|
| 2162 |
+
"reward": 1.2051593363285065,
|
| 2163 |
+
"reward_std": 0.4200758896768093,
|
| 2164 |
+
"rewards/cosine_scaled_reward": 1.205159269273281,
|
| 2165 |
+
"rewards/format_reward": 0.0,
|
| 2166 |
+
"step": 154
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"clip_ratio": 0.0,
|
| 2170 |
+
"completion_length": 710.2265853881836,
|
| 2171 |
+
"epoch": 1.5779467680608366,
|
| 2172 |
+
"grad_norm": 0.1381485015153885,
|
| 2173 |
+
"learning_rate": 1.280297186660752e-07,
|
| 2174 |
+
"loss": 0.0181,
|
| 2175 |
+
"num_tokens": 125290220.0,
|
| 2176 |
+
"reward": 1.2739061638712883,
|
| 2177 |
+
"reward_std": 0.4085957519710064,
|
| 2178 |
+
"rewards/cosine_scaled_reward": 1.2739060744643211,
|
| 2179 |
+
"rewards/format_reward": 0.0,
|
| 2180 |
+
"step": 155
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"clip_ratio": 0.0,
|
| 2184 |
+
"completion_length": 808.0145492553711,
|
| 2185 |
+
"epoch": 1.58808618504436,
|
| 2186 |
+
"grad_norm": 0.13849686086177826,
|
| 2187 |
+
"learning_rate": 1.2212521282287093e-07,
|
| 2188 |
+
"loss": -0.005,
|
| 2189 |
+
"num_tokens": 126134129.0,
|
| 2190 |
+
"reward": 1.175368033349514,
|
| 2191 |
+
"reward_std": 0.3809916824102402,
|
| 2192 |
+
"rewards/cosine_scaled_reward": 1.1753679811954498,
|
| 2193 |
+
"rewards/format_reward": 0.0,
|
| 2194 |
+
"step": 156
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"clip_ratio": 0.0,
|
| 2198 |
+
"completion_length": 754.8638763427734,
|
| 2199 |
+
"epoch": 1.5982256020278833,
|
| 2200 |
+
"grad_norm": 0.14109836518764496,
|
| 2201 |
+
"learning_rate": 1.1634110262918717e-07,
|
| 2202 |
+
"loss": -0.0008,
|
| 2203 |
+
"num_tokens": 126954495.0,
|
| 2204 |
+
"reward": 1.2158635556697845,
|
| 2205 |
+
"reward_std": 0.34099204279482365,
|
| 2206 |
+
"rewards/cosine_scaled_reward": 1.2158635258674622,
|
| 2207 |
+
"rewards/format_reward": 0.0,
|
| 2208 |
+
"step": 157
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"clip_ratio": 0.0,
|
| 2212 |
+
"completion_length": 775.9598541259766,
|
| 2213 |
+
"epoch": 1.6083650190114067,
|
| 2214 |
+
"grad_norm": 0.12850748002529144,
|
| 2215 |
+
"learning_rate": 1.1067923097512255e-07,
|
| 2216 |
+
"loss": -0.0005,
|
| 2217 |
+
"num_tokens": 127780467.0,
|
| 2218 |
+
"reward": 1.1476298496127129,
|
| 2219 |
+
"reward_std": 0.4081256799399853,
|
| 2220 |
+
"rewards/cosine_scaled_reward": 1.1476298123598099,
|
| 2221 |
+
"rewards/format_reward": 0.0,
|
| 2222 |
+
"step": 158
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"clip_ratio": 0.0,
|
| 2226 |
+
"completion_length": 753.6797256469727,
|
| 2227 |
+
"epoch": 1.6185044359949303,
|
| 2228 |
+
"grad_norm": 0.36824139952659607,
|
| 2229 |
+
"learning_rate": 1.0514140180404202e-07,
|
| 2230 |
+
"loss": 0.0054,
|
| 2231 |
+
"num_tokens": 128584820.0,
|
| 2232 |
+
"reward": 1.0943407788872719,
|
| 2233 |
+
"reward_std": 0.41071633249521255,
|
| 2234 |
+
"rewards/cosine_scaled_reward": 1.0943407267332077,
|
| 2235 |
+
"rewards/format_reward": 0.0,
|
| 2236 |
+
"step": 159
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"clip_ratio": 0.0,
|
| 2240 |
+
"completion_length": 821.4553909301758,
|
| 2241 |
+
"epoch": 1.6286438529784537,
|
| 2242 |
+
"grad_norm": 0.15713122487068176,
|
| 2243 |
+
"learning_rate": 9.972937953781984e-08,
|
| 2244 |
+
"loss": 0.0185,
|
| 2245 |
+
"num_tokens": 129467812.0,
|
| 2246 |
+
"reward": 1.086461715400219,
|
| 2247 |
+
"reward_std": 0.4641183950006962,
|
| 2248 |
+
"rewards/cosine_scaled_reward": 1.0864616632461548,
|
| 2249 |
+
"rewards/format_reward": 0.0,
|
| 2250 |
+
"step": 160
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"clip_ratio": 0.0,
|
| 2254 |
+
"completion_length": 705.2288208007812,
|
| 2255 |
+
"epoch": 1.6387832699619773,
|
| 2256 |
+
"grad_norm": 0.13201545178890228,
|
| 2257 |
+
"learning_rate": 9.444488851467041e-08,
|
| 2258 |
+
"loss": 0.0098,
|
| 2259 |
+
"num_tokens": 130217369.0,
|
| 2260 |
+
"reward": 1.2702905237674713,
|
| 2261 |
+
"reward_std": 0.3846561126410961,
|
| 2262 |
+
"rewards/cosine_scaled_reward": 1.2702905088663101,
|
| 2263 |
+
"rewards/format_reward": 0.0,
|
| 2264 |
+
"step": 161
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"clip_ratio": 0.0,
|
| 2268 |
+
"completion_length": 732.4542770385742,
|
| 2269 |
+
"epoch": 1.6489226869455007,
|
| 2270 |
+
"grad_norm": 0.12551482021808624,
|
| 2271 |
+
"learning_rate": 8.928961243975436e-08,
|
| 2272 |
+
"loss": 0.0025,
|
| 2273 |
+
"num_tokens": 131006504.0,
|
| 2274 |
+
"reward": 1.1392531916499138,
|
| 2275 |
+
"reward_std": 0.4202544465661049,
|
| 2276 |
+
"rewards/cosine_scaled_reward": 1.150568701326847,
|
| 2277 |
+
"rewards/format_reward": 0.0,
|
| 2278 |
+
"step": 162
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"clip_ratio": 0.0,
|
| 2282 |
+
"completion_length": 690.4821739196777,
|
| 2283 |
+
"epoch": 1.659062103929024,
|
| 2284 |
+
"grad_norm": 0.14846356213092804,
|
| 2285 |
+
"learning_rate": 8.426519384872732e-08,
|
| 2286 |
+
"loss": 0.0151,
|
| 2287 |
+
"num_tokens": 131747312.0,
|
| 2288 |
+
"reward": 1.2898453325033188,
|
| 2289 |
+
"reward_std": 0.3518759645521641,
|
| 2290 |
+
"rewards/cosine_scaled_reward": 1.2898452877998352,
|
| 2291 |
+
"rewards/format_reward": 0.0,
|
| 2292 |
+
"step": 163
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"clip_ratio": 0.0,
|
| 2296 |
+
"completion_length": 749.0937881469727,
|
| 2297 |
+
"epoch": 1.6692015209125475,
|
| 2298 |
+
"grad_norm": 0.13933837413787842,
|
| 2299 |
+
"learning_rate": 7.937323358440934e-08,
|
| 2300 |
+
"loss": 0.028,
|
| 2301 |
+
"num_tokens": 132549532.0,
|
| 2302 |
+
"reward": 1.1977392584085464,
|
| 2303 |
+
"reward_std": 0.43290739692747593,
|
| 2304 |
+
"rewards/cosine_scaled_reward": 1.1977391764521599,
|
| 2305 |
+
"rewards/format_reward": 0.0,
|
| 2306 |
+
"step": 164
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"clip_ratio": 0.0,
|
| 2310 |
+
"completion_length": 750.6361999511719,
|
| 2311 |
+
"epoch": 1.6793409378960709,
|
| 2312 |
+
"grad_norm": 0.1263691633939743,
|
| 2313 |
+
"learning_rate": 7.461529028673463e-08,
|
| 2314 |
+
"loss": 0.0414,
|
| 2315 |
+
"num_tokens": 133354910.0,
|
| 2316 |
+
"reward": 1.2933627367019653,
|
| 2317 |
+
"reward_std": 0.41812827065587044,
|
| 2318 |
+
"rewards/cosine_scaled_reward": 1.2933626621961594,
|
| 2319 |
+
"rewards/format_reward": 0.0,
|
| 2320 |
+
"step": 165
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"clip_ratio": 0.0,
|
| 2324 |
+
"completion_length": 748.4531555175781,
|
| 2325 |
+
"epoch": 1.6894803548795945,
|
| 2326 |
+
"grad_norm": 0.1337745487689972,
|
| 2327 |
+
"learning_rate": 6.999287989614971e-08,
|
| 2328 |
+
"loss": 0.0003,
|
| 2329 |
+
"num_tokens": 134157964.0,
|
| 2330 |
+
"reward": 1.1261718794703484,
|
| 2331 |
+
"reward_std": 0.43879349157214165,
|
| 2332 |
+
"rewards/cosine_scaled_reward": 1.1261718571186066,
|
| 2333 |
+
"rewards/format_reward": 0.0,
|
| 2334 |
+
"step": 166
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"clip_ratio": 0.0,
|
| 2338 |
+
"completion_length": 684.155158996582,
|
| 2339 |
+
"epoch": 1.6996197718631179,
|
| 2340 |
+
"grad_norm": 0.14675858616828918,
|
| 2341 |
+
"learning_rate": 6.550747517061656e-08,
|
| 2342 |
+
"loss": 0.0135,
|
| 2343 |
+
"num_tokens": 134897511.0,
|
| 2344 |
+
"reward": 1.31067955493927,
|
| 2345 |
+
"reward_std": 0.4250149428844452,
|
| 2346 |
+
"rewards/cosine_scaled_reward": 1.3106794953346252,
|
| 2347 |
+
"rewards/format_reward": 0.0,
|
| 2348 |
+
"step": 167
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"clip_ratio": 0.0,
|
| 2352 |
+
"completion_length": 708.9955825805664,
|
| 2353 |
+
"epoch": 1.7097591888466415,
|
| 2354 |
+
"grad_norm": 0.1361108422279358,
|
| 2355 |
+
"learning_rate": 6.116050521637218e-08,
|
| 2356 |
+
"loss": 0.0403,
|
| 2357 |
+
"num_tokens": 135661099.0,
|
| 2358 |
+
"reward": 1.2582149505615234,
|
| 2359 |
+
"reward_std": 0.4354391284286976,
|
| 2360 |
+
"rewards/cosine_scaled_reward": 1.2582148760557175,
|
| 2361 |
+
"rewards/format_reward": 0.0,
|
| 2362 |
+
"step": 168
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"clip_ratio": 0.0,
|
| 2366 |
+
"completion_length": 708.2154312133789,
|
| 2367 |
+
"epoch": 1.7198986058301649,
|
| 2368 |
+
"grad_norm": 0.13875119388103485,
|
| 2369 |
+
"learning_rate": 5.6953355032598795e-08,
|
| 2370 |
+
"loss": -0.0346,
|
| 2371 |
+
"num_tokens": 136432732.0,
|
| 2372 |
+
"reward": 1.1267788708209991,
|
| 2373 |
+
"reward_std": 0.43002479895949364,
|
| 2374 |
+
"rewards/cosine_scaled_reward": 1.126778818666935,
|
| 2375 |
+
"rewards/format_reward": 0.0,
|
| 2376 |
+
"step": 169
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"clip_ratio": 0.0,
|
| 2380 |
+
"completion_length": 729.7221374511719,
|
| 2381 |
+
"epoch": 1.7300380228136882,
|
| 2382 |
+
"grad_norm": 0.136752188205719,
|
| 2383 |
+
"learning_rate": 5.288736507014435e-08,
|
| 2384 |
+
"loss": -0.0081,
|
| 2385 |
+
"num_tokens": 137210563.0,
|
| 2386 |
+
"reward": 1.251162275671959,
|
| 2387 |
+
"reward_std": 0.36924222111701965,
|
| 2388 |
+
"rewards/cosine_scaled_reward": 1.2511622309684753,
|
| 2389 |
+
"rewards/format_reward": 0.0,
|
| 2390 |
+
"step": 170
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"clip_ratio": 0.0,
|
| 2394 |
+
"completion_length": 707.0078430175781,
|
| 2395 |
+
"epoch": 1.7401774397972116,
|
| 2396 |
+
"grad_norm": 0.132461279630661,
|
| 2397 |
+
"learning_rate": 4.896383080443933e-08,
|
| 2398 |
+
"loss": 0.0231,
|
| 2399 |
+
"num_tokens": 137968794.0,
|
| 2400 |
+
"reward": 1.2026441097259521,
|
| 2401 |
+
"reward_std": 0.42860128730535507,
|
| 2402 |
+
"rewards/cosine_scaled_reward": 1.202644057571888,
|
| 2403 |
+
"rewards/format_reward": 0.0,
|
| 2404 |
+
"step": 171
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"clip_ratio": 0.0,
|
| 2408 |
+
"completion_length": 690.8460159301758,
|
| 2409 |
+
"epoch": 1.750316856780735,
|
| 2410 |
+
"grad_norm": 0.14912405610084534,
|
| 2411 |
+
"learning_rate": 4.518400232274078e-08,
|
| 2412 |
+
"loss": 0.0096,
|
| 2413 |
+
"num_tokens": 138718296.0,
|
| 2414 |
+
"reward": 1.188114494085312,
|
| 2415 |
+
"reward_std": 0.39624364115297794,
|
| 2416 |
+
"rewards/cosine_scaled_reward": 1.2003108784556389,
|
| 2417 |
+
"rewards/format_reward": 0.0,
|
| 2418 |
+
"step": 172
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"clip_ratio": 0.0,
|
| 2422 |
+
"completion_length": 684.9765930175781,
|
| 2423 |
+
"epoch": 1.7604562737642584,
|
| 2424 |
+
"grad_norm": 0.14246663451194763,
|
| 2425 |
+
"learning_rate": 4.1549083925840165e-08,
|
| 2426 |
+
"loss": 0.0125,
|
| 2427 |
+
"num_tokens": 139449483.0,
|
| 2428 |
+
"reward": 1.2953269332647324,
|
| 2429 |
+
"reward_std": 0.4603438973426819,
|
| 2430 |
+
"rewards/cosine_scaled_reward": 1.29532690346241,
|
| 2431 |
+
"rewards/format_reward": 0.0,
|
| 2432 |
+
"step": 173
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"clip_ratio": 0.0,
|
| 2436 |
+
"completion_length": 685.9721298217773,
|
| 2437 |
+
"epoch": 1.770595690747782,
|
| 2438 |
+
"grad_norm": 0.1809549778699875,
|
| 2439 |
+
"learning_rate": 3.806023374435663e-08,
|
| 2440 |
+
"loss": -0.0066,
|
| 2441 |
+
"num_tokens": 140192642.0,
|
| 2442 |
+
"reward": 1.18511962890625,
|
| 2443 |
+
"reward_std": 0.365385384298861,
|
| 2444 |
+
"rewards/cosine_scaled_reward": 1.1851196065545082,
|
| 2445 |
+
"rewards/format_reward": 0.0,
|
| 2446 |
+
"step": 174
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"clip_ratio": 0.0,
|
| 2450 |
+
"completion_length": 718.2578353881836,
|
| 2451 |
+
"epoch": 1.7807351077313056,
|
| 2452 |
+
"grad_norm": 2.9148614406585693,
|
| 2453 |
+
"learning_rate": 3.4718563369743213e-08,
|
| 2454 |
+
"loss": 0.0077,
|
| 2455 |
+
"num_tokens": 140969529.0,
|
| 2456 |
+
"reward": 1.1897177621722221,
|
| 2457 |
+
"reward_std": 0.44410305470228195,
|
| 2458 |
+
"rewards/cosine_scaled_reward": 1.1897177323698997,
|
| 2459 |
+
"rewards/format_reward": 0.0,
|
| 2460 |
+
"step": 175
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"clip_ratio": 0.0,
|
| 2464 |
+
"completion_length": 672.2578392028809,
|
| 2465 |
+
"epoch": 1.790874524714829,
|
| 2466 |
+
"grad_norm": 0.14049698412418365,
|
| 2467 |
+
"learning_rate": 3.15251375001192e-08,
|
| 2468 |
+
"loss": 0.0167,
|
| 2469 |
+
"num_tokens": 141711976.0,
|
| 2470 |
+
"reward": 1.2624182030558586,
|
| 2471 |
+
"reward_std": 0.45537084713578224,
|
| 2472 |
+
"rewards/cosine_scaled_reward": 1.2624181509017944,
|
| 2473 |
+
"rewards/format_reward": 0.0,
|
| 2474 |
+
"step": 176
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"clip_ratio": 0.0,
|
| 2478 |
+
"completion_length": 697.0145378112793,
|
| 2479 |
+
"epoch": 1.8010139416983524,
|
| 2480 |
+
"grad_norm": 0.22702006995677948,
|
| 2481 |
+
"learning_rate": 2.8480973601043955e-08,
|
| 2482 |
+
"loss": -0.01,
|
| 2483 |
+
"num_tokens": 142471309.0,
|
| 2484 |
+
"reward": 1.249974675476551,
|
| 2485 |
+
"reward_std": 0.3576275184750557,
|
| 2486 |
+
"rewards/cosine_scaled_reward": 1.2499746307730675,
|
| 2487 |
+
"rewards/format_reward": 0.0,
|
| 2488 |
+
"step": 177
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"clip_ratio": 0.0,
|
| 2492 |
+
"completion_length": 744.4810409545898,
|
| 2493 |
+
"epoch": 1.8111533586818758,
|
| 2494 |
+
"grad_norm": 0.12813909351825714,
|
| 2495 |
+
"learning_rate": 2.558704158134023e-08,
|
| 2496 |
+
"loss": -0.0078,
|
| 2497 |
+
"num_tokens": 143278388.0,
|
| 2498 |
+
"reward": 1.1319350376725197,
|
| 2499 |
+
"reward_std": 0.41413314267992973,
|
| 2500 |
+
"rewards/cosine_scaled_reward": 1.1319349706172943,
|
| 2501 |
+
"rewards/format_reward": 0.0,
|
| 2502 |
+
"step": 178
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"clip_ratio": 0.0,
|
| 2506 |
+
"completion_length": 750.3560638427734,
|
| 2507 |
+
"epoch": 1.8212927756653992,
|
| 2508 |
+
"grad_norm": 0.13594254851341248,
|
| 2509 |
+
"learning_rate": 2.2844263484068093e-08,
|
| 2510 |
+
"loss": 0.0003,
|
| 2511 |
+
"num_tokens": 144077707.0,
|
| 2512 |
+
"reward": 1.2112557888031006,
|
| 2513 |
+
"reward_std": 0.3645152598619461,
|
| 2514 |
+
"rewards/cosine_scaled_reward": 1.2112557291984558,
|
| 2515 |
+
"rewards/format_reward": 0.0,
|
| 2516 |
+
"step": 179
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"clip_ratio": 0.0,
|
| 2520 |
+
"completion_length": 682.1417694091797,
|
| 2521 |
+
"epoch": 1.8314321926489225,
|
| 2522 |
+
"grad_norm": 0.15148992836475372,
|
| 2523 |
+
"learning_rate": 2.025351319275137e-08,
|
| 2524 |
+
"loss": -0.0027,
|
| 2525 |
+
"num_tokens": 144819874.0,
|
| 2526 |
+
"reward": 1.2360253632068634,
|
| 2527 |
+
"reward_std": 0.43914780393242836,
|
| 2528 |
+
"rewards/cosine_scaled_reward": 1.2360252812504768,
|
| 2529 |
+
"rewards/format_reward": 0.0,
|
| 2530 |
+
"step": 180
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"clip_ratio": 0.0,
|
| 2534 |
+
"completion_length": 724.106071472168,
|
| 2535 |
+
"epoch": 1.8415716096324461,
|
| 2536 |
+
"grad_norm": 0.14325113594532013,
|
| 2537 |
+
"learning_rate": 1.781561615294652e-08,
|
| 2538 |
+
"loss": 0.0052,
|
| 2539 |
+
"num_tokens": 145603137.0,
|
| 2540 |
+
"reward": 1.226213201880455,
|
| 2541 |
+
"reward_std": 0.4393893778324127,
|
| 2542 |
+
"rewards/cosine_scaled_reward": 1.2262131571769714,
|
| 2543 |
+
"rewards/format_reward": 0.0,
|
| 2544 |
+
"step": 181
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"clip_ratio": 0.0,
|
| 2548 |
+
"completion_length": 782.7399749755859,
|
| 2549 |
+
"epoch": 1.8517110266159695,
|
| 2550 |
+
"grad_norm": 0.1272156685590744,
|
| 2551 |
+
"learning_rate": 1.553134910924636e-08,
|
| 2552 |
+
"loss": -0.0251,
|
| 2553 |
+
"num_tokens": 146440576.0,
|
| 2554 |
+
"reward": 1.0694795548915863,
|
| 2555 |
+
"reward_std": 0.3888841047883034,
|
| 2556 |
+
"rewards/cosine_scaled_reward": 1.0694795101881027,
|
| 2557 |
+
"rewards/format_reward": 0.0,
|
| 2558 |
+
"step": 182
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"clip_ratio": 0.0,
|
| 2562 |
+
"completion_length": 677.8895492553711,
|
| 2563 |
+
"epoch": 1.8618504435994931,
|
| 2564 |
+
"grad_norm": 0.14028111100196838,
|
| 2565 |
+
"learning_rate": 1.340143985779829e-08,
|
| 2566 |
+
"loss": 0.0087,
|
| 2567 |
+
"num_tokens": 147174725.0,
|
| 2568 |
+
"reward": 1.2450221478939056,
|
| 2569 |
+
"reward_std": 0.3928499836474657,
|
| 2570 |
+
"rewards/cosine_scaled_reward": 1.2450221329927444,
|
| 2571 |
+
"rewards/format_reward": 0.0,
|
| 2572 |
+
"step": 183
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"clip_ratio": 0.0,
|
| 2576 |
+
"completion_length": 698.4096298217773,
|
| 2577 |
+
"epoch": 1.8719898605830165,
|
| 2578 |
+
"grad_norm": 0.1386420577764511,
|
| 2579 |
+
"learning_rate": 1.1426567014420297e-08,
|
| 2580 |
+
"loss": -0.001,
|
| 2581 |
+
"num_tokens": 147927204.0,
|
| 2582 |
+
"reward": 1.2862959504127502,
|
| 2583 |
+
"reward_std": 0.39938454143702984,
|
| 2584 |
+
"rewards/cosine_scaled_reward": 1.2862958759069443,
|
| 2585 |
+
"rewards/format_reward": 0.0,
|
| 2586 |
+
"step": 184
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"clip_ratio": 0.0,
|
| 2590 |
+
"completion_length": 749.4877471923828,
|
| 2591 |
+
"epoch": 1.88212927756654,
|
| 2592 |
+
"grad_norm": 0.26094314455986023,
|
| 2593 |
+
"learning_rate": 9.607359798384784e-09,
|
| 2594 |
+
"loss": 0.0099,
|
| 2595 |
+
"num_tokens": 148714225.0,
|
| 2596 |
+
"reward": 1.15062565356493,
|
| 2597 |
+
"reward_std": 0.49165763705968857,
|
| 2598 |
+
"rewards/cosine_scaled_reward": 1.150625616312027,
|
| 2599 |
+
"rewards/format_reward": 0.0,
|
| 2600 |
+
"step": 185
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"clip_ratio": 0.0,
|
| 2604 |
+
"completion_length": 701.1071624755859,
|
| 2605 |
+
"epoch": 1.8922686945500633,
|
| 2606 |
+
"grad_norm": 0.14008425176143646,
|
| 2607 |
+
"learning_rate": 7.944397831941951e-09,
|
| 2608 |
+
"loss": 0.0034,
|
| 2609 |
+
"num_tokens": 149488913.0,
|
| 2610 |
+
"reward": 1.181966707110405,
|
| 2611 |
+
"reward_std": 0.4252682514488697,
|
| 2612 |
+
"rewards/cosine_scaled_reward": 1.181966632604599,
|
| 2613 |
+
"rewards/format_reward": 0.0,
|
| 2614 |
+
"step": 186
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"clip_ratio": 0.0,
|
| 2618 |
+
"completion_length": 779.9766082763672,
|
| 2619 |
+
"epoch": 1.9024081115335867,
|
| 2620 |
+
"grad_norm": 0.16207562386989594,
|
| 2621 |
+
"learning_rate": 6.438210955644452e-09,
|
| 2622 |
+
"loss": -0.0089,
|
| 2623 |
+
"num_tokens": 150330196.0,
|
| 2624 |
+
"reward": 1.0697861537337303,
|
| 2625 |
+
"reward_std": 0.3918624483048916,
|
| 2626 |
+
"rewards/cosine_scaled_reward": 1.0697861164808273,
|
| 2627 |
+
"rewards/format_reward": 0.0,
|
| 2628 |
+
"step": 187
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"clip_ratio": 0.0,
|
| 2632 |
+
"completion_length": 725.6529235839844,
|
| 2633 |
+
"epoch": 1.9125475285171103,
|
| 2634 |
+
"grad_norm": 0.1371779590845108,
|
| 2635 |
+
"learning_rate": 5.0892790595336575e-09,
|
| 2636 |
+
"loss": 0.0236,
|
| 2637 |
+
"num_tokens": 151117397.0,
|
| 2638 |
+
"reward": 1.2266732975840569,
|
| 2639 |
+
"reward_std": 0.4197930619120598,
|
| 2640 |
+
"rewards/cosine_scaled_reward": 1.226673237979412,
|
| 2641 |
+
"rewards/format_reward": 0.0,
|
| 2642 |
+
"step": 188
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"clip_ratio": 0.0,
|
| 2646 |
+
"completion_length": 735.1306228637695,
|
| 2647 |
+
"epoch": 1.9226869455006337,
|
| 2648 |
+
"grad_norm": 0.14734813570976257,
|
| 2649 |
+
"learning_rate": 3.898031930240797e-09,
|
| 2650 |
+
"loss": -0.013,
|
| 2651 |
+
"num_tokens": 151901074.0,
|
| 2652 |
+
"reward": 1.1564877182245255,
|
| 2653 |
+
"reward_std": 0.4156847130507231,
|
| 2654 |
+
"rewards/cosine_scaled_reward": 1.1564876809716225,
|
| 2655 |
+
"rewards/format_reward": 0.0,
|
| 2656 |
+
"step": 189
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"clip_ratio": 0.0,
|
| 2660 |
+
"completion_length": 774.4129943847656,
|
| 2661 |
+
"epoch": 1.9328263624841573,
|
| 2662 |
+
"grad_norm": 0.12307488918304443,
|
| 2663 |
+
"learning_rate": 2.8648491140513264e-09,
|
| 2664 |
+
"loss": -0.0197,
|
| 2665 |
+
"num_tokens": 152721396.0,
|
| 2666 |
+
"reward": 1.2314427196979523,
|
| 2667 |
+
"reward_std": 0.39738379418849945,
|
| 2668 |
+
"rewards/cosine_scaled_reward": 1.2314426600933075,
|
| 2669 |
+
"rewards/format_reward": 0.0,
|
| 2670 |
+
"step": 190
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"clip_ratio": 0.0,
|
| 2674 |
+
"completion_length": 639.6127433776855,
|
| 2675 |
+
"epoch": 1.9429657794676807,
|
| 2676 |
+
"grad_norm": 0.15224085748195648,
|
| 2677 |
+
"learning_rate": 1.9900597959770505e-09,
|
| 2678 |
+
"loss": 0.0385,
|
| 2679 |
+
"num_tokens": 153419617.0,
|
| 2680 |
+
"reward": 1.3687696307897568,
|
| 2681 |
+
"reward_std": 0.3240409158170223,
|
| 2682 |
+
"rewards/cosine_scaled_reward": 1.3687695413827896,
|
| 2683 |
+
"rewards/format_reward": 0.0,
|
| 2684 |
+
"step": 191
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"clip_ratio": 0.0,
|
| 2688 |
+
"completion_length": 727.8538208007812,
|
| 2689 |
+
"epoch": 1.953105196451204,
|
| 2690 |
+
"grad_norm": 0.12927082180976868,
|
| 2691 |
+
"learning_rate": 1.2739426948732424e-09,
|
| 2692 |
+
"loss": 0.002,
|
| 2693 |
+
"num_tokens": 154205742.0,
|
| 2694 |
+
"reward": 1.2560684382915497,
|
| 2695 |
+
"reward_std": 0.3877686746418476,
|
| 2696 |
+
"rewards/cosine_scaled_reward": 1.256068378686905,
|
| 2697 |
+
"rewards/format_reward": 0.0,
|
| 2698 |
+
"step": 192
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"clip_ratio": 0.0,
|
| 2702 |
+
"completion_length": 720.2232513427734,
|
| 2703 |
+
"epoch": 1.9632446134347274,
|
| 2704 |
+
"grad_norm": 0.5261313915252686,
|
| 2705 |
+
"learning_rate": 7.16725974635568e-10,
|
| 2706 |
+
"loss": 0.0203,
|
| 2707 |
+
"num_tokens": 154976902.0,
|
| 2708 |
+
"reward": 1.1852427944540977,
|
| 2709 |
+
"reward_std": 0.4709174670279026,
|
| 2710 |
+
"rewards/cosine_scaled_reward": 1.1852427273988724,
|
| 2711 |
+
"rewards/format_reward": 0.0,
|
| 2712 |
+
"step": 193
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"clip_ratio": 0.0,
|
| 2716 |
+
"completion_length": 744.1283721923828,
|
| 2717 |
+
"epoch": 1.9733840304182508,
|
| 2718 |
+
"grad_norm": 0.13184396922588348,
|
| 2719 |
+
"learning_rate": 3.185871715041255e-10,
|
| 2720 |
+
"loss": -0.0006,
|
| 2721 |
+
"num_tokens": 155776489.0,
|
| 2722 |
+
"reward": 1.153962902724743,
|
| 2723 |
+
"reward_std": 0.35575347393751144,
|
| 2724 |
+
"rewards/cosine_scaled_reward": 1.1539628580212593,
|
| 2725 |
+
"rewards/format_reward": 0.0,
|
| 2726 |
+
"step": 194
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"clip_ratio": 0.0,
|
| 2730 |
+
"completion_length": 681.7768096923828,
|
| 2731 |
+
"epoch": 1.9835234474017744,
|
| 2732 |
+
"grad_norm": 0.14284567534923553,
|
| 2733 |
+
"learning_rate": 7.96531374983589e-11,
|
| 2734 |
+
"loss": 0.0074,
|
| 2735 |
+
"num_tokens": 156507841.0,
|
| 2736 |
+
"reward": 1.263557255268097,
|
| 2737 |
+
"reward_std": 0.42133441381156445,
|
| 2738 |
+
"rewards/cosine_scaled_reward": 1.263557218015194,
|
| 2739 |
+
"rewards/format_reward": 0.0,
|
| 2740 |
+
"step": 195
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"clip_ratio": 0.0,
|
| 2744 |
+
"completion_length": 712.19921875,
|
| 2745 |
+
"epoch": 1.9936628643852978,
|
| 2746 |
+
"grad_norm": 0.12404794245958328,
|
| 2747 |
+
"learning_rate": 0.0,
|
| 2748 |
+
"loss": -0.0205,
|
| 2749 |
+
"num_tokens": 157287049.0,
|
| 2750 |
+
"reward": 1.1777000427246094,
|
| 2751 |
+
"reward_std": 0.4066210947930813,
|
| 2752 |
+
"rewards/cosine_scaled_reward": 1.1776999980211258,
|
| 2753 |
+
"rewards/format_reward": 0.0,
|
| 2754 |
+
"step": 196
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 1.9936628643852978,
|
| 2758 |
+
"step": 196,
|
| 2759 |
+
"total_flos": 0.0,
|
| 2760 |
+
"train_loss": -0.009652293185873921,
|
| 2761 |
+
"train_runtime": 33755.1654,
|
| 2762 |
+
"train_samples_per_second": 0.654,
|
| 2763 |
+
"train_steps_per_second": 0.006
|
| 2764 |
+
}
|
| 2765 |
+
],
|
| 2766 |
+
"logging_steps": 1,
|
| 2767 |
+
"max_steps": 196,
|
| 2768 |
+
"num_input_tokens_seen": 0,
|
| 2769 |
+
"num_train_epochs": 2,
|
| 2770 |
+
"save_steps": 500,
|
| 2771 |
+
"stateful_callbacks": {
|
| 2772 |
+
"TrainerControl": {
|
| 2773 |
+
"args": {
|
| 2774 |
+
"should_epoch_stop": false,
|
| 2775 |
+
"should_evaluate": false,
|
| 2776 |
+
"should_log": false,
|
| 2777 |
+
"should_save": true,
|
| 2778 |
+
"should_training_stop": true
|
| 2779 |
+
},
|
| 2780 |
+
"attributes": {}
|
| 2781 |
+
}
|
| 2782 |
+
},
|
| 2783 |
+
"total_flos": 0.0,
|
| 2784 |
+
"train_batch_size": 16,
|
| 2785 |
+
"trial_name": null,
|
| 2786 |
+
"trial_params": null
|
| 2787 |
+
}
|