Model save
Browse files- README.md +68 -0
- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +1943 -0
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: DAPO
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- dapo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for DAPO
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="kangdawei/DAPO", 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 |
+
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
This model was trained with DAPO, a method introduced in [DAPO: An Open-Source LLM Reinforcement Learning System at Scale](https://huggingface.co/papers/2503.14476).
|
| 34 |
+
|
| 35 |
+
### Framework versions
|
| 36 |
+
|
| 37 |
+
- TRL: 0.16.0.dev0
|
| 38 |
+
- Transformers: 4.57.1
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.2.0
|
| 41 |
+
- Tokenizers: 0.22.1
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite DAPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@article{yu2025dapo,
|
| 49 |
+
title = {{DAPO: An Open-Source LLM Reinforcement Learning System at Scale}},
|
| 50 |
+
author = {Qiying Yu and Zheng Zhang and others},
|
| 51 |
+
year = 2025,
|
| 52 |
+
eprint = {arXiv:2503.14476},
|
| 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": 0.06302435559220612,
|
| 4 |
+
"train_runtime": 83388.0569,
|
| 5 |
+
"train_samples": 7000,
|
| 6 |
+
"train_samples_per_second": 0.058,
|
| 7 |
+
"train_steps_per_second": 0.001
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 0.06302435559220612,
|
| 4 |
+
"train_runtime": 83388.0569,
|
| 5 |
+
"train_samples": 7000,
|
| 6 |
+
"train_samples_per_second": 0.058,
|
| 7 |
+
"train_steps_per_second": 0.001
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1943 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.11428571428571428,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 100,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"clip_fraction": 0.0,
|
| 14 |
+
"completion_length": 2216.625045776367,
|
| 15 |
+
"dapo/avg_reward_std": 0.23920068350331536,
|
| 16 |
+
"dapo/filter_reward_index": 0.0,
|
| 17 |
+
"dapo/kept_prompts_ratio": 0.3735632248993578,
|
| 18 |
+
"dapo/num_sampling_attempts": 3.625,
|
| 19 |
+
"dapo/sampling_efficiency": 33.86904761904762,
|
| 20 |
+
"dapo/total_prompts_processed": 21.75,
|
| 21 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 22 |
+
"epoch": 0.001142857142857143,
|
| 23 |
+
"grad_norm": 0.10874509066343307,
|
| 24 |
+
"kl": 0.0,
|
| 25 |
+
"learning_rate": 0.0,
|
| 26 |
+
"loss": 0.0468,
|
| 27 |
+
"reward": 0.6486758906394243,
|
| 28 |
+
"reward_std": 0.9342863708734512,
|
| 29 |
+
"step": 1
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"clip_fraction": 0.0,
|
| 33 |
+
"completion_length": 2926.4757690429688,
|
| 34 |
+
"dapo/avg_reward_std": 0.24011585204040303,
|
| 35 |
+
"dapo/filter_reward_index": 0.0,
|
| 36 |
+
"dapo/kept_prompts_ratio": 0.3284313836518456,
|
| 37 |
+
"dapo/num_sampling_attempts": 4.25,
|
| 38 |
+
"dapo/sampling_efficiency": 26.874999999999993,
|
| 39 |
+
"dapo/total_prompts_processed": 25.5,
|
| 40 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 41 |
+
"epoch": 0.002285714285714286,
|
| 42 |
+
"grad_norm": 0.12814132869243622,
|
| 43 |
+
"kl": 0.0,
|
| 44 |
+
"learning_rate": 1e-07,
|
| 45 |
+
"loss": 0.0508,
|
| 46 |
+
"reward": 0.2922485675662756,
|
| 47 |
+
"reward_std": 0.9327598959207535,
|
| 48 |
+
"step": 2
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"clip_fraction": 0.0,
|
| 52 |
+
"completion_length": 2888.1527709960938,
|
| 53 |
+
"dapo/avg_reward_std": 0.2903491040070852,
|
| 54 |
+
"dapo/filter_reward_index": 0.0,
|
| 55 |
+
"dapo/kept_prompts_ratio": 0.36111111839612325,
|
| 56 |
+
"dapo/num_sampling_attempts": 3.75,
|
| 57 |
+
"dapo/sampling_efficiency": 36.875,
|
| 58 |
+
"dapo/total_prompts_processed": 22.5,
|
| 59 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 60 |
+
"epoch": 0.0034285714285714284,
|
| 61 |
+
"grad_norm": 0.1155443787574768,
|
| 62 |
+
"kl": 2.9146671295166016e-05,
|
| 63 |
+
"learning_rate": 2e-07,
|
| 64 |
+
"loss": 0.0647,
|
| 65 |
+
"reward": 0.3509849710389972,
|
| 66 |
+
"reward_std": 0.9315856546163559,
|
| 67 |
+
"step": 3
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"clip_fraction": 0.0,
|
| 71 |
+
"completion_length": 2535.718734741211,
|
| 72 |
+
"dapo/avg_reward_std": 0.25628158891642533,
|
| 73 |
+
"dapo/filter_reward_index": 0.0,
|
| 74 |
+
"dapo/kept_prompts_ratio": 0.35802469595714853,
|
| 75 |
+
"dapo/num_sampling_attempts": 3.375,
|
| 76 |
+
"dapo/sampling_efficiency": 41.56249999999999,
|
| 77 |
+
"dapo/total_prompts_processed": 20.25,
|
| 78 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 79 |
+
"epoch": 0.004571428571428572,
|
| 80 |
+
"grad_norm": 0.14338600635528564,
|
| 81 |
+
"kl": 2.1044164896011353e-05,
|
| 82 |
+
"learning_rate": 3e-07,
|
| 83 |
+
"loss": 0.0536,
|
| 84 |
+
"reward": 0.5615630690008402,
|
| 85 |
+
"reward_std": 0.9670609682798386,
|
| 86 |
+
"step": 4
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"clip_fraction": 0.0,
|
| 90 |
+
"completion_length": 2548.916702270508,
|
| 91 |
+
"dapo/avg_reward_std": 0.2889887053391029,
|
| 92 |
+
"dapo/filter_reward_index": 0.0,
|
| 93 |
+
"dapo/kept_prompts_ratio": 0.40804598814454573,
|
| 94 |
+
"dapo/num_sampling_attempts": 3.625,
|
| 95 |
+
"dapo/sampling_efficiency": 36.875,
|
| 96 |
+
"dapo/total_prompts_processed": 21.75,
|
| 97 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 98 |
+
"epoch": 0.005714285714285714,
|
| 99 |
+
"grad_norm": 0.10121661424636841,
|
| 100 |
+
"kl": 2.7820467948913574e-05,
|
| 101 |
+
"learning_rate": 4e-07,
|
| 102 |
+
"loss": 0.0263,
|
| 103 |
+
"reward": 0.5986085031181574,
|
| 104 |
+
"reward_std": 0.9444186091423035,
|
| 105 |
+
"step": 5
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"clip_fraction": 0.0,
|
| 109 |
+
"completion_length": 2357.579864501953,
|
| 110 |
+
"dapo/avg_reward_std": 0.30308351665735245,
|
| 111 |
+
"dapo/filter_reward_index": 0.0,
|
| 112 |
+
"dapo/kept_prompts_ratio": 0.36309524306229185,
|
| 113 |
+
"dapo/num_sampling_attempts": 3.5,
|
| 114 |
+
"dapo/sampling_efficiency": 37.5,
|
| 115 |
+
"dapo/total_prompts_processed": 21.0,
|
| 116 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 117 |
+
"epoch": 0.006857142857142857,
|
| 118 |
+
"grad_norm": 0.171969935297966,
|
| 119 |
+
"kl": 2.6032328605651855e-05,
|
| 120 |
+
"learning_rate": 5e-07,
|
| 121 |
+
"loss": 0.0906,
|
| 122 |
+
"reward": 0.4527070773765445,
|
| 123 |
+
"reward_std": 0.9109365493059158,
|
| 124 |
+
"step": 6
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"clip_fraction": 0.0,
|
| 128 |
+
"completion_length": 2404.2534790039062,
|
| 129 |
+
"dapo/avg_reward_std": 0.3077041815828394,
|
| 130 |
+
"dapo/filter_reward_index": 0.0,
|
| 131 |
+
"dapo/kept_prompts_ratio": 0.41975309506610586,
|
| 132 |
+
"dapo/num_sampling_attempts": 3.375,
|
| 133 |
+
"dapo/sampling_efficiency": 37.916666666666664,
|
| 134 |
+
"dapo/total_prompts_processed": 20.25,
|
| 135 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 136 |
+
"epoch": 0.008,
|
| 137 |
+
"grad_norm": 0.12406504899263382,
|
| 138 |
+
"kl": 1.9066035747528076e-05,
|
| 139 |
+
"learning_rate": 6e-07,
|
| 140 |
+
"loss": 0.0645,
|
| 141 |
+
"reward": 0.5808906648308039,
|
| 142 |
+
"reward_std": 0.9664968773722649,
|
| 143 |
+
"step": 7
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"clip_fraction": 0.0,
|
| 147 |
+
"completion_length": 2833.3056030273438,
|
| 148 |
+
"dapo/avg_reward_std": 0.2214778729023472,
|
| 149 |
+
"dapo/filter_reward_index": 0.0,
|
| 150 |
+
"dapo/kept_prompts_ratio": 0.31182796435971416,
|
| 151 |
+
"dapo/num_sampling_attempts": 3.875,
|
| 152 |
+
"dapo/sampling_efficiency": 36.577380952380956,
|
| 153 |
+
"dapo/total_prompts_processed": 23.25,
|
| 154 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 155 |
+
"epoch": 0.009142857142857144,
|
| 156 |
+
"grad_norm": 0.13480524718761444,
|
| 157 |
+
"kl": 3.4965574741363525e-05,
|
| 158 |
+
"learning_rate": 7e-07,
|
| 159 |
+
"loss": 0.0738,
|
| 160 |
+
"reward": 0.5177570842206478,
|
| 161 |
+
"reward_std": 0.9147621840238571,
|
| 162 |
+
"step": 8
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"clip_fraction": 0.0,
|
| 166 |
+
"completion_length": 2965.6736450195312,
|
| 167 |
+
"dapo/avg_reward_std": 0.2788830002148946,
|
| 168 |
+
"dapo/filter_reward_index": 0.0,
|
| 169 |
+
"dapo/kept_prompts_ratio": 0.3888888966154169,
|
| 170 |
+
"dapo/num_sampling_attempts": 3.375,
|
| 171 |
+
"dapo/sampling_efficiency": 46.36904761904761,
|
| 172 |
+
"dapo/total_prompts_processed": 20.25,
|
| 173 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 174 |
+
"epoch": 0.010285714285714285,
|
| 175 |
+
"grad_norm": 0.08226096630096436,
|
| 176 |
+
"kl": 1.4536082744598389e-05,
|
| 177 |
+
"learning_rate": 8e-07,
|
| 178 |
+
"loss": 0.0316,
|
| 179 |
+
"reward": 0.5644797384738922,
|
| 180 |
+
"reward_std": 0.9423079788684845,
|
| 181 |
+
"step": 9
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"clip_fraction": 0.0,
|
| 185 |
+
"completion_length": 2574.461814880371,
|
| 186 |
+
"dapo/avg_reward_std": 0.3602010520065532,
|
| 187 |
+
"dapo/filter_reward_index": 0.0,
|
| 188 |
+
"dapo/kept_prompts_ratio": 0.588235302883036,
|
| 189 |
+
"dapo/num_sampling_attempts": 2.125,
|
| 190 |
+
"dapo/sampling_efficiency": 61.45833333333333,
|
| 191 |
+
"dapo/total_prompts_processed": 12.75,
|
| 192 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 193 |
+
"epoch": 0.011428571428571429,
|
| 194 |
+
"grad_norm": 0.1667146533727646,
|
| 195 |
+
"kl": 2.9319897294044495e-05,
|
| 196 |
+
"learning_rate": 9e-07,
|
| 197 |
+
"loss": 0.0894,
|
| 198 |
+
"reward": 0.6415909845381975,
|
| 199 |
+
"reward_std": 0.9869548827409744,
|
| 200 |
+
"step": 10
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"clip_fraction": 0.0,
|
| 204 |
+
"completion_length": 2798.982666015625,
|
| 205 |
+
"dapo/avg_reward_std": 0.15393146287117684,
|
| 206 |
+
"dapo/filter_reward_index": 0.0,
|
| 207 |
+
"dapo/kept_prompts_ratio": 0.14880952797830105,
|
| 208 |
+
"dapo/num_sampling_attempts": 7.0,
|
| 209 |
+
"dapo/sampling_efficiency": 15.882936507936506,
|
| 210 |
+
"dapo/total_prompts_processed": 42.0,
|
| 211 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 212 |
+
"epoch": 0.012571428571428572,
|
| 213 |
+
"grad_norm": 0.1166534572839737,
|
| 214 |
+
"kl": 2.0567327737808228e-05,
|
| 215 |
+
"learning_rate": 1e-06,
|
| 216 |
+
"loss": 0.0207,
|
| 217 |
+
"reward": 0.2987014357931912,
|
| 218 |
+
"reward_std": 0.868266686797142,
|
| 219 |
+
"step": 11
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"clip_fraction": 0.0,
|
| 223 |
+
"completion_length": 2377.555595397949,
|
| 224 |
+
"dapo/avg_reward_std": 0.21645361091941595,
|
| 225 |
+
"dapo/filter_reward_index": 0.0,
|
| 226 |
+
"dapo/kept_prompts_ratio": 0.2968750037252903,
|
| 227 |
+
"dapo/num_sampling_attempts": 4.0,
|
| 228 |
+
"dapo/sampling_efficiency": 38.125,
|
| 229 |
+
"dapo/total_prompts_processed": 24.0,
|
| 230 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 231 |
+
"epoch": 0.013714285714285714,
|
| 232 |
+
"grad_norm": 0.23483960330486298,
|
| 233 |
+
"kl": 3.6854296922683716e-05,
|
| 234 |
+
"learning_rate": 9.997258721585931e-07,
|
| 235 |
+
"loss": 0.0491,
|
| 236 |
+
"reward": 0.6348252706229687,
|
| 237 |
+
"reward_std": 0.9863902181386948,
|
| 238 |
+
"step": 12
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"clip_fraction": 0.0,
|
| 242 |
+
"completion_length": 2688.1111755371094,
|
| 243 |
+
"dapo/avg_reward_std": 0.34906478971242905,
|
| 244 |
+
"dapo/filter_reward_index": 0.0,
|
| 245 |
+
"dapo/kept_prompts_ratio": 0.49166667461395264,
|
| 246 |
+
"dapo/num_sampling_attempts": 2.5,
|
| 247 |
+
"dapo/sampling_efficiency": 52.08333333333333,
|
| 248 |
+
"dapo/total_prompts_processed": 15.0,
|
| 249 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 250 |
+
"epoch": 0.014857142857142857,
|
| 251 |
+
"grad_norm": 0.09364266693592072,
|
| 252 |
+
"kl": 3.152713179588318e-05,
|
| 253 |
+
"learning_rate": 9.989038226169207e-07,
|
| 254 |
+
"loss": 0.0431,
|
| 255 |
+
"reward": 0.5878111608326435,
|
| 256 |
+
"reward_std": 0.9752944633364677,
|
| 257 |
+
"step": 13
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"clip_fraction": 0.0,
|
| 261 |
+
"completion_length": 2029.9132270812988,
|
| 262 |
+
"dapo/avg_reward_std": 0.25792322993278505,
|
| 263 |
+
"dapo/filter_reward_index": 0.0,
|
| 264 |
+
"dapo/kept_prompts_ratio": 0.36666666984558105,
|
| 265 |
+
"dapo/num_sampling_attempts": 3.125,
|
| 266 |
+
"dapo/sampling_efficiency": 57.5,
|
| 267 |
+
"dapo/total_prompts_processed": 18.75,
|
| 268 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 269 |
+
"epoch": 0.016,
|
| 270 |
+
"grad_norm": 0.13894271850585938,
|
| 271 |
+
"kl": 4.156678915023804e-05,
|
| 272 |
+
"learning_rate": 9.975348529157229e-07,
|
| 273 |
+
"loss": 0.0279,
|
| 274 |
+
"reward": 0.5834919223561883,
|
| 275 |
+
"reward_std": 0.9710095003247261,
|
| 276 |
+
"step": 14
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"clip_fraction": 0.0,
|
| 280 |
+
"completion_length": 2817.8576583862305,
|
| 281 |
+
"dapo/avg_reward_std": 0.3106007158756256,
|
| 282 |
+
"dapo/filter_reward_index": 0.0,
|
| 283 |
+
"dapo/kept_prompts_ratio": 0.5333333484828472,
|
| 284 |
+
"dapo/num_sampling_attempts": 2.5,
|
| 285 |
+
"dapo/sampling_efficiency": 52.08333333333333,
|
| 286 |
+
"dapo/total_prompts_processed": 15.0,
|
| 287 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 288 |
+
"epoch": 0.017142857142857144,
|
| 289 |
+
"grad_norm": 0.08778129518032074,
|
| 290 |
+
"kl": 3.078579902648926e-05,
|
| 291 |
+
"learning_rate": 9.956206309337066e-07,
|
| 292 |
+
"loss": 0.0343,
|
| 293 |
+
"reward": 0.6716702915728092,
|
| 294 |
+
"reward_std": 0.99223193526268,
|
| 295 |
+
"step": 15
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"clip_fraction": 0.0,
|
| 299 |
+
"completion_length": 2570.2500076293945,
|
| 300 |
+
"dapo/avg_reward_std": 0.244095021715531,
|
| 301 |
+
"dapo/filter_reward_index": 0.0,
|
| 302 |
+
"dapo/kept_prompts_ratio": 0.35897436336829114,
|
| 303 |
+
"dapo/num_sampling_attempts": 3.25,
|
| 304 |
+
"dapo/sampling_efficiency": 44.49404761904762,
|
| 305 |
+
"dapo/total_prompts_processed": 19.5,
|
| 306 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 307 |
+
"epoch": 0.018285714285714287,
|
| 308 |
+
"grad_norm": 0.07460447400808334,
|
| 309 |
+
"kl": 0.00025935471057891846,
|
| 310 |
+
"learning_rate": 9.931634888554935e-07,
|
| 311 |
+
"loss": 0.0146,
|
| 312 |
+
"reward": 0.7213943339884281,
|
| 313 |
+
"reward_std": 0.9671430364251137,
|
| 314 |
+
"step": 16
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"clip_fraction": 0.0,
|
| 318 |
+
"completion_length": 2483.413215637207,
|
| 319 |
+
"dapo/avg_reward_std": 0.2672279636065165,
|
| 320 |
+
"dapo/filter_reward_index": 0.0,
|
| 321 |
+
"dapo/kept_prompts_ratio": 0.35000000496705375,
|
| 322 |
+
"dapo/num_sampling_attempts": 3.75,
|
| 323 |
+
"dapo/sampling_efficiency": 29.166666666666664,
|
| 324 |
+
"dapo/total_prompts_processed": 22.5,
|
| 325 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 326 |
+
"epoch": 0.019428571428571427,
|
| 327 |
+
"grad_norm": 0.12397046387195587,
|
| 328 |
+
"kl": 0.00022289156913757324,
|
| 329 |
+
"learning_rate": 9.901664203302124e-07,
|
| 330 |
+
"loss": 0.0624,
|
| 331 |
+
"reward": 0.4952134042978287,
|
| 332 |
+
"reward_std": 0.9074268043041229,
|
| 333 |
+
"step": 17
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"clip_fraction": 0.0,
|
| 337 |
+
"completion_length": 2537.8194580078125,
|
| 338 |
+
"dapo/avg_reward_std": 0.34170445956681905,
|
| 339 |
+
"dapo/filter_reward_index": 0.0,
|
| 340 |
+
"dapo/kept_prompts_ratio": 0.5438596567040995,
|
| 341 |
+
"dapo/num_sampling_attempts": 2.375,
|
| 342 |
+
"dapo/sampling_efficiency": 48.95833333333333,
|
| 343 |
+
"dapo/total_prompts_processed": 14.25,
|
| 344 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 345 |
+
"epoch": 0.02057142857142857,
|
| 346 |
+
"grad_norm": 0.1614188253879547,
|
| 347 |
+
"kl": 0.0003694295883178711,
|
| 348 |
+
"learning_rate": 9.866330768241983e-07,
|
| 349 |
+
"loss": 0.1136,
|
| 350 |
+
"reward": 0.6263789646327496,
|
| 351 |
+
"reward_std": 0.9367138147354126,
|
| 352 |
+
"step": 18
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"clip_fraction": 0.0,
|
| 356 |
+
"completion_length": 2041.2916984558105,
|
| 357 |
+
"dapo/avg_reward_std": 0.23441629879402393,
|
| 358 |
+
"dapo/filter_reward_index": 0.0,
|
| 359 |
+
"dapo/kept_prompts_ratio": 0.31818182224577124,
|
| 360 |
+
"dapo/num_sampling_attempts": 4.125,
|
| 361 |
+
"dapo/sampling_efficiency": 38.36805555555556,
|
| 362 |
+
"dapo/total_prompts_processed": 24.75,
|
| 363 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 364 |
+
"epoch": 0.021714285714285714,
|
| 365 |
+
"grad_norm": 0.2115960717201233,
|
| 366 |
+
"kl": 0.0005898326635360718,
|
| 367 |
+
"learning_rate": 9.825677631722435e-07,
|
| 368 |
+
"loss": 0.0603,
|
| 369 |
+
"reward": 0.6228582374751568,
|
| 370 |
+
"reward_std": 0.9455358982086182,
|
| 371 |
+
"step": 19
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"clip_fraction": 0.0,
|
| 375 |
+
"completion_length": 2392.7882385253906,
|
| 376 |
+
"dapo/avg_reward_std": 0.22908216629709516,
|
| 377 |
+
"dapo/filter_reward_index": 0.0,
|
| 378 |
+
"dapo/kept_prompts_ratio": 0.2952381010566439,
|
| 379 |
+
"dapo/num_sampling_attempts": 4.375,
|
| 380 |
+
"dapo/sampling_efficiency": 33.541666666666664,
|
| 381 |
+
"dapo/total_prompts_processed": 26.25,
|
| 382 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 383 |
+
"epoch": 0.022857142857142857,
|
| 384 |
+
"grad_norm": 0.20383711159229279,
|
| 385 |
+
"kl": 0.0008958578109741211,
|
| 386 |
+
"learning_rate": 9.779754323328192e-07,
|
| 387 |
+
"loss": 0.1313,
|
| 388 |
+
"reward": 0.41653589624911547,
|
| 389 |
+
"reward_std": 0.9027180448174477,
|
| 390 |
+
"step": 20
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"clip_fraction": 0.0,
|
| 394 |
+
"completion_length": 2966.260452270508,
|
| 395 |
+
"dapo/avg_reward_std": 0.16204138861762152,
|
| 396 |
+
"dapo/filter_reward_index": 0.0,
|
| 397 |
+
"dapo/kept_prompts_ratio": 0.25555555986033546,
|
| 398 |
+
"dapo/num_sampling_attempts": 5.625,
|
| 399 |
+
"dapo/sampling_efficiency": 22.84722222222222,
|
| 400 |
+
"dapo/total_prompts_processed": 33.75,
|
| 401 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 402 |
+
"epoch": 0.024,
|
| 403 |
+
"grad_norm": 0.1098903939127922,
|
| 404 |
+
"kl": 0.0002017766237258911,
|
| 405 |
+
"learning_rate": 9.728616793536587e-07,
|
| 406 |
+
"loss": 0.0825,
|
| 407 |
+
"reward": 0.43902475386857986,
|
| 408 |
+
"reward_std": 0.9111825451254845,
|
| 409 |
+
"step": 21
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"clip_fraction": 0.0,
|
| 413 |
+
"completion_length": 3016.357696533203,
|
| 414 |
+
"dapo/avg_reward_std": 0.28799043401427893,
|
| 415 |
+
"dapo/filter_reward_index": 0.0,
|
| 416 |
+
"dapo/kept_prompts_ratio": 0.42028986371081806,
|
| 417 |
+
"dapo/num_sampling_attempts": 2.875,
|
| 418 |
+
"dapo/sampling_efficiency": 52.20238095238095,
|
| 419 |
+
"dapo/total_prompts_processed": 17.25,
|
| 420 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 421 |
+
"epoch": 0.025142857142857144,
|
| 422 |
+
"grad_norm": 0.1315963715314865,
|
| 423 |
+
"kl": 0.0005468130111694336,
|
| 424 |
+
"learning_rate": 9.672327345550543e-07,
|
| 425 |
+
"loss": 0.0657,
|
| 426 |
+
"reward": 0.5281127206981182,
|
| 427 |
+
"reward_std": 0.9846171587705612,
|
| 428 |
+
"step": 22
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"clip_fraction": 0.0,
|
| 432 |
+
"completion_length": 2408.8333282470703,
|
| 433 |
+
"dapo/avg_reward_std": 0.24506365811383282,
|
| 434 |
+
"dapo/filter_reward_index": 0.0,
|
| 435 |
+
"dapo/kept_prompts_ratio": 0.3703703780968984,
|
| 436 |
+
"dapo/num_sampling_attempts": 3.375,
|
| 437 |
+
"dapo/sampling_efficiency": 40.74404761904761,
|
| 438 |
+
"dapo/total_prompts_processed": 20.25,
|
| 439 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 440 |
+
"epoch": 0.026285714285714287,
|
| 441 |
+
"grad_norm": 0.12457310408353806,
|
| 442 |
+
"kl": 0.001109391450881958,
|
| 443 |
+
"learning_rate": 9.610954559391704e-07,
|
| 444 |
+
"loss": 0.0304,
|
| 445 |
+
"reward": 0.6419337540864944,
|
| 446 |
+
"reward_std": 0.9689808040857315,
|
| 447 |
+
"step": 23
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"clip_fraction": 0.0,
|
| 451 |
+
"completion_length": 2891.7777709960938,
|
| 452 |
+
"dapo/avg_reward_std": 0.2580765459848487,
|
| 453 |
+
"dapo/filter_reward_index": 0.0,
|
| 454 |
+
"dapo/kept_prompts_ratio": 0.4420289954413538,
|
| 455 |
+
"dapo/num_sampling_attempts": 2.875,
|
| 456 |
+
"dapo/sampling_efficiency": 51.785714285714285,
|
| 457 |
+
"dapo/total_prompts_processed": 17.25,
|
| 458 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 459 |
+
"epoch": 0.027428571428571427,
|
| 460 |
+
"grad_norm": 0.09673310071229935,
|
| 461 |
+
"kl": 0.0006018728017807007,
|
| 462 |
+
"learning_rate": 9.54457320834625e-07,
|
| 463 |
+
"loss": 0.0143,
|
| 464 |
+
"reward": 0.4589955974370241,
|
| 465 |
+
"reward_std": 0.9405186697840691,
|
| 466 |
+
"step": 24
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"clip_fraction": 0.0,
|
| 470 |
+
"completion_length": 2994.8159790039062,
|
| 471 |
+
"dapo/avg_reward_std": 0.24148962597052256,
|
| 472 |
+
"dapo/filter_reward_index": 0.0,
|
| 473 |
+
"dapo/kept_prompts_ratio": 0.3777777850627899,
|
| 474 |
+
"dapo/num_sampling_attempts": 3.75,
|
| 475 |
+
"dapo/sampling_efficiency": 34.99999999999999,
|
| 476 |
+
"dapo/total_prompts_processed": 22.5,
|
| 477 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 478 |
+
"epoch": 0.02857142857142857,
|
| 479 |
+
"grad_norm": 0.12189235538244247,
|
| 480 |
+
"kl": 0.0021944642066955566,
|
| 481 |
+
"learning_rate": 9.473264167865171e-07,
|
| 482 |
+
"loss": 0.0869,
|
| 483 |
+
"reward": 0.4214114509522915,
|
| 484 |
+
"reward_std": 0.918621838092804,
|
| 485 |
+
"step": 25
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"clip_fraction": 0.0,
|
| 489 |
+
"completion_length": 3106.2743530273438,
|
| 490 |
+
"dapo/avg_reward_std": 0.21211836412549018,
|
| 491 |
+
"dapo/filter_reward_index": 0.0,
|
| 492 |
+
"dapo/kept_prompts_ratio": 0.23750000558793544,
|
| 493 |
+
"dapo/num_sampling_attempts": 5.0,
|
| 494 |
+
"dapo/sampling_efficiency": 25.729166666666664,
|
| 495 |
+
"dapo/total_prompts_processed": 30.0,
|
| 496 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 497 |
+
"epoch": 0.029714285714285714,
|
| 498 |
+
"grad_norm": 0.11006143689155579,
|
| 499 |
+
"kl": 0.002092994749546051,
|
| 500 |
+
"learning_rate": 9.397114317029974e-07,
|
| 501 |
+
"loss": 0.0617,
|
| 502 |
+
"reward": 0.4296974149765447,
|
| 503 |
+
"reward_std": 0.9136241301894188,
|
| 504 |
+
"step": 26
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"clip_fraction": 0.0,
|
| 508 |
+
"completion_length": 2601.388946533203,
|
| 509 |
+
"dapo/avg_reward_std": 0.24121128850513035,
|
| 510 |
+
"dapo/filter_reward_index": 0.0,
|
| 511 |
+
"dapo/kept_prompts_ratio": 0.28240741416811943,
|
| 512 |
+
"dapo/num_sampling_attempts": 4.5,
|
| 513 |
+
"dapo/sampling_efficiency": 25.76388888888889,
|
| 514 |
+
"dapo/total_prompts_processed": 27.0,
|
| 515 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 516 |
+
"epoch": 0.030857142857142857,
|
| 517 |
+
"grad_norm": 0.11345893889665604,
|
| 518 |
+
"kl": 0.003206908702850342,
|
| 519 |
+
"learning_rate": 9.316216432703916e-07,
|
| 520 |
+
"loss": 0.0926,
|
| 521 |
+
"reward": 0.5876726619899273,
|
| 522 |
+
"reward_std": 0.9382903277873993,
|
| 523 |
+
"step": 27
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"clip_fraction": 0.0,
|
| 527 |
+
"completion_length": 2861.6180839538574,
|
| 528 |
+
"dapo/avg_reward_std": 0.23961352888080809,
|
| 529 |
+
"dapo/filter_reward_index": 0.0,
|
| 530 |
+
"dapo/kept_prompts_ratio": 0.3055555605226093,
|
| 531 |
+
"dapo/num_sampling_attempts": 4.5,
|
| 532 |
+
"dapo/sampling_efficiency": 27.94642857142857,
|
| 533 |
+
"dapo/total_prompts_processed": 27.0,
|
| 534 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 535 |
+
"epoch": 0.032,
|
| 536 |
+
"grad_norm": 0.1445908397436142,
|
| 537 |
+
"kl": 0.0031346678733825684,
|
| 538 |
+
"learning_rate": 9.230669076497687e-07,
|
| 539 |
+
"loss": 0.0852,
|
| 540 |
+
"reward": 0.40619770623743534,
|
| 541 |
+
"reward_std": 0.9506878778338432,
|
| 542 |
+
"step": 28
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"clip_fraction": 0.0,
|
| 546 |
+
"completion_length": 2729.1875,
|
| 547 |
+
"dapo/avg_reward_std": 0.24243796567122142,
|
| 548 |
+
"dapo/filter_reward_index": 0.0,
|
| 549 |
+
"dapo/kept_prompts_ratio": 0.35555555919806164,
|
| 550 |
+
"dapo/num_sampling_attempts": 3.75,
|
| 551 |
+
"dapo/sampling_efficiency": 39.93055555555555,
|
| 552 |
+
"dapo/total_prompts_processed": 22.5,
|
| 553 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 554 |
+
"epoch": 0.03314285714285714,
|
| 555 |
+
"grad_norm": 0.11093314737081528,
|
| 556 |
+
"kl": 0.0027089565992355347,
|
| 557 |
+
"learning_rate": 9.140576474687263e-07,
|
| 558 |
+
"loss": 0.0604,
|
| 559 |
+
"reward": 0.6693072468042374,
|
| 560 |
+
"reward_std": 0.9926005378365517,
|
| 561 |
+
"step": 29
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"clip_fraction": 0.0,
|
| 565 |
+
"completion_length": 3155.7083740234375,
|
| 566 |
+
"dapo/avg_reward_std": 0.222336781601752,
|
| 567 |
+
"dapo/filter_reward_index": 0.0,
|
| 568 |
+
"dapo/kept_prompts_ratio": 0.311827961956301,
|
| 569 |
+
"dapo/num_sampling_attempts": 3.875,
|
| 570 |
+
"dapo/sampling_efficiency": 42.93154761904761,
|
| 571 |
+
"dapo/total_prompts_processed": 23.25,
|
| 572 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 573 |
+
"epoch": 0.03428571428571429,
|
| 574 |
+
"grad_norm": 0.08208812773227692,
|
| 575 |
+
"kl": 0.001552581787109375,
|
| 576 |
+
"learning_rate": 9.046048391230247e-07,
|
| 577 |
+
"loss": 0.0268,
|
| 578 |
+
"reward": 0.521108225453645,
|
| 579 |
+
"reward_std": 0.9469912871718407,
|
| 580 |
+
"step": 30
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"clip_fraction": 0.0,
|
| 584 |
+
"completion_length": 2657.559036254883,
|
| 585 |
+
"dapo/avg_reward_std": 0.1865689324008094,
|
| 586 |
+
"dapo/filter_reward_index": 0.0,
|
| 587 |
+
"dapo/kept_prompts_ratio": 0.259259263260497,
|
| 588 |
+
"dapo/num_sampling_attempts": 4.5,
|
| 589 |
+
"dapo/sampling_efficiency": 51.076388888888886,
|
| 590 |
+
"dapo/total_prompts_processed": 27.0,
|
| 591 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 592 |
+
"epoch": 0.03542857142857143,
|
| 593 |
+
"grad_norm": 0.1316680908203125,
|
| 594 |
+
"kl": 0.009428024291992188,
|
| 595 |
+
"learning_rate": 8.9471999940354e-07,
|
| 596 |
+
"loss": 0.0745,
|
| 597 |
+
"reward": 0.6315789166837931,
|
| 598 |
+
"reward_std": 0.9327967762947083,
|
| 599 |
+
"step": 31
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"clip_fraction": 0.0,
|
| 603 |
+
"completion_length": 3071.7535095214844,
|
| 604 |
+
"dapo/avg_reward_std": 0.3048748767375946,
|
| 605 |
+
"dapo/filter_reward_index": 0.0,
|
| 606 |
+
"dapo/kept_prompts_ratio": 0.40000000298023225,
|
| 607 |
+
"dapo/num_sampling_attempts": 3.125,
|
| 608 |
+
"dapo/sampling_efficiency": 55.104166666666664,
|
| 609 |
+
"dapo/total_prompts_processed": 18.75,
|
| 610 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 611 |
+
"epoch": 0.036571428571428574,
|
| 612 |
+
"grad_norm": 0.10442829132080078,
|
| 613 |
+
"kl": 0.0021753311157226562,
|
| 614 |
+
"learning_rate": 8.844151714648274e-07,
|
| 615 |
+
"loss": 0.0567,
|
| 616 |
+
"reward": 0.5447857324033976,
|
| 617 |
+
"reward_std": 0.921301856637001,
|
| 618 |
+
"step": 32
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"clip_fraction": 0.0,
|
| 622 |
+
"completion_length": 3025.826416015625,
|
| 623 |
+
"dapo/avg_reward_std": 0.23097028769552708,
|
| 624 |
+
"dapo/filter_reward_index": 0.0,
|
| 625 |
+
"dapo/kept_prompts_ratio": 0.3385416748933494,
|
| 626 |
+
"dapo/num_sampling_attempts": 4.0,
|
| 627 |
+
"dapo/sampling_efficiency": 38.95833333333333,
|
| 628 |
+
"dapo/total_prompts_processed": 24.0,
|
| 629 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 630 |
+
"epoch": 0.037714285714285714,
|
| 631 |
+
"grad_norm": 0.09167502820491791,
|
| 632 |
+
"kl": 0.003194093704223633,
|
| 633 |
+
"learning_rate": 8.737029101523929e-07,
|
| 634 |
+
"loss": 0.0612,
|
| 635 |
+
"reward": 0.5547973131760955,
|
| 636 |
+
"reward_std": 0.9730775579810143,
|
| 637 |
+
"step": 33
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"clip_fraction": 0.0,
|
| 641 |
+
"completion_length": 2558.7812423706055,
|
| 642 |
+
"dapo/avg_reward_std": 0.2557758816650936,
|
| 643 |
+
"dapo/filter_reward_index": 0.0,
|
| 644 |
+
"dapo/kept_prompts_ratio": 0.3452381007373333,
|
| 645 |
+
"dapo/num_sampling_attempts": 3.5,
|
| 646 |
+
"dapo/sampling_efficiency": 42.113095238095234,
|
| 647 |
+
"dapo/total_prompts_processed": 21.0,
|
| 648 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 649 |
+
"epoch": 0.038857142857142854,
|
| 650 |
+
"grad_norm": 0.11055821925401688,
|
| 651 |
+
"kl": 0.019285082817077637,
|
| 652 |
+
"learning_rate": 8.625962667065487e-07,
|
| 653 |
+
"loss": 0.0831,
|
| 654 |
+
"reward": 0.5826370492577553,
|
| 655 |
+
"reward_std": 0.9168377369642258,
|
| 656 |
+
"step": 34
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"clip_fraction": 0.0,
|
| 660 |
+
"completion_length": 2909.2361602783203,
|
| 661 |
+
"dapo/avg_reward_std": 0.22593376713414345,
|
| 662 |
+
"dapo/filter_reward_index": 0.0,
|
| 663 |
+
"dapo/kept_prompts_ratio": 0.30645161626800416,
|
| 664 |
+
"dapo/num_sampling_attempts": 3.875,
|
| 665 |
+
"dapo/sampling_efficiency": 49.598214285714285,
|
| 666 |
+
"dapo/total_prompts_processed": 23.25,
|
| 667 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 668 |
+
"epoch": 0.04,
|
| 669 |
+
"grad_norm": 0.09941194951534271,
|
| 670 |
+
"kl": 0.004673004150390625,
|
| 671 |
+
"learning_rate": 8.511087728614862e-07,
|
| 672 |
+
"loss": 0.0581,
|
| 673 |
+
"reward": 0.5392080545425415,
|
| 674 |
+
"reward_std": 0.9793680757284164,
|
| 675 |
+
"step": 35
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"clip_fraction": 0.0,
|
| 679 |
+
"completion_length": 2629.3333435058594,
|
| 680 |
+
"dapo/avg_reward_std": 0.2632370889186859,
|
| 681 |
+
"dapo/filter_reward_index": 0.0,
|
| 682 |
+
"dapo/kept_prompts_ratio": 0.338541675824672,
|
| 683 |
+
"dapo/num_sampling_attempts": 4.0,
|
| 684 |
+
"dapo/sampling_efficiency": 29.513888888888886,
|
| 685 |
+
"dapo/total_prompts_processed": 24.0,
|
| 686 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 687 |
+
"epoch": 0.04114285714285714,
|
| 688 |
+
"grad_norm": 0.17353901267051697,
|
| 689 |
+
"kl": 0.010207176208496094,
|
| 690 |
+
"learning_rate": 8.392544243589427e-07,
|
| 691 |
+
"loss": 0.0623,
|
| 692 |
+
"reward": 0.5811682712519541,
|
| 693 |
+
"reward_std": 0.9331383407115936,
|
| 694 |
+
"step": 36
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"clip_fraction": 0.0,
|
| 698 |
+
"completion_length": 3220.9409790039062,
|
| 699 |
+
"dapo/avg_reward_std": 0.2187359256403787,
|
| 700 |
+
"dapo/filter_reward_index": 0.0,
|
| 701 |
+
"dapo/kept_prompts_ratio": 0.29047619913305556,
|
| 702 |
+
"dapo/num_sampling_attempts": 4.375,
|
| 703 |
+
"dapo/sampling_efficiency": 36.25,
|
| 704 |
+
"dapo/total_prompts_processed": 26.25,
|
| 705 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 706 |
+
"epoch": 0.04228571428571429,
|
| 707 |
+
"grad_norm": 0.10708803683519363,
|
| 708 |
+
"kl": 0.0023801326751708984,
|
| 709 |
+
"learning_rate": 8.270476638965461e-07,
|
| 710 |
+
"loss": 0.0657,
|
| 711 |
+
"reward": 0.48440539091825485,
|
| 712 |
+
"reward_std": 0.9014616012573242,
|
| 713 |
+
"step": 37
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"clip_fraction": 0.0,
|
| 717 |
+
"completion_length": 3233.420135498047,
|
| 718 |
+
"dapo/avg_reward_std": 0.2624325007200241,
|
| 719 |
+
"dapo/filter_reward_index": 0.0,
|
| 720 |
+
"dapo/kept_prompts_ratio": 0.30208333721384406,
|
| 721 |
+
"dapo/num_sampling_attempts": 4.0,
|
| 722 |
+
"dapo/sampling_efficiency": 30.119047619047617,
|
| 723 |
+
"dapo/total_prompts_processed": 24.0,
|
| 724 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 725 |
+
"epoch": 0.04342857142857143,
|
| 726 |
+
"grad_norm": 0.0923333689570427,
|
| 727 |
+
"kl": 0.0031156539916992188,
|
| 728 |
+
"learning_rate": 8.145033635316128e-07,
|
| 729 |
+
"loss": 0.053,
|
| 730 |
+
"reward": 0.45120809972286224,
|
| 731 |
+
"reward_std": 0.9732232913374901,
|
| 732 |
+
"step": 38
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"clip_fraction": 0.0,
|
| 736 |
+
"completion_length": 2787.031280517578,
|
| 737 |
+
"dapo/avg_reward_std": 0.1930955442644301,
|
| 738 |
+
"dapo/filter_reward_index": 0.0,
|
| 739 |
+
"dapo/kept_prompts_ratio": 0.23412698933056422,
|
| 740 |
+
"dapo/num_sampling_attempts": 5.25,
|
| 741 |
+
"dapo/sampling_efficiency": 23.244047619047617,
|
| 742 |
+
"dapo/total_prompts_processed": 31.5,
|
| 743 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 744 |
+
"epoch": 0.044571428571428574,
|
| 745 |
+
"grad_norm": 0.12707453966140747,
|
| 746 |
+
"kl": 0.006325244903564453,
|
| 747 |
+
"learning_rate": 8.01636806561836e-07,
|
| 748 |
+
"loss": 0.0905,
|
| 749 |
+
"reward": 0.5048832832835615,
|
| 750 |
+
"reward_std": 0.9330806732177734,
|
| 751 |
+
"step": 39
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"clip_fraction": 0.0,
|
| 755 |
+
"completion_length": 2921.6180572509766,
|
| 756 |
+
"dapo/avg_reward_std": 0.25906160804960465,
|
| 757 |
+
"dapo/filter_reward_index": 0.0,
|
| 758 |
+
"dapo/kept_prompts_ratio": 0.3009259340663751,
|
| 759 |
+
"dapo/num_sampling_attempts": 4.5,
|
| 760 |
+
"dapo/sampling_efficiency": 31.562499999999996,
|
| 761 |
+
"dapo/total_prompts_processed": 27.0,
|
| 762 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 763 |
+
"epoch": 0.045714285714285714,
|
| 764 |
+
"grad_norm": 0.1152920126914978,
|
| 765 |
+
"kl": 0.004504203796386719,
|
| 766 |
+
"learning_rate": 7.884636689049422e-07,
|
| 767 |
+
"loss": 0.0443,
|
| 768 |
+
"reward": 0.3671413380652666,
|
| 769 |
+
"reward_std": 0.9126428663730621,
|
| 770 |
+
"step": 40
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"clip_fraction": 0.0,
|
| 774 |
+
"completion_length": 3100.8194732666016,
|
| 775 |
+
"dapo/avg_reward_std": 0.26266304695087933,
|
| 776 |
+
"dapo/filter_reward_index": 0.0,
|
| 777 |
+
"dapo/kept_prompts_ratio": 0.3985507280930229,
|
| 778 |
+
"dapo/num_sampling_attempts": 2.875,
|
| 779 |
+
"dapo/sampling_efficiency": 59.895833333333336,
|
| 780 |
+
"dapo/total_prompts_processed": 17.25,
|
| 781 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 782 |
+
"epoch": 0.046857142857142854,
|
| 783 |
+
"grad_norm": 0.1462322324514389,
|
| 784 |
+
"kl": 0.0058536529541015625,
|
| 785 |
+
"learning_rate": 7.75e-07,
|
| 786 |
+
"loss": 0.0836,
|
| 787 |
+
"reward": 0.6537042334675789,
|
| 788 |
+
"reward_std": 0.9643120691180229,
|
| 789 |
+
"step": 41
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"clip_fraction": 0.0,
|
| 793 |
+
"completion_length": 3083.8541870117188,
|
| 794 |
+
"dapo/avg_reward_std": 0.2028282030540354,
|
| 795 |
+
"dapo/filter_reward_index": 0.0,
|
| 796 |
+
"dapo/kept_prompts_ratio": 0.27941177215646296,
|
| 797 |
+
"dapo/num_sampling_attempts": 4.25,
|
| 798 |
+
"dapo/sampling_efficiency": 34.61309523809524,
|
| 799 |
+
"dapo/total_prompts_processed": 25.5,
|
| 800 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 801 |
+
"epoch": 0.048,
|
| 802 |
+
"grad_norm": 0.11620575189590454,
|
| 803 |
+
"kl": 0.005963563919067383,
|
| 804 |
+
"learning_rate": 7.612622032536507e-07,
|
| 805 |
+
"loss": 0.0756,
|
| 806 |
+
"reward": 0.6132493373006582,
|
| 807 |
+
"reward_std": 0.9271278157830238,
|
| 808 |
+
"step": 42
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"clip_fraction": 0.0,
|
| 812 |
+
"completion_length": 2860.6840209960938,
|
| 813 |
+
"dapo/avg_reward_std": 0.2537354379892349,
|
| 814 |
+
"dapo/filter_reward_index": 0.0,
|
| 815 |
+
"dapo/kept_prompts_ratio": 0.31666667262713116,
|
| 816 |
+
"dapo/num_sampling_attempts": 3.75,
|
| 817 |
+
"dapo/sampling_efficiency": 27.916666666666664,
|
| 818 |
+
"dapo/total_prompts_processed": 22.5,
|
| 819 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 820 |
+
"epoch": 0.04914285714285714,
|
| 821 |
+
"grad_norm": 0.15706917643547058,
|
| 822 |
+
"kl": 0.012288570404052734,
|
| 823 |
+
"learning_rate": 7.472670160550848e-07,
|
| 824 |
+
"loss": 0.0864,
|
| 825 |
+
"reward": 0.4896182883530855,
|
| 826 |
+
"reward_std": 0.9406783953309059,
|
| 827 |
+
"step": 43
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"clip_fraction": 0.0,
|
| 831 |
+
"completion_length": 3230.951416015625,
|
| 832 |
+
"dapo/avg_reward_std": 0.2785276919603348,
|
| 833 |
+
"dapo/filter_reward_index": 0.0,
|
| 834 |
+
"dapo/kept_prompts_ratio": 0.4047619104385376,
|
| 835 |
+
"dapo/num_sampling_attempts": 3.5,
|
| 836 |
+
"dapo/sampling_efficiency": 35.20833333333333,
|
| 837 |
+
"dapo/total_prompts_processed": 21.0,
|
| 838 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 839 |
+
"epoch": 0.05028571428571429,
|
| 840 |
+
"grad_norm": 0.10281670838594437,
|
| 841 |
+
"kl": 0.0028905868530273438,
|
| 842 |
+
"learning_rate": 7.330314893841101e-07,
|
| 843 |
+
"loss": 0.0474,
|
| 844 |
+
"reward": 0.5266857808455825,
|
| 845 |
+
"reward_std": 0.9769049882888794,
|
| 846 |
+
"step": 44
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"clip_fraction": 0.0,
|
| 850 |
+
"completion_length": 2598.888885498047,
|
| 851 |
+
"dapo/avg_reward_std": 0.25520460651471066,
|
| 852 |
+
"dapo/filter_reward_index": 0.0,
|
| 853 |
+
"dapo/kept_prompts_ratio": 0.3205128231873879,
|
| 854 |
+
"dapo/num_sampling_attempts": 3.25,
|
| 855 |
+
"dapo/sampling_efficiency": 54.61309523809524,
|
| 856 |
+
"dapo/total_prompts_processed": 19.5,
|
| 857 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 858 |
+
"epoch": 0.05142857142857143,
|
| 859 |
+
"grad_norm": 0.20818237960338593,
|
| 860 |
+
"kl": 0.0046825408935546875,
|
| 861 |
+
"learning_rate": 7.185729670371604e-07,
|
| 862 |
+
"loss": 0.111,
|
| 863 |
+
"reward": 0.8208948634564877,
|
| 864 |
+
"reward_std": 0.9365335553884506,
|
| 865 |
+
"step": 45
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"clip_fraction": 0.0,
|
| 869 |
+
"completion_length": 2529.66316986084,
|
| 870 |
+
"dapo/avg_reward_std": 0.23859836988978916,
|
| 871 |
+
"dapo/filter_reward_index": 0.0,
|
| 872 |
+
"dapo/kept_prompts_ratio": 0.2916666753590107,
|
| 873 |
+
"dapo/num_sampling_attempts": 4.5,
|
| 874 |
+
"dapo/sampling_efficiency": 25.535714285714285,
|
| 875 |
+
"dapo/total_prompts_processed": 27.0,
|
| 876 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 877 |
+
"epoch": 0.052571428571428575,
|
| 878 |
+
"grad_norm": 0.12924660742282867,
|
| 879 |
+
"kl": 0.05440711975097656,
|
| 880 |
+
"learning_rate": 7.039090644965509e-07,
|
| 881 |
+
"loss": 0.058,
|
| 882 |
+
"reward": 0.5307688321918249,
|
| 883 |
+
"reward_std": 0.9391194358468056,
|
| 884 |
+
"step": 46
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"clip_fraction": 0.0,
|
| 888 |
+
"completion_length": 2737.288230895996,
|
| 889 |
+
"dapo/avg_reward_std": 0.25754969901052016,
|
| 890 |
+
"dapo/filter_reward_index": 0.0,
|
| 891 |
+
"dapo/kept_prompts_ratio": 0.3678160998327979,
|
| 892 |
+
"dapo/num_sampling_attempts": 3.625,
|
| 893 |
+
"dapo/sampling_efficiency": 41.14583333333333,
|
| 894 |
+
"dapo/total_prompts_processed": 21.75,
|
| 895 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 896 |
+
"epoch": 0.053714285714285714,
|
| 897 |
+
"grad_norm": 0.1452113687992096,
|
| 898 |
+
"kl": 0.01877737045288086,
|
| 899 |
+
"learning_rate": 6.890576474687263e-07,
|
| 900 |
+
"loss": 0.0601,
|
| 901 |
+
"reward": 0.5596560873091221,
|
| 902 |
+
"reward_std": 0.9911476969718933,
|
| 903 |
+
"step": 47
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"clip_fraction": 0.0,
|
| 907 |
+
"completion_length": 2543.0694885253906,
|
| 908 |
+
"dapo/avg_reward_std": 0.2434165603839434,
|
| 909 |
+
"dapo/filter_reward_index": 0.0,
|
| 910 |
+
"dapo/kept_prompts_ratio": 0.3717948794364929,
|
| 911 |
+
"dapo/num_sampling_attempts": 3.25,
|
| 912 |
+
"dapo/sampling_efficiency": 34.37499999999999,
|
| 913 |
+
"dapo/total_prompts_processed": 19.5,
|
| 914 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 915 |
+
"epoch": 0.054857142857142854,
|
| 916 |
+
"grad_norm": 0.15664616227149963,
|
| 917 |
+
"kl": 0.008816719055175781,
|
| 918 |
+
"learning_rate": 6.740368101176495e-07,
|
| 919 |
+
"loss": 0.0783,
|
| 920 |
+
"reward": 0.7667456082999706,
|
| 921 |
+
"reward_std": 0.9330208897590637,
|
| 922 |
+
"step": 48
|
| 923 |
+
},
|
| 924 |
+
{
|
| 925 |
+
"clip_fraction": 0.0,
|
| 926 |
+
"completion_length": 3054.357666015625,
|
| 927 |
+
"dapo/avg_reward_std": 0.16933719928448016,
|
| 928 |
+
"dapo/filter_reward_index": 0.0,
|
| 929 |
+
"dapo/kept_prompts_ratio": 0.22222222693455526,
|
| 930 |
+
"dapo/num_sampling_attempts": 4.875,
|
| 931 |
+
"dapo/sampling_efficiency": 26.5625,
|
| 932 |
+
"dapo/total_prompts_processed": 29.25,
|
| 933 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 934 |
+
"epoch": 0.056,
|
| 935 |
+
"grad_norm": 0.13884593546390533,
|
| 936 |
+
"kl": 0.00569915771484375,
|
| 937 |
+
"learning_rate": 6.588648530198504e-07,
|
| 938 |
+
"loss": 0.0645,
|
| 939 |
+
"reward": 0.7750914767384529,
|
| 940 |
+
"reward_std": 0.9781928732991219,
|
| 941 |
+
"step": 49
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"clip_fraction": 0.0,
|
| 945 |
+
"completion_length": 3030.9652709960938,
|
| 946 |
+
"dapo/avg_reward_std": 0.2089548914721518,
|
| 947 |
+
"dapo/filter_reward_index": 0.0,
|
| 948 |
+
"dapo/kept_prompts_ratio": 0.28282828629016876,
|
| 949 |
+
"dapo/num_sampling_attempts": 4.125,
|
| 950 |
+
"dapo/sampling_efficiency": 33.779761904761905,
|
| 951 |
+
"dapo/total_prompts_processed": 24.75,
|
| 952 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 953 |
+
"epoch": 0.05714285714285714,
|
| 954 |
+
"grad_norm": 0.13095000386238098,
|
| 955 |
+
"kl": 0.005908966064453125,
|
| 956 |
+
"learning_rate": 6.435602608679916e-07,
|
| 957 |
+
"loss": 0.0854,
|
| 958 |
+
"reward": 0.7626989148557186,
|
| 959 |
+
"reward_std": 0.9684056863188744,
|
| 960 |
+
"step": 50
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"clip_fraction": 0.0,
|
| 964 |
+
"completion_length": 3176.8819274902344,
|
| 965 |
+
"dapo/avg_reward_std": 0.2258962235516972,
|
| 966 |
+
"dapo/filter_reward_index": 0.0,
|
| 967 |
+
"dapo/kept_prompts_ratio": 0.29629630057348144,
|
| 968 |
+
"dapo/num_sampling_attempts": 4.5,
|
| 969 |
+
"dapo/sampling_efficiency": 33.25892857142857,
|
| 970 |
+
"dapo/total_prompts_processed": 27.0,
|
| 971 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 972 |
+
"epoch": 0.05828571428571429,
|
| 973 |
+
"grad_norm": 0.11041354387998581,
|
| 974 |
+
"kl": 0.002262115478515625,
|
| 975 |
+
"learning_rate": 6.281416799501187e-07,
|
| 976 |
+
"loss": 0.0892,
|
| 977 |
+
"reward": 0.6493857521563768,
|
| 978 |
+
"reward_std": 0.9608959034085274,
|
| 979 |
+
"step": 51
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"clip_fraction": 0.0,
|
| 983 |
+
"completion_length": 2991.208366394043,
|
| 984 |
+
"dapo/avg_reward_std": 0.23346692004374095,
|
| 985 |
+
"dapo/filter_reward_index": 0.0,
|
| 986 |
+
"dapo/kept_prompts_ratio": 0.3392857201397419,
|
| 987 |
+
"dapo/num_sampling_attempts": 3.5,
|
| 988 |
+
"dapo/sampling_efficiency": 52.70833333333333,
|
| 989 |
+
"dapo/total_prompts_processed": 21.0,
|
| 990 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 991 |
+
"epoch": 0.05942857142857143,
|
| 992 |
+
"grad_norm": 0.13827170431613922,
|
| 993 |
+
"kl": 0.014558792114257812,
|
| 994 |
+
"learning_rate": 6.126278954320294e-07,
|
| 995 |
+
"loss": 0.0435,
|
| 996 |
+
"reward": 0.5274152141064405,
|
| 997 |
+
"reward_std": 0.9937505125999451,
|
| 998 |
+
"step": 52
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"clip_fraction": 0.0,
|
| 1002 |
+
"completion_length": 2921.013946533203,
|
| 1003 |
+
"dapo/avg_reward_std": 0.2715419438378564,
|
| 1004 |
+
"dapo/filter_reward_index": 0.0,
|
| 1005 |
+
"dapo/kept_prompts_ratio": 0.3390804626818361,
|
| 1006 |
+
"dapo/num_sampling_attempts": 3.625,
|
| 1007 |
+
"dapo/sampling_efficiency": 51.5625,
|
| 1008 |
+
"dapo/total_prompts_processed": 21.75,
|
| 1009 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1010 |
+
"epoch": 0.060571428571428575,
|
| 1011 |
+
"grad_norm": 0.09735170006752014,
|
| 1012 |
+
"kl": 0.009172439575195312,
|
| 1013 |
+
"learning_rate": 5.97037808470444e-07,
|
| 1014 |
+
"loss": 0.0541,
|
| 1015 |
+
"reward": 0.7217882052063942,
|
| 1016 |
+
"reward_std": 0.9594404622912407,
|
| 1017 |
+
"step": 53
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"clip_fraction": 0.0,
|
| 1021 |
+
"completion_length": 3133.46875,
|
| 1022 |
+
"dapo/avg_reward_std": 0.2624934350068753,
|
| 1023 |
+
"dapo/filter_reward_index": 0.0,
|
| 1024 |
+
"dapo/kept_prompts_ratio": 0.35256411077884525,
|
| 1025 |
+
"dapo/num_sampling_attempts": 3.25,
|
| 1026 |
+
"dapo/sampling_efficiency": 41.041666666666664,
|
| 1027 |
+
"dapo/total_prompts_processed": 19.5,
|
| 1028 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1029 |
+
"epoch": 0.061714285714285715,
|
| 1030 |
+
"grad_norm": 0.10414379835128784,
|
| 1031 |
+
"kl": 0.010915756225585938,
|
| 1032 |
+
"learning_rate": 5.813904131848564e-07,
|
| 1033 |
+
"loss": 0.061,
|
| 1034 |
+
"reward": 0.5302782151848078,
|
| 1035 |
+
"reward_std": 0.9707583636045456,
|
| 1036 |
+
"step": 54
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"clip_fraction": 0.0,
|
| 1040 |
+
"completion_length": 3010.5938110351562,
|
| 1041 |
+
"dapo/avg_reward_std": 0.21664191484451295,
|
| 1042 |
+
"dapo/filter_reward_index": 0.0,
|
| 1043 |
+
"dapo/kept_prompts_ratio": 0.24444444941149818,
|
| 1044 |
+
"dapo/num_sampling_attempts": 5.625,
|
| 1045 |
+
"dapo/sampling_efficiency": 19.791666666666664,
|
| 1046 |
+
"dapo/total_prompts_processed": 33.75,
|
| 1047 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1048 |
+
"epoch": 0.06285714285714286,
|
| 1049 |
+
"grad_norm": 0.11232081800699234,
|
| 1050 |
+
"kl": 0.012262344360351562,
|
| 1051 |
+
"learning_rate": 5.657047735161255e-07,
|
| 1052 |
+
"loss": 0.0561,
|
| 1053 |
+
"reward": 0.5284321270883083,
|
| 1054 |
+
"reward_std": 0.9165859594941139,
|
| 1055 |
+
"step": 55
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"clip_fraction": 0.0,
|
| 1059 |
+
"completion_length": 3144.951416015625,
|
| 1060 |
+
"dapo/avg_reward_std": 0.2279102834207671,
|
| 1061 |
+
"dapo/filter_reward_index": 0.0,
|
| 1062 |
+
"dapo/kept_prompts_ratio": 0.34523809807641165,
|
| 1063 |
+
"dapo/num_sampling_attempts": 3.5,
|
| 1064 |
+
"dapo/sampling_efficiency": 45.32738095238095,
|
| 1065 |
+
"dapo/total_prompts_processed": 21.0,
|
| 1066 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1067 |
+
"epoch": 0.064,
|
| 1068 |
+
"grad_norm": 0.13161872327327728,
|
| 1069 |
+
"kl": 0.007735252380371094,
|
| 1070 |
+
"learning_rate": 5.5e-07,
|
| 1071 |
+
"loss": 0.0717,
|
| 1072 |
+
"reward": 0.6519734226167202,
|
| 1073 |
+
"reward_std": 0.9642440155148506,
|
| 1074 |
+
"step": 56
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"clip_fraction": 0.0,
|
| 1078 |
+
"completion_length": 3222.6111450195312,
|
| 1079 |
+
"dapo/avg_reward_std": 0.2675224413042483,
|
| 1080 |
+
"dapo/filter_reward_index": 0.0,
|
| 1081 |
+
"dapo/kept_prompts_ratio": 0.4492753724689069,
|
| 1082 |
+
"dapo/num_sampling_attempts": 2.875,
|
| 1083 |
+
"dapo/sampling_efficiency": 45.535714285714285,
|
| 1084 |
+
"dapo/total_prompts_processed": 17.25,
|
| 1085 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1086 |
+
"epoch": 0.06514285714285714,
|
| 1087 |
+
"grad_norm": 0.09332293272018433,
|
| 1088 |
+
"kl": 0.0064525604248046875,
|
| 1089 |
+
"learning_rate": 5.342952264838747e-07,
|
| 1090 |
+
"loss": 0.0302,
|
| 1091 |
+
"reward": 0.5501165799796581,
|
| 1092 |
+
"reward_std": 0.9585564360022545,
|
| 1093 |
+
"step": 57
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"clip_fraction": 0.0,
|
| 1097 |
+
"completion_length": 2679.9236907958984,
|
| 1098 |
+
"dapo/avg_reward_std": 0.17708626160254845,
|
| 1099 |
+
"dapo/filter_reward_index": 0.0,
|
| 1100 |
+
"dapo/kept_prompts_ratio": 0.24358974741055414,
|
| 1101 |
+
"dapo/num_sampling_attempts": 4.875,
|
| 1102 |
+
"dapo/sampling_efficiency": 28.91865079365079,
|
| 1103 |
+
"dapo/total_prompts_processed": 29.25,
|
| 1104 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1105 |
+
"epoch": 0.06628571428571428,
|
| 1106 |
+
"grad_norm": 0.16309793293476105,
|
| 1107 |
+
"kl": 0.01690673828125,
|
| 1108 |
+
"learning_rate": 5.186095868151436e-07,
|
| 1109 |
+
"loss": 0.0846,
|
| 1110 |
+
"reward": 0.8469000309705734,
|
| 1111 |
+
"reward_std": 0.9497043192386627,
|
| 1112 |
+
"step": 58
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"clip_fraction": 0.0,
|
| 1116 |
+
"completion_length": 2847.048629760742,
|
| 1117 |
+
"dapo/avg_reward_std": 0.2622834824282548,
|
| 1118 |
+
"dapo/filter_reward_index": 0.0,
|
| 1119 |
+
"dapo/kept_prompts_ratio": 0.3563218476443455,
|
| 1120 |
+
"dapo/num_sampling_attempts": 3.625,
|
| 1121 |
+
"dapo/sampling_efficiency": 29.999999999999993,
|
| 1122 |
+
"dapo/total_prompts_processed": 21.75,
|
| 1123 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1124 |
+
"epoch": 0.06742857142857143,
|
| 1125 |
+
"grad_norm": 0.09638360142707825,
|
| 1126 |
+
"kl": 0.0057086944580078125,
|
| 1127 |
+
"learning_rate": 5.02962191529556e-07,
|
| 1128 |
+
"loss": 0.0634,
|
| 1129 |
+
"reward": 0.6089529246091843,
|
| 1130 |
+
"reward_std": 0.9450863003730774,
|
| 1131 |
+
"step": 59
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"clip_fraction": 0.0,
|
| 1135 |
+
"completion_length": 3119.9132385253906,
|
| 1136 |
+
"dapo/avg_reward_std": 0.19833819533503333,
|
| 1137 |
+
"dapo/filter_reward_index": 0.0,
|
| 1138 |
+
"dapo/kept_prompts_ratio": 0.2674418656631958,
|
| 1139 |
+
"dapo/num_sampling_attempts": 5.375,
|
| 1140 |
+
"dapo/sampling_efficiency": 29.563492063492063,
|
| 1141 |
+
"dapo/total_prompts_processed": 32.25,
|
| 1142 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1143 |
+
"epoch": 0.06857142857142857,
|
| 1144 |
+
"grad_norm": 0.1252850890159607,
|
| 1145 |
+
"kl": 0.008715629577636719,
|
| 1146 |
+
"learning_rate": 4.873721045679706e-07,
|
| 1147 |
+
"loss": 0.0666,
|
| 1148 |
+
"reward": 0.5249154977500439,
|
| 1149 |
+
"reward_std": 0.947566568851471,
|
| 1150 |
+
"step": 60
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"clip_fraction": 0.0,
|
| 1154 |
+
"completion_length": 2844.795181274414,
|
| 1155 |
+
"dapo/avg_reward_std": 0.2648707001373686,
|
| 1156 |
+
"dapo/filter_reward_index": 0.0,
|
| 1157 |
+
"dapo/kept_prompts_ratio": 0.35632184610284606,
|
| 1158 |
+
"dapo/num_sampling_attempts": 3.625,
|
| 1159 |
+
"dapo/sampling_efficiency": 39.791666666666664,
|
| 1160 |
+
"dapo/total_prompts_processed": 21.75,
|
| 1161 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1162 |
+
"epoch": 0.06971428571428571,
|
| 1163 |
+
"grad_norm": 0.10366301238536835,
|
| 1164 |
+
"kl": 0.056069374084472656,
|
| 1165 |
+
"learning_rate": 4.7185832004988133e-07,
|
| 1166 |
+
"loss": 0.037,
|
| 1167 |
+
"reward": 0.5161248315125704,
|
| 1168 |
+
"reward_std": 0.9692364558577538,
|
| 1169 |
+
"step": 61
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"clip_fraction": 0.0,
|
| 1173 |
+
"completion_length": 3053.951446533203,
|
| 1174 |
+
"dapo/avg_reward_std": 0.21576767837679064,
|
| 1175 |
+
"dapo/filter_reward_index": 0.0,
|
| 1176 |
+
"dapo/kept_prompts_ratio": 0.25225225574261434,
|
| 1177 |
+
"dapo/num_sampling_attempts": 4.625,
|
| 1178 |
+
"dapo/sampling_efficiency": 37.013888888888886,
|
| 1179 |
+
"dapo/total_prompts_processed": 27.75,
|
| 1180 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1181 |
+
"epoch": 0.07085714285714285,
|
| 1182 |
+
"grad_norm": 0.14441759884357452,
|
| 1183 |
+
"kl": 0.009164810180664062,
|
| 1184 |
+
"learning_rate": 4.5643973913200837e-07,
|
| 1185 |
+
"loss": 0.0609,
|
| 1186 |
+
"reward": 0.6510533541440964,
|
| 1187 |
+
"reward_std": 0.9361515268683434,
|
| 1188 |
+
"step": 62
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"clip_fraction": 0.0,
|
| 1192 |
+
"completion_length": 3326.781280517578,
|
| 1193 |
+
"dapo/avg_reward_std": 0.2158982500885472,
|
| 1194 |
+
"dapo/filter_reward_index": 0.0,
|
| 1195 |
+
"dapo/kept_prompts_ratio": 0.3181818254066236,
|
| 1196 |
+
"dapo/num_sampling_attempts": 4.125,
|
| 1197 |
+
"dapo/sampling_efficiency": 44.49404761904761,
|
| 1198 |
+
"dapo/total_prompts_processed": 24.75,
|
| 1199 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1200 |
+
"epoch": 0.072,
|
| 1201 |
+
"grad_norm": 0.12127737700939178,
|
| 1202 |
+
"kl": 0.031108856201171875,
|
| 1203 |
+
"learning_rate": 4.4113514698014953e-07,
|
| 1204 |
+
"loss": 0.0463,
|
| 1205 |
+
"reward": 0.45860649459064007,
|
| 1206 |
+
"reward_std": 0.9209225550293922,
|
| 1207 |
+
"step": 63
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"clip_fraction": 0.0,
|
| 1211 |
+
"completion_length": 3208.6319885253906,
|
| 1212 |
+
"dapo/avg_reward_std": 0.28419332668699065,
|
| 1213 |
+
"dapo/filter_reward_index": 0.0,
|
| 1214 |
+
"dapo/kept_prompts_ratio": 0.3563218440475135,
|
| 1215 |
+
"dapo/num_sampling_attempts": 3.625,
|
| 1216 |
+
"dapo/sampling_efficiency": 53.591269841269835,
|
| 1217 |
+
"dapo/total_prompts_processed": 21.75,
|
| 1218 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1219 |
+
"epoch": 0.07314285714285715,
|
| 1220 |
+
"grad_norm": 0.13326792418956757,
|
| 1221 |
+
"kl": 0.0061321258544921875,
|
| 1222 |
+
"learning_rate": 4.2596318988235037e-07,
|
| 1223 |
+
"loss": 0.0614,
|
| 1224 |
+
"reward": 0.5644803196191788,
|
| 1225 |
+
"reward_std": 0.9919605851173401,
|
| 1226 |
+
"step": 64
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"clip_fraction": 0.0,
|
| 1230 |
+
"completion_length": 2597.437530517578,
|
| 1231 |
+
"dapo/avg_reward_std": 0.2766759342380932,
|
| 1232 |
+
"dapo/filter_reward_index": 0.0,
|
| 1233 |
+
"dapo/kept_prompts_ratio": 0.3928571529686451,
|
| 1234 |
+
"dapo/num_sampling_attempts": 3.5,
|
| 1235 |
+
"dapo/sampling_efficiency": 32.08333333333333,
|
| 1236 |
+
"dapo/total_prompts_processed": 21.0,
|
| 1237 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1238 |
+
"epoch": 0.07428571428571429,
|
| 1239 |
+
"grad_norm": 0.10434358566999435,
|
| 1240 |
+
"kl": 0.049472808837890625,
|
| 1241 |
+
"learning_rate": 4.1094235253127374e-07,
|
| 1242 |
+
"loss": 0.0312,
|
| 1243 |
+
"reward": 0.393868962302804,
|
| 1244 |
+
"reward_std": 0.9459580257534981,
|
| 1245 |
+
"step": 65
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"clip_fraction": 0.0,
|
| 1249 |
+
"completion_length": 2630.0833587646484,
|
| 1250 |
+
"dapo/avg_reward_std": 0.25837596147148695,
|
| 1251 |
+
"dapo/filter_reward_index": 0.0,
|
| 1252 |
+
"dapo/kept_prompts_ratio": 0.35802469595714853,
|
| 1253 |
+
"dapo/num_sampling_attempts": 3.375,
|
| 1254 |
+
"dapo/sampling_efficiency": 31.666666666666664,
|
| 1255 |
+
"dapo/total_prompts_processed": 20.25,
|
| 1256 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1257 |
+
"epoch": 0.07542857142857143,
|
| 1258 |
+
"grad_norm": 0.11327924579381943,
|
| 1259 |
+
"kl": 0.23560714721679688,
|
| 1260 |
+
"learning_rate": 3.9609093550344907e-07,
|
| 1261 |
+
"loss": 0.0563,
|
| 1262 |
+
"reward": 0.674448698759079,
|
| 1263 |
+
"reward_std": 0.9591537117958069,
|
| 1264 |
+
"step": 66
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"clip_fraction": 0.0,
|
| 1268 |
+
"completion_length": 3343.3159790039062,
|
| 1269 |
+
"dapo/avg_reward_std": 0.2785816714167595,
|
| 1270 |
+
"dapo/filter_reward_index": 0.0,
|
| 1271 |
+
"dapo/kept_prompts_ratio": 0.41666667101283866,
|
| 1272 |
+
"dapo/num_sampling_attempts": 3.0,
|
| 1273 |
+
"dapo/sampling_efficiency": 42.08333333333333,
|
| 1274 |
+
"dapo/total_prompts_processed": 18.0,
|
| 1275 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1276 |
+
"epoch": 0.07657142857142857,
|
| 1277 |
+
"grad_norm": 0.10341926664113998,
|
| 1278 |
+
"kl": 0.005463600158691406,
|
| 1279 |
+
"learning_rate": 3.8142703296283953e-07,
|
| 1280 |
+
"loss": 0.0653,
|
| 1281 |
+
"reward": 0.42072685062885284,
|
| 1282 |
+
"reward_std": 0.9649706333875656,
|
| 1283 |
+
"step": 67
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"clip_fraction": 0.0,
|
| 1287 |
+
"completion_length": 2880.0590438842773,
|
| 1288 |
+
"dapo/avg_reward_std": 0.2447407204243872,
|
| 1289 |
+
"dapo/filter_reward_index": 0.0,
|
| 1290 |
+
"dapo/kept_prompts_ratio": 0.2916666728754838,
|
| 1291 |
+
"dapo/num_sampling_attempts": 4.5,
|
| 1292 |
+
"dapo/sampling_efficiency": 28.591269841269842,
|
| 1293 |
+
"dapo/total_prompts_processed": 27.0,
|
| 1294 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1295 |
+
"epoch": 0.07771428571428571,
|
| 1296 |
+
"grad_norm": 0.15764088928699493,
|
| 1297 |
+
"kl": 0.011991500854492188,
|
| 1298 |
+
"learning_rate": 3.6696851061588994e-07,
|
| 1299 |
+
"loss": 0.1004,
|
| 1300 |
+
"reward": 0.537701515480876,
|
| 1301 |
+
"reward_std": 0.9107673466205597,
|
| 1302 |
+
"step": 68
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"clip_fraction": 0.0,
|
| 1306 |
+
"completion_length": 2839.0069580078125,
|
| 1307 |
+
"dapo/avg_reward_std": 0.21828406437849388,
|
| 1308 |
+
"dapo/filter_reward_index": 0.0,
|
| 1309 |
+
"dapo/kept_prompts_ratio": 0.26495726865071517,
|
| 1310 |
+
"dapo/num_sampling_attempts": 4.875,
|
| 1311 |
+
"dapo/sampling_efficiency": 32.39583333333333,
|
| 1312 |
+
"dapo/total_prompts_processed": 29.25,
|
| 1313 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1314 |
+
"epoch": 0.07885714285714286,
|
| 1315 |
+
"grad_norm": 0.1426348239183426,
|
| 1316 |
+
"kl": 0.16588592529296875,
|
| 1317 |
+
"learning_rate": 3.5273298394491515e-07,
|
| 1318 |
+
"loss": 0.065,
|
| 1319 |
+
"reward": 0.5752462260425091,
|
| 1320 |
+
"reward_std": 0.9265653118491173,
|
| 1321 |
+
"step": 69
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"clip_fraction": 0.0,
|
| 1325 |
+
"completion_length": 3154.9479370117188,
|
| 1326 |
+
"dapo/avg_reward_std": 0.24686445650600253,
|
| 1327 |
+
"dapo/filter_reward_index": 0.0,
|
| 1328 |
+
"dapo/kept_prompts_ratio": 0.40476191185769583,
|
| 1329 |
+
"dapo/num_sampling_attempts": 2.625,
|
| 1330 |
+
"dapo/sampling_efficiency": 58.75,
|
| 1331 |
+
"dapo/total_prompts_processed": 15.75,
|
| 1332 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1333 |
+
"epoch": 0.08,
|
| 1334 |
+
"grad_norm": 0.09042708575725555,
|
| 1335 |
+
"kl": 0.015224456787109375,
|
| 1336 |
+
"learning_rate": 3.387377967463493e-07,
|
| 1337 |
+
"loss": 0.0278,
|
| 1338 |
+
"reward": 0.5091124139726162,
|
| 1339 |
+
"reward_std": 0.9951601624488831,
|
| 1340 |
+
"step": 70
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"clip_fraction": 0.0,
|
| 1344 |
+
"completion_length": 2558.7118377685547,
|
| 1345 |
+
"dapo/avg_reward_std": 0.24922772922686168,
|
| 1346 |
+
"dapo/filter_reward_index": 0.0,
|
| 1347 |
+
"dapo/kept_prompts_ratio": 0.35119048452803064,
|
| 1348 |
+
"dapo/num_sampling_attempts": 3.5,
|
| 1349 |
+
"dapo/sampling_efficiency": 41.979166666666664,
|
| 1350 |
+
"dapo/total_prompts_processed": 21.0,
|
| 1351 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1352 |
+
"epoch": 0.08114285714285714,
|
| 1353 |
+
"grad_norm": 0.18424691259860992,
|
| 1354 |
+
"kl": 0.012338638305664062,
|
| 1355 |
+
"learning_rate": 3.250000000000001e-07,
|
| 1356 |
+
"loss": 0.135,
|
| 1357 |
+
"reward": 0.80832345969975,
|
| 1358 |
+
"reward_std": 0.9256910160183907,
|
| 1359 |
+
"step": 71
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"clip_fraction": 0.0,
|
| 1363 |
+
"completion_length": 2797.5659790039062,
|
| 1364 |
+
"dapo/avg_reward_std": 0.3421325541677929,
|
| 1365 |
+
"dapo/filter_reward_index": 0.0,
|
| 1366 |
+
"dapo/kept_prompts_ratio": 0.4603174655210404,
|
| 1367 |
+
"dapo/num_sampling_attempts": 2.625,
|
| 1368 |
+
"dapo/sampling_efficiency": 52.916666666666664,
|
| 1369 |
+
"dapo/total_prompts_processed": 15.75,
|
| 1370 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1371 |
+
"epoch": 0.08228571428571428,
|
| 1372 |
+
"grad_norm": 0.10505988448858261,
|
| 1373 |
+
"kl": 0.027385711669921875,
|
| 1374 |
+
"learning_rate": 3.115363310950578e-07,
|
| 1375 |
+
"loss": 0.0435,
|
| 1376 |
+
"reward": 0.5198174491524696,
|
| 1377 |
+
"reward_std": 0.932801865041256,
|
| 1378 |
+
"step": 72
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"clip_fraction": 0.0,
|
| 1382 |
+
"completion_length": 3024.5243225097656,
|
| 1383 |
+
"dapo/avg_reward_std": 0.26287247288611626,
|
| 1384 |
+
"dapo/filter_reward_index": 0.0,
|
| 1385 |
+
"dapo/kept_prompts_ratio": 0.33333334038334506,
|
| 1386 |
+
"dapo/num_sampling_attempts": 3.875,
|
| 1387 |
+
"dapo/sampling_efficiency": 40.0297619047619,
|
| 1388 |
+
"dapo/total_prompts_processed": 23.25,
|
| 1389 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1390 |
+
"epoch": 0.08342857142857144,
|
| 1391 |
+
"grad_norm": 0.09084703773260117,
|
| 1392 |
+
"kl": 0.09223747253417969,
|
| 1393 |
+
"learning_rate": 2.9836319343816397e-07,
|
| 1394 |
+
"loss": 0.0314,
|
| 1395 |
+
"reward": 0.3449883237481117,
|
| 1396 |
+
"reward_std": 0.9521737843751907,
|
| 1397 |
+
"step": 73
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"clip_fraction": 0.0,
|
| 1401 |
+
"completion_length": 2648.7257080078125,
|
| 1402 |
+
"dapo/avg_reward_std": 0.2678213362340574,
|
| 1403 |
+
"dapo/filter_reward_index": 0.0,
|
| 1404 |
+
"dapo/kept_prompts_ratio": 0.38271605582148943,
|
| 1405 |
+
"dapo/num_sampling_attempts": 3.375,
|
| 1406 |
+
"dapo/sampling_efficiency": 40.0,
|
| 1407 |
+
"dapo/total_prompts_processed": 20.25,
|
| 1408 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1409 |
+
"epoch": 0.08457142857142858,
|
| 1410 |
+
"grad_norm": 0.15155129134655,
|
| 1411 |
+
"kl": 1.0743579864501953,
|
| 1412 |
+
"learning_rate": 2.854966364683872e-07,
|
| 1413 |
+
"loss": 0.0851,
|
| 1414 |
+
"reward": 0.7227161657065153,
|
| 1415 |
+
"reward_std": 0.9239719212055206,
|
| 1416 |
+
"step": 74
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"clip_fraction": 0.0,
|
| 1420 |
+
"completion_length": 2659.388900756836,
|
| 1421 |
+
"dapo/avg_reward_std": 0.28101804742106684,
|
| 1422 |
+
"dapo/filter_reward_index": 0.0,
|
| 1423 |
+
"dapo/kept_prompts_ratio": 0.37037037699310865,
|
| 1424 |
+
"dapo/num_sampling_attempts": 3.375,
|
| 1425 |
+
"dapo/sampling_efficiency": 34.791666666666664,
|
| 1426 |
+
"dapo/total_prompts_processed": 20.25,
|
| 1427 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1428 |
+
"epoch": 0.08571428571428572,
|
| 1429 |
+
"grad_norm": 0.1127755343914032,
|
| 1430 |
+
"kl": 0.02587890625,
|
| 1431 |
+
"learning_rate": 2.729523361034538e-07,
|
| 1432 |
+
"loss": 0.0523,
|
| 1433 |
+
"reward": 0.7372388476505876,
|
| 1434 |
+
"reward_std": 0.918749064207077,
|
| 1435 |
+
"step": 75
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"clip_fraction": 0.0,
|
| 1439 |
+
"completion_length": 2402.364585876465,
|
| 1440 |
+
"dapo/avg_reward_std": 0.26893362632164586,
|
| 1441 |
+
"dapo/filter_reward_index": 0.0,
|
| 1442 |
+
"dapo/kept_prompts_ratio": 0.36538462111583125,
|
| 1443 |
+
"dapo/num_sampling_attempts": 3.25,
|
| 1444 |
+
"dapo/sampling_efficiency": 48.854166666666664,
|
| 1445 |
+
"dapo/total_prompts_processed": 19.5,
|
| 1446 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1447 |
+
"epoch": 0.08685714285714285,
|
| 1448 |
+
"grad_norm": 0.14693324267864227,
|
| 1449 |
+
"kl": 0.12501144409179688,
|
| 1450 |
+
"learning_rate": 2.6074557564105724e-07,
|
| 1451 |
+
"loss": 0.0747,
|
| 1452 |
+
"reward": 0.6182113699615002,
|
| 1453 |
+
"reward_std": 0.9421844929456711,
|
| 1454 |
+
"step": 76
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"clip_fraction": 0.0,
|
| 1458 |
+
"completion_length": 2970.1146392822266,
|
| 1459 |
+
"dapo/avg_reward_std": 0.2118390180170536,
|
| 1460 |
+
"dapo/filter_reward_index": 0.0,
|
| 1461 |
+
"dapo/kept_prompts_ratio": 0.25000000521540644,
|
| 1462 |
+
"dapo/num_sampling_attempts": 5.0,
|
| 1463 |
+
"dapo/sampling_efficiency": 30.53571428571428,
|
| 1464 |
+
"dapo/total_prompts_processed": 30.0,
|
| 1465 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1466 |
+
"epoch": 0.088,
|
| 1467 |
+
"grad_norm": 0.12072475999593735,
|
| 1468 |
+
"kl": 0.05495643615722656,
|
| 1469 |
+
"learning_rate": 2.488912271385139e-07,
|
| 1470 |
+
"loss": 0.0498,
|
| 1471 |
+
"reward": 0.46035338938236237,
|
| 1472 |
+
"reward_std": 0.9146044701337814,
|
| 1473 |
+
"step": 77
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"clip_fraction": 0.0,
|
| 1477 |
+
"completion_length": 2959.0972442626953,
|
| 1478 |
+
"dapo/avg_reward_std": 0.13832776496807733,
|
| 1479 |
+
"dapo/filter_reward_index": 0.0,
|
| 1480 |
+
"dapo/kept_prompts_ratio": 0.1631944477558136,
|
| 1481 |
+
"dapo/num_sampling_attempts": 6.0,
|
| 1482 |
+
"dapo/sampling_efficiency": 30.868055555555557,
|
| 1483 |
+
"dapo/total_prompts_processed": 36.0,
|
| 1484 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1485 |
+
"epoch": 0.08914285714285715,
|
| 1486 |
+
"grad_norm": 0.14289411902427673,
|
| 1487 |
+
"kl": 0.23297691345214844,
|
| 1488 |
+
"learning_rate": 2.374037332934512e-07,
|
| 1489 |
+
"loss": 0.0742,
|
| 1490 |
+
"reward": 0.49553669430315495,
|
| 1491 |
+
"reward_std": 0.9023259580135345,
|
| 1492 |
+
"step": 78
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"clip_fraction": 0.0,
|
| 1496 |
+
"completion_length": 2935.8159942626953,
|
| 1497 |
+
"dapo/avg_reward_std": 0.2931290553374724,
|
| 1498 |
+
"dapo/filter_reward_index": 0.0,
|
| 1499 |
+
"dapo/kept_prompts_ratio": 0.44696970080787485,
|
| 1500 |
+
"dapo/num_sampling_attempts": 2.75,
|
| 1501 |
+
"dapo/sampling_efficiency": 58.854166666666664,
|
| 1502 |
+
"dapo/total_prompts_processed": 16.5,
|
| 1503 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1504 |
+
"epoch": 0.09028571428571429,
|
| 1505 |
+
"grad_norm": 0.13638050854206085,
|
| 1506 |
+
"kl": 0.03482818603515625,
|
| 1507 |
+
"learning_rate": 2.2629708984760706e-07,
|
| 1508 |
+
"loss": 0.0609,
|
| 1509 |
+
"reward": 0.4563083341345191,
|
| 1510 |
+
"reward_std": 0.9425384849309921,
|
| 1511 |
+
"step": 79
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"clip_fraction": 0.0,
|
| 1515 |
+
"completion_length": 3111.340301513672,
|
| 1516 |
+
"dapo/avg_reward_std": 0.22562272967518987,
|
| 1517 |
+
"dapo/filter_reward_index": 0.0,
|
| 1518 |
+
"dapo/kept_prompts_ratio": 0.3063063154349456,
|
| 1519 |
+
"dapo/num_sampling_attempts": 4.625,
|
| 1520 |
+
"dapo/sampling_efficiency": 28.819444444444446,
|
| 1521 |
+
"dapo/total_prompts_processed": 27.75,
|
| 1522 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1523 |
+
"epoch": 0.09142857142857143,
|
| 1524 |
+
"grad_norm": 0.10739335417747498,
|
| 1525 |
+
"kl": 0.008031845092773438,
|
| 1526 |
+
"learning_rate": 2.1558482853517253e-07,
|
| 1527 |
+
"loss": 0.0574,
|
| 1528 |
+
"reward": 0.6980459969490767,
|
| 1529 |
+
"reward_std": 0.9673654958605766,
|
| 1530 |
+
"step": 80
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"clip_fraction": 0.0,
|
| 1534 |
+
"completion_length": 2921.6111450195312,
|
| 1535 |
+
"dapo/avg_reward_std": 0.2788313144239886,
|
| 1536 |
+
"dapo/filter_reward_index": 0.0,
|
| 1537 |
+
"dapo/kept_prompts_ratio": 0.3333333386429425,
|
| 1538 |
+
"dapo/num_sampling_attempts": 3.625,
|
| 1539 |
+
"dapo/sampling_efficiency": 34.27083333333333,
|
| 1540 |
+
"dapo/total_prompts_processed": 21.75,
|
| 1541 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1542 |
+
"epoch": 0.09257142857142857,
|
| 1543 |
+
"grad_norm": 0.18038466572761536,
|
| 1544 |
+
"kl": 0.016963958740234375,
|
| 1545 |
+
"learning_rate": 2.0528000059645995e-07,
|
| 1546 |
+
"loss": 0.0958,
|
| 1547 |
+
"reward": 0.6405055914074183,
|
| 1548 |
+
"reward_std": 0.9560460075736046,
|
| 1549 |
+
"step": 81
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"clip_fraction": 0.0,
|
| 1553 |
+
"completion_length": 3220.687530517578,
|
| 1554 |
+
"dapo/avg_reward_std": 0.1744266465688363,
|
| 1555 |
+
"dapo/filter_reward_index": 0.0,
|
| 1556 |
+
"dapo/kept_prompts_ratio": 0.2307692349721224,
|
| 1557 |
+
"dapo/num_sampling_attempts": 4.875,
|
| 1558 |
+
"dapo/sampling_efficiency": 26.666666666666664,
|
| 1559 |
+
"dapo/total_prompts_processed": 29.25,
|
| 1560 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1561 |
+
"epoch": 0.09371428571428571,
|
| 1562 |
+
"grad_norm": 0.12377161532640457,
|
| 1563 |
+
"kl": 0.009552001953125,
|
| 1564 |
+
"learning_rate": 1.9539516087697517e-07,
|
| 1565 |
+
"loss": 0.061,
|
| 1566 |
+
"reward": 0.5073397234082222,
|
| 1567 |
+
"reward_std": 0.9641925543546677,
|
| 1568 |
+
"step": 82
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"clip_fraction": 0.0,
|
| 1572 |
+
"completion_length": 2663.1597442626953,
|
| 1573 |
+
"dapo/avg_reward_std": 0.2496542421079451,
|
| 1574 |
+
"dapo/filter_reward_index": 0.0,
|
| 1575 |
+
"dapo/kept_prompts_ratio": 0.3333333413447103,
|
| 1576 |
+
"dapo/num_sampling_attempts": 3.875,
|
| 1577 |
+
"dapo/sampling_efficiency": 40.451388888888886,
|
| 1578 |
+
"dapo/total_prompts_processed": 23.25,
|
| 1579 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1580 |
+
"epoch": 0.09485714285714286,
|
| 1581 |
+
"grad_norm": 0.1273493468761444,
|
| 1582 |
+
"kl": 0.04001617431640625,
|
| 1583 |
+
"learning_rate": 1.8594235253127372e-07,
|
| 1584 |
+
"loss": 0.0521,
|
| 1585 |
+
"reward": 0.49824655149132013,
|
| 1586 |
+
"reward_std": 0.9464590474963188,
|
| 1587 |
+
"step": 83
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"clip_fraction": 0.0,
|
| 1591 |
+
"completion_length": 3073.2986450195312,
|
| 1592 |
+
"dapo/avg_reward_std": 0.27911247177557513,
|
| 1593 |
+
"dapo/filter_reward_index": 0.0,
|
| 1594 |
+
"dapo/kept_prompts_ratio": 0.4318181892687624,
|
| 1595 |
+
"dapo/num_sampling_attempts": 2.75,
|
| 1596 |
+
"dapo/sampling_efficiency": 60.3125,
|
| 1597 |
+
"dapo/total_prompts_processed": 16.5,
|
| 1598 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1599 |
+
"epoch": 0.096,
|
| 1600 |
+
"grad_norm": 0.14399568736553192,
|
| 1601 |
+
"kl": 0.010408401489257812,
|
| 1602 |
+
"learning_rate": 1.7693309235023127e-07,
|
| 1603 |
+
"loss": 0.0657,
|
| 1604 |
+
"reward": 0.624765045940876,
|
| 1605 |
+
"reward_std": 0.954634428024292,
|
| 1606 |
+
"step": 84
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"clip_fraction": 0.0,
|
| 1610 |
+
"completion_length": 3073.7535095214844,
|
| 1611 |
+
"dapo/avg_reward_std": 0.17655213298024358,
|
| 1612 |
+
"dapo/filter_reward_index": 0.0,
|
| 1613 |
+
"dapo/kept_prompts_ratio": 0.24324324888152046,
|
| 1614 |
+
"dapo/num_sampling_attempts": 4.625,
|
| 1615 |
+
"dapo/sampling_efficiency": 29.82142857142857,
|
| 1616 |
+
"dapo/total_prompts_processed": 27.75,
|
| 1617 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1618 |
+
"epoch": 0.09714285714285714,
|
| 1619 |
+
"grad_norm": 0.12462300807237625,
|
| 1620 |
+
"kl": 0.007053375244140625,
|
| 1621 |
+
"learning_rate": 1.6837835672960831e-07,
|
| 1622 |
+
"loss": 0.062,
|
| 1623 |
+
"reward": 0.6820014184340835,
|
| 1624 |
+
"reward_std": 0.8695997595787048,
|
| 1625 |
+
"step": 85
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"clip_fraction": 0.0,
|
| 1629 |
+
"completion_length": 2741.204849243164,
|
| 1630 |
+
"dapo/avg_reward_std": 0.21997538357973098,
|
| 1631 |
+
"dapo/filter_reward_index": 0.0,
|
| 1632 |
+
"dapo/kept_prompts_ratio": 0.2666666731238365,
|
| 1633 |
+
"dapo/num_sampling_attempts": 5.0,
|
| 1634 |
+
"dapo/sampling_efficiency": 27.896825396825395,
|
| 1635 |
+
"dapo/total_prompts_processed": 30.0,
|
| 1636 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1637 |
+
"epoch": 0.09828571428571428,
|
| 1638 |
+
"grad_norm": 0.14978615939617157,
|
| 1639 |
+
"kl": 0.025630950927734375,
|
| 1640 |
+
"learning_rate": 1.6028856829700258e-07,
|
| 1641 |
+
"loss": 0.0585,
|
| 1642 |
+
"reward": 0.5304304007440805,
|
| 1643 |
+
"reward_std": 0.9523463025689125,
|
| 1644 |
+
"step": 86
|
| 1645 |
+
},
|
| 1646 |
+
{
|
| 1647 |
+
"clip_fraction": 0.0,
|
| 1648 |
+
"completion_length": 3223.7257080078125,
|
| 1649 |
+
"dapo/avg_reward_std": 0.27104776600996655,
|
| 1650 |
+
"dapo/filter_reward_index": 0.0,
|
| 1651 |
+
"dapo/kept_prompts_ratio": 0.3666666716337204,
|
| 1652 |
+
"dapo/num_sampling_attempts": 3.75,
|
| 1653 |
+
"dapo/sampling_efficiency": 43.333333333333336,
|
| 1654 |
+
"dapo/total_prompts_processed": 22.5,
|
| 1655 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1656 |
+
"epoch": 0.09942857142857142,
|
| 1657 |
+
"grad_norm": 0.1086694523692131,
|
| 1658 |
+
"kl": 0.009660720825195312,
|
| 1659 |
+
"learning_rate": 1.5267358321348285e-07,
|
| 1660 |
+
"loss": 0.058,
|
| 1661 |
+
"reward": 0.5936380252242088,
|
| 1662 |
+
"reward_std": 0.919317290186882,
|
| 1663 |
+
"step": 87
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"clip_fraction": 0.0,
|
| 1667 |
+
"completion_length": 2934.5833740234375,
|
| 1668 |
+
"dapo/avg_reward_std": 0.23462909049001232,
|
| 1669 |
+
"dapo/filter_reward_index": 0.0,
|
| 1670 |
+
"dapo/kept_prompts_ratio": 0.3333333376152762,
|
| 1671 |
+
"dapo/num_sampling_attempts": 3.625,
|
| 1672 |
+
"dapo/sampling_efficiency": 52.84722222222222,
|
| 1673 |
+
"dapo/total_prompts_processed": 21.75,
|
| 1674 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1675 |
+
"epoch": 0.10057142857142858,
|
| 1676 |
+
"grad_norm": 0.14571106433868408,
|
| 1677 |
+
"kl": 0.02588653564453125,
|
| 1678 |
+
"learning_rate": 1.4554267916537495e-07,
|
| 1679 |
+
"loss": 0.0741,
|
| 1680 |
+
"reward": 0.5716092269867659,
|
| 1681 |
+
"reward_std": 0.9475584626197815,
|
| 1682 |
+
"step": 88
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"clip_fraction": 0.0,
|
| 1686 |
+
"completion_length": 3017.2673950195312,
|
| 1687 |
+
"dapo/avg_reward_std": 0.22858241697152457,
|
| 1688 |
+
"dapo/filter_reward_index": 0.0,
|
| 1689 |
+
"dapo/kept_prompts_ratio": 0.3333333383003871,
|
| 1690 |
+
"dapo/num_sampling_attempts": 3.75,
|
| 1691 |
+
"dapo/sampling_efficiency": 45.416666666666664,
|
| 1692 |
+
"dapo/total_prompts_processed": 22.5,
|
| 1693 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1694 |
+
"epoch": 0.10171428571428572,
|
| 1695 |
+
"grad_norm": 0.10647116601467133,
|
| 1696 |
+
"kl": 0.034389495849609375,
|
| 1697 |
+
"learning_rate": 1.3890454406082956e-07,
|
| 1698 |
+
"loss": 0.0586,
|
| 1699 |
+
"reward": 0.5356123449746519,
|
| 1700 |
+
"reward_std": 0.9426311627030373,
|
| 1701 |
+
"step": 89
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"clip_fraction": 0.0,
|
| 1705 |
+
"completion_length": 2540.9548950195312,
|
| 1706 |
+
"dapo/avg_reward_std": 0.16863613526026408,
|
| 1707 |
+
"dapo/filter_reward_index": 0.0,
|
| 1708 |
+
"dapo/kept_prompts_ratio": 0.22592593100335862,
|
| 1709 |
+
"dapo/num_sampling_attempts": 5.625,
|
| 1710 |
+
"dapo/sampling_efficiency": 28.75,
|
| 1711 |
+
"dapo/total_prompts_processed": 33.75,
|
| 1712 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1713 |
+
"epoch": 0.10285714285714286,
|
| 1714 |
+
"grad_norm": 0.1207195371389389,
|
| 1715 |
+
"kl": 0.7361793518066406,
|
| 1716 |
+
"learning_rate": 1.3276726544494571e-07,
|
| 1717 |
+
"loss": 0.0349,
|
| 1718 |
+
"reward": 0.750616230070591,
|
| 1719 |
+
"reward_std": 1.0088519006967545,
|
| 1720 |
+
"step": 90
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"clip_fraction": 0.0,
|
| 1724 |
+
"completion_length": 3054.5833435058594,
|
| 1725 |
+
"dapo/avg_reward_std": 0.2058313423767686,
|
| 1726 |
+
"dapo/filter_reward_index": 0.0,
|
| 1727 |
+
"dapo/kept_prompts_ratio": 0.28645834047347307,
|
| 1728 |
+
"dapo/num_sampling_attempts": 4.0,
|
| 1729 |
+
"dapo/sampling_efficiency": 41.36904761904762,
|
| 1730 |
+
"dapo/total_prompts_processed": 24.0,
|
| 1731 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1732 |
+
"epoch": 0.104,
|
| 1733 |
+
"grad_norm": 0.10659411549568176,
|
| 1734 |
+
"kl": 0.009166717529296875,
|
| 1735 |
+
"learning_rate": 1.2713832064634125e-07,
|
| 1736 |
+
"loss": 0.06,
|
| 1737 |
+
"reward": 0.49192704539746046,
|
| 1738 |
+
"reward_std": 0.8957021087408066,
|
| 1739 |
+
"step": 91
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"clip_fraction": 0.0,
|
| 1743 |
+
"completion_length": 2958.43408203125,
|
| 1744 |
+
"dapo/avg_reward_std": 0.317311546076899,
|
| 1745 |
+
"dapo/filter_reward_index": 0.0,
|
| 1746 |
+
"dapo/kept_prompts_ratio": 0.47101450160793634,
|
| 1747 |
+
"dapo/num_sampling_attempts": 2.875,
|
| 1748 |
+
"dapo/sampling_efficiency": 44.166666666666664,
|
| 1749 |
+
"dapo/total_prompts_processed": 17.25,
|
| 1750 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1751 |
+
"epoch": 0.10514285714285715,
|
| 1752 |
+
"grad_norm": 0.1002211645245552,
|
| 1753 |
+
"kl": 0.00801849365234375,
|
| 1754 |
+
"learning_rate": 1.220245676671809e-07,
|
| 1755 |
+
"loss": 0.0508,
|
| 1756 |
+
"reward": 0.7598672257736325,
|
| 1757 |
+
"reward_std": 0.9218961223959923,
|
| 1758 |
+
"step": 92
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"clip_fraction": 0.0,
|
| 1762 |
+
"completion_length": 3257.7881774902344,
|
| 1763 |
+
"dapo/avg_reward_std": 0.2586492033941405,
|
| 1764 |
+
"dapo/filter_reward_index": 0.0,
|
| 1765 |
+
"dapo/kept_prompts_ratio": 0.36309524678758215,
|
| 1766 |
+
"dapo/num_sampling_attempts": 3.5,
|
| 1767 |
+
"dapo/sampling_efficiency": 41.5625,
|
| 1768 |
+
"dapo/total_prompts_processed": 21.0,
|
| 1769 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1770 |
+
"epoch": 0.10628571428571429,
|
| 1771 |
+
"grad_norm": 0.12036111950874329,
|
| 1772 |
+
"kl": 0.01373291015625,
|
| 1773 |
+
"learning_rate": 1.1743223682775649e-07,
|
| 1774 |
+
"loss": 0.0459,
|
| 1775 |
+
"reward": 0.5575436241924763,
|
| 1776 |
+
"reward_std": 0.9431066736578941,
|
| 1777 |
+
"step": 93
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"clip_fraction": 0.0,
|
| 1781 |
+
"completion_length": 2740.1284942626953,
|
| 1782 |
+
"dapo/avg_reward_std": 0.2375115204241968,
|
| 1783 |
+
"dapo/filter_reward_index": 0.0,
|
| 1784 |
+
"dapo/kept_prompts_ratio": 0.35483871688765867,
|
| 1785 |
+
"dapo/num_sampling_attempts": 3.875,
|
| 1786 |
+
"dapo/sampling_efficiency": 29.999999999999996,
|
| 1787 |
+
"dapo/total_prompts_processed": 23.25,
|
| 1788 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1789 |
+
"epoch": 0.10742857142857143,
|
| 1790 |
+
"grad_norm": 0.14863841235637665,
|
| 1791 |
+
"kl": 0.032642364501953125,
|
| 1792 |
+
"learning_rate": 1.1336692317580158e-07,
|
| 1793 |
+
"loss": 0.0742,
|
| 1794 |
+
"reward": 0.5738632343709469,
|
| 1795 |
+
"reward_std": 0.9468542039394379,
|
| 1796 |
+
"step": 94
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"clip_fraction": 0.0,
|
| 1800 |
+
"completion_length": 2899.937515258789,
|
| 1801 |
+
"dapo/avg_reward_std": 0.2901096656208947,
|
| 1802 |
+
"dapo/filter_reward_index": 0.0,
|
| 1803 |
+
"dapo/kept_prompts_ratio": 0.4206349246558689,
|
| 1804 |
+
"dapo/num_sampling_attempts": 2.625,
|
| 1805 |
+
"dapo/sampling_efficiency": 64.58333333333333,
|
| 1806 |
+
"dapo/total_prompts_processed": 15.75,
|
| 1807 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1808 |
+
"epoch": 0.10857142857142857,
|
| 1809 |
+
"grad_norm": 0.13841120898723602,
|
| 1810 |
+
"kl": 0.012683868408203125,
|
| 1811 |
+
"learning_rate": 1.0983357966978745e-07,
|
| 1812 |
+
"loss": 0.0653,
|
| 1813 |
+
"reward": 0.6555321607738733,
|
| 1814 |
+
"reward_std": 0.9674765914678574,
|
| 1815 |
+
"step": 95
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"clip_fraction": 0.0,
|
| 1819 |
+
"completion_length": 2926.1910247802734,
|
| 1820 |
+
"dapo/avg_reward_std": 0.18252932499436772,
|
| 1821 |
+
"dapo/filter_reward_index": 0.0,
|
| 1822 |
+
"dapo/kept_prompts_ratio": 0.2696078485425781,
|
| 1823 |
+
"dapo/num_sampling_attempts": 4.25,
|
| 1824 |
+
"dapo/sampling_efficiency": 37.82738095238095,
|
| 1825 |
+
"dapo/total_prompts_processed": 25.5,
|
| 1826 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1827 |
+
"epoch": 0.10971428571428571,
|
| 1828 |
+
"grad_norm": 0.13530230522155762,
|
| 1829 |
+
"kl": 0.05282402038574219,
|
| 1830 |
+
"learning_rate": 1.068365111445064e-07,
|
| 1831 |
+
"loss": 0.0762,
|
| 1832 |
+
"reward": 0.5449853939935565,
|
| 1833 |
+
"reward_std": 0.952080488204956,
|
| 1834 |
+
"step": 96
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"clip_fraction": 0.0,
|
| 1838 |
+
"completion_length": 2798.031280517578,
|
| 1839 |
+
"dapo/avg_reward_std": 0.23633464597738707,
|
| 1840 |
+
"dapo/filter_reward_index": 0.0,
|
| 1841 |
+
"dapo/kept_prompts_ratio": 0.3461538478732109,
|
| 1842 |
+
"dapo/num_sampling_attempts": 3.25,
|
| 1843 |
+
"dapo/sampling_efficiency": 38.541666666666664,
|
| 1844 |
+
"dapo/total_prompts_processed": 19.5,
|
| 1845 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1846 |
+
"epoch": 0.11085714285714286,
|
| 1847 |
+
"grad_norm": 0.1648494303226471,
|
| 1848 |
+
"kl": 0.025691986083984375,
|
| 1849 |
+
"learning_rate": 1.0437936906629334e-07,
|
| 1850 |
+
"loss": 0.0939,
|
| 1851 |
+
"reward": 0.673285935074091,
|
| 1852 |
+
"reward_std": 0.979133740067482,
|
| 1853 |
+
"step": 97
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"clip_fraction": 0.0,
|
| 1857 |
+
"completion_length": 3240.7361450195312,
|
| 1858 |
+
"dapo/avg_reward_std": 0.2805523918225215,
|
| 1859 |
+
"dapo/filter_reward_index": 0.0,
|
| 1860 |
+
"dapo/kept_prompts_ratio": 0.3653846193964665,
|
| 1861 |
+
"dapo/num_sampling_attempts": 3.25,
|
| 1862 |
+
"dapo/sampling_efficiency": 54.513888888888886,
|
| 1863 |
+
"dapo/total_prompts_processed": 19.5,
|
| 1864 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1865 |
+
"epoch": 0.112,
|
| 1866 |
+
"grad_norm": 0.12132810056209564,
|
| 1867 |
+
"kl": 0.01453399658203125,
|
| 1868 |
+
"learning_rate": 1.0246514708427701e-07,
|
| 1869 |
+
"loss": 0.0557,
|
| 1870 |
+
"reward": 0.5335402796044946,
|
| 1871 |
+
"reward_std": 0.9456770345568657,
|
| 1872 |
+
"step": 98
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"clip_fraction": 0.0,
|
| 1876 |
+
"completion_length": 2497.9132499694824,
|
| 1877 |
+
"dapo/avg_reward_std": 0.2488528937101364,
|
| 1878 |
+
"dapo/filter_reward_index": 0.0,
|
| 1879 |
+
"dapo/kept_prompts_ratio": 0.3888888942698638,
|
| 1880 |
+
"dapo/num_sampling_attempts": 3.0,
|
| 1881 |
+
"dapo/sampling_efficiency": 58.05555555555555,
|
| 1882 |
+
"dapo/total_prompts_processed": 18.0,
|
| 1883 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1884 |
+
"epoch": 0.11314285714285714,
|
| 1885 |
+
"grad_norm": 0.24999241530895233,
|
| 1886 |
+
"kl": 0.028301239013671875,
|
| 1887 |
+
"learning_rate": 1.0109617738307911e-07,
|
| 1888 |
+
"loss": 0.1037,
|
| 1889 |
+
"reward": 0.785055335611105,
|
| 1890 |
+
"reward_std": 0.9553829357028008,
|
| 1891 |
+
"step": 99
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"clip_fraction": 0.0,
|
| 1895 |
+
"completion_length": 3039.6284790039062,
|
| 1896 |
+
"dapo/avg_reward_std": 0.2903642791012923,
|
| 1897 |
+
"dapo/filter_reward_index": 0.0,
|
| 1898 |
+
"dapo/kept_prompts_ratio": 0.38194445086022216,
|
| 1899 |
+
"dapo/num_sampling_attempts": 3.0,
|
| 1900 |
+
"dapo/sampling_efficiency": 46.24999999999999,
|
| 1901 |
+
"dapo/total_prompts_processed": 18.0,
|
| 1902 |
+
"dapo/valid_prompts_collected": 6.0,
|
| 1903 |
+
"epoch": 0.11428571428571428,
|
| 1904 |
+
"grad_norm": 0.14126254618167877,
|
| 1905 |
+
"kl": 0.014410018920898438,
|
| 1906 |
+
"learning_rate": 1.002741278414069e-07,
|
| 1907 |
+
"loss": 0.0643,
|
| 1908 |
+
"reward": 0.4948624651879072,
|
| 1909 |
+
"reward_std": 0.9704382866621017,
|
| 1910 |
+
"step": 100
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 0.11428571428571428,
|
| 1914 |
+
"step": 100,
|
| 1915 |
+
"total_flos": 0.0,
|
| 1916 |
+
"train_loss": 0.06302435559220612,
|
| 1917 |
+
"train_runtime": 83388.0569,
|
| 1918 |
+
"train_samples_per_second": 0.058,
|
| 1919 |
+
"train_steps_per_second": 0.001
|
| 1920 |
+
}
|
| 1921 |
+
],
|
| 1922 |
+
"logging_steps": 1,
|
| 1923 |
+
"max_steps": 100,
|
| 1924 |
+
"num_input_tokens_seen": 0,
|
| 1925 |
+
"num_train_epochs": 1,
|
| 1926 |
+
"save_steps": 10,
|
| 1927 |
+
"stateful_callbacks": {
|
| 1928 |
+
"TrainerControl": {
|
| 1929 |
+
"args": {
|
| 1930 |
+
"should_epoch_stop": false,
|
| 1931 |
+
"should_evaluate": false,
|
| 1932 |
+
"should_log": false,
|
| 1933 |
+
"should_save": true,
|
| 1934 |
+
"should_training_stop": true
|
| 1935 |
+
},
|
| 1936 |
+
"attributes": {}
|
| 1937 |
+
}
|
| 1938 |
+
},
|
| 1939 |
+
"total_flos": 0.0,
|
| 1940 |
+
"train_batch_size": 6,
|
| 1941 |
+
"trial_name": null,
|
| 1942 |
+
"trial_params": null
|
| 1943 |
+
}
|