Model save
Browse files- README.md +68 -0
- all_results.json +8 -0
- config.json +1 -1
- generation_config.json +14 -0
- train_results.json +8 -0
- trainer_state.json +1242 -0
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: Qwen/Qwen2.5-3B-Instruct
|
| 3 |
+
library_name: transformers
|
| 4 |
+
model_name: qwen-2.5-3b-r1-c13k-individual
|
| 5 |
+
tags:
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
- trl
|
| 8 |
+
- grpo
|
| 9 |
+
licence: license
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Model Card for qwen-2.5-3b-r1-c13k-individual
|
| 13 |
+
|
| 14 |
+
This model is a fine-tuned version of [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct).
|
| 15 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 16 |
+
|
| 17 |
+
## Quick start
|
| 18 |
+
|
| 19 |
+
```python
|
| 20 |
+
from transformers import pipeline
|
| 21 |
+
|
| 22 |
+
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
|
| 23 |
+
generator = pipeline("text-generation", model="xhb120633/qwen-2.5-3b-r1-c13k-individual", 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 GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
|
| 34 |
+
|
| 35 |
+
### Framework versions
|
| 36 |
+
|
| 37 |
+
- TRL: 0.15.1
|
| 38 |
+
- Transformers: 4.48.1
|
| 39 |
+
- Pytorch: 2.5.1
|
| 40 |
+
- Datasets: 3.1.0
|
| 41 |
+
- Tokenizers: 0.21.0
|
| 42 |
+
|
| 43 |
+
## Citations
|
| 44 |
+
|
| 45 |
+
Cite GRPO as:
|
| 46 |
+
|
| 47 |
+
```bibtex
|
| 48 |
+
@article{zhihong2024deepseekmath,
|
| 49 |
+
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
|
| 50 |
+
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
|
| 51 |
+
year = 2024,
|
| 52 |
+
eprint = {arXiv:2402.03300},
|
| 53 |
+
}
|
| 54 |
+
|
| 55 |
+
```
|
| 56 |
+
|
| 57 |
+
Cite TRL as:
|
| 58 |
+
|
| 59 |
+
```bibtex
|
| 60 |
+
@misc{vonwerra2022trl,
|
| 61 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 62 |
+
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
|
| 63 |
+
year = 2020,
|
| 64 |
+
journal = {GitHub repository},
|
| 65 |
+
publisher = {GitHub},
|
| 66 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 67 |
+
}
|
| 68 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 7.065190569642255e-05,
|
| 4 |
+
"train_runtime": 24090.8459,
|
| 5 |
+
"train_samples": 3176,
|
| 6 |
+
"train_samples_per_second": 1.992,
|
| 7 |
+
"train_steps_per_second": 0.021
|
| 8 |
+
}
|
config.json
CHANGED
|
@@ -23,7 +23,7 @@
|
|
| 23 |
"tie_word_embeddings": true,
|
| 24 |
"torch_dtype": "bfloat16",
|
| 25 |
"transformers_version": "4.48.1",
|
| 26 |
-
"use_cache":
|
| 27 |
"use_sliding_window": false,
|
| 28 |
"vocab_size": 151936
|
| 29 |
}
|
|
|
|
| 23 |
"tie_word_embeddings": true,
|
| 24 |
"torch_dtype": "bfloat16",
|
| 25 |
"transformers_version": "4.48.1",
|
| 26 |
+
"use_cache": true,
|
| 27 |
"use_sliding_window": false,
|
| 28 |
"vocab_size": 151936
|
| 29 |
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151643,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
151645,
|
| 6 |
+
151643
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 151643,
|
| 9 |
+
"repetition_penalty": 1.05,
|
| 10 |
+
"temperature": 0.7,
|
| 11 |
+
"top_k": 20,
|
| 12 |
+
"top_p": 0.8,
|
| 13 |
+
"transformers_version": "4.48.1"
|
| 14 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 7.065190569642255e-05,
|
| 4 |
+
"train_runtime": 24090.8459,
|
| 5 |
+
"train_samples": 3176,
|
| 6 |
+
"train_samples_per_second": 1.992,
|
| 7 |
+
"train_steps_per_second": 0.021
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1242 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.2594458438287153,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 500,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"completion_length": 722.9521026611328,
|
| 13 |
+
"epoch": 0.012594458438287154,
|
| 14 |
+
"grad_norm": 0.07664977191949898,
|
| 15 |
+
"kl": 0.00014810562133789064,
|
| 16 |
+
"learning_rate": 3.333333333333333e-07,
|
| 17 |
+
"loss": 0.0,
|
| 18 |
+
"reward": 0.46451360136270525,
|
| 19 |
+
"reward_std": 0.10854418443050236,
|
| 20 |
+
"rewards/wrapped_prediction_reward_func": 0.46451360136270525,
|
| 21 |
+
"step": 5
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"completion_length": 730.1583602905273,
|
| 25 |
+
"epoch": 0.02518891687657431,
|
| 26 |
+
"grad_norm": 0.08074394124833935,
|
| 27 |
+
"kl": 0.00022151470184326172,
|
| 28 |
+
"learning_rate": 6.666666666666666e-07,
|
| 29 |
+
"loss": 0.0,
|
| 30 |
+
"reward": 0.46406558752059934,
|
| 31 |
+
"reward_std": 0.09366982954088598,
|
| 32 |
+
"rewards/wrapped_prediction_reward_func": 0.46406558752059934,
|
| 33 |
+
"step": 10
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"completion_length": 732.6021026611328,
|
| 37 |
+
"epoch": 0.037783375314861464,
|
| 38 |
+
"grad_norm": 0.07971439169776136,
|
| 39 |
+
"kl": 0.00023424625396728516,
|
| 40 |
+
"learning_rate": 1e-06,
|
| 41 |
+
"loss": 0.0,
|
| 42 |
+
"reward": 0.4597253814339638,
|
| 43 |
+
"reward_std": 0.09797948987688869,
|
| 44 |
+
"rewards/wrapped_prediction_reward_func": 0.4597253814339638,
|
| 45 |
+
"step": 15
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"completion_length": 706.5271057128906,
|
| 49 |
+
"epoch": 0.05037783375314862,
|
| 50 |
+
"grad_norm": 0.07675378344787744,
|
| 51 |
+
"kl": 0.00025036334991455076,
|
| 52 |
+
"learning_rate": 9.997377845227574e-07,
|
| 53 |
+
"loss": 0.0,
|
| 54 |
+
"reward": 0.47573257163167,
|
| 55 |
+
"reward_std": 0.09649912532186136,
|
| 56 |
+
"rewards/wrapped_prediction_reward_func": 0.47573257163167,
|
| 57 |
+
"step": 20
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"completion_length": 709.5146011352539,
|
| 61 |
+
"epoch": 0.06297229219143577,
|
| 62 |
+
"grad_norm": 0.07792838886271755,
|
| 63 |
+
"kl": 0.00035452842712402344,
|
| 64 |
+
"learning_rate": 9.989514131188558e-07,
|
| 65 |
+
"loss": 0.0,
|
| 66 |
+
"reward": 0.4748247370123863,
|
| 67 |
+
"reward_std": 0.08561795827117749,
|
| 68 |
+
"rewards/wrapped_prediction_reward_func": 0.4748247370123863,
|
| 69 |
+
"step": 25
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"completion_length": 746.320849609375,
|
| 73 |
+
"epoch": 0.07556675062972293,
|
| 74 |
+
"grad_norm": 0.0755043003664804,
|
| 75 |
+
"kl": 0.0006094932556152344,
|
| 76 |
+
"learning_rate": 9.97641710583307e-07,
|
| 77 |
+
"loss": 0.0,
|
| 78 |
+
"reward": 0.459003009647131,
|
| 79 |
+
"reward_std": 0.09646977222291753,
|
| 80 |
+
"rewards/wrapped_prediction_reward_func": 0.459003009647131,
|
| 81 |
+
"step": 30
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"completion_length": 739.545849609375,
|
| 85 |
+
"epoch": 0.08816120906801007,
|
| 86 |
+
"grad_norm": 0.06831186280995111,
|
| 87 |
+
"kl": 0.00098114013671875,
|
| 88 |
+
"learning_rate": 9.958100506132126e-07,
|
| 89 |
+
"loss": 0.0,
|
| 90 |
+
"reward": 0.491700629144907,
|
| 91 |
+
"reward_std": 0.08609421639703214,
|
| 92 |
+
"rewards/wrapped_prediction_reward_func": 0.491700629144907,
|
| 93 |
+
"step": 35
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"completion_length": 735.9604385375976,
|
| 97 |
+
"epoch": 0.10075566750629723,
|
| 98 |
+
"grad_norm": 0.06820840594292561,
|
| 99 |
+
"kl": 0.00146484375,
|
| 100 |
+
"learning_rate": 9.934583543669453e-07,
|
| 101 |
+
"loss": 0.0,
|
| 102 |
+
"reward": 0.4918279483914375,
|
| 103 |
+
"reward_std": 0.10381489984574728,
|
| 104 |
+
"rewards/wrapped_prediction_reward_func": 0.4918279483914375,
|
| 105 |
+
"step": 40
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"completion_length": 761.6375183105469,
|
| 109 |
+
"epoch": 0.11335012594458438,
|
| 110 |
+
"grad_norm": 0.07306576306189429,
|
| 111 |
+
"kl": 0.0023061752319335936,
|
| 112 |
+
"learning_rate": 9.905890884491194e-07,
|
| 113 |
+
"loss": 0.0,
|
| 114 |
+
"reward": 0.4687023714184761,
|
| 115 |
+
"reward_std": 0.10100088955368847,
|
| 116 |
+
"rewards/wrapped_prediction_reward_func": 0.4687023714184761,
|
| 117 |
+
"step": 45
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"completion_length": 762.1875213623047,
|
| 121 |
+
"epoch": 0.12594458438287154,
|
| 122 |
+
"grad_norm": 0.0803081141044505,
|
| 123 |
+
"kl": 0.0034528732299804687,
|
| 124 |
+
"learning_rate": 9.872052623234631e-07,
|
| 125 |
+
"loss": 0.0,
|
| 126 |
+
"reward": 0.47996846288442613,
|
| 127 |
+
"reward_std": 0.09802622515708208,
|
| 128 |
+
"rewards/wrapped_prediction_reward_func": 0.47996846288442613,
|
| 129 |
+
"step": 50
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"completion_length": 790.0812713623047,
|
| 133 |
+
"epoch": 0.1385390428211587,
|
| 134 |
+
"grad_norm": 0.0704359174357588,
|
| 135 |
+
"kl": 0.003958320617675782,
|
| 136 |
+
"learning_rate": 9.833104251563055e-07,
|
| 137 |
+
"loss": 0.0,
|
| 138 |
+
"reward": 0.48190433755517004,
|
| 139 |
+
"reward_std": 0.08568599077407271,
|
| 140 |
+
"rewards/wrapped_prediction_reward_func": 0.48190433755517004,
|
| 141 |
+
"step": 55
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"completion_length": 768.5521057128906,
|
| 145 |
+
"epoch": 0.15113350125944586,
|
| 146 |
+
"grad_norm": 0.06457778900067923,
|
| 147 |
+
"kl": 0.005392837524414063,
|
| 148 |
+
"learning_rate": 9.789086620939935e-07,
|
| 149 |
+
"loss": 0.0,
|
| 150 |
+
"reward": 0.48327849954366686,
|
| 151 |
+
"reward_std": 0.09457279161288171,
|
| 152 |
+
"rewards/wrapped_prediction_reward_func": 0.48327849954366686,
|
| 153 |
+
"step": 60
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"completion_length": 770.9166870117188,
|
| 157 |
+
"epoch": 0.163727959697733,
|
| 158 |
+
"grad_norm": 0.06812633405378012,
|
| 159 |
+
"kl": 0.006523513793945312,
|
| 160 |
+
"learning_rate": 9.740045899781352e-07,
|
| 161 |
+
"loss": 0.0,
|
| 162 |
+
"reward": 0.5019405260682106,
|
| 163 |
+
"reward_std": 0.07568035275908187,
|
| 164 |
+
"rewards/wrapped_prediction_reward_func": 0.5019405260682106,
|
| 165 |
+
"step": 65
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"completion_length": 749.1146057128906,
|
| 169 |
+
"epoch": 0.17632241813602015,
|
| 170 |
+
"grad_norm": 0.06685526422767288,
|
| 171 |
+
"kl": 0.007244110107421875,
|
| 172 |
+
"learning_rate": 9.686033525031719e-07,
|
| 173 |
+
"loss": 0.0,
|
| 174 |
+
"reward": 0.501540158689022,
|
| 175 |
+
"reward_std": 0.07553914005402476,
|
| 176 |
+
"rewards/wrapped_prediction_reward_func": 0.501540158689022,
|
| 177 |
+
"step": 70
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"completion_length": 761.0937683105469,
|
| 181 |
+
"epoch": 0.1889168765743073,
|
| 182 |
+
"grad_norm": 0.0722156044539115,
|
| 183 |
+
"kl": 0.007537460327148438,
|
| 184 |
+
"learning_rate": 9.62710614821352e-07,
|
| 185 |
+
"loss": 0.0,
|
| 186 |
+
"reward": 0.4996995717287064,
|
| 187 |
+
"reward_std": 0.0968896305654198,
|
| 188 |
+
"rewards/wrapped_prediction_reward_func": 0.4996995717287064,
|
| 189 |
+
"step": 75
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"completion_length": 767.5000213623047,
|
| 193 |
+
"epoch": 0.20151133501259447,
|
| 194 |
+
"grad_norm": 0.078058173076431,
|
| 195 |
+
"kl": 0.0085418701171875,
|
| 196 |
+
"learning_rate": 9.5633255760077e-07,
|
| 197 |
+
"loss": 0.0,
|
| 198 |
+
"reward": 0.4846237309277058,
|
| 199 |
+
"reward_std": 0.10277664107270539,
|
| 200 |
+
"rewards/wrapped_prediction_reward_func": 0.4846237309277058,
|
| 201 |
+
"step": 80
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"completion_length": 737.7312728881836,
|
| 205 |
+
"epoch": 0.2141057934508816,
|
| 206 |
+
"grad_norm": 0.07298119554807302,
|
| 207 |
+
"kl": 0.010308837890625,
|
| 208 |
+
"learning_rate": 9.494758705426976e-07,
|
| 209 |
+
"loss": 0.0,
|
| 210 |
+
"reward": 0.5145397499203682,
|
| 211 |
+
"reward_std": 0.09095026951981708,
|
| 212 |
+
"rewards/wrapped_prediction_reward_func": 0.5145397499203682,
|
| 213 |
+
"step": 85
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"completion_length": 780.7896041870117,
|
| 217 |
+
"epoch": 0.22670025188916876,
|
| 218 |
+
"grad_norm": 0.07249771083699665,
|
| 219 |
+
"kl": 0.0119293212890625,
|
| 220 |
+
"learning_rate": 9.421477453650117e-07,
|
| 221 |
+
"loss": 0.0,
|
| 222 |
+
"reward": 0.5147036135196685,
|
| 223 |
+
"reward_std": 0.09051896830787882,
|
| 224 |
+
"rewards/wrapped_prediction_reward_func": 0.5147036135196685,
|
| 225 |
+
"step": 90
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"completion_length": 761.1833541870117,
|
| 229 |
+
"epoch": 0.23929471032745592,
|
| 230 |
+
"grad_norm": 0.07426543152283624,
|
| 231 |
+
"kl": 0.01405029296875,
|
| 232 |
+
"learning_rate": 9.343558682590755e-07,
|
| 233 |
+
"loss": 0.0,
|
| 234 |
+
"reward": 0.5229489721357823,
|
| 235 |
+
"reward_std": 0.1045097156893462,
|
| 236 |
+
"rewards/wrapped_prediction_reward_func": 0.5229489721357823,
|
| 237 |
+
"step": 95
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"completion_length": 747.8270980834961,
|
| 241 |
+
"epoch": 0.2518891687657431,
|
| 242 |
+
"grad_norm": 0.08009584068360569,
|
| 243 |
+
"kl": 0.016656494140625,
|
| 244 |
+
"learning_rate": 9.261084118279846e-07,
|
| 245 |
+
"loss": 0.0,
|
| 246 |
+
"reward": 0.5308140270411968,
|
| 247 |
+
"reward_std": 0.10999625454423949,
|
| 248 |
+
"rewards/wrapped_prediction_reward_func": 0.5308140270411968,
|
| 249 |
+
"step": 100
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"completion_length": 755.6771011352539,
|
| 253 |
+
"epoch": 0.26448362720403024,
|
| 254 |
+
"grad_norm": 0.0717796741561669,
|
| 255 |
+
"kl": 0.01963958740234375,
|
| 256 |
+
"learning_rate": 9.174140265146355e-07,
|
| 257 |
+
"loss": 0.0,
|
| 258 |
+
"reward": 0.5279295884072781,
|
| 259 |
+
"reward_std": 0.12174326665699482,
|
| 260 |
+
"rewards/wrapped_prediction_reward_func": 0.5279295884072781,
|
| 261 |
+
"step": 105
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"completion_length": 741.0396072387696,
|
| 265 |
+
"epoch": 0.2770780856423174,
|
| 266 |
+
"grad_norm": 0.06788337142864205,
|
| 267 |
+
"kl": 0.0238006591796875,
|
| 268 |
+
"learning_rate": 9.082818315286054e-07,
|
| 269 |
+
"loss": 0.0,
|
| 270 |
+
"reward": 0.5241287641227246,
|
| 271 |
+
"reward_std": 0.08864832047838718,
|
| 272 |
+
"rewards/wrapped_prediction_reward_func": 0.5241287641227246,
|
| 273 |
+
"step": 110
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"completion_length": 785.7208526611328,
|
| 277 |
+
"epoch": 0.28967254408060455,
|
| 278 |
+
"grad_norm": 0.0701122609050844,
|
| 279 |
+
"kl": 0.025714111328125,
|
| 280 |
+
"learning_rate": 8.987214052813603e-07,
|
| 281 |
+
"loss": 0.0,
|
| 282 |
+
"reward": 0.5118412531912326,
|
| 283 |
+
"reward_std": 0.11687530118506402,
|
| 284 |
+
"rewards/wrapped_prediction_reward_func": 0.5118412531912326,
|
| 285 |
+
"step": 115
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"completion_length": 785.070849609375,
|
| 289 |
+
"epoch": 0.3022670025188917,
|
| 290 |
+
"grad_norm": 0.07472239714054472,
|
| 291 |
+
"kl": 0.0284881591796875,
|
| 292 |
+
"learning_rate": 8.887427753398247e-07,
|
| 293 |
+
"loss": 0.0,
|
| 294 |
+
"reward": 0.5511950686573982,
|
| 295 |
+
"reward_std": 0.07816179413348437,
|
| 296 |
+
"rewards/wrapped_prediction_reward_func": 0.5511950686573982,
|
| 297 |
+
"step": 120
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"completion_length": 833.4291870117188,
|
| 301 |
+
"epoch": 0.3148614609571788,
|
| 302 |
+
"grad_norm": 0.07422495442339502,
|
| 303 |
+
"kl": 0.027734375,
|
| 304 |
+
"learning_rate": 8.783564079088476e-07,
|
| 305 |
+
"loss": 0.0,
|
| 306 |
+
"reward": 0.5402487128973007,
|
| 307 |
+
"reward_std": 0.13062875589821488,
|
| 308 |
+
"rewards/wrapped_prediction_reward_func": 0.5402487128973007,
|
| 309 |
+
"step": 125
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"completion_length": 814.3541854858398,
|
| 313 |
+
"epoch": 0.327455919395466,
|
| 314 |
+
"grad_norm": 0.06823337736452399,
|
| 315 |
+
"kl": 0.02915191650390625,
|
| 316 |
+
"learning_rate": 8.675731968536002e-07,
|
| 317 |
+
"loss": 0.0,
|
| 318 |
+
"reward": 0.530899728089571,
|
| 319 |
+
"reward_std": 0.11218150332570076,
|
| 320 |
+
"rewards/wrapped_prediction_reward_func": 0.530899728089571,
|
| 321 |
+
"step": 130
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"completion_length": 812.7791900634766,
|
| 325 |
+
"epoch": 0.34005037783375314,
|
| 326 |
+
"grad_norm": 0.07003585561947805,
|
| 327 |
+
"kl": 0.03168487548828125,
|
| 328 |
+
"learning_rate": 8.564044522734146e-07,
|
| 329 |
+
"loss": 0.0,
|
| 330 |
+
"reward": 0.5358106784522534,
|
| 331 |
+
"reward_std": 0.10542880366556347,
|
| 332 |
+
"rewards/wrapped_prediction_reward_func": 0.5358106784522534,
|
| 333 |
+
"step": 135
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"completion_length": 817.2354385375977,
|
| 337 |
+
"epoch": 0.3526448362720403,
|
| 338 |
+
"grad_norm": 0.07278698573023756,
|
| 339 |
+
"kl": 0.03175811767578125,
|
| 340 |
+
"learning_rate": 8.448618886390521e-07,
|
| 341 |
+
"loss": 0.0,
|
| 342 |
+
"reward": 0.549003791064024,
|
| 343 |
+
"reward_std": 0.0949756694608368,
|
| 344 |
+
"rewards/wrapped_prediction_reward_func": 0.549003791064024,
|
| 345 |
+
"step": 140
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"completion_length": 761.7271057128906,
|
| 349 |
+
"epoch": 0.36523929471032746,
|
| 350 |
+
"grad_norm": 0.0726454943691525,
|
| 351 |
+
"kl": 0.0360321044921875,
|
| 352 |
+
"learning_rate": 8.329576125058405e-07,
|
| 353 |
+
"loss": 0.0,
|
| 354 |
+
"reward": 0.5476619251072407,
|
| 355 |
+
"reward_std": 0.08561794870765879,
|
| 356 |
+
"rewards/wrapped_prediction_reward_func": 0.5476619251072407,
|
| 357 |
+
"step": 145
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"completion_length": 784.9791854858398,
|
| 361 |
+
"epoch": 0.3778337531486146,
|
| 362 |
+
"grad_norm": 0.0734552135965303,
|
| 363 |
+
"kl": 0.03544921875,
|
| 364 |
+
"learning_rate": 8.207041098155699e-07,
|
| 365 |
+
"loss": 0.0,
|
| 366 |
+
"reward": 0.5536449149250984,
|
| 367 |
+
"reward_std": 0.09187358255730942,
|
| 368 |
+
"rewards/wrapped_prediction_reward_func": 0.5536449149250984,
|
| 369 |
+
"step": 150
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"completion_length": 750.6729354858398,
|
| 373 |
+
"epoch": 0.3904282115869018,
|
| 374 |
+
"grad_norm": 0.08269464276856019,
|
| 375 |
+
"kl": 0.0421600341796875,
|
| 376 |
+
"learning_rate": 8.081142328004636e-07,
|
| 377 |
+
"loss": 0.0,
|
| 378 |
+
"reward": 0.560660557448864,
|
| 379 |
+
"reward_std": 0.08419233104214072,
|
| 380 |
+
"rewards/wrapped_prediction_reward_func": 0.560660557448864,
|
| 381 |
+
"step": 155
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"completion_length": 736.0833541870118,
|
| 385 |
+
"epoch": 0.40302267002518893,
|
| 386 |
+
"grad_norm": 0.07934493494179942,
|
| 387 |
+
"kl": 0.2113433837890625,
|
| 388 |
+
"learning_rate": 7.952011865029613e-07,
|
| 389 |
+
"loss": 0.0001,
|
| 390 |
+
"reward": 0.5728602230548858,
|
| 391 |
+
"reward_std": 0.08108794330037199,
|
| 392 |
+
"rewards/wrapped_prediction_reward_func": 0.5728602230548858,
|
| 393 |
+
"step": 160
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"completion_length": 751.6396011352539,
|
| 397 |
+
"epoch": 0.4156171284634761,
|
| 398 |
+
"grad_norm": 0.077727529432095,
|
| 399 |
+
"kl": 0.0558685302734375,
|
| 400 |
+
"learning_rate": 7.819785149254532e-07,
|
| 401 |
+
"loss": 0.0,
|
| 402 |
+
"reward": 0.5558124162256718,
|
| 403 |
+
"reward_std": 0.07910064500756561,
|
| 404 |
+
"rewards/wrapped_prediction_reward_func": 0.5558124162256718,
|
| 405 |
+
"step": 165
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"completion_length": 794.9833511352539,
|
| 409 |
+
"epoch": 0.4282115869017632,
|
| 410 |
+
"grad_norm": 0.08198281407723804,
|
| 411 |
+
"kl": 0.063677978515625,
|
| 412 |
+
"learning_rate": 7.684600868244919e-07,
|
| 413 |
+
"loss": 0.0001,
|
| 414 |
+
"reward": 0.5688013978302479,
|
| 415 |
+
"reward_std": 0.08710555551806465,
|
| 416 |
+
"rewards/wrapped_prediction_reward_func": 0.5688013978302479,
|
| 417 |
+
"step": 170
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"completion_length": 765.7625213623047,
|
| 421 |
+
"epoch": 0.44080604534005036,
|
| 422 |
+
"grad_norm": 0.07454384571252022,
|
| 423 |
+
"kl": 0.0582550048828125,
|
| 424 |
+
"learning_rate": 7.546600811643816e-07,
|
| 425 |
+
"loss": 0.0001,
|
| 426 |
+
"reward": 0.5501547582447529,
|
| 427 |
+
"reward_std": 0.06000117462826893,
|
| 428 |
+
"rewards/wrapped_prediction_reward_func": 0.5501547582447529,
|
| 429 |
+
"step": 175
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"completion_length": 804.5687713623047,
|
| 433 |
+
"epoch": 0.4534005037783375,
|
| 434 |
+
"grad_norm": 0.0759962714852466,
|
| 435 |
+
"kl": 0.051287841796875,
|
| 436 |
+
"learning_rate": 7.405929722454025e-07,
|
| 437 |
+
"loss": 0.0,
|
| 438 |
+
"reward": 0.5598295979201794,
|
| 439 |
+
"reward_std": 0.08419673591852188,
|
| 440 |
+
"rewards/wrapped_prediction_reward_func": 0.5598295979201794,
|
| 441 |
+
"step": 180
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"completion_length": 762.9896026611328,
|
| 445 |
+
"epoch": 0.4659949622166247,
|
| 446 |
+
"grad_norm": 0.05921819318149447,
|
| 447 |
+
"kl": 0.059259033203125,
|
| 448 |
+
"learning_rate": 7.262735145222695e-07,
|
| 449 |
+
"loss": 0.0001,
|
| 450 |
+
"reward": 0.5768641114234925,
|
| 451 |
+
"reward_std": 0.0767961086006835,
|
| 452 |
+
"rewards/wrapped_prediction_reward_func": 0.5768641114234925,
|
| 453 |
+
"step": 185
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"completion_length": 756.3416885375976,
|
| 457 |
+
"epoch": 0.47858942065491183,
|
| 458 |
+
"grad_norm": 0.07564514416368207,
|
| 459 |
+
"kl": 0.0628173828125,
|
| 460 |
+
"learning_rate": 7.117167271287452e-07,
|
| 461 |
+
"loss": 0.0,
|
| 462 |
+
"reward": 0.5682952009141445,
|
| 463 |
+
"reward_std": 0.07079082436393946,
|
| 464 |
+
"rewards/wrapped_prediction_reward_func": 0.5682952009141445,
|
| 465 |
+
"step": 190
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"completion_length": 792.5271087646485,
|
| 469 |
+
"epoch": 0.491183879093199,
|
| 470 |
+
"grad_norm": 0.05939427024749797,
|
| 471 |
+
"kl": 0.0619049072265625,
|
| 472 |
+
"learning_rate": 6.969378781246436e-07,
|
| 473 |
+
"loss": 0.0,
|
| 474 |
+
"reward": 0.555230350792408,
|
| 475 |
+
"reward_std": 0.05828617985825986,
|
| 476 |
+
"rewards/wrapped_prediction_reward_func": 0.555230350792408,
|
| 477 |
+
"step": 195
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"completion_length": 805.2312698364258,
|
| 481 |
+
"epoch": 0.5037783375314862,
|
| 482 |
+
"grad_norm": 0.053418514095113,
|
| 483 |
+
"kl": 0.0571746826171875,
|
| 484 |
+
"learning_rate": 6.819524684817438e-07,
|
| 485 |
+
"loss": 0.0001,
|
| 486 |
+
"reward": 0.5742501869797707,
|
| 487 |
+
"reward_std": 0.04643813910661265,
|
| 488 |
+
"rewards/wrapped_prediction_reward_func": 0.5742501869797707,
|
| 489 |
+
"step": 200
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"completion_length": 817.6541931152344,
|
| 493 |
+
"epoch": 0.5163727959697733,
|
| 494 |
+
"grad_norm": 0.059367612614114815,
|
| 495 |
+
"kl": 0.0566925048828125,
|
| 496 |
+
"learning_rate": 6.667762158254103e-07,
|
| 497 |
+
"loss": 0.0,
|
| 498 |
+
"reward": 0.5665707983076572,
|
| 499 |
+
"reward_std": 0.05952281908830628,
|
| 500 |
+
"rewards/wrapped_prediction_reward_func": 0.5665707983076572,
|
| 501 |
+
"step": 205
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"completion_length": 785.7229415893555,
|
| 505 |
+
"epoch": 0.5289672544080605,
|
| 506 |
+
"grad_norm": 0.06832063353240915,
|
| 507 |
+
"kl": 0.063702392578125,
|
| 508 |
+
"learning_rate": 6.514250379489753e-07,
|
| 509 |
+
"loss": 0.0001,
|
| 510 |
+
"reward": 0.5705230697989464,
|
| 511 |
+
"reward_std": 0.055411939066834746,
|
| 512 |
+
"rewards/wrapped_prediction_reward_func": 0.5705230697989464,
|
| 513 |
+
"step": 210
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"completion_length": 845.8625198364258,
|
| 517 |
+
"epoch": 0.5415617128463476,
|
| 518 |
+
"grad_norm": 0.059630081848652705,
|
| 519 |
+
"kl": 0.055462646484375,
|
| 520 |
+
"learning_rate": 6.359150361181714e-07,
|
| 521 |
+
"loss": 0.0001,
|
| 522 |
+
"reward": 0.5717524968087673,
|
| 523 |
+
"reward_std": 0.058589562051929535,
|
| 524 |
+
"rewards/wrapped_prediction_reward_func": 0.5717524968087673,
|
| 525 |
+
"step": 215
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"completion_length": 842.5708572387696,
|
| 529 |
+
"epoch": 0.5541561712846348,
|
| 530 |
+
"grad_norm": 0.06370639502021583,
|
| 531 |
+
"kl": 0.0563568115234375,
|
| 532 |
+
"learning_rate": 6.202624781831268e-07,
|
| 533 |
+
"loss": 0.0001,
|
| 534 |
+
"reward": 0.5911780953407287,
|
| 535 |
+
"reward_std": 0.05995338945649564,
|
| 536 |
+
"rewards/wrapped_prediction_reward_func": 0.5911780953407287,
|
| 537 |
+
"step": 220
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"completion_length": 819.0875198364258,
|
| 541 |
+
"epoch": 0.5667506297229219,
|
| 542 |
+
"grad_norm": 0.05972945064618038,
|
| 543 |
+
"kl": 0.1108428955078125,
|
| 544 |
+
"learning_rate": 6.044837815156376e-07,
|
| 545 |
+
"loss": 0.0001,
|
| 546 |
+
"reward": 0.5578226193785667,
|
| 547 |
+
"reward_std": 0.04124137028120458,
|
| 548 |
+
"rewards/wrapped_prediction_reward_func": 0.5578226193785667,
|
| 549 |
+
"step": 225
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"completion_length": 815.4666931152344,
|
| 553 |
+
"epoch": 0.5793450881612091,
|
| 554 |
+
"grad_norm": 0.05137674910063656,
|
| 555 |
+
"kl": 0.0625457763671875,
|
| 556 |
+
"learning_rate": 5.885954957896115e-07,
|
| 557 |
+
"loss": 0.0001,
|
| 558 |
+
"reward": 0.5861903376877308,
|
| 559 |
+
"reward_std": 0.04839982387493365,
|
| 560 |
+
"rewards/wrapped_prediction_reward_func": 0.5861903376877308,
|
| 561 |
+
"step": 230
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"completion_length": 802.9166900634766,
|
| 565 |
+
"epoch": 0.5919395465994962,
|
| 566 |
+
"grad_norm": 0.052644946129698264,
|
| 567 |
+
"kl": 0.065142822265625,
|
| 568 |
+
"learning_rate": 5.726142856227452e-07,
|
| 569 |
+
"loss": 0.0001,
|
| 570 |
+
"reward": 0.5952158778905868,
|
| 571 |
+
"reward_std": 0.05475405912147835,
|
| 572 |
+
"rewards/wrapped_prediction_reward_func": 0.5952158778905868,
|
| 573 |
+
"step": 235
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"completion_length": 839.5187728881835,
|
| 577 |
+
"epoch": 0.6045340050377834,
|
| 578 |
+
"grad_norm": 0.0590405309579095,
|
| 579 |
+
"kl": 0.0682373046875,
|
| 580 |
+
"learning_rate": 5.565569130976422e-07,
|
| 581 |
+
"loss": 0.0,
|
| 582 |
+
"reward": 0.5894757807254791,
|
| 583 |
+
"reward_std": 0.06357808914035559,
|
| 584 |
+
"rewards/wrapped_prediction_reward_func": 0.5894757807254791,
|
| 585 |
+
"step": 240
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"completion_length": 835.6958541870117,
|
| 589 |
+
"epoch": 0.6171284634760705,
|
| 590 |
+
"grad_norm": 0.047712245488407155,
|
| 591 |
+
"kl": 0.064862060546875,
|
| 592 |
+
"learning_rate": 5.404402201807021e-07,
|
| 593 |
+
"loss": 0.0001,
|
| 594 |
+
"reward": 0.5752220213413238,
|
| 595 |
+
"reward_std": 0.04643567528109997,
|
| 596 |
+
"rewards/wrapped_prediction_reward_func": 0.5752220213413238,
|
| 597 |
+
"step": 245
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"completion_length": 821.4750274658203,
|
| 601 |
+
"epoch": 0.6297229219143576,
|
| 602 |
+
"grad_norm": 0.05442474889712706,
|
| 603 |
+
"kl": 0.0702850341796875,
|
| 604 |
+
"learning_rate": 5.242811110572242e-07,
|
| 605 |
+
"loss": 0.0001,
|
| 606 |
+
"reward": 0.5673072084784507,
|
| 607 |
+
"reward_std": 0.03941638254909776,
|
| 608 |
+
"rewards/wrapped_prediction_reward_func": 0.5673072084784507,
|
| 609 |
+
"step": 250
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"completion_length": 781.9187728881836,
|
| 613 |
+
"epoch": 0.6423173803526449,
|
| 614 |
+
"grad_norm": 0.05809885115112598,
|
| 615 |
+
"kl": 0.070941162109375,
|
| 616 |
+
"learning_rate": 5.080965344012508e-07,
|
| 617 |
+
"loss": 0.0001,
|
| 618 |
+
"reward": 0.5991133064031601,
|
| 619 |
+
"reward_std": 0.04335737858200446,
|
| 620 |
+
"rewards/wrapped_prediction_reward_func": 0.5991133064031601,
|
| 621 |
+
"step": 255
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"completion_length": 817.3083541870117,
|
| 625 |
+
"epoch": 0.654911838790932,
|
| 626 |
+
"grad_norm": 0.03682776713225834,
|
| 627 |
+
"kl": 0.0657196044921875,
|
| 628 |
+
"learning_rate": 4.919034655987492e-07,
|
| 629 |
+
"loss": 0.0001,
|
| 630 |
+
"reward": 0.5929536901414394,
|
| 631 |
+
"reward_std": 0.06003379854373634,
|
| 632 |
+
"rewards/wrapped_prediction_reward_func": 0.5929536901414394,
|
| 633 |
+
"step": 260
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"completion_length": 808.1771072387695,
|
| 637 |
+
"epoch": 0.6675062972292192,
|
| 638 |
+
"grad_norm": 0.05594643798861689,
|
| 639 |
+
"kl": 0.071002197265625,
|
| 640 |
+
"learning_rate": 4.75718888942776e-07,
|
| 641 |
+
"loss": 0.0001,
|
| 642 |
+
"reward": 0.5693795874714851,
|
| 643 |
+
"reward_std": 0.04352557165548206,
|
| 644 |
+
"rewards/wrapped_prediction_reward_func": 0.5693795874714851,
|
| 645 |
+
"step": 265
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"completion_length": 850.0666854858398,
|
| 649 |
+
"epoch": 0.6801007556675063,
|
| 650 |
+
"grad_norm": 0.05396780654610799,
|
| 651 |
+
"kl": 0.066058349609375,
|
| 652 |
+
"learning_rate": 4.595597798192979e-07,
|
| 653 |
+
"loss": 0.0001,
|
| 654 |
+
"reward": 0.5841031737625599,
|
| 655 |
+
"reward_std": 0.054474068916169925,
|
| 656 |
+
"rewards/wrapped_prediction_reward_func": 0.5841031737625599,
|
| 657 |
+
"step": 270
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"completion_length": 807.5771041870117,
|
| 661 |
+
"epoch": 0.6926952141057935,
|
| 662 |
+
"grad_norm": 0.05692079461397134,
|
| 663 |
+
"kl": 0.072412109375,
|
| 664 |
+
"learning_rate": 4.434430869023579e-07,
|
| 665 |
+
"loss": 0.0001,
|
| 666 |
+
"reward": 0.6022952854633331,
|
| 667 |
+
"reward_std": 0.05330510977655649,
|
| 668 |
+
"rewards/wrapped_prediction_reward_func": 0.6022952854633331,
|
| 669 |
+
"step": 275
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"completion_length": 846.0312774658203,
|
| 673 |
+
"epoch": 0.7052896725440806,
|
| 674 |
+
"grad_norm": 0.04955339683095041,
|
| 675 |
+
"kl": 0.065411376953125,
|
| 676 |
+
"learning_rate": 4.2738571437725496e-07,
|
| 677 |
+
"loss": 0.0001,
|
| 678 |
+
"reward": 0.5892672084271908,
|
| 679 |
+
"reward_std": 0.05133460226934403,
|
| 680 |
+
"rewards/wrapped_prediction_reward_func": 0.5892672084271908,
|
| 681 |
+
"step": 280
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"completion_length": 781.9396057128906,
|
| 685 |
+
"epoch": 0.7178841309823678,
|
| 686 |
+
"grad_norm": 0.056799969581110395,
|
| 687 |
+
"kl": 0.071551513671875,
|
| 688 |
+
"learning_rate": 4.1140450421038866e-07,
|
| 689 |
+
"loss": 0.0001,
|
| 690 |
+
"reward": 0.5798937246203423,
|
| 691 |
+
"reward_std": 0.051263501844368876,
|
| 692 |
+
"rewards/wrapped_prediction_reward_func": 0.5798937246203423,
|
| 693 |
+
"step": 285
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"completion_length": 797.3041900634765,
|
| 697 |
+
"epoch": 0.7304785894206549,
|
| 698 |
+
"grad_norm": 0.050561172649658176,
|
| 699 |
+
"kl": 0.077020263671875,
|
| 700 |
+
"learning_rate": 3.955162184843624e-07,
|
| 701 |
+
"loss": 0.0001,
|
| 702 |
+
"reward": 0.5860190108418465,
|
| 703 |
+
"reward_std": 0.04162131273187697,
|
| 704 |
+
"rewards/wrapped_prediction_reward_func": 0.5860190108418465,
|
| 705 |
+
"step": 290
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"completion_length": 822.4833526611328,
|
| 709 |
+
"epoch": 0.743073047858942,
|
| 710 |
+
"grad_norm": 0.0434111466318861,
|
| 711 |
+
"kl": 0.0822509765625,
|
| 712 |
+
"learning_rate": 3.7973752181687327e-07,
|
| 713 |
+
"loss": 0.0001,
|
| 714 |
+
"reward": 0.5851313889026641,
|
| 715 |
+
"reward_std": 0.05500690509798005,
|
| 716 |
+
"rewards/wrapped_prediction_reward_func": 0.5851313889026641,
|
| 717 |
+
"step": 295
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"completion_length": 855.5312805175781,
|
| 721 |
+
"epoch": 0.7556675062972292,
|
| 722 |
+
"grad_norm": 0.051422594050945035,
|
| 723 |
+
"kl": 0.0636383056640625,
|
| 724 |
+
"learning_rate": 3.640849638818285e-07,
|
| 725 |
+
"loss": 0.0001,
|
| 726 |
+
"reward": 0.5980591103434563,
|
| 727 |
+
"reward_std": 0.036254264996387064,
|
| 728 |
+
"rewards/wrapped_prediction_reward_func": 0.5980591103434563,
|
| 729 |
+
"step": 300
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"completion_length": 792.4396087646485,
|
| 733 |
+
"epoch": 0.7682619647355163,
|
| 734 |
+
"grad_norm": 0.03997352927393124,
|
| 735 |
+
"kl": 0.0724853515625,
|
| 736 |
+
"learning_rate": 3.485749620510247e-07,
|
| 737 |
+
"loss": 0.0001,
|
| 738 |
+
"reward": 0.6093361288309097,
|
| 739 |
+
"reward_std": 0.03319716795813292,
|
| 740 |
+
"rewards/wrapped_prediction_reward_func": 0.6093361288309097,
|
| 741 |
+
"step": 305
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"completion_length": 812.2625213623047,
|
| 745 |
+
"epoch": 0.7808564231738035,
|
| 746 |
+
"grad_norm": 0.058958040086389545,
|
| 747 |
+
"kl": 0.072906494140625,
|
| 748 |
+
"learning_rate": 3.3322378417458977e-07,
|
| 749 |
+
"loss": 0.0001,
|
| 750 |
+
"reward": 0.613701456785202,
|
| 751 |
+
"reward_std": 0.041399752866709605,
|
| 752 |
+
"rewards/wrapped_prediction_reward_func": 0.613701456785202,
|
| 753 |
+
"step": 310
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"completion_length": 830.6354431152344,
|
| 757 |
+
"epoch": 0.7934508816120907,
|
| 758 |
+
"grad_norm": 0.03534980235662457,
|
| 759 |
+
"kl": 0.067303466796875,
|
| 760 |
+
"learning_rate": 3.1804753151825627e-07,
|
| 761 |
+
"loss": 0.0001,
|
| 762 |
+
"reward": 0.5734647884964943,
|
| 763 |
+
"reward_std": 0.051193205296294765,
|
| 764 |
+
"rewards/wrapped_prediction_reward_func": 0.5734647884964943,
|
| 765 |
+
"step": 315
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"completion_length": 813.2833557128906,
|
| 769 |
+
"epoch": 0.8060453400503779,
|
| 770 |
+
"grad_norm": 0.06257699195249379,
|
| 771 |
+
"kl": 0.068646240234375,
|
| 772 |
+
"learning_rate": 3.030621218753565e-07,
|
| 773 |
+
"loss": 0.0001,
|
| 774 |
+
"reward": 0.6155773043632508,
|
| 775 |
+
"reward_std": 0.039483394497074185,
|
| 776 |
+
"rewards/wrapped_prediction_reward_func": 0.6155773043632508,
|
| 777 |
+
"step": 320
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"completion_length": 798.3354370117188,
|
| 781 |
+
"epoch": 0.818639798488665,
|
| 782 |
+
"grad_norm": 0.047751764206928074,
|
| 783 |
+
"kl": 0.0704833984375,
|
| 784 |
+
"learning_rate": 2.8828327287125507e-07,
|
| 785 |
+
"loss": 0.0001,
|
| 786 |
+
"reward": 0.608567351102829,
|
| 787 |
+
"reward_std": 0.03854468032368459,
|
| 788 |
+
"rewards/wrapped_prediction_reward_func": 0.608567351102829,
|
| 789 |
+
"step": 325
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"completion_length": 828.9812713623047,
|
| 793 |
+
"epoch": 0.8312342569269522,
|
| 794 |
+
"grad_norm": 0.04585758374215142,
|
| 795 |
+
"kl": 0.0710540771484375,
|
| 796 |
+
"learning_rate": 2.7372648547773056e-07,
|
| 797 |
+
"loss": 0.0001,
|
| 798 |
+
"reward": 0.5888618856668473,
|
| 799 |
+
"reward_std": 0.03441393570974469,
|
| 800 |
+
"rewards/wrapped_prediction_reward_func": 0.5888618856668473,
|
| 801 |
+
"step": 330
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"completion_length": 821.6791931152344,
|
| 805 |
+
"epoch": 0.8438287153652393,
|
| 806 |
+
"grad_norm": 0.04800095851199432,
|
| 807 |
+
"kl": 0.070928955078125,
|
| 808 |
+
"learning_rate": 2.5940702775459744e-07,
|
| 809 |
+
"loss": 0.0001,
|
| 810 |
+
"reward": 0.5711221612989903,
|
| 811 |
+
"reward_std": 0.037968299351632595,
|
| 812 |
+
"rewards/wrapped_prediction_reward_func": 0.5711221612989903,
|
| 813 |
+
"step": 335
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"completion_length": 825.0750167846679,
|
| 817 |
+
"epoch": 0.8564231738035264,
|
| 818 |
+
"grad_norm": 0.032868154882953755,
|
| 819 |
+
"kl": 0.074005126953125,
|
| 820 |
+
"learning_rate": 2.4533991883561867e-07,
|
| 821 |
+
"loss": 0.0002,
|
| 822 |
+
"reward": 0.5961314618587494,
|
| 823 |
+
"reward_std": 0.02036965051665902,
|
| 824 |
+
"rewards/wrapped_prediction_reward_func": 0.5961314618587494,
|
| 825 |
+
"step": 340
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"completion_length": 789.3958572387695,
|
| 829 |
+
"epoch": 0.8690176322418136,
|
| 830 |
+
"grad_norm": 0.04730084864351471,
|
| 831 |
+
"kl": 0.0718597412109375,
|
| 832 |
+
"learning_rate": 2.3153991317550808e-07,
|
| 833 |
+
"loss": 0.0001,
|
| 834 |
+
"reward": 0.5693817652761937,
|
| 835 |
+
"reward_std": 0.027137306291842835,
|
| 836 |
+
"rewards/wrapped_prediction_reward_func": 0.5693817652761937,
|
| 837 |
+
"step": 345
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"completion_length": 825.4916915893555,
|
| 841 |
+
"epoch": 0.8816120906801007,
|
| 842 |
+
"grad_norm": 0.05929565464749555,
|
| 843 |
+
"kl": 0.0830902099609375,
|
| 844 |
+
"learning_rate": 2.180214850745467e-07,
|
| 845 |
+
"loss": 0.0001,
|
| 846 |
+
"reward": 0.5720083937048912,
|
| 847 |
+
"reward_std": 0.04108033460797742,
|
| 848 |
+
"rewards/wrapped_prediction_reward_func": 0.5720083937048912,
|
| 849 |
+
"step": 350
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"completion_length": 794.3854446411133,
|
| 853 |
+
"epoch": 0.8942065491183879,
|
| 854 |
+
"grad_norm": 0.042508843892672864,
|
| 855 |
+
"kl": 0.0805755615234375,
|
| 856 |
+
"learning_rate": 2.0479881349703882e-07,
|
| 857 |
+
"loss": 0.0001,
|
| 858 |
+
"reward": 0.6149648398160934,
|
| 859 |
+
"reward_std": 0.028531424299580978,
|
| 860 |
+
"rewards/wrapped_prediction_reward_func": 0.6149648398160934,
|
| 861 |
+
"step": 355
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"completion_length": 828.1271102905273,
|
| 865 |
+
"epoch": 0.906801007556675,
|
| 866 |
+
"grad_norm": 0.13929773259081887,
|
| 867 |
+
"kl": 0.1125732421875,
|
| 868 |
+
"learning_rate": 1.918857671995363e-07,
|
| 869 |
+
"loss": 0.0001,
|
| 870 |
+
"reward": 0.5847930148243904,
|
| 871 |
+
"reward_std": 0.044799246825277805,
|
| 872 |
+
"rewards/wrapped_prediction_reward_func": 0.5847930148243904,
|
| 873 |
+
"step": 360
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"completion_length": 843.9896087646484,
|
| 877 |
+
"epoch": 0.9193954659949622,
|
| 878 |
+
"grad_norm": 0.044992966442867634,
|
| 879 |
+
"kl": 0.0656494140625,
|
| 880 |
+
"learning_rate": 1.7929589018443014e-07,
|
| 881 |
+
"loss": 0.0001,
|
| 882 |
+
"reward": 0.5992885082960129,
|
| 883 |
+
"reward_std": 0.029869268811307848,
|
| 884 |
+
"rewards/wrapped_prediction_reward_func": 0.5992885082960129,
|
| 885 |
+
"step": 365
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"completion_length": 791.6791900634765,
|
| 889 |
+
"epoch": 0.9319899244332494,
|
| 890 |
+
"grad_norm": 0.03456117446845568,
|
| 891 |
+
"kl": 0.07545166015625,
|
| 892 |
+
"learning_rate": 1.6704238749415955e-07,
|
| 893 |
+
"loss": 0.0002,
|
| 894 |
+
"reward": 0.5825795724987983,
|
| 895 |
+
"reward_std": 0.021053510857746004,
|
| 896 |
+
"rewards/wrapped_prediction_reward_func": 0.5825795724987983,
|
| 897 |
+
"step": 370
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"completion_length": 795.2250274658203,
|
| 901 |
+
"epoch": 0.9445843828715366,
|
| 902 |
+
"grad_norm": 0.057925315822869305,
|
| 903 |
+
"kl": 0.083660888671875,
|
| 904 |
+
"learning_rate": 1.5513811136094785e-07,
|
| 905 |
+
"loss": 0.0001,
|
| 906 |
+
"reward": 0.5829600065946579,
|
| 907 |
+
"reward_std": 0.034487396385520695,
|
| 908 |
+
"rewards/wrapped_prediction_reward_func": 0.5829600065946579,
|
| 909 |
+
"step": 375
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"completion_length": 846.0875274658204,
|
| 913 |
+
"epoch": 0.9571788413098237,
|
| 914 |
+
"grad_norm": 0.038214011850815244,
|
| 915 |
+
"kl": 0.072222900390625,
|
| 916 |
+
"learning_rate": 1.435955477265855e-07,
|
| 917 |
+
"loss": 0.0001,
|
| 918 |
+
"reward": 0.6045917004346848,
|
| 919 |
+
"reward_std": 0.05368394823744893,
|
| 920 |
+
"rewards/wrapped_prediction_reward_func": 0.6045917004346848,
|
| 921 |
+
"step": 380
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"completion_length": 815.0458526611328,
|
| 925 |
+
"epoch": 0.9697732997481109,
|
| 926 |
+
"grad_norm": 0.030377786493206754,
|
| 927 |
+
"kl": 0.078546142578125,
|
| 928 |
+
"learning_rate": 1.3242680314639993e-07,
|
| 929 |
+
"loss": 0.0001,
|
| 930 |
+
"reward": 0.6047316044569016,
|
| 931 |
+
"reward_std": 0.024706879258155824,
|
| 932 |
+
"rewards/wrapped_prediction_reward_func": 0.6047316044569016,
|
| 933 |
+
"step": 385
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"completion_length": 823.7146057128906,
|
| 937 |
+
"epoch": 0.982367758186398,
|
| 938 |
+
"grad_norm": 0.04345292422364211,
|
| 939 |
+
"kl": 0.07333984375,
|
| 940 |
+
"learning_rate": 1.2164359209115232e-07,
|
| 941 |
+
"loss": 0.0001,
|
| 942 |
+
"reward": 0.5892785102128982,
|
| 943 |
+
"reward_std": 0.03435427148942836,
|
| 944 |
+
"rewards/wrapped_prediction_reward_func": 0.5892785102128982,
|
| 945 |
+
"step": 390
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"completion_length": 854.6458587646484,
|
| 949 |
+
"epoch": 0.9949622166246851,
|
| 950 |
+
"grad_norm": 0.05065816791479697,
|
| 951 |
+
"kl": 0.0732177734375,
|
| 952 |
+
"learning_rate": 1.1125722466017545e-07,
|
| 953 |
+
"loss": 0.0001,
|
| 954 |
+
"reward": 0.5809489727020264,
|
| 955 |
+
"reward_std": 0.052034072624519465,
|
| 956 |
+
"rewards/wrapped_prediction_reward_func": 0.5809489727020264,
|
| 957 |
+
"step": 395
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"completion_length": 810.3968933105468,
|
| 961 |
+
"epoch": 1.0075566750629723,
|
| 962 |
+
"grad_norm": 0.03509442740105049,
|
| 963 |
+
"kl": 0.075762939453125,
|
| 964 |
+
"learning_rate": 1.0127859471863969e-07,
|
| 965 |
+
"loss": 0.0001,
|
| 966 |
+
"reward": 0.6079857856035232,
|
| 967 |
+
"reward_std": 0.02203769111074507,
|
| 968 |
+
"rewards/wrapped_prediction_reward_func": 0.6079857856035232,
|
| 969 |
+
"step": 400
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"completion_length": 790.1479385375976,
|
| 973 |
+
"epoch": 1.0201511335012594,
|
| 974 |
+
"grad_norm": 0.03530726181544221,
|
| 975 |
+
"kl": 0.0824920654296875,
|
| 976 |
+
"learning_rate": 9.171816847139447e-08,
|
| 977 |
+
"loss": 0.0001,
|
| 978 |
+
"reward": 0.6192175649106503,
|
| 979 |
+
"reward_std": 0.0369229513395112,
|
| 980 |
+
"rewards/wrapped_prediction_reward_func": 0.6192175649106503,
|
| 981 |
+
"step": 405
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"completion_length": 793.627099609375,
|
| 985 |
+
"epoch": 1.0327455919395465,
|
| 986 |
+
"grad_norm": 0.032582398864671,
|
| 987 |
+
"kl": 0.077081298828125,
|
| 988 |
+
"learning_rate": 8.258597348536451e-08,
|
| 989 |
+
"loss": 0.0001,
|
| 990 |
+
"reward": 0.6033682949841023,
|
| 991 |
+
"reward_std": 0.034924795664846894,
|
| 992 |
+
"rewards/wrapped_prediction_reward_func": 0.6033682949841023,
|
| 993 |
+
"step": 410
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"completion_length": 820.5625213623047,
|
| 997 |
+
"epoch": 1.0453400503778338,
|
| 998 |
+
"grad_norm": 0.03594193252721311,
|
| 999 |
+
"kl": 0.0728424072265625,
|
| 1000 |
+
"learning_rate": 7.389158817201541e-08,
|
| 1001 |
+
"loss": 0.0001,
|
| 1002 |
+
"reward": 0.5938129395246505,
|
| 1003 |
+
"reward_std": 0.025039249239489435,
|
| 1004 |
+
"rewards/wrapped_prediction_reward_func": 0.5938129395246505,
|
| 1005 |
+
"step": 415
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"completion_length": 791.6812713623046,
|
| 1009 |
+
"epoch": 1.057934508816121,
|
| 1010 |
+
"grad_norm": 0.2660556227962935,
|
| 1011 |
+
"kl": 0.178399658203125,
|
| 1012 |
+
"learning_rate": 6.564413174092443e-08,
|
| 1013 |
+
"loss": 0.0001,
|
| 1014 |
+
"reward": 0.5894127897918224,
|
| 1015 |
+
"reward_std": 0.026753197464859112,
|
| 1016 |
+
"rewards/wrapped_prediction_reward_func": 0.5894127897918224,
|
| 1017 |
+
"step": 420
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"completion_length": 816.7833587646485,
|
| 1021 |
+
"epoch": 1.070528967254408,
|
| 1022 |
+
"grad_norm": 0.04448961501333409,
|
| 1023 |
+
"kl": 0.074652099609375,
|
| 1024 |
+
"learning_rate": 5.785225463498828e-08,
|
| 1025 |
+
"loss": 0.0002,
|
| 1026 |
+
"reward": 0.6132952854037285,
|
| 1027 |
+
"reward_std": 0.04002904643421061,
|
| 1028 |
+
"rewards/wrapped_prediction_reward_func": 0.6132952854037285,
|
| 1029 |
+
"step": 425
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"completion_length": 848.4500198364258,
|
| 1033 |
+
"epoch": 1.0831234256926952,
|
| 1034 |
+
"grad_norm": 0.048148418977785004,
|
| 1035 |
+
"kl": 0.073956298828125,
|
| 1036 |
+
"learning_rate": 5.052412945730239e-08,
|
| 1037 |
+
"loss": 0.0001,
|
| 1038 |
+
"reward": 0.5875438123941421,
|
| 1039 |
+
"reward_std": 0.043150549242272976,
|
| 1040 |
+
"rewards/wrapped_prediction_reward_func": 0.5875438123941421,
|
| 1041 |
+
"step": 430
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"completion_length": 830.6750244140625,
|
| 1045 |
+
"epoch": 1.0957178841309823,
|
| 1046 |
+
"grad_norm": 0.05194640005960778,
|
| 1047 |
+
"kl": 0.082000732421875,
|
| 1048 |
+
"learning_rate": 4.366744239922998e-08,
|
| 1049 |
+
"loss": 0.0002,
|
| 1050 |
+
"reward": 0.6154551565647125,
|
| 1051 |
+
"reward_std": 0.040564915537834166,
|
| 1052 |
+
"rewards/wrapped_prediction_reward_func": 0.6154551565647125,
|
| 1053 |
+
"step": 435
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"completion_length": 836.8979400634765,
|
| 1057 |
+
"epoch": 1.1083123425692696,
|
| 1058 |
+
"grad_norm": 0.01997423624486904,
|
| 1059 |
+
"kl": 0.080609130859375,
|
| 1060 |
+
"learning_rate": 3.7289385178647935e-08,
|
| 1061 |
+
"loss": 0.0001,
|
| 1062 |
+
"reward": 0.60478435754776,
|
| 1063 |
+
"reward_std": 0.026485644932836293,
|
| 1064 |
+
"rewards/wrapped_prediction_reward_func": 0.60478435754776,
|
| 1065 |
+
"step": 440
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"completion_length": 832.4625228881836,
|
| 1069 |
+
"epoch": 1.1209068010075567,
|
| 1070 |
+
"grad_norm": 0.017870734412291134,
|
| 1071 |
+
"kl": 0.0710479736328125,
|
| 1072 |
+
"learning_rate": 3.1396647496828244e-08,
|
| 1073 |
+
"loss": 0.0001,
|
| 1074 |
+
"reward": 0.6108729064464569,
|
| 1075 |
+
"reward_std": 0.03767378572374582,
|
| 1076 |
+
"rewards/wrapped_prediction_reward_func": 0.6108729064464569,
|
| 1077 |
+
"step": 445
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"completion_length": 882.7062759399414,
|
| 1081 |
+
"epoch": 1.1335012594458438,
|
| 1082 |
+
"grad_norm": 0.05015364292113255,
|
| 1083 |
+
"kl": 0.0659332275390625,
|
| 1084 |
+
"learning_rate": 2.5995410021864783e-08,
|
| 1085 |
+
"loss": 0.0001,
|
| 1086 |
+
"reward": 0.5940894782543182,
|
| 1087 |
+
"reward_std": 0.04104239339358173,
|
| 1088 |
+
"rewards/wrapped_prediction_reward_func": 0.5940894782543182,
|
| 1089 |
+
"step": 450
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"completion_length": 820.8708557128906,
|
| 1093 |
+
"epoch": 1.146095717884131,
|
| 1094 |
+
"grad_norm": 0.03890549939275878,
|
| 1095 |
+
"kl": 0.079840087890625,
|
| 1096 |
+
"learning_rate": 2.109133790600648e-08,
|
| 1097 |
+
"loss": 0.0002,
|
| 1098 |
+
"reward": 0.6024712681770324,
|
| 1099 |
+
"reward_std": 0.03324083940824494,
|
| 1100 |
+
"rewards/wrapped_prediction_reward_func": 0.6024712681770324,
|
| 1101 |
+
"step": 455
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"completion_length": 816.364598083496,
|
| 1105 |
+
"epoch": 1.1586901763224182,
|
| 1106 |
+
"grad_norm": 0.03882316928789839,
|
| 1107 |
+
"kl": 0.07708740234375,
|
| 1108 |
+
"learning_rate": 1.6689574843694432e-08,
|
| 1109 |
+
"loss": 0.0001,
|
| 1110 |
+
"reward": 0.575566939264536,
|
| 1111 |
+
"reward_std": 0.049419730342924593,
|
| 1112 |
+
"rewards/wrapped_prediction_reward_func": 0.575566939264536,
|
| 1113 |
+
"step": 460
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"completion_length": 854.1916915893555,
|
| 1117 |
+
"epoch": 1.1712846347607053,
|
| 1118 |
+
"grad_norm": 0.018046225222633556,
|
| 1119 |
+
"kl": 0.06839599609375,
|
| 1120 |
+
"learning_rate": 1.2794737676536993e-08,
|
| 1121 |
+
"loss": 0.0001,
|
| 1122 |
+
"reward": 0.5933746129274369,
|
| 1123 |
+
"reward_std": 0.02758802007883787,
|
| 1124 |
+
"rewards/wrapped_prediction_reward_func": 0.5933746129274369,
|
| 1125 |
+
"step": 465
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"completion_length": 877.6416931152344,
|
| 1129 |
+
"epoch": 1.1838790931989924,
|
| 1130 |
+
"grad_norm": 0.04393883375775292,
|
| 1131 |
+
"kl": 0.064520263671875,
|
| 1132 |
+
"learning_rate": 9.410911550880474e-09,
|
| 1133 |
+
"loss": 0.0001,
|
| 1134 |
+
"reward": 0.5791490338742733,
|
| 1135 |
+
"reward_std": 0.02540514359716326,
|
| 1136 |
+
"rewards/wrapped_prediction_reward_func": 0.5791490338742733,
|
| 1137 |
+
"step": 470
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"completion_length": 823.2521118164062,
|
| 1141 |
+
"epoch": 1.1964735516372795,
|
| 1142 |
+
"grad_norm": 0.019200187518650894,
|
| 1143 |
+
"kl": 0.072186279296875,
|
| 1144 |
+
"learning_rate": 6.541645633054649e-09,
|
| 1145 |
+
"loss": 0.0002,
|
| 1146 |
+
"reward": 0.577340692281723,
|
| 1147 |
+
"reward_std": 0.024969300418160856,
|
| 1148 |
+
"rewards/wrapped_prediction_reward_func": 0.577340692281723,
|
| 1149 |
+
"step": 475
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"completion_length": 835.3208541870117,
|
| 1153 |
+
"epoch": 1.2090680100755669,
|
| 1154 |
+
"grad_norm": 0.03179886042365874,
|
| 1155 |
+
"kl": 0.07593994140625,
|
| 1156 |
+
"learning_rate": 4.189949386787462e-09,
|
| 1157 |
+
"loss": 0.0,
|
| 1158 |
+
"reward": 0.6053503692150116,
|
| 1159 |
+
"reward_std": 0.05023272330872715,
|
| 1160 |
+
"rewards/wrapped_prediction_reward_func": 0.6053503692150116,
|
| 1161 |
+
"step": 480
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"completion_length": 800.9750183105468,
|
| 1165 |
+
"epoch": 1.221662468513854,
|
| 1166 |
+
"grad_norm": 0.031246606223002214,
|
| 1167 |
+
"kl": 0.090142822265625,
|
| 1168 |
+
"learning_rate": 2.3582894166930267e-09,
|
| 1169 |
+
"loss": 0.0001,
|
| 1170 |
+
"reward": 0.6070939235389232,
|
| 1171 |
+
"reward_std": 0.03144086834508926,
|
| 1172 |
+
"rewards/wrapped_prediction_reward_func": 0.6070939235389232,
|
| 1173 |
+
"step": 485
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"completion_length": 848.7771072387695,
|
| 1177 |
+
"epoch": 1.234256926952141,
|
| 1178 |
+
"grad_norm": 0.05601067005700773,
|
| 1179 |
+
"kl": 0.07252197265625,
|
| 1180 |
+
"learning_rate": 1.0485868811441756e-09,
|
| 1181 |
+
"loss": 0.0001,
|
| 1182 |
+
"reward": 0.5712633952498436,
|
| 1183 |
+
"reward_std": 0.04291044136043638,
|
| 1184 |
+
"rewards/wrapped_prediction_reward_func": 0.5712633952498436,
|
| 1185 |
+
"step": 490
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"completion_length": 771.6208572387695,
|
| 1189 |
+
"epoch": 1.2468513853904282,
|
| 1190 |
+
"grad_norm": 0.03724004112267365,
|
| 1191 |
+
"kl": 0.085308837890625,
|
| 1192 |
+
"learning_rate": 2.6221547724253333e-10,
|
| 1193 |
+
"loss": 0.0001,
|
| 1194 |
+
"reward": 0.6077594101428986,
|
| 1195 |
+
"reward_std": 0.026009649853222072,
|
| 1196 |
+
"rewards/wrapped_prediction_reward_func": 0.6077594101428986,
|
| 1197 |
+
"step": 495
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"completion_length": 796.8458557128906,
|
| 1201 |
+
"epoch": 1.2594458438287153,
|
| 1202 |
+
"grad_norm": 0.053721210517970415,
|
| 1203 |
+
"kl": 0.0768798828125,
|
| 1204 |
+
"learning_rate": 0.0,
|
| 1205 |
+
"loss": 0.0002,
|
| 1206 |
+
"reward": 0.601429545879364,
|
| 1207 |
+
"reward_std": 0.024498459114693106,
|
| 1208 |
+
"rewards/wrapped_prediction_reward_func": 0.601429545879364,
|
| 1209 |
+
"step": 500
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 1.2594458438287153,
|
| 1213 |
+
"step": 500,
|
| 1214 |
+
"total_flos": 0.0,
|
| 1215 |
+
"train_loss": 7.065190569642255e-05,
|
| 1216 |
+
"train_runtime": 24090.8459,
|
| 1217 |
+
"train_samples_per_second": 1.992,
|
| 1218 |
+
"train_steps_per_second": 0.021
|
| 1219 |
+
}
|
| 1220 |
+
],
|
| 1221 |
+
"logging_steps": 5,
|
| 1222 |
+
"max_steps": 500,
|
| 1223 |
+
"num_input_tokens_seen": 0,
|
| 1224 |
+
"num_train_epochs": 2,
|
| 1225 |
+
"save_steps": 100,
|
| 1226 |
+
"stateful_callbacks": {
|
| 1227 |
+
"TrainerControl": {
|
| 1228 |
+
"args": {
|
| 1229 |
+
"should_epoch_stop": false,
|
| 1230 |
+
"should_evaluate": false,
|
| 1231 |
+
"should_log": false,
|
| 1232 |
+
"should_save": true,
|
| 1233 |
+
"should_training_stop": true
|
| 1234 |
+
},
|
| 1235 |
+
"attributes": {}
|
| 1236 |
+
}
|
| 1237 |
+
},
|
| 1238 |
+
"total_flos": 0.0,
|
| 1239 |
+
"train_batch_size": 2,
|
| 1240 |
+
"trial_name": null,
|
| 1241 |
+
"trial_params": null
|
| 1242 |
+
}
|