Model save
Browse files- README.md +67 -0
- all_results.json +8 -0
- config.json +1 -1
- generation_config.json +6 -0
- train_results.json +8 -0
- trainer_state.json +2239 -0
README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
model_name: CodeLlama-7b-hf_1000rl_cpp
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
- trl
|
| 7 |
+
- grpo
|
| 8 |
+
licence: license
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Model Card for CodeLlama-7b-hf_1000rl_cpp
|
| 12 |
+
|
| 13 |
+
This model is a fine-tuned version of [None](https://huggingface.co/None).
|
| 14 |
+
It has been trained using [TRL](https://github.com/huggingface/trl).
|
| 15 |
+
|
| 16 |
+
## Quick start
|
| 17 |
+
|
| 18 |
+
```python
|
| 19 |
+
from transformers import pipeline
|
| 20 |
+
|
| 21 |
+
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?"
|
| 22 |
+
generator = pipeline("text-generation", model="zerozeroz/CodeLlama-7b-hf_1000rl_cpp", device="cuda")
|
| 23 |
+
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
|
| 24 |
+
print(output["generated_text"])
|
| 25 |
+
```
|
| 26 |
+
|
| 27 |
+
## Training procedure
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
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).
|
| 33 |
+
|
| 34 |
+
### Framework versions
|
| 35 |
+
|
| 36 |
+
- TRL: 0.14.0
|
| 37 |
+
- Transformers: 4.48.1
|
| 38 |
+
- Pytorch: 2.5.1+cu121
|
| 39 |
+
- Datasets: 3.1.0
|
| 40 |
+
- Tokenizers: 0.21.0
|
| 41 |
+
|
| 42 |
+
## Citations
|
| 43 |
+
|
| 44 |
+
Cite GRPO as:
|
| 45 |
+
|
| 46 |
+
```bibtex
|
| 47 |
+
@article{zhihong2024deepseekmath,
|
| 48 |
+
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
|
| 49 |
+
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},
|
| 50 |
+
year = 2024,
|
| 51 |
+
eprint = {arXiv:2402.03300},
|
| 52 |
+
}
|
| 53 |
+
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
Cite TRL as:
|
| 57 |
+
|
| 58 |
+
```bibtex
|
| 59 |
+
@misc{vonwerra2022trl,
|
| 60 |
+
title = {{TRL: Transformer Reinforcement Learning}},
|
| 61 |
+
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},
|
| 62 |
+
year = 2020,
|
| 63 |
+
journal = {GitHub repository},
|
| 64 |
+
publisher = {GitHub},
|
| 65 |
+
howpublished = {\url{https://github.com/huggingface/trl}}
|
| 66 |
+
}
|
| 67 |
+
```
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 6.70051409522318e-05,
|
| 4 |
+
"train_runtime": 7347.2678,
|
| 5 |
+
"train_samples": 338,
|
| 6 |
+
"train_samples_per_second": 0.138,
|
| 7 |
+
"train_steps_per_second": 0.023
|
| 8 |
+
}
|
config.json
CHANGED
|
@@ -25,6 +25,6 @@
|
|
| 25 |
"tie_word_embeddings": false,
|
| 26 |
"torch_dtype": "bfloat16",
|
| 27 |
"transformers_version": "4.48.1",
|
| 28 |
-
"use_cache":
|
| 29 |
"vocab_size": 32016
|
| 30 |
}
|
|
|
|
| 25 |
"tie_word_embeddings": false,
|
| 26 |
"torch_dtype": "bfloat16",
|
| 27 |
"transformers_version": "4.48.1",
|
| 28 |
+
"use_cache": true,
|
| 29 |
"vocab_size": 32016
|
| 30 |
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.48.1"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"total_flos": 0.0,
|
| 3 |
+
"train_loss": 6.70051409522318e-05,
|
| 4 |
+
"train_runtime": 7347.2678,
|
| 5 |
+
"train_samples": 338,
|
| 6 |
+
"train_samples_per_second": 0.138,
|
| 7 |
+
"train_steps_per_second": 0.023
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2239 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.9734513274336285,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 169,
|
| 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": 186.12500381469727,
|
| 13 |
+
"epoch": 0.017699115044247787,
|
| 14 |
+
"grad_norm": 2.5752858647583037,
|
| 15 |
+
"kl": 0.0,
|
| 16 |
+
"learning_rate": 8.333333333333333e-08,
|
| 17 |
+
"loss": 0.0,
|
| 18 |
+
"reward": 0.48444899916648865,
|
| 19 |
+
"reward_std": 0.46400773525238037,
|
| 20 |
+
"rewards/correct_code_reward_func": 0.3125,
|
| 21 |
+
"rewards/len_reward_func": 0.17194896191358566,
|
| 22 |
+
"step": 1
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"completion_length": 164.375,
|
| 26 |
+
"epoch": 0.035398230088495575,
|
| 27 |
+
"grad_norm": 10.223894549317365,
|
| 28 |
+
"kl": 0.0,
|
| 29 |
+
"learning_rate": 1.6666666666666665e-07,
|
| 30 |
+
"loss": 0.0,
|
| 31 |
+
"reward": 0.47450903058052063,
|
| 32 |
+
"reward_std": 0.40167468786239624,
|
| 33 |
+
"rewards/correct_code_reward_func": 0.2500000149011612,
|
| 34 |
+
"rewards/len_reward_func": 0.22450900077819824,
|
| 35 |
+
"step": 2
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"completion_length": 164.4791717529297,
|
| 39 |
+
"epoch": 0.05309734513274336,
|
| 40 |
+
"grad_norm": 2.2899995314753827,
|
| 41 |
+
"kl": 4.023313522338867e-05,
|
| 42 |
+
"learning_rate": 2.5e-07,
|
| 43 |
+
"loss": 0.0,
|
| 44 |
+
"reward": 0.37769100069999695,
|
| 45 |
+
"reward_std": 0.5044284015893936,
|
| 46 |
+
"rewards/correct_code_reward_func": 0.2083333432674408,
|
| 47 |
+
"rewards/len_reward_func": 0.16935766488313675,
|
| 48 |
+
"step": 3
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"completion_length": 190.12500762939453,
|
| 52 |
+
"epoch": 0.07079646017699115,
|
| 53 |
+
"grad_norm": 6.009103649538728,
|
| 54 |
+
"kl": 8.815526962280273e-05,
|
| 55 |
+
"learning_rate": 3.333333333333333e-07,
|
| 56 |
+
"loss": 0.0,
|
| 57 |
+
"reward": 0.37105801701545715,
|
| 58 |
+
"reward_std": 0.47604119777679443,
|
| 59 |
+
"rewards/correct_code_reward_func": 0.2083333395421505,
|
| 60 |
+
"rewards/len_reward_func": 0.16272468864917755,
|
| 61 |
+
"step": 4
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"completion_length": 126.10416793823242,
|
| 65 |
+
"epoch": 0.08849557522123894,
|
| 66 |
+
"grad_norm": 2.3931977443517964,
|
| 67 |
+
"kl": 3.49879264831543e-05,
|
| 68 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 69 |
+
"loss": 0.0,
|
| 70 |
+
"reward": 0.5486156344413757,
|
| 71 |
+
"reward_std": 0.40741610527038574,
|
| 72 |
+
"rewards/correct_code_reward_func": 0.3958333358168602,
|
| 73 |
+
"rewards/len_reward_func": 0.15278229862451553,
|
| 74 |
+
"step": 5
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"completion_length": 166.45833587646484,
|
| 78 |
+
"epoch": 0.10619469026548672,
|
| 79 |
+
"grad_norm": 1.5486991749328145,
|
| 80 |
+
"kl": 5.221366882324219e-05,
|
| 81 |
+
"learning_rate": 5e-07,
|
| 82 |
+
"loss": 0.0,
|
| 83 |
+
"reward": 0.5515251755714417,
|
| 84 |
+
"reward_std": 0.47483181953430176,
|
| 85 |
+
"rewards/correct_code_reward_func": 0.2916666716337204,
|
| 86 |
+
"rewards/len_reward_func": 0.25985850393772125,
|
| 87 |
+
"step": 6
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"completion_length": 161.58333587646484,
|
| 91 |
+
"epoch": 0.12389380530973451,
|
| 92 |
+
"grad_norm": 5.714527026195051,
|
| 93 |
+
"kl": 5.5789947509765625e-05,
|
| 94 |
+
"learning_rate": 4.999535676028337e-07,
|
| 95 |
+
"loss": 0.0,
|
| 96 |
+
"reward": 0.46977874636650085,
|
| 97 |
+
"reward_std": 0.461082324385643,
|
| 98 |
+
"rewards/correct_code_reward_func": 0.229166679084301,
|
| 99 |
+
"rewards/len_reward_func": 0.24061208218336105,
|
| 100 |
+
"step": 7
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"completion_length": 161.89583587646484,
|
| 104 |
+
"epoch": 0.1415929203539823,
|
| 105 |
+
"grad_norm": 3.6484747523940286,
|
| 106 |
+
"kl": 0.00010442733764648438,
|
| 107 |
+
"learning_rate": 4.998142876590749e-07,
|
| 108 |
+
"loss": 0.0,
|
| 109 |
+
"reward": 0.2943377196788788,
|
| 110 |
+
"reward_std": 0.3771718442440033,
|
| 111 |
+
"rewards/correct_code_reward_func": 0.1666666679084301,
|
| 112 |
+
"rewards/len_reward_func": 0.12767105549573898,
|
| 113 |
+
"step": 8
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"completion_length": 147.7916717529297,
|
| 117 |
+
"epoch": 0.1592920353982301,
|
| 118 |
+
"grad_norm": 14.356223088928283,
|
| 119 |
+
"kl": 0.00017464160919189453,
|
| 120 |
+
"learning_rate": 4.99582211905537e-07,
|
| 121 |
+
"loss": 0.0,
|
| 122 |
+
"reward": 0.6274224072694778,
|
| 123 |
+
"reward_std": 0.43979279696941376,
|
| 124 |
+
"rewards/correct_code_reward_func": 0.4166666865348816,
|
| 125 |
+
"rewards/len_reward_func": 0.21075571328401566,
|
| 126 |
+
"step": 9
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"completion_length": 188.2916717529297,
|
| 130 |
+
"epoch": 0.17699115044247787,
|
| 131 |
+
"grad_norm": 3.0299854001506494,
|
| 132 |
+
"kl": 8.487701416015625e-05,
|
| 133 |
+
"learning_rate": 4.992574265488882e-07,
|
| 134 |
+
"loss": 0.0,
|
| 135 |
+
"reward": 0.502624899148941,
|
| 136 |
+
"reward_std": 0.35237132012844086,
|
| 137 |
+
"rewards/correct_code_reward_func": 0.3125000149011612,
|
| 138 |
+
"rewards/len_reward_func": 0.19012490659952164,
|
| 139 |
+
"step": 10
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"completion_length": 109.1875,
|
| 143 |
+
"epoch": 0.19469026548672566,
|
| 144 |
+
"grad_norm": 1.8891984084011832,
|
| 145 |
+
"kl": 0.00060272216796875,
|
| 146 |
+
"learning_rate": 4.988400522336303e-07,
|
| 147 |
+
"loss": 0.0,
|
| 148 |
+
"reward": 0.6938449144363403,
|
| 149 |
+
"reward_std": 0.5790197551250458,
|
| 150 |
+
"rewards/correct_code_reward_func": 0.5208333432674408,
|
| 151 |
+
"rewards/len_reward_func": 0.1730116307735443,
|
| 152 |
+
"step": 11
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"completion_length": 171.2916717529297,
|
| 156 |
+
"epoch": 0.21238938053097345,
|
| 157 |
+
"grad_norm": 2.7849518039157664,
|
| 158 |
+
"kl": 0.0006923675537109375,
|
| 159 |
+
"learning_rate": 4.983302439972828e-07,
|
| 160 |
+
"loss": 0.0,
|
| 161 |
+
"reward": 0.4239259362220764,
|
| 162 |
+
"reward_std": 0.48799367249011993,
|
| 163 |
+
"rewards/correct_code_reward_func": 0.20833333395421505,
|
| 164 |
+
"rewards/len_reward_func": 0.21559256315231323,
|
| 165 |
+
"step": 12
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"completion_length": 196.2291717529297,
|
| 169 |
+
"epoch": 0.23008849557522124,
|
| 170 |
+
"grad_norm": 1.392005725111761,
|
| 171 |
+
"kl": 0.000652313232421875,
|
| 172 |
+
"learning_rate": 4.977281912127938e-07,
|
| 173 |
+
"loss": 0.0,
|
| 174 |
+
"reward": 0.5957696735858917,
|
| 175 |
+
"reward_std": 0.5921532511711121,
|
| 176 |
+
"rewards/correct_code_reward_func": 0.4166666716337204,
|
| 177 |
+
"rewards/len_reward_func": 0.17910300940275192,
|
| 178 |
+
"step": 13
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"completion_length": 158.64583587646484,
|
| 182 |
+
"epoch": 0.24778761061946902,
|
| 183 |
+
"grad_norm": 1.5701172166666861,
|
| 184 |
+
"kl": 0.000614166259765625,
|
| 185 |
+
"learning_rate": 4.970341175181955e-07,
|
| 186 |
+
"loss": 0.0,
|
| 187 |
+
"reward": 0.46809761226177216,
|
| 188 |
+
"reward_std": 0.4979168623685837,
|
| 189 |
+
"rewards/correct_code_reward_func": 0.2708333432674408,
|
| 190 |
+
"rewards/len_reward_func": 0.19726425409317017,
|
| 191 |
+
"step": 14
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"completion_length": 147.1041717529297,
|
| 195 |
+
"epoch": 0.26548672566371684,
|
| 196 |
+
"grad_norm": 2.032633618646926,
|
| 197 |
+
"kl": 0.0015106201171875,
|
| 198 |
+
"learning_rate": 4.962482807335314e-07,
|
| 199 |
+
"loss": 0.0,
|
| 200 |
+
"reward": 0.5865300893783569,
|
| 201 |
+
"reward_std": 0.5591276288032532,
|
| 202 |
+
"rewards/correct_code_reward_func": 0.4166666716337204,
|
| 203 |
+
"rewards/len_reward_func": 0.16986342146992683,
|
| 204 |
+
"step": 15
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"completion_length": 177.9791717529297,
|
| 208 |
+
"epoch": 0.2831858407079646,
|
| 209 |
+
"grad_norm": 53.72073405864756,
|
| 210 |
+
"kl": 0.018054962158203125,
|
| 211 |
+
"learning_rate": 4.95370972765087e-07,
|
| 212 |
+
"loss": 0.0,
|
| 213 |
+
"reward": 0.8554503321647644,
|
| 214 |
+
"reward_std": 0.48426124453544617,
|
| 215 |
+
"rewards/correct_code_reward_func": 0.5625000298023224,
|
| 216 |
+
"rewards/len_reward_func": 0.2929503321647644,
|
| 217 |
+
"step": 16
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"completion_length": 187.31250762939453,
|
| 221 |
+
"epoch": 0.3008849557522124,
|
| 222 |
+
"grad_norm": 2.4065750408913034,
|
| 223 |
+
"kl": 0.001964569091796875,
|
| 224 |
+
"learning_rate": 4.944025194969586e-07,
|
| 225 |
+
"loss": 0.0,
|
| 226 |
+
"reward": 0.36077703535556793,
|
| 227 |
+
"reward_std": 0.35280461609363556,
|
| 228 |
+
"rewards/correct_code_reward_func": 0.1041666716337204,
|
| 229 |
+
"rewards/len_reward_func": 0.25661035627126694,
|
| 230 |
+
"step": 17
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"completion_length": 186.14583587646484,
|
| 234 |
+
"epoch": 0.3185840707964602,
|
| 235 |
+
"grad_norm": 20.459541697505053,
|
| 236 |
+
"kl": 0.021259307861328125,
|
| 237 |
+
"learning_rate": 4.933432806700003e-07,
|
| 238 |
+
"loss": 0.0,
|
| 239 |
+
"reward": 0.5004578679800034,
|
| 240 |
+
"reward_std": 0.42624031007289886,
|
| 241 |
+
"rewards/correct_code_reward_func": 0.2708333432674408,
|
| 242 |
+
"rewards/len_reward_func": 0.22962453216314316,
|
| 243 |
+
"step": 18
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"completion_length": 184.2916717529297,
|
| 247 |
+
"epoch": 0.336283185840708,
|
| 248 |
+
"grad_norm": 3.4167543405430436,
|
| 249 |
+
"kl": 0.008358001708984375,
|
| 250 |
+
"learning_rate": 4.921936497481956e-07,
|
| 251 |
+
"loss": 0.0,
|
| 252 |
+
"reward": 0.3476429730653763,
|
| 253 |
+
"reward_std": 0.43119728565216064,
|
| 254 |
+
"rewards/correct_code_reward_func": 0.0833333358168602,
|
| 255 |
+
"rewards/len_reward_func": 0.2643096297979355,
|
| 256 |
+
"step": 19
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"completion_length": 179.87500762939453,
|
| 260 |
+
"epoch": 0.35398230088495575,
|
| 261 |
+
"grad_norm": 2.8250290981037574,
|
| 262 |
+
"kl": 0.0052490234375,
|
| 263 |
+
"learning_rate": 4.909540537725006e-07,
|
| 264 |
+
"loss": 0.0,
|
| 265 |
+
"reward": 0.476805180311203,
|
| 266 |
+
"reward_std": 0.36166805028915405,
|
| 267 |
+
"rewards/correct_code_reward_func": 0.2708333432674408,
|
| 268 |
+
"rewards/len_reward_func": 0.2059718370437622,
|
| 269 |
+
"step": 20
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"completion_length": 148.9375,
|
| 273 |
+
"epoch": 0.37168141592920356,
|
| 274 |
+
"grad_norm": 1.9459373869466972,
|
| 275 |
+
"kl": 0.002933502197265625,
|
| 276 |
+
"learning_rate": 4.896249532022171e-07,
|
| 277 |
+
"loss": 0.0,
|
| 278 |
+
"reward": 0.5851579010486603,
|
| 279 |
+
"reward_std": 0.3626042455434799,
|
| 280 |
+
"rewards/correct_code_reward_func": 0.3333333358168602,
|
| 281 |
+
"rewards/len_reward_func": 0.2518245801329613,
|
| 282 |
+
"step": 21
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"completion_length": 213.8541717529297,
|
| 286 |
+
"epoch": 0.3893805309734513,
|
| 287 |
+
"grad_norm": 0.5828214607615373,
|
| 288 |
+
"kl": 0.0029754638671875,
|
| 289 |
+
"learning_rate": 4.882068417439492e-07,
|
| 290 |
+
"loss": 0.0,
|
| 291 |
+
"reward": 0.5553887784481049,
|
| 292 |
+
"reward_std": 0.5428541898727417,
|
| 293 |
+
"rewards/correct_code_reward_func": 0.291666679084301,
|
| 294 |
+
"rewards/len_reward_func": 0.26372211426496506,
|
| 295 |
+
"step": 22
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"completion_length": 113.43750381469727,
|
| 299 |
+
"epoch": 0.40707964601769914,
|
| 300 |
+
"grad_norm": 1.0373945570716487,
|
| 301 |
+
"kl": 0.0034637451171875,
|
| 302 |
+
"learning_rate": 4.867002461682128e-07,
|
| 303 |
+
"loss": 0.0,
|
| 304 |
+
"reward": 0.49457328021526337,
|
| 305 |
+
"reward_std": 0.38711363077163696,
|
| 306 |
+
"rewards/correct_code_reward_func": 0.2291666716337204,
|
| 307 |
+
"rewards/len_reward_func": 0.26540662348270416,
|
| 308 |
+
"step": 23
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"completion_length": 145.5416717529297,
|
| 312 |
+
"epoch": 0.4247787610619469,
|
| 313 |
+
"grad_norm": 0.8868802158083684,
|
| 314 |
+
"kl": 0.00489044189453125,
|
| 315 |
+
"learning_rate": 4.851057261137608e-07,
|
| 316 |
+
"loss": 0.0,
|
| 317 |
+
"reward": 0.7876249551773071,
|
| 318 |
+
"reward_std": 0.4593852013349533,
|
| 319 |
+
"rewards/correct_code_reward_func": 0.5208333432674408,
|
| 320 |
+
"rewards/len_reward_func": 0.26679155230522156,
|
| 321 |
+
"step": 24
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"completion_length": 223.50000762939453,
|
| 325 |
+
"epoch": 0.4424778761061947,
|
| 326 |
+
"grad_norm": 2.7373544353808152,
|
| 327 |
+
"kl": 0.005214691162109375,
|
| 328 |
+
"learning_rate": 4.83423873879701e-07,
|
| 329 |
+
"loss": 0.0,
|
| 330 |
+
"reward": 0.45352864265441895,
|
| 331 |
+
"reward_std": 0.5192392915487289,
|
| 332 |
+
"rewards/correct_code_reward_func": 0.2291666679084301,
|
| 333 |
+
"rewards/len_reward_func": 0.22436197102069855,
|
| 334 |
+
"step": 25
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"completion_length": 172.56250762939453,
|
| 338 |
+
"epoch": 0.46017699115044247,
|
| 339 |
+
"grad_norm": 1.5744378292792975,
|
| 340 |
+
"kl": 0.0035400390625,
|
| 341 |
+
"learning_rate": 4.816553142054805e-07,
|
| 342 |
+
"loss": 0.0,
|
| 343 |
+
"reward": 0.4348207265138626,
|
| 344 |
+
"reward_std": 0.40823744237422943,
|
| 345 |
+
"rewards/correct_code_reward_func": 0.14583333395421505,
|
| 346 |
+
"rewards/len_reward_func": 0.2889874130487442,
|
| 347 |
+
"step": 26
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"completion_length": 163.5416717529297,
|
| 351 |
+
"epoch": 0.4778761061946903,
|
| 352 |
+
"grad_norm": 37.42131429307232,
|
| 353 |
+
"kl": 0.0279998779296875,
|
| 354 |
+
"learning_rate": 4.798007040388211e-07,
|
| 355 |
+
"loss": 0.0,
|
| 356 |
+
"reward": 0.44616006314754486,
|
| 357 |
+
"reward_std": 0.41116151213645935,
|
| 358 |
+
"rewards/correct_code_reward_func": 0.18750000558793545,
|
| 359 |
+
"rewards/len_reward_func": 0.25866006314754486,
|
| 360 |
+
"step": 27
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"completion_length": 147.1458396911621,
|
| 364 |
+
"epoch": 0.49557522123893805,
|
| 365 |
+
"grad_norm": 3.226533563373661,
|
| 366 |
+
"kl": 0.013214111328125,
|
| 367 |
+
"learning_rate": 4.778607322916895e-07,
|
| 368 |
+
"loss": 0.0,
|
| 369 |
+
"reward": 0.5665057748556137,
|
| 370 |
+
"reward_std": 0.3880573436617851,
|
| 371 |
+
"rewards/correct_code_reward_func": 0.2708333432674408,
|
| 372 |
+
"rewards/len_reward_func": 0.29567238688468933,
|
| 373 |
+
"step": 28
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"completion_length": 127.56250381469727,
|
| 377 |
+
"epoch": 0.5132743362831859,
|
| 378 |
+
"grad_norm": 4.362166865135135,
|
| 379 |
+
"kl": 0.01708984375,
|
| 380 |
+
"learning_rate": 4.758361195843951e-07,
|
| 381 |
+
"loss": 0.0,
|
| 382 |
+
"reward": 0.8003825545310974,
|
| 383 |
+
"reward_std": 0.510110080242157,
|
| 384 |
+
"rewards/correct_code_reward_func": 0.4583333432674408,
|
| 385 |
+
"rewards/len_reward_func": 0.34204918146133423,
|
| 386 |
+
"step": 29
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"completion_length": 173.37500381469727,
|
| 390 |
+
"epoch": 0.5309734513274337,
|
| 391 |
+
"grad_norm": 0.8500372624160994,
|
| 392 |
+
"kl": 0.00797271728515625,
|
| 393 |
+
"learning_rate": 4.7372761797790825e-07,
|
| 394 |
+
"loss": 0.0,
|
| 395 |
+
"reward": 0.37487032264471054,
|
| 396 |
+
"reward_std": 0.32572413980960846,
|
| 397 |
+
"rewards/correct_code_reward_func": 0.1458333432674408,
|
| 398 |
+
"rewards/len_reward_func": 0.22903696447610855,
|
| 399 |
+
"step": 30
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"completion_length": 119.14583587646484,
|
| 403 |
+
"epoch": 0.5486725663716814,
|
| 404 |
+
"grad_norm": 2.2352274664870624,
|
| 405 |
+
"kl": 0.014129638671875,
|
| 406 |
+
"learning_rate": 4.7153601069450146e-07,
|
| 407 |
+
"loss": 0.0,
|
| 408 |
+
"reward": 0.5202069282531738,
|
| 409 |
+
"reward_std": 0.5186183750629425,
|
| 410 |
+
"rewards/correct_code_reward_func": 0.291666679084301,
|
| 411 |
+
"rewards/len_reward_func": 0.22854027152061462,
|
| 412 |
+
"step": 31
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"completion_length": 111.31250381469727,
|
| 416 |
+
"epoch": 0.5663716814159292,
|
| 417 |
+
"grad_norm": 1.1283808844332623,
|
| 418 |
+
"kl": 0.015625,
|
| 419 |
+
"learning_rate": 4.6926211182681294e-07,
|
| 420 |
+
"loss": 0.0,
|
| 421 |
+
"reward": 0.4815867692232132,
|
| 422 |
+
"reward_std": 0.294177770614624,
|
| 423 |
+
"rewards/correct_code_reward_func": 0.1458333432674408,
|
| 424 |
+
"rewards/len_reward_func": 0.3357534259557724,
|
| 425 |
+
"step": 32
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"completion_length": 111.87500381469727,
|
| 429 |
+
"epoch": 0.584070796460177,
|
| 430 |
+
"grad_norm": 15.382447167413178,
|
| 431 |
+
"kl": 0.0740966796875,
|
| 432 |
+
"learning_rate": 4.6690676603544553e-07,
|
| 433 |
+
"loss": 0.0001,
|
| 434 |
+
"reward": 0.30563782900571823,
|
| 435 |
+
"reward_std": 0.5501309335231781,
|
| 436 |
+
"rewards/correct_code_reward_func": 0.1458333395421505,
|
| 437 |
+
"rewards/len_reward_func": 0.15980449318885803,
|
| 438 |
+
"step": 33
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"completion_length": 74.31250190734863,
|
| 442 |
+
"epoch": 0.6017699115044248,
|
| 443 |
+
"grad_norm": 6.16835237121908,
|
| 444 |
+
"kl": 0.038818359375,
|
| 445 |
+
"learning_rate": 4.644708482352092e-07,
|
| 446 |
+
"loss": 0.0,
|
| 447 |
+
"reward": 0.7065277397632599,
|
| 448 |
+
"reward_std": 0.4479583501815796,
|
| 449 |
+
"rewards/correct_code_reward_func": 0.3958333358168602,
|
| 450 |
+
"rewards/len_reward_func": 0.3106943815946579,
|
| 451 |
+
"step": 34
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"completion_length": 96.45833587646484,
|
| 455 |
+
"epoch": 0.6194690265486725,
|
| 456 |
+
"grad_norm": 6.369953012669968,
|
| 457 |
+
"kl": 0.035186767578125,
|
| 458 |
+
"learning_rate": 4.6195526327012623e-07,
|
| 459 |
+
"loss": 0.0,
|
| 460 |
+
"reward": 0.3565668910741806,
|
| 461 |
+
"reward_std": 0.39397968351840973,
|
| 462 |
+
"rewards/correct_code_reward_func": 0.08333333395421505,
|
| 463 |
+
"rewards/len_reward_func": 0.273233562707901,
|
| 464 |
+
"step": 35
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"completion_length": 112.14583587646484,
|
| 468 |
+
"epoch": 0.6371681415929203,
|
| 469 |
+
"grad_norm": 3.3487214552142164,
|
| 470 |
+
"kl": 0.0289306640625,
|
| 471 |
+
"learning_rate": 4.5936094557731806e-07,
|
| 472 |
+
"loss": 0.0,
|
| 473 |
+
"reward": 0.7201341390609741,
|
| 474 |
+
"reward_std": 0.3427349328994751,
|
| 475 |
+
"rewards/correct_code_reward_func": 0.3541666865348816,
|
| 476 |
+
"rewards/len_reward_func": 0.3659674823284149,
|
| 477 |
+
"step": 36
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"completion_length": 101.64583587646484,
|
| 481 |
+
"epoch": 0.6548672566371682,
|
| 482 |
+
"grad_norm": 5.120880381398817,
|
| 483 |
+
"kl": 0.05126953125,
|
| 484 |
+
"learning_rate": 4.566888588399006e-07,
|
| 485 |
+
"loss": 0.0001,
|
| 486 |
+
"reward": 0.5204495191574097,
|
| 487 |
+
"reward_std": 0.3199263662099838,
|
| 488 |
+
"rewards/correct_code_reward_func": 0.1250000037252903,
|
| 489 |
+
"rewards/len_reward_func": 0.39544953405857086,
|
| 490 |
+
"step": 37
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"completion_length": 120.5625,
|
| 494 |
+
"epoch": 0.672566371681416,
|
| 495 |
+
"grad_norm": 10.359850576348157,
|
| 496 |
+
"kl": 0.09923553466796875,
|
| 497 |
+
"learning_rate": 4.5393999562901517e-07,
|
| 498 |
+
"loss": 0.0001,
|
| 499 |
+
"reward": 0.6964526474475861,
|
| 500 |
+
"reward_std": 0.5385148227214813,
|
| 501 |
+
"rewards/correct_code_reward_func": 0.4166666865348816,
|
| 502 |
+
"rewards/len_reward_func": 0.27978597581386566,
|
| 503 |
+
"step": 38
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"completion_length": 110.85417175292969,
|
| 507 |
+
"epoch": 0.6902654867256637,
|
| 508 |
+
"grad_norm": 12.273703285519217,
|
| 509 |
+
"kl": 0.0859375,
|
| 510 |
+
"learning_rate": 4.511153770351288e-07,
|
| 511 |
+
"loss": 0.0001,
|
| 512 |
+
"reward": 0.551011860370636,
|
| 513 |
+
"reward_std": 0.36342084407806396,
|
| 514 |
+
"rewards/correct_code_reward_func": 0.1875000111758709,
|
| 515 |
+
"rewards/len_reward_func": 0.3635118901729584,
|
| 516 |
+
"step": 39
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"completion_length": 68.77083587646484,
|
| 520 |
+
"epoch": 0.7079646017699115,
|
| 521 |
+
"grad_norm": 2.2127761337332514,
|
| 522 |
+
"kl": 0.03765869140625,
|
| 523 |
+
"learning_rate": 4.482160522887403e-07,
|
| 524 |
+
"loss": 0.0,
|
| 525 |
+
"reward": 0.7378278374671936,
|
| 526 |
+
"reward_std": 0.39128121733665466,
|
| 527 |
+
"rewards/correct_code_reward_func": 0.4375000149011612,
|
| 528 |
+
"rewards/len_reward_func": 0.3003278076648712,
|
| 529 |
+
"step": 40
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"completion_length": 135.33333587646484,
|
| 533 |
+
"epoch": 0.7256637168141593,
|
| 534 |
+
"grad_norm": 7.600755671861986,
|
| 535 |
+
"kl": 0.1025390625,
|
| 536 |
+
"learning_rate": 4.4524309837063504e-07,
|
| 537 |
+
"loss": 0.0001,
|
| 538 |
+
"reward": 0.6550733149051666,
|
| 539 |
+
"reward_std": 0.5409570336341858,
|
| 540 |
+
"rewards/correct_code_reward_func": 0.3333333432674408,
|
| 541 |
+
"rewards/len_reward_func": 0.32173997163772583,
|
| 542 |
+
"step": 41
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"completion_length": 87.35416793823242,
|
| 546 |
+
"epoch": 0.7433628318584071,
|
| 547 |
+
"grad_norm": 1.6503719126782672,
|
| 548 |
+
"kl": 0.0264892578125,
|
| 549 |
+
"learning_rate": 4.4219761961182965e-07,
|
| 550 |
+
"loss": 0.0,
|
| 551 |
+
"reward": 0.6292450726032257,
|
| 552 |
+
"reward_std": 0.4537566006183624,
|
| 553 |
+
"rewards/correct_code_reward_func": 0.2708333432674408,
|
| 554 |
+
"rewards/len_reward_func": 0.3584117591381073,
|
| 555 |
+
"step": 42
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"completion_length": 65.29166793823242,
|
| 559 |
+
"epoch": 0.7610619469026548,
|
| 560 |
+
"grad_norm": 4.327974078868625,
|
| 561 |
+
"kl": 0.1019287109375,
|
| 562 |
+
"learning_rate": 4.390807472833584e-07,
|
| 563 |
+
"loss": 0.0001,
|
| 564 |
+
"reward": 0.5142460465431213,
|
| 565 |
+
"reward_std": 0.41541849076747894,
|
| 566 |
+
"rewards/correct_code_reward_func": 0.1875000074505806,
|
| 567 |
+
"rewards/len_reward_func": 0.32674603164196014,
|
| 568 |
+
"step": 43
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"completion_length": 88.62500381469727,
|
| 572 |
+
"epoch": 0.7787610619469026,
|
| 573 |
+
"grad_norm": 2.8856078275542814,
|
| 574 |
+
"kl": 0.068115234375,
|
| 575 |
+
"learning_rate": 4.3589363917605234e-07,
|
| 576 |
+
"loss": 0.0001,
|
| 577 |
+
"reward": 0.48976023495197296,
|
| 578 |
+
"reward_std": 0.39692020416259766,
|
| 579 |
+
"rewards/correct_code_reward_func": 0.1875,
|
| 580 |
+
"rewards/len_reward_func": 0.30226024240255356,
|
| 581 |
+
"step": 44
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"completion_length": 78.75000381469727,
|
| 585 |
+
"epoch": 0.7964601769911505,
|
| 586 |
+
"grad_norm": 1.6857493365819893,
|
| 587 |
+
"kl": 0.04052734375,
|
| 588 |
+
"learning_rate": 4.3263747917046697e-07,
|
| 589 |
+
"loss": 0.0,
|
| 590 |
+
"reward": 0.5636806637048721,
|
| 591 |
+
"reward_std": 0.5081076323986053,
|
| 592 |
+
"rewards/correct_code_reward_func": 0.2500000111758709,
|
| 593 |
+
"rewards/len_reward_func": 0.3136806935071945,
|
| 594 |
+
"step": 45
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"completion_length": 78.50000190734863,
|
| 598 |
+
"epoch": 0.8141592920353983,
|
| 599 |
+
"grad_norm": 1.2755984783224172,
|
| 600 |
+
"kl": 0.0218505859375,
|
| 601 |
+
"learning_rate": 4.2931347679711924e-07,
|
| 602 |
+
"loss": 0.0,
|
| 603 |
+
"reward": 0.6265330016613007,
|
| 604 |
+
"reward_std": 0.38107292354106903,
|
| 605 |
+
"rewards/correct_code_reward_func": 0.2916666716337204,
|
| 606 |
+
"rewards/len_reward_func": 0.33486635982990265,
|
| 607 |
+
"step": 46
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"completion_length": 80.29166793823242,
|
| 611 |
+
"epoch": 0.831858407079646,
|
| 612 |
+
"grad_norm": 1.1240225012044924,
|
| 613 |
+
"kl": 0.0335693359375,
|
| 614 |
+
"learning_rate": 4.259228667871962e-07,
|
| 615 |
+
"loss": 0.0,
|
| 616 |
+
"reward": 0.5936128199100494,
|
| 617 |
+
"reward_std": 0.4675884544849396,
|
| 618 |
+
"rewards/correct_code_reward_func": 0.25,
|
| 619 |
+
"rewards/len_reward_func": 0.34361281991004944,
|
| 620 |
+
"step": 47
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"completion_length": 50.0,
|
| 624 |
+
"epoch": 0.8495575221238938,
|
| 625 |
+
"grad_norm": 8.693611957424073,
|
| 626 |
+
"kl": 0.2470703125,
|
| 627 |
+
"learning_rate": 4.224669086139029e-07,
|
| 628 |
+
"loss": 0.0002,
|
| 629 |
+
"reward": 0.49966277182102203,
|
| 630 |
+
"reward_std": 0.3804493248462677,
|
| 631 |
+
"rewards/correct_code_reward_func": 0.08333333395421505,
|
| 632 |
+
"rewards/len_reward_func": 0.41632944345474243,
|
| 633 |
+
"step": 48
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"completion_length": 92.20833587646484,
|
| 637 |
+
"epoch": 0.8672566371681416,
|
| 638 |
+
"grad_norm": 1.2744927278879081,
|
| 639 |
+
"kl": 0.02978515625,
|
| 640 |
+
"learning_rate": 4.1894688602461913e-07,
|
| 641 |
+
"loss": 0.0,
|
| 642 |
+
"reward": 0.7709261626005173,
|
| 643 |
+
"reward_std": 0.3614697605371475,
|
| 644 |
+
"rewards/correct_code_reward_func": 0.3958333544433117,
|
| 645 |
+
"rewards/len_reward_func": 0.3750927746295929,
|
| 646 |
+
"step": 49
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"completion_length": 127.62500381469727,
|
| 650 |
+
"epoch": 0.8849557522123894,
|
| 651 |
+
"grad_norm": 0.566625300559752,
|
| 652 |
+
"kl": 0.016326904296875,
|
| 653 |
+
"learning_rate": 4.153641065640402e-07,
|
| 654 |
+
"loss": 0.0,
|
| 655 |
+
"reward": 0.5525736510753632,
|
| 656 |
+
"reward_std": 0.42078813910484314,
|
| 657 |
+
"rewards/correct_code_reward_func": 0.20833333395421505,
|
| 658 |
+
"rewards/len_reward_func": 0.34424029290676117,
|
| 659 |
+
"step": 50
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"completion_length": 65.375,
|
| 663 |
+
"epoch": 0.9026548672566371,
|
| 664 |
+
"grad_norm": 12.046039464547933,
|
| 665 |
+
"kl": 0.081298828125,
|
| 666 |
+
"learning_rate": 4.11719901088477e-07,
|
| 667 |
+
"loss": 0.0001,
|
| 668 |
+
"reward": 0.6643719673156738,
|
| 669 |
+
"reward_std": 0.4581465870141983,
|
| 670 |
+
"rewards/correct_code_reward_func": 0.3333333432674408,
|
| 671 |
+
"rewards/len_reward_func": 0.3310386538505554,
|
| 672 |
+
"step": 51
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"completion_length": 103.58333587646484,
|
| 676 |
+
"epoch": 0.9203539823008849,
|
| 677 |
+
"grad_norm": 10.540579232965062,
|
| 678 |
+
"kl": 0.0799560546875,
|
| 679 |
+
"learning_rate": 4.080156232714976e-07,
|
| 680 |
+
"loss": 0.0001,
|
| 681 |
+
"reward": 0.6433776319026947,
|
| 682 |
+
"reward_std": 0.3912515640258789,
|
| 683 |
+
"rewards/correct_code_reward_func": 0.29166667722165585,
|
| 684 |
+
"rewards/len_reward_func": 0.3517109453678131,
|
| 685 |
+
"step": 52
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"completion_length": 62.375003814697266,
|
| 689 |
+
"epoch": 0.9380530973451328,
|
| 690 |
+
"grad_norm": 2.819698340947531,
|
| 691 |
+
"kl": 0.115966796875,
|
| 692 |
+
"learning_rate": 4.0425264910109233e-07,
|
| 693 |
+
"loss": 0.0001,
|
| 694 |
+
"reward": 0.7070481181144714,
|
| 695 |
+
"reward_std": 0.44986239075660706,
|
| 696 |
+
"rewards/correct_code_reward_func": 0.3125,
|
| 697 |
+
"rewards/len_reward_func": 0.39454813301563263,
|
| 698 |
+
"step": 53
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"completion_length": 59.0625,
|
| 702 |
+
"epoch": 0.9557522123893806,
|
| 703 |
+
"grad_norm": 3.22095354432732,
|
| 704 |
+
"kl": 0.1131591796875,
|
| 705 |
+
"learning_rate": 4.004323763685511e-07,
|
| 706 |
+
"loss": 0.0001,
|
| 707 |
+
"reward": 0.5839295089244843,
|
| 708 |
+
"reward_std": 0.26084955781698227,
|
| 709 |
+
"rewards/correct_code_reward_func": 0.14583333395421505,
|
| 710 |
+
"rewards/len_reward_func": 0.43809613585472107,
|
| 711 |
+
"step": 54
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"completion_length": 125.58333969116211,
|
| 715 |
+
"epoch": 0.9734513274336283,
|
| 716 |
+
"grad_norm": 8.551645640447115,
|
| 717 |
+
"kl": 0.15509033203125,
|
| 718 |
+
"learning_rate": 3.9655622414924007e-07,
|
| 719 |
+
"loss": 0.0002,
|
| 720 |
+
"reward": 0.528711274266243,
|
| 721 |
+
"reward_std": 0.3454602509737015,
|
| 722 |
+
"rewards/correct_code_reward_func": 0.1666666716337204,
|
| 723 |
+
"rewards/len_reward_func": 0.3620445877313614,
|
| 724 |
+
"step": 55
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"completion_length": 60.000003814697266,
|
| 728 |
+
"epoch": 0.9911504424778761,
|
| 729 |
+
"grad_norm": 38.727062970707685,
|
| 730 |
+
"kl": 0.20166015625,
|
| 731 |
+
"learning_rate": 3.92625632275474e-07,
|
| 732 |
+
"loss": 0.0002,
|
| 733 |
+
"reward": 0.891117125749588,
|
| 734 |
+
"reward_std": 0.36153580248355865,
|
| 735 |
+
"rewards/correct_code_reward_func": 0.5000000149011612,
|
| 736 |
+
"rewards/len_reward_func": 0.39111708104610443,
|
| 737 |
+
"step": 56
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"completion_length": 91.54167175292969,
|
| 741 |
+
"epoch": 1.0,
|
| 742 |
+
"grad_norm": 38.727062970707685,
|
| 743 |
+
"kl": 0.1865234375,
|
| 744 |
+
"learning_rate": 3.886420608016766e-07,
|
| 745 |
+
"loss": 0.0001,
|
| 746 |
+
"reward": 0.5094150304794312,
|
| 747 |
+
"reward_std": 0.4194638133049011,
|
| 748 |
+
"rewards/correct_code_reward_func": 0.1666666716337204,
|
| 749 |
+
"rewards/len_reward_func": 0.3427482843399048,
|
| 750 |
+
"step": 57
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"completion_length": 96.41666793823242,
|
| 754 |
+
"epoch": 1.0176991150442478,
|
| 755 |
+
"grad_norm": 13.425409986987928,
|
| 756 |
+
"kl": 0.04998779296875,
|
| 757 |
+
"learning_rate": 3.846069894620305e-07,
|
| 758 |
+
"loss": 0.0001,
|
| 759 |
+
"reward": 0.5195242762565613,
|
| 760 |
+
"reward_std": 0.3198210597038269,
|
| 761 |
+
"rewards/correct_code_reward_func": 0.125,
|
| 762 |
+
"rewards/len_reward_func": 0.3945242464542389,
|
| 763 |
+
"step": 58
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"completion_length": 79.3125,
|
| 767 |
+
"epoch": 1.0353982300884956,
|
| 768 |
+
"grad_norm": 1.7429704138905477,
|
| 769 |
+
"kl": 0.0509033203125,
|
| 770 |
+
"learning_rate": 3.8052191712081595e-07,
|
| 771 |
+
"loss": 0.0001,
|
| 772 |
+
"reward": 0.655539482831955,
|
| 773 |
+
"reward_std": 0.2987876646220684,
|
| 774 |
+
"rewards/correct_code_reward_func": 0.1875,
|
| 775 |
+
"rewards/len_reward_func": 0.46803948283195496,
|
| 776 |
+
"step": 59
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"completion_length": 61.187503814697266,
|
| 780 |
+
"epoch": 1.0530973451327434,
|
| 781 |
+
"grad_norm": 7.538322601947877,
|
| 782 |
+
"kl": 0.1343994140625,
|
| 783 |
+
"learning_rate": 3.763883612156441e-07,
|
| 784 |
+
"loss": 0.0001,
|
| 785 |
+
"reward": 0.7049243450164795,
|
| 786 |
+
"reward_std": 0.5336888134479523,
|
| 787 |
+
"rewards/correct_code_reward_func": 0.3125000149011612,
|
| 788 |
+
"rewards/len_reward_func": 0.3924243599176407,
|
| 789 |
+
"step": 60
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"completion_length": 71.45833587646484,
|
| 793 |
+
"epoch": 1.0707964601769913,
|
| 794 |
+
"grad_norm": 1.628031440232102,
|
| 795 |
+
"kl": 0.053955078125,
|
| 796 |
+
"learning_rate": 3.7220785719379095e-07,
|
| 797 |
+
"loss": 0.0001,
|
| 798 |
+
"reward": 0.5856566280126572,
|
| 799 |
+
"reward_std": 0.31656239926815033,
|
| 800 |
+
"rewards/correct_code_reward_func": 0.2083333395421505,
|
| 801 |
+
"rewards/len_reward_func": 0.37732329964637756,
|
| 802 |
+
"step": 61
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"completion_length": 80.64583587646484,
|
| 806 |
+
"epoch": 1.0884955752212389,
|
| 807 |
+
"grad_norm": 0.8807120675882707,
|
| 808 |
+
"kl": 0.05035400390625,
|
| 809 |
+
"learning_rate": 3.6798195794184134e-07,
|
| 810 |
+
"loss": 0.0001,
|
| 811 |
+
"reward": 0.5311767309904099,
|
| 812 |
+
"reward_std": 0.2926383763551712,
|
| 813 |
+
"rewards/correct_code_reward_func": 0.1041666716337204,
|
| 814 |
+
"rewards/len_reward_func": 0.42701007425785065,
|
| 815 |
+
"step": 62
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"completion_length": 112.29167175292969,
|
| 819 |
+
"epoch": 1.1061946902654867,
|
| 820 |
+
"grad_norm": 1.0053639016562932,
|
| 821 |
+
"kl": 0.07916259765625,
|
| 822 |
+
"learning_rate": 3.6371223320885484e-07,
|
| 823 |
+
"loss": 0.0001,
|
| 824 |
+
"reward": 0.4587331861257553,
|
| 825 |
+
"reward_std": 0.33583496510982513,
|
| 826 |
+
"rewards/correct_code_reward_func": 0.08333333395421505,
|
| 827 |
+
"rewards/len_reward_func": 0.3753998577594757,
|
| 828 |
+
"step": 63
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"completion_length": 83.20833587646484,
|
| 832 |
+
"epoch": 1.1238938053097345,
|
| 833 |
+
"grad_norm": 1.7713543449970612,
|
| 834 |
+
"kl": 0.08245849609375,
|
| 835 |
+
"learning_rate": 3.594002690232682e-07,
|
| 836 |
+
"loss": 0.0001,
|
| 837 |
+
"reward": 0.6741357147693634,
|
| 838 |
+
"reward_std": 0.43141913414001465,
|
| 839 |
+
"rewards/correct_code_reward_func": 0.2708333432674408,
|
| 840 |
+
"rewards/len_reward_func": 0.4033023715019226,
|
| 841 |
+
"step": 64
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"completion_length": 84.10416984558105,
|
| 845 |
+
"epoch": 1.1415929203539823,
|
| 846 |
+
"grad_norm": 25.050556888869668,
|
| 847 |
+
"kl": 0.12774658203125,
|
| 848 |
+
"learning_rate": 3.5504766710375047e-07,
|
| 849 |
+
"loss": 0.0001,
|
| 850 |
+
"reward": 0.6237881779670715,
|
| 851 |
+
"reward_std": 0.28890369832515717,
|
| 852 |
+
"rewards/correct_code_reward_func": 0.2083333432674408,
|
| 853 |
+
"rewards/len_reward_func": 0.41545483469963074,
|
| 854 |
+
"step": 65
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"completion_length": 82.27083587646484,
|
| 858 |
+
"epoch": 1.1592920353982301,
|
| 859 |
+
"grad_norm": 1.3229069926746502,
|
| 860 |
+
"kl": 0.0810546875,
|
| 861 |
+
"learning_rate": 3.506560442642299e-07,
|
| 862 |
+
"loss": 0.0001,
|
| 863 |
+
"reward": 0.7232638597488403,
|
| 864 |
+
"reward_std": 0.37123511731624603,
|
| 865 |
+
"rewards/correct_code_reward_func": 0.3541666716337204,
|
| 866 |
+
"rewards/len_reward_func": 0.36909720301628113,
|
| 867 |
+
"step": 66
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"completion_length": 58.00000190734863,
|
| 871 |
+
"epoch": 1.176991150442478,
|
| 872 |
+
"grad_norm": 14.439064098343355,
|
| 873 |
+
"kl": 0.1383056640625,
|
| 874 |
+
"learning_rate": 3.462270318133136e-07,
|
| 875 |
+
"loss": 0.0001,
|
| 876 |
+
"reward": 0.9435772597789764,
|
| 877 |
+
"reward_std": 0.2863286882638931,
|
| 878 |
+
"rewards/correct_code_reward_func": 0.520833358168602,
|
| 879 |
+
"rewards/len_reward_func": 0.42274393141269684,
|
| 880 |
+
"step": 67
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"completion_length": 71.10416984558105,
|
| 884 |
+
"epoch": 1.1946902654867257,
|
| 885 |
+
"grad_norm": 7.987124238678158,
|
| 886 |
+
"kl": 0.122314453125,
|
| 887 |
+
"learning_rate": 3.41762274948323e-07,
|
| 888 |
+
"loss": 0.0001,
|
| 889 |
+
"reward": 0.7681821584701538,
|
| 890 |
+
"reward_std": 0.3728492707014084,
|
| 891 |
+
"rewards/correct_code_reward_func": 0.3541666865348816,
|
| 892 |
+
"rewards/len_reward_func": 0.41401544213294983,
|
| 893 |
+
"step": 68
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"completion_length": 60.66666793823242,
|
| 897 |
+
"epoch": 1.2123893805309733,
|
| 898 |
+
"grad_norm": 2.9983488159761817,
|
| 899 |
+
"kl": 0.0537109375,
|
| 900 |
+
"learning_rate": 3.372634321441702e-07,
|
| 901 |
+
"loss": 0.0001,
|
| 902 |
+
"reward": 0.7418330609798431,
|
| 903 |
+
"reward_std": 0.38279734551906586,
|
| 904 |
+
"rewards/correct_code_reward_func": 0.3541666865348816,
|
| 905 |
+
"rewards/len_reward_func": 0.38766637444496155,
|
| 906 |
+
"step": 69
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"completion_length": 80.62500381469727,
|
| 910 |
+
"epoch": 1.2300884955752212,
|
| 911 |
+
"grad_norm": 3.023771746536167,
|
| 912 |
+
"kl": 0.08154296875,
|
| 913 |
+
"learning_rate": 3.3273217453730205e-07,
|
| 914 |
+
"loss": 0.0001,
|
| 915 |
+
"reward": 0.6804832518100739,
|
| 916 |
+
"reward_std": 0.4174923449754715,
|
| 917 |
+
"rewards/correct_code_reward_func": 0.2708333432674408,
|
| 918 |
+
"rewards/len_reward_func": 0.40964992344379425,
|
| 919 |
+
"step": 70
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"completion_length": 69.5,
|
| 923 |
+
"epoch": 1.247787610619469,
|
| 924 |
+
"grad_norm": 1.4050656113052118,
|
| 925 |
+
"kl": 0.0853271484375,
|
| 926 |
+
"learning_rate": 3.281701853049416e-07,
|
| 927 |
+
"loss": 0.0001,
|
| 928 |
+
"reward": 0.7291666865348816,
|
| 929 |
+
"reward_std": 0.32618680596351624,
|
| 930 |
+
"rewards/correct_code_reward_func": 0.2708333432674408,
|
| 931 |
+
"rewards/len_reward_func": 0.4583333432674408,
|
| 932 |
+
"step": 71
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"completion_length": 74.47916793823242,
|
| 936 |
+
"epoch": 1.2654867256637168,
|
| 937 |
+
"grad_norm": 2.8179580865899894,
|
| 938 |
+
"kl": 0.0390625,
|
| 939 |
+
"learning_rate": 3.23579159039856e-07,
|
| 940 |
+
"loss": 0.0,
|
| 941 |
+
"reward": 0.75091952085495,
|
| 942 |
+
"reward_std": 0.4079275578260422,
|
| 943 |
+
"rewards/correct_code_reward_func": 0.2916666716337204,
|
| 944 |
+
"rewards/len_reward_func": 0.45925286412239075,
|
| 945 |
+
"step": 72
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"completion_length": 65.52083587646484,
|
| 949 |
+
"epoch": 1.2831858407079646,
|
| 950 |
+
"grad_norm": 1.1299435463555585,
|
| 951 |
+
"kl": 0.0399169921875,
|
| 952 |
+
"learning_rate": 3.1896080112088477e-07,
|
| 953 |
+
"loss": 0.0,
|
| 954 |
+
"reward": 0.7291666865348816,
|
| 955 |
+
"reward_std": 0.40629878640174866,
|
| 956 |
+
"rewards/correct_code_reward_func": 0.2708333395421505,
|
| 957 |
+
"rewards/len_reward_func": 0.4583333432674408,
|
| 958 |
+
"step": 73
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"completion_length": 57.000003814697266,
|
| 962 |
+
"epoch": 1.3008849557522124,
|
| 963 |
+
"grad_norm": 24.705642459397488,
|
| 964 |
+
"kl": 0.17333984375,
|
| 965 |
+
"learning_rate": 3.1431682707946117e-07,
|
| 966 |
+
"loss": 0.0002,
|
| 967 |
+
"reward": 0.5623736083507538,
|
| 968 |
+
"reward_std": 0.24504420161247253,
|
| 969 |
+
"rewards/correct_code_reward_func": 0.12500000558793545,
|
| 970 |
+
"rewards/len_reward_func": 0.4373735785484314,
|
| 971 |
+
"step": 74
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"completion_length": 78.33333587646484,
|
| 975 |
+
"epoch": 1.3185840707964602,
|
| 976 |
+
"grad_norm": 0.7830442669221305,
|
| 977 |
+
"kl": 0.0224609375,
|
| 978 |
+
"learning_rate": 3.096489619623621e-07,
|
| 979 |
+
"loss": 0.0,
|
| 980 |
+
"reward": 0.7488380074501038,
|
| 981 |
+
"reward_std": 0.30931201577186584,
|
| 982 |
+
"rewards/correct_code_reward_func": 0.2500000149011612,
|
| 983 |
+
"rewards/len_reward_func": 0.49883799254894257,
|
| 984 |
+
"step": 75
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"completion_length": 87.60417175292969,
|
| 988 |
+
"epoch": 1.336283185840708,
|
| 989 |
+
"grad_norm": 2.1746791001290044,
|
| 990 |
+
"kl": 0.04168701171875,
|
| 991 |
+
"learning_rate": 3.049589396909239e-07,
|
| 992 |
+
"loss": 0.0,
|
| 993 |
+
"reward": 0.5558712333440781,
|
| 994 |
+
"reward_std": 0.20058585330843925,
|
| 995 |
+
"rewards/correct_code_reward_func": 0.1041666716337204,
|
| 996 |
+
"rewards/len_reward_func": 0.45170454680919647,
|
| 997 |
+
"step": 76
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"completion_length": 72.14583587646484,
|
| 1001 |
+
"epoch": 1.3539823008849559,
|
| 1002 |
+
"grad_norm": 0.41651138876768207,
|
| 1003 |
+
"kl": 0.0919189453125,
|
| 1004 |
+
"learning_rate": 3.0024850241696126e-07,
|
| 1005 |
+
"loss": 0.0001,
|
| 1006 |
+
"reward": 0.6021915674209595,
|
| 1007 |
+
"reward_std": 0.22243696451187134,
|
| 1008 |
+
"rewards/correct_code_reward_func": 0.1458333358168602,
|
| 1009 |
+
"rewards/len_reward_func": 0.4563582092523575,
|
| 1010 |
+
"step": 77
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"completion_length": 54.916666984558105,
|
| 1014 |
+
"epoch": 1.3716814159292037,
|
| 1015 |
+
"grad_norm": 8.88612982008444,
|
| 1016 |
+
"kl": 0.106689453125,
|
| 1017 |
+
"learning_rate": 2.9551939987562865e-07,
|
| 1018 |
+
"loss": 0.0001,
|
| 1019 |
+
"reward": 0.6803071200847626,
|
| 1020 |
+
"reward_std": 0.2964301370084286,
|
| 1021 |
+
"rewards/correct_code_reward_func": 0.2291666679084301,
|
| 1022 |
+
"rewards/len_reward_func": 0.4511404484510422,
|
| 1023 |
+
"step": 78
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"completion_length": 93.91666793823242,
|
| 1027 |
+
"epoch": 1.3893805309734513,
|
| 1028 |
+
"grad_norm": 3.7969039322604865,
|
| 1029 |
+
"kl": 0.086761474609375,
|
| 1030 |
+
"learning_rate": 2.9077338873546564e-07,
|
| 1031 |
+
"loss": 0.0001,
|
| 1032 |
+
"reward": 0.7916666865348816,
|
| 1033 |
+
"reward_std": 0.34503278136253357,
|
| 1034 |
+
"rewards/correct_code_reward_func": 0.2916666716337204,
|
| 1035 |
+
"rewards/len_reward_func": 0.5,
|
| 1036 |
+
"step": 79
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"completion_length": 56.08333396911621,
|
| 1040 |
+
"epoch": 1.407079646017699,
|
| 1041 |
+
"grad_norm": 3.0759876398616743,
|
| 1042 |
+
"kl": 0.076171875,
|
| 1043 |
+
"learning_rate": 2.860122319458661e-07,
|
| 1044 |
+
"loss": 0.0001,
|
| 1045 |
+
"reward": 0.8750000596046448,
|
| 1046 |
+
"reward_std": 0.36751919984817505,
|
| 1047 |
+
"rewards/correct_code_reward_func": 0.3958333432674408,
|
| 1048 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1049 |
+
"step": 80
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"completion_length": 77.58333587646484,
|
| 1053 |
+
"epoch": 1.424778761061947,
|
| 1054 |
+
"grad_norm": 5.267849973823214,
|
| 1055 |
+
"kl": 0.0604248046875,
|
| 1056 |
+
"learning_rate": 2.8123769808221403e-07,
|
| 1057 |
+
"loss": 0.0001,
|
| 1058 |
+
"reward": 0.9166666865348816,
|
| 1059 |
+
"reward_std": 0.4720035046339035,
|
| 1060 |
+
"rewards/correct_code_reward_func": 0.4375000298023224,
|
| 1061 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1062 |
+
"step": 81
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"completion_length": 86.18750190734863,
|
| 1066 |
+
"epoch": 1.4424778761061947,
|
| 1067 |
+
"grad_norm": 16.256704067320886,
|
| 1068 |
+
"kl": 0.05712890625,
|
| 1069 |
+
"learning_rate": 2.764515606889307e-07,
|
| 1070 |
+
"loss": 0.0001,
|
| 1071 |
+
"reward": 0.8077563941478729,
|
| 1072 |
+
"reward_std": 0.3568233400583267,
|
| 1073 |
+
"rewards/correct_code_reward_func": 0.3958333432674408,
|
| 1074 |
+
"rewards/len_reward_func": 0.41192303597927094,
|
| 1075 |
+
"step": 82
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"completion_length": 106.89583587646484,
|
| 1079 |
+
"epoch": 1.4601769911504425,
|
| 1080 |
+
"grad_norm": 2.2341312424056214,
|
| 1081 |
+
"kl": 0.021087646484375,
|
| 1082 |
+
"learning_rate": 2.716555976206747e-07,
|
| 1083 |
+
"loss": 0.0,
|
| 1084 |
+
"reward": 0.7582251131534576,
|
| 1085 |
+
"reward_std": 0.3638576865196228,
|
| 1086 |
+
"rewards/correct_code_reward_func": 0.2916666716337204,
|
| 1087 |
+
"rewards/len_reward_func": 0.46655844151973724,
|
| 1088 |
+
"step": 83
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"completion_length": 48.41666793823242,
|
| 1092 |
+
"epoch": 1.4778761061946903,
|
| 1093 |
+
"grad_norm": 1.3490761289725859,
|
| 1094 |
+
"kl": 0.05279541015625,
|
| 1095 |
+
"learning_rate": 2.66851590381942e-07,
|
| 1096 |
+
"loss": 0.0001,
|
| 1097 |
+
"reward": 0.7612736523151398,
|
| 1098 |
+
"reward_std": 0.23646536469459534,
|
| 1099 |
+
"rewards/correct_code_reward_func": 0.2708333358168602,
|
| 1100 |
+
"rewards/len_reward_func": 0.490440309047699,
|
| 1101 |
+
"step": 84
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"completion_length": 60.187503814697266,
|
| 1105 |
+
"epoch": 1.495575221238938,
|
| 1106 |
+
"grad_norm": 4.538439276716702,
|
| 1107 |
+
"kl": 0.0347900390625,
|
| 1108 |
+
"learning_rate": 2.620413234653093e-07,
|
| 1109 |
+
"loss": 0.0,
|
| 1110 |
+
"reward": 0.8333333432674408,
|
| 1111 |
+
"reward_std": 0.438393235206604,
|
| 1112 |
+
"rewards/correct_code_reward_func": 0.3541666716337204,
|
| 1113 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1114 |
+
"step": 85
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"completion_length": 138.50000762939453,
|
| 1118 |
+
"epoch": 1.5132743362831858,
|
| 1119 |
+
"grad_norm": 0.9112971760232713,
|
| 1120 |
+
"kl": 0.021392822265625,
|
| 1121 |
+
"learning_rate": 2.5722658368856814e-07,
|
| 1122 |
+
"loss": 0.0,
|
| 1123 |
+
"reward": 0.6596881151199341,
|
| 1124 |
+
"reward_std": 0.4098925292491913,
|
| 1125 |
+
"rewards/correct_code_reward_func": 0.1875000074505806,
|
| 1126 |
+
"rewards/len_reward_func": 0.4721881300210953,
|
| 1127 |
+
"step": 86
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"completion_length": 67.89583587646484,
|
| 1131 |
+
"epoch": 1.5309734513274336,
|
| 1132 |
+
"grad_norm": 1.5750222049346805,
|
| 1133 |
+
"kl": 0.033447265625,
|
| 1134 |
+
"learning_rate": 2.5240915953099515e-07,
|
| 1135 |
+
"loss": 0.0,
|
| 1136 |
+
"reward": 0.9335185885429382,
|
| 1137 |
+
"reward_std": 0.32536589354276657,
|
| 1138 |
+
"rewards/correct_code_reward_func": 0.4375000149011612,
|
| 1139 |
+
"rewards/len_reward_func": 0.49601852893829346,
|
| 1140 |
+
"step": 87
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"completion_length": 56.395835876464844,
|
| 1144 |
+
"epoch": 1.5486725663716814,
|
| 1145 |
+
"grad_norm": 0.8224337332016152,
|
| 1146 |
+
"kl": 0.0401611328125,
|
| 1147 |
+
"learning_rate": 2.4759084046900483e-07,
|
| 1148 |
+
"loss": 0.0,
|
| 1149 |
+
"reward": 0.7500000298023224,
|
| 1150 |
+
"reward_std": 0.2342708334326744,
|
| 1151 |
+
"rewards/correct_code_reward_func": 0.3125000149011612,
|
| 1152 |
+
"rewards/len_reward_func": 0.4375000149011612,
|
| 1153 |
+
"step": 88
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"completion_length": 55.250003814697266,
|
| 1157 |
+
"epoch": 1.5663716814159292,
|
| 1158 |
+
"grad_norm": 1.7919585934332407,
|
| 1159 |
+
"kl": 0.046142578125,
|
| 1160 |
+
"learning_rate": 2.427734163114319e-07,
|
| 1161 |
+
"loss": 0.0,
|
| 1162 |
+
"reward": 0.8125000298023224,
|
| 1163 |
+
"reward_std": 0.3177132308483124,
|
| 1164 |
+
"rewards/correct_code_reward_func": 0.3125000149011612,
|
| 1165 |
+
"rewards/len_reward_func": 0.5,
|
| 1166 |
+
"step": 89
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"completion_length": 78.27083587646484,
|
| 1170 |
+
"epoch": 1.584070796460177,
|
| 1171 |
+
"grad_norm": 1.6894819978109317,
|
| 1172 |
+
"kl": 0.0758056640625,
|
| 1173 |
+
"learning_rate": 2.3795867653469067e-07,
|
| 1174 |
+
"loss": 0.0001,
|
| 1175 |
+
"reward": 0.8326875865459442,
|
| 1176 |
+
"reward_std": 0.27240147814154625,
|
| 1177 |
+
"rewards/correct_code_reward_func": 0.35416667722165585,
|
| 1178 |
+
"rewards/len_reward_func": 0.47852087020874023,
|
| 1179 |
+
"step": 90
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"completion_length": 102.83333969116211,
|
| 1183 |
+
"epoch": 1.6017699115044248,
|
| 1184 |
+
"grad_norm": 22.765338134366964,
|
| 1185 |
+
"kl": 0.07025146484375,
|
| 1186 |
+
"learning_rate": 2.3314840961805802e-07,
|
| 1187 |
+
"loss": 0.0001,
|
| 1188 |
+
"reward": 0.686832070350647,
|
| 1189 |
+
"reward_std": 0.2984095811843872,
|
| 1190 |
+
"rewards/correct_code_reward_func": 0.2083333432674408,
|
| 1191 |
+
"rewards/len_reward_func": 0.47849874198436737,
|
| 1192 |
+
"step": 91
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"completion_length": 54.31250190734863,
|
| 1196 |
+
"epoch": 1.6194690265486726,
|
| 1197 |
+
"grad_norm": 1.6877966926685302,
|
| 1198 |
+
"kl": 0.0550537109375,
|
| 1199 |
+
"learning_rate": 2.2834440237932533e-07,
|
| 1200 |
+
"loss": 0.0001,
|
| 1201 |
+
"reward": 0.8750000298023224,
|
| 1202 |
+
"reward_std": 0.38857419788837433,
|
| 1203 |
+
"rewards/correct_code_reward_func": 0.4166666865348816,
|
| 1204 |
+
"rewards/len_reward_func": 0.4583333432674408,
|
| 1205 |
+
"step": 92
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"completion_length": 120.70833969116211,
|
| 1209 |
+
"epoch": 1.6371681415929205,
|
| 1210 |
+
"grad_norm": 10.817524146996126,
|
| 1211 |
+
"kl": 0.0389404296875,
|
| 1212 |
+
"learning_rate": 2.2354843931106932e-07,
|
| 1213 |
+
"loss": 0.0,
|
| 1214 |
+
"reward": 0.7042862474918365,
|
| 1215 |
+
"reward_std": 0.3615289777517319,
|
| 1216 |
+
"rewards/correct_code_reward_func": 0.2083333432674408,
|
| 1217 |
+
"rewards/len_reward_func": 0.49595288932323456,
|
| 1218 |
+
"step": 93
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"completion_length": 44.250000953674316,
|
| 1222 |
+
"epoch": 1.6548672566371683,
|
| 1223 |
+
"grad_norm": 28.86285310594792,
|
| 1224 |
+
"kl": 0.064697265625,
|
| 1225 |
+
"learning_rate": 2.1876230191778597e-07,
|
| 1226 |
+
"loss": 0.0001,
|
| 1227 |
+
"reward": 0.8750000298023224,
|
| 1228 |
+
"reward_std": 0.3857583850622177,
|
| 1229 |
+
"rewards/correct_code_reward_func": 0.3750000149011612,
|
| 1230 |
+
"rewards/len_reward_func": 0.5,
|
| 1231 |
+
"step": 94
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"completion_length": 53.62500190734863,
|
| 1235 |
+
"epoch": 1.672566371681416,
|
| 1236 |
+
"grad_norm": 3.9097872367150397,
|
| 1237 |
+
"kl": 0.0594482421875,
|
| 1238 |
+
"learning_rate": 2.1398776805413398e-07,
|
| 1239 |
+
"loss": 0.0001,
|
| 1240 |
+
"reward": 0.8750000298023224,
|
| 1241 |
+
"reward_std": 0.3233579397201538,
|
| 1242 |
+
"rewards/correct_code_reward_func": 0.3750000149011612,
|
| 1243 |
+
"rewards/len_reward_func": 0.5,
|
| 1244 |
+
"step": 95
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"completion_length": 65.00000190734863,
|
| 1248 |
+
"epoch": 1.6902654867256637,
|
| 1249 |
+
"grad_norm": 1.5650440517517894,
|
| 1250 |
+
"kl": 0.0361328125,
|
| 1251 |
+
"learning_rate": 2.092266112645343e-07,
|
| 1252 |
+
"loss": 0.0,
|
| 1253 |
+
"reward": 0.8589357733726501,
|
| 1254 |
+
"reward_std": 0.3421449810266495,
|
| 1255 |
+
"rewards/correct_code_reward_func": 0.3750000149011612,
|
| 1256 |
+
"rewards/len_reward_func": 0.48393575847148895,
|
| 1257 |
+
"step": 96
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"completion_length": 68.06250381469727,
|
| 1261 |
+
"epoch": 1.7079646017699115,
|
| 1262 |
+
"grad_norm": 2.91895596774148,
|
| 1263 |
+
"kl": 0.0269775390625,
|
| 1264 |
+
"learning_rate": 2.044806001243714e-07,
|
| 1265 |
+
"loss": 0.0,
|
| 1266 |
+
"reward": 0.9035088419914246,
|
| 1267 |
+
"reward_std": 0.48539453744888306,
|
| 1268 |
+
"rewards/correct_code_reward_func": 0.4375000298023224,
|
| 1269 |
+
"rewards/len_reward_func": 0.466008797287941,
|
| 1270 |
+
"step": 97
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"completion_length": 73.70833396911621,
|
| 1274 |
+
"epoch": 1.7256637168141593,
|
| 1275 |
+
"grad_norm": 24.67219025250006,
|
| 1276 |
+
"kl": 0.0465087890625,
|
| 1277 |
+
"learning_rate": 1.997514975830388e-07,
|
| 1278 |
+
"loss": 0.0,
|
| 1279 |
+
"reward": 0.8952526450157166,
|
| 1280 |
+
"reward_std": 0.29202311858534813,
|
| 1281 |
+
"rewards/correct_code_reward_func": 0.3958333432674408,
|
| 1282 |
+
"rewards/len_reward_func": 0.49941930174827576,
|
| 1283 |
+
"step": 98
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"completion_length": 97.75000381469727,
|
| 1287 |
+
"epoch": 1.7433628318584071,
|
| 1288 |
+
"grad_norm": 1.6430763507769786,
|
| 1289 |
+
"kl": 0.0406494140625,
|
| 1290 |
+
"learning_rate": 1.9504106030907603e-07,
|
| 1291 |
+
"loss": 0.0,
|
| 1292 |
+
"reward": 0.7537775337696075,
|
| 1293 |
+
"reward_std": 0.3211555927991867,
|
| 1294 |
+
"rewards/correct_code_reward_func": 0.2916666865348816,
|
| 1295 |
+
"rewards/len_reward_func": 0.46211084723472595,
|
| 1296 |
+
"step": 99
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"completion_length": 52.62500190734863,
|
| 1300 |
+
"epoch": 1.7610619469026547,
|
| 1301 |
+
"grad_norm": 3.9570608772498286,
|
| 1302 |
+
"kl": 0.0634765625,
|
| 1303 |
+
"learning_rate": 1.9035103803763792e-07,
|
| 1304 |
+
"loss": 0.0001,
|
| 1305 |
+
"reward": 0.8125,
|
| 1306 |
+
"reward_std": 0.3857453167438507,
|
| 1307 |
+
"rewards/correct_code_reward_func": 0.3125,
|
| 1308 |
+
"rewards/len_reward_func": 0.5,
|
| 1309 |
+
"step": 100
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"completion_length": 81.35416793823242,
|
| 1313 |
+
"epoch": 1.7787610619469025,
|
| 1314 |
+
"grad_norm": 2.731852842110385,
|
| 1315 |
+
"kl": 0.033447265625,
|
| 1316 |
+
"learning_rate": 1.8568317292053891e-07,
|
| 1317 |
+
"loss": 0.0,
|
| 1318 |
+
"reward": 0.9219505190849304,
|
| 1319 |
+
"reward_std": 0.2703188881278038,
|
| 1320 |
+
"rewards/correct_code_reward_func": 0.4375000149011612,
|
| 1321 |
+
"rewards/len_reward_func": 0.48445048928260803,
|
| 1322 |
+
"step": 101
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"completion_length": 83.31250381469727,
|
| 1326 |
+
"epoch": 1.7964601769911503,
|
| 1327 |
+
"grad_norm": 2.4613714386546595,
|
| 1328 |
+
"kl": 0.022613525390625,
|
| 1329 |
+
"learning_rate": 1.8103919887911523e-07,
|
| 1330 |
+
"loss": 0.0,
|
| 1331 |
+
"reward": 0.7291666865348816,
|
| 1332 |
+
"reward_std": 0.30859364569187164,
|
| 1333 |
+
"rewards/correct_code_reward_func": 0.2291666716337204,
|
| 1334 |
+
"rewards/len_reward_func": 0.5,
|
| 1335 |
+
"step": 102
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"completion_length": 79.33333587646484,
|
| 1339 |
+
"epoch": 1.8141592920353982,
|
| 1340 |
+
"grad_norm": 4.215262416117561,
|
| 1341 |
+
"kl": 0.040771484375,
|
| 1342 |
+
"learning_rate": 1.7642084096014402e-07,
|
| 1343 |
+
"loss": 0.0,
|
| 1344 |
+
"reward": 0.7859848439693451,
|
| 1345 |
+
"reward_std": 0.4127810299396515,
|
| 1346 |
+
"rewards/correct_code_reward_func": 0.291666679084301,
|
| 1347 |
+
"rewards/len_reward_func": 0.4943181872367859,
|
| 1348 |
+
"step": 103
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"completion_length": 58.52083396911621,
|
| 1352 |
+
"epoch": 1.831858407079646,
|
| 1353 |
+
"grad_norm": 4.745442984631991,
|
| 1354 |
+
"kl": 0.08740234375,
|
| 1355 |
+
"learning_rate": 1.7182981469505847e-07,
|
| 1356 |
+
"loss": 0.0001,
|
| 1357 |
+
"reward": 0.8333333730697632,
|
| 1358 |
+
"reward_std": 0.45660629868507385,
|
| 1359 |
+
"rewards/correct_code_reward_func": 0.354166679084301,
|
| 1360 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1361 |
+
"step": 104
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"completion_length": 112.64583587646484,
|
| 1365 |
+
"epoch": 1.8495575221238938,
|
| 1366 |
+
"grad_norm": 1.7418907068386247,
|
| 1367 |
+
"kl": 0.0325927734375,
|
| 1368 |
+
"learning_rate": 1.672678254626979e-07,
|
| 1369 |
+
"loss": 0.0,
|
| 1370 |
+
"reward": 0.6364757716655731,
|
| 1371 |
+
"reward_std": 0.3013785034418106,
|
| 1372 |
+
"rewards/correct_code_reward_func": 0.18750000558793545,
|
| 1373 |
+
"rewards/len_reward_func": 0.4489757716655731,
|
| 1374 |
+
"step": 105
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"completion_length": 57.5,
|
| 1378 |
+
"epoch": 1.8672566371681416,
|
| 1379 |
+
"grad_norm": 3.579303393703891,
|
| 1380 |
+
"kl": 0.0543212890625,
|
| 1381 |
+
"learning_rate": 1.6273656785582984e-07,
|
| 1382 |
+
"loss": 0.0001,
|
| 1383 |
+
"reward": 0.8536822199821472,
|
| 1384 |
+
"reward_std": 0.45756329596042633,
|
| 1385 |
+
"rewards/correct_code_reward_func": 0.3958333432674408,
|
| 1386 |
+
"rewards/len_reward_func": 0.45784883201122284,
|
| 1387 |
+
"step": 106
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"completion_length": 83.52083587646484,
|
| 1391 |
+
"epoch": 1.8849557522123894,
|
| 1392 |
+
"grad_norm": 2.1418835681881925,
|
| 1393 |
+
"kl": 0.0758056640625,
|
| 1394 |
+
"learning_rate": 1.5823772505167698e-07,
|
| 1395 |
+
"loss": 0.0001,
|
| 1396 |
+
"reward": 0.8958333432674408,
|
| 1397 |
+
"reward_std": 0.49633626639842987,
|
| 1398 |
+
"rewards/correct_code_reward_func": 0.4375000149011612,
|
| 1399 |
+
"rewards/len_reward_func": 0.4583333432674408,
|
| 1400 |
+
"step": 107
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"completion_length": 94.20833587646484,
|
| 1404 |
+
"epoch": 1.9026548672566372,
|
| 1405 |
+
"grad_norm": 32.45107688241668,
|
| 1406 |
+
"kl": 0.025146484375,
|
| 1407 |
+
"learning_rate": 1.5377296818668635e-07,
|
| 1408 |
+
"loss": 0.0,
|
| 1409 |
+
"reward": 0.854166716337204,
|
| 1410 |
+
"reward_std": 0.37177951633930206,
|
| 1411 |
+
"rewards/correct_code_reward_func": 0.3958333432674408,
|
| 1412 |
+
"rewards/len_reward_func": 0.4583333432674408,
|
| 1413 |
+
"step": 108
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"completion_length": 72.89583396911621,
|
| 1417 |
+
"epoch": 1.920353982300885,
|
| 1418 |
+
"grad_norm": 1.7762554154885124,
|
| 1419 |
+
"kl": 0.04364013671875,
|
| 1420 |
+
"learning_rate": 1.4934395573577013e-07,
|
| 1421 |
+
"loss": 0.0,
|
| 1422 |
+
"reward": 0.7892157137393951,
|
| 1423 |
+
"reward_std": 0.3116895519196987,
|
| 1424 |
+
"rewards/correct_code_reward_func": 0.3125000149011612,
|
| 1425 |
+
"rewards/len_reward_func": 0.47671568393707275,
|
| 1426 |
+
"step": 109
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"completion_length": 66.39583587646484,
|
| 1430 |
+
"epoch": 1.9380530973451329,
|
| 1431 |
+
"grad_norm": 1.3218901012489044,
|
| 1432 |
+
"kl": 0.02496337890625,
|
| 1433 |
+
"learning_rate": 1.4495233289624958e-07,
|
| 1434 |
+
"loss": 0.0,
|
| 1435 |
+
"reward": 0.7881554365158081,
|
| 1436 |
+
"reward_std": 0.3200981765985489,
|
| 1437 |
+
"rewards/correct_code_reward_func": 0.2916666716337204,
|
| 1438 |
+
"rewards/len_reward_func": 0.4964887797832489,
|
| 1439 |
+
"step": 110
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"completion_length": 58.291669845581055,
|
| 1443 |
+
"epoch": 1.9557522123893807,
|
| 1444 |
+
"grad_norm": 0.9195423141452592,
|
| 1445 |
+
"kl": 0.066162109375,
|
| 1446 |
+
"learning_rate": 1.4059973097673185e-07,
|
| 1447 |
+
"loss": 0.0001,
|
| 1448 |
+
"reward": 0.7083333432674408,
|
| 1449 |
+
"reward_std": 0.3382536321878433,
|
| 1450 |
+
"rewards/correct_code_reward_func": 0.2500000111758709,
|
| 1451 |
+
"rewards/len_reward_func": 0.4583333432674408,
|
| 1452 |
+
"step": 111
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"completion_length": 94.41667175292969,
|
| 1456 |
+
"epoch": 1.9734513274336283,
|
| 1457 |
+
"grad_norm": 89.52879883380525,
|
| 1458 |
+
"kl": 0.0601806640625,
|
| 1459 |
+
"learning_rate": 1.3628776679114514e-07,
|
| 1460 |
+
"loss": 0.0001,
|
| 1461 |
+
"reward": 0.7916666865348816,
|
| 1462 |
+
"reward_std": 0.31285394728183746,
|
| 1463 |
+
"rewards/correct_code_reward_func": 0.3125,
|
| 1464 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1465 |
+
"step": 112
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"completion_length": 100.27083587646484,
|
| 1469 |
+
"epoch": 1.991150442477876,
|
| 1470 |
+
"grad_norm": 2.390953296031483,
|
| 1471 |
+
"kl": 0.056396484375,
|
| 1472 |
+
"learning_rate": 1.320180420581587e-07,
|
| 1473 |
+
"loss": 0.0001,
|
| 1474 |
+
"reward": 0.6097994744777679,
|
| 1475 |
+
"reward_std": 0.18817735463380814,
|
| 1476 |
+
"rewards/correct_code_reward_func": 0.12500000558793545,
|
| 1477 |
+
"rewards/len_reward_func": 0.48479947447776794,
|
| 1478 |
+
"step": 113
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"completion_length": 63.875,
|
| 1482 |
+
"epoch": 2.0,
|
| 1483 |
+
"grad_norm": 3.601264581967262,
|
| 1484 |
+
"kl": 0.02001953125,
|
| 1485 |
+
"learning_rate": 1.2779214280620908e-07,
|
| 1486 |
+
"loss": 0.0,
|
| 1487 |
+
"reward": 0.6666666865348816,
|
| 1488 |
+
"reward_std": 0.17817416787147522,
|
| 1489 |
+
"rewards/correct_code_reward_func": 0.1666666716337204,
|
| 1490 |
+
"rewards/len_reward_func": 0.5,
|
| 1491 |
+
"step": 114
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"completion_length": 48.64583396911621,
|
| 1495 |
+
"epoch": 2.017699115044248,
|
| 1496 |
+
"grad_norm": 3.9192662388145028,
|
| 1497 |
+
"kl": 0.05908203125,
|
| 1498 |
+
"learning_rate": 1.2361163878435594e-07,
|
| 1499 |
+
"loss": 0.0001,
|
| 1500 |
+
"reward": 0.6176470816135406,
|
| 1501 |
+
"reward_std": 0.2277354598045349,
|
| 1502 |
+
"rewards/correct_code_reward_func": 0.12500000558793545,
|
| 1503 |
+
"rewards/len_reward_func": 0.49264705181121826,
|
| 1504 |
+
"step": 115
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"completion_length": 51.41666793823242,
|
| 1508 |
+
"epoch": 2.0353982300884956,
|
| 1509 |
+
"grad_norm": 2.2242727716589967,
|
| 1510 |
+
"kl": 0.04486083984375,
|
| 1511 |
+
"learning_rate": 1.1947808287918403e-07,
|
| 1512 |
+
"loss": 0.0,
|
| 1513 |
+
"reward": 0.7083333730697632,
|
| 1514 |
+
"reward_std": 0.2994871214032173,
|
| 1515 |
+
"rewards/correct_code_reward_func": 0.2083333395421505,
|
| 1516 |
+
"rewards/len_reward_func": 0.5,
|
| 1517 |
+
"step": 116
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"completion_length": 69.04166793823242,
|
| 1521 |
+
"epoch": 2.0530973451327434,
|
| 1522 |
+
"grad_norm": 5.407021499840827,
|
| 1523 |
+
"kl": 0.0469970703125,
|
| 1524 |
+
"learning_rate": 1.1539301053796949e-07,
|
| 1525 |
+
"loss": 0.0,
|
| 1526 |
+
"reward": 0.9791666865348816,
|
| 1527 |
+
"reward_std": 0.2946278229355812,
|
| 1528 |
+
"rewards/correct_code_reward_func": 0.4791666865348816,
|
| 1529 |
+
"rewards/len_reward_func": 0.5,
|
| 1530 |
+
"step": 117
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"completion_length": 47.83333396911621,
|
| 1534 |
+
"epoch": 2.0707964601769913,
|
| 1535 |
+
"grad_norm": 29.2301422488625,
|
| 1536 |
+
"kl": 0.08349609375,
|
| 1537 |
+
"learning_rate": 1.1135793919832334e-07,
|
| 1538 |
+
"loss": 0.0001,
|
| 1539 |
+
"reward": 0.9467871785163879,
|
| 1540 |
+
"reward_std": 0.43978360295295715,
|
| 1541 |
+
"rewards/correct_code_reward_func": 0.4791666716337204,
|
| 1542 |
+
"rewards/len_reward_func": 0.46762049198150635,
|
| 1543 |
+
"step": 118
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"completion_length": 90.41666793823242,
|
| 1547 |
+
"epoch": 2.088495575221239,
|
| 1548 |
+
"grad_norm": 1.804477992356325,
|
| 1549 |
+
"kl": 0.02587890625,
|
| 1550 |
+
"learning_rate": 1.0737436772452602e-07,
|
| 1551 |
+
"loss": 0.0,
|
| 1552 |
+
"reward": 0.7708333432674408,
|
| 1553 |
+
"reward_std": 0.3584126979112625,
|
| 1554 |
+
"rewards/correct_code_reward_func": 0.2708333358168602,
|
| 1555 |
+
"rewards/len_reward_func": 0.5,
|
| 1556 |
+
"step": 119
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"completion_length": 150.75000762939453,
|
| 1560 |
+
"epoch": 2.106194690265487,
|
| 1561 |
+
"grad_norm": 24.510631619520034,
|
| 1562 |
+
"kl": 0.0252685546875,
|
| 1563 |
+
"learning_rate": 1.0344377585075997e-07,
|
| 1564 |
+
"loss": 0.0,
|
| 1565 |
+
"reward": 0.684719979763031,
|
| 1566 |
+
"reward_std": 0.30992700159549713,
|
| 1567 |
+
"rewards/correct_code_reward_func": 0.1875000074505806,
|
| 1568 |
+
"rewards/len_reward_func": 0.4972199648618698,
|
| 1569 |
+
"step": 120
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"completion_length": 79.85416793823242,
|
| 1573 |
+
"epoch": 2.1238938053097347,
|
| 1574 |
+
"grad_norm": 2.667553033080654,
|
| 1575 |
+
"kl": 0.0318603515625,
|
| 1576 |
+
"learning_rate": 9.95676236314489e-08,
|
| 1577 |
+
"loss": 0.0,
|
| 1578 |
+
"reward": 0.8333333432674408,
|
| 1579 |
+
"reward_std": 0.31142252683639526,
|
| 1580 |
+
"rewards/correct_code_reward_func": 0.3333333432674408,
|
| 1581 |
+
"rewards/len_reward_func": 0.5,
|
| 1582 |
+
"step": 121
|
| 1583 |
+
},
|
| 1584 |
+
{
|
| 1585 |
+
"completion_length": 91.56250381469727,
|
| 1586 |
+
"epoch": 2.1415929203539825,
|
| 1587 |
+
"grad_norm": 100.44717449922435,
|
| 1588 |
+
"kl": 0.2076416015625,
|
| 1589 |
+
"learning_rate": 9.574735089890764e-08,
|
| 1590 |
+
"loss": 0.0002,
|
| 1591 |
+
"reward": 0.7708333432674408,
|
| 1592 |
+
"reward_std": 0.30231600999832153,
|
| 1593 |
+
"rewards/correct_code_reward_func": 0.2916666716337204,
|
| 1594 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1595 |
+
"step": 122
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"completion_length": 87.70833396911621,
|
| 1599 |
+
"epoch": 2.15929203539823,
|
| 1600 |
+
"grad_norm": 77.23064615733736,
|
| 1601 |
+
"kl": 0.22283935546875,
|
| 1602 |
+
"learning_rate": 9.198437672850248e-08,
|
| 1603 |
+
"loss": 0.0002,
|
| 1604 |
+
"reward": 0.770833358168602,
|
| 1605 |
+
"reward_std": 0.3205290399491787,
|
| 1606 |
+
"rewards/correct_code_reward_func": 0.2916666865348816,
|
| 1607 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1608 |
+
"step": 123
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"completion_length": 62.312503814697266,
|
| 1612 |
+
"epoch": 2.1769911504424777,
|
| 1613 |
+
"grad_norm": 17.734240204261667,
|
| 1614 |
+
"kl": 0.0535888671875,
|
| 1615 |
+
"learning_rate": 8.8280098911523e-08,
|
| 1616 |
+
"loss": 0.0001,
|
| 1617 |
+
"reward": 0.875,
|
| 1618 |
+
"reward_std": 0.2994871288537979,
|
| 1619 |
+
"rewards/correct_code_reward_func": 0.3958333432674408,
|
| 1620 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1621 |
+
"step": 124
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"completion_length": 65.41666793823242,
|
| 1625 |
+
"epoch": 2.1946902654867255,
|
| 1626 |
+
"grad_norm": 28.039058681720235,
|
| 1627 |
+
"kl": 0.0321044921875,
|
| 1628 |
+
"learning_rate": 8.463589343595976e-08,
|
| 1629 |
+
"loss": 0.0,
|
| 1630 |
+
"reward": 1.0833333730697632,
|
| 1631 |
+
"reward_std": 0.5049939155578613,
|
| 1632 |
+
"rewards/correct_code_reward_func": 0.5833333432674408,
|
| 1633 |
+
"rewards/len_reward_func": 0.5,
|
| 1634 |
+
"step": 125
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"completion_length": 98.64583396911621,
|
| 1638 |
+
"epoch": 2.2123893805309733,
|
| 1639 |
+
"grad_norm": 6.872794457305926,
|
| 1640 |
+
"kl": 0.0435791015625,
|
| 1641 |
+
"learning_rate": 8.105311397538084e-08,
|
| 1642 |
+
"loss": 0.0,
|
| 1643 |
+
"reward": 0.7077258378267288,
|
| 1644 |
+
"reward_std": 0.37921955436468124,
|
| 1645 |
+
"rewards/correct_code_reward_func": 0.25000000558793545,
|
| 1646 |
+
"rewards/len_reward_func": 0.4577258378267288,
|
| 1647 |
+
"step": 126
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"completion_length": 71.22916984558105,
|
| 1651 |
+
"epoch": 2.230088495575221,
|
| 1652 |
+
"grad_norm": 2.9985031046208857,
|
| 1653 |
+
"kl": 0.0357666015625,
|
| 1654 |
+
"learning_rate": 7.753309138609704e-08,
|
| 1655 |
+
"loss": 0.0,
|
| 1656 |
+
"reward": 0.7608647048473358,
|
| 1657 |
+
"reward_std": 0.35784420371055603,
|
| 1658 |
+
"rewards/correct_code_reward_func": 0.2708333358168602,
|
| 1659 |
+
"rewards/len_reward_func": 0.490031361579895,
|
| 1660 |
+
"step": 127
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"completion_length": 72.87500190734863,
|
| 1664 |
+
"epoch": 2.247787610619469,
|
| 1665 |
+
"grad_norm": 1.1881439057403123,
|
| 1666 |
+
"kl": 0.03509521484375,
|
| 1667 |
+
"learning_rate": 7.407713321280375e-08,
|
| 1668 |
+
"loss": 0.0,
|
| 1669 |
+
"reward": 0.854166716337204,
|
| 1670 |
+
"reward_std": 0.30231600999832153,
|
| 1671 |
+
"rewards/correct_code_reward_func": 0.354166679084301,
|
| 1672 |
+
"rewards/len_reward_func": 0.5,
|
| 1673 |
+
"step": 128
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"completion_length": 95.39583587646484,
|
| 1677 |
+
"epoch": 2.265486725663717,
|
| 1678 |
+
"grad_norm": 2.071026140631432,
|
| 1679 |
+
"kl": 0.0494384765625,
|
| 1680 |
+
"learning_rate": 7.06865232028808e-08,
|
| 1681 |
+
"loss": 0.0,
|
| 1682 |
+
"reward": 0.8162688612937927,
|
| 1683 |
+
"reward_std": 0.29165610671043396,
|
| 1684 |
+
"rewards/correct_code_reward_func": 0.3541666716337204,
|
| 1685 |
+
"rewards/len_reward_func": 0.46210217475891113,
|
| 1686 |
+
"step": 129
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"completion_length": 78.43750381469727,
|
| 1690 |
+
"epoch": 2.2831858407079646,
|
| 1691 |
+
"grad_norm": 0.7368071897691737,
|
| 1692 |
+
"kl": 0.027099609375,
|
| 1693 |
+
"learning_rate": 6.736252082953306e-08,
|
| 1694 |
+
"loss": 0.0,
|
| 1695 |
+
"reward": 0.8333333432674408,
|
| 1696 |
+
"reward_std": 0.2903675436973572,
|
| 1697 |
+
"rewards/correct_code_reward_func": 0.3333333358168602,
|
| 1698 |
+
"rewards/len_reward_func": 0.5,
|
| 1699 |
+
"step": 130
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"completion_length": 73.45833587646484,
|
| 1703 |
+
"epoch": 2.3008849557522124,
|
| 1704 |
+
"grad_norm": 2.1125816683606726,
|
| 1705 |
+
"kl": 0.0389404296875,
|
| 1706 |
+
"learning_rate": 6.410636082394771e-08,
|
| 1707 |
+
"loss": 0.0,
|
| 1708 |
+
"reward": 0.5384955406188965,
|
| 1709 |
+
"reward_std": 0.272975854575634,
|
| 1710 |
+
"rewards/correct_code_reward_func": 0.0833333358168602,
|
| 1711 |
+
"rewards/len_reward_func": 0.4551621973514557,
|
| 1712 |
+
"step": 131
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"completion_length": 101.02083587646484,
|
| 1716 |
+
"epoch": 2.3185840707964602,
|
| 1717 |
+
"grad_norm": 3.1749370092384415,
|
| 1718 |
+
"kl": 0.0380859375,
|
| 1719 |
+
"learning_rate": 6.091925271664156e-08,
|
| 1720 |
+
"loss": 0.0,
|
| 1721 |
+
"reward": 0.6666666716337204,
|
| 1722 |
+
"reward_std": 0.2903675250709057,
|
| 1723 |
+
"rewards/correct_code_reward_func": 0.1875,
|
| 1724 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1725 |
+
"step": 132
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"completion_length": 64.22916793823242,
|
| 1729 |
+
"epoch": 2.336283185840708,
|
| 1730 |
+
"grad_norm": 2.773648686745547,
|
| 1731 |
+
"kl": 0.11944580078125,
|
| 1732 |
+
"learning_rate": 5.7802380388170344e-08,
|
| 1733 |
+
"loss": 0.0001,
|
| 1734 |
+
"reward": 0.8125,
|
| 1735 |
+
"reward_std": 0.37034809589385986,
|
| 1736 |
+
"rewards/correct_code_reward_func": 0.3333333358168602,
|
| 1737 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1738 |
+
"step": 133
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"completion_length": 56.64583396911621,
|
| 1742 |
+
"epoch": 2.353982300884956,
|
| 1743 |
+
"grad_norm": 3.7650155223115793,
|
| 1744 |
+
"kl": 0.0516357421875,
|
| 1745 |
+
"learning_rate": 5.4756901629364886e-08,
|
| 1746 |
+
"loss": 0.0001,
|
| 1747 |
+
"reward": 1.129325270652771,
|
| 1748 |
+
"reward_std": 0.32795289158821106,
|
| 1749 |
+
"rewards/correct_code_reward_func": 0.6458333432674408,
|
| 1750 |
+
"rewards/len_reward_func": 0.4834919422864914,
|
| 1751 |
+
"step": 134
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"completion_length": 70.1875,
|
| 1755 |
+
"epoch": 2.3716814159292037,
|
| 1756 |
+
"grad_norm": 1.0135949886271995,
|
| 1757 |
+
"kl": 0.042236328125,
|
| 1758 |
+
"learning_rate": 5.178394771125968e-08,
|
| 1759 |
+
"loss": 0.0,
|
| 1760 |
+
"reward": 0.7218915224075317,
|
| 1761 |
+
"reward_std": 0.2874666675925255,
|
| 1762 |
+
"rewards/correct_code_reward_func": 0.2291666716337204,
|
| 1763 |
+
"rewards/len_reward_func": 0.49272486567497253,
|
| 1764 |
+
"step": 135
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"completion_length": 132.7916717529297,
|
| 1768 |
+
"epoch": 2.3893805309734515,
|
| 1769 |
+
"grad_norm": 1.3477438258822867,
|
| 1770 |
+
"kl": 0.03106689453125,
|
| 1771 |
+
"learning_rate": 4.888462296487128e-08,
|
| 1772 |
+
"loss": 0.0,
|
| 1773 |
+
"reward": 0.9166666865348816,
|
| 1774 |
+
"reward_std": 0.42218445241451263,
|
| 1775 |
+
"rewards/correct_code_reward_func": 0.4166666865348816,
|
| 1776 |
+
"rewards/len_reward_func": 0.5,
|
| 1777 |
+
"step": 136
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"completion_length": 60.22916793823242,
|
| 1781 |
+
"epoch": 2.4070796460176993,
|
| 1782 |
+
"grad_norm": 1.2650991231847002,
|
| 1783 |
+
"kl": 0.058837890625,
|
| 1784 |
+
"learning_rate": 4.606000437098476e-08,
|
| 1785 |
+
"loss": 0.0001,
|
| 1786 |
+
"reward": 0.8958333730697632,
|
| 1787 |
+
"reward_std": 0.2041093409061432,
|
| 1788 |
+
"rewards/correct_code_reward_func": 0.4166666716337204,
|
| 1789 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1790 |
+
"step": 137
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"completion_length": 64.6875,
|
| 1794 |
+
"epoch": 2.4247787610619467,
|
| 1795 |
+
"grad_norm": 2.901815806581681,
|
| 1796 |
+
"kl": 0.03662109375,
|
| 1797 |
+
"learning_rate": 4.3311141160099377e-08,
|
| 1798 |
+
"loss": 0.0,
|
| 1799 |
+
"reward": 0.9150572419166565,
|
| 1800 |
+
"reward_std": 0.3222523480653763,
|
| 1801 |
+
"rewards/correct_code_reward_func": 0.4166666716337204,
|
| 1802 |
+
"rewards/len_reward_func": 0.4983905553817749,
|
| 1803 |
+
"step": 138
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"completion_length": 92.79166984558105,
|
| 1807 |
+
"epoch": 2.442477876106195,
|
| 1808 |
+
"grad_norm": 0.6516445128630077,
|
| 1809 |
+
"kl": 0.03466796875,
|
| 1810 |
+
"learning_rate": 4.063905442268201e-08,
|
| 1811 |
+
"loss": 0.0,
|
| 1812 |
+
"reward": 0.6041666865348816,
|
| 1813 |
+
"reward_std": 0.16340987384319305,
|
| 1814 |
+
"rewards/correct_code_reward_func": 0.1041666679084301,
|
| 1815 |
+
"rewards/len_reward_func": 0.5,
|
| 1816 |
+
"step": 139
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"completion_length": 67.60416793823242,
|
| 1820 |
+
"epoch": 2.4601769911504423,
|
| 1821 |
+
"grad_norm": 1.8554639671156326,
|
| 1822 |
+
"kl": 0.0423583984375,
|
| 1823 |
+
"learning_rate": 3.8044736729873795e-08,
|
| 1824 |
+
"loss": 0.0,
|
| 1825 |
+
"reward": 0.9469756484031677,
|
| 1826 |
+
"reward_std": 0.43488872051239014,
|
| 1827 |
+
"rewards/correct_code_reward_func": 0.479166679084301,
|
| 1828 |
+
"rewards/len_reward_func": 0.46780893206596375,
|
| 1829 |
+
"step": 140
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"completion_length": 77.45833396911621,
|
| 1833 |
+
"epoch": 2.47787610619469,
|
| 1834 |
+
"grad_norm": 3.8300963552021754,
|
| 1835 |
+
"kl": 0.04541015625,
|
| 1836 |
+
"learning_rate": 3.552915176479071e-08,
|
| 1837 |
+
"loss": 0.0,
|
| 1838 |
+
"reward": 1.0416666865348816,
|
| 1839 |
+
"reward_std": 0.320542111992836,
|
| 1840 |
+
"rewards/correct_code_reward_func": 0.5416666716337204,
|
| 1841 |
+
"rewards/len_reward_func": 0.5,
|
| 1842 |
+
"step": 141
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"completion_length": 58.708335876464844,
|
| 1846 |
+
"epoch": 2.495575221238938,
|
| 1847 |
+
"grad_norm": 1.6137020616307847,
|
| 1848 |
+
"kl": 0.06591796875,
|
| 1849 |
+
"learning_rate": 3.309323396455446e-08,
|
| 1850 |
+
"loss": 0.0001,
|
| 1851 |
+
"reward": 0.9375000596046448,
|
| 1852 |
+
"reward_std": 0.2658637687563896,
|
| 1853 |
+
"rewards/correct_code_reward_func": 0.4583333432674408,
|
| 1854 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1855 |
+
"step": 142
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"completion_length": 86.12500381469727,
|
| 1859 |
+
"epoch": 2.5132743362831858,
|
| 1860 |
+
"grad_norm": 3.390075353900997,
|
| 1861 |
+
"kl": 0.0819091796875,
|
| 1862 |
+
"learning_rate": 3.073788817318707e-08,
|
| 1863 |
+
"loss": 0.0001,
|
| 1864 |
+
"reward": 0.8744058609008789,
|
| 1865 |
+
"reward_std": 0.429522842168808,
|
| 1866 |
+
"rewards/correct_code_reward_func": 0.3750000149011612,
|
| 1867 |
+
"rewards/len_reward_func": 0.4994058310985565,
|
| 1868 |
+
"step": 143
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"completion_length": 76.89583587646484,
|
| 1872 |
+
"epoch": 2.5309734513274336,
|
| 1873 |
+
"grad_norm": 17.278809046793295,
|
| 1874 |
+
"kl": 0.2576904296875,
|
| 1875 |
+
"learning_rate": 2.846398930549859e-08,
|
| 1876 |
+
"loss": 0.0003,
|
| 1877 |
+
"reward": 0.7083333432674408,
|
| 1878 |
+
"reward_std": 0.1178511306643486,
|
| 1879 |
+
"rewards/correct_code_reward_func": 0.20833333395421505,
|
| 1880 |
+
"rewards/len_reward_func": 0.5,
|
| 1881 |
+
"step": 144
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"completion_length": 86.625,
|
| 1885 |
+
"epoch": 2.5486725663716814,
|
| 1886 |
+
"grad_norm": 4.630135346925831,
|
| 1887 |
+
"kl": 0.0435791015625,
|
| 1888 |
+
"learning_rate": 2.62723820220917e-08,
|
| 1889 |
+
"loss": 0.0,
|
| 1890 |
+
"reward": 0.7428782880306244,
|
| 1891 |
+
"reward_std": 0.3244539424777031,
|
| 1892 |
+
"rewards/correct_code_reward_func": 0.2708333432674408,
|
| 1893 |
+
"rewards/len_reward_func": 0.4720449447631836,
|
| 1894 |
+
"step": 145
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"completion_length": 60.54166793823242,
|
| 1898 |
+
"epoch": 2.566371681415929,
|
| 1899 |
+
"grad_norm": 1.7403036344539455,
|
| 1900 |
+
"kl": 0.067626953125,
|
| 1901 |
+
"learning_rate": 2.416388041560491e-08,
|
| 1902 |
+
"loss": 0.0001,
|
| 1903 |
+
"reward": 1.0208333730697632,
|
| 1904 |
+
"reward_std": 0.24056155234575272,
|
| 1905 |
+
"rewards/correct_code_reward_func": 0.520833358168602,
|
| 1906 |
+
"rewards/len_reward_func": 0.5,
|
| 1907 |
+
"step": 146
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"completion_length": 72.68750190734863,
|
| 1911 |
+
"epoch": 2.584070796460177,
|
| 1912 |
+
"grad_norm": 8.835725978147865,
|
| 1913 |
+
"kl": 0.03240966796875,
|
| 1914 |
+
"learning_rate": 2.2139267708310454e-08,
|
| 1915 |
+
"loss": 0.0,
|
| 1916 |
+
"reward": 1.0000000596046448,
|
| 1917 |
+
"reward_std": 0.42218445241451263,
|
| 1918 |
+
"rewards/correct_code_reward_func": 0.5000000149011612,
|
| 1919 |
+
"rewards/len_reward_func": 0.5,
|
| 1920 |
+
"step": 147
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"completion_length": 65.31250381469727,
|
| 1924 |
+
"epoch": 2.601769911504425,
|
| 1925 |
+
"grad_norm": 21.996556093676393,
|
| 1926 |
+
"kl": 0.144287109375,
|
| 1927 |
+
"learning_rate": 2.019929596117889e-08,
|
| 1928 |
+
"loss": 0.0001,
|
| 1929 |
+
"reward": 0.9583333730697632,
|
| 1930 |
+
"reward_std": 0.4446708858013153,
|
| 1931 |
+
"rewards/correct_code_reward_func": 0.4583333432674408,
|
| 1932 |
+
"rewards/len_reward_func": 0.5,
|
| 1933 |
+
"step": 148
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"completion_length": 54.270835876464844,
|
| 1937 |
+
"epoch": 2.6194690265486726,
|
| 1938 |
+
"grad_norm": 4.1936490058839695,
|
| 1939 |
+
"kl": 0.099609375,
|
| 1940 |
+
"learning_rate": 1.8344685794519504e-08,
|
| 1941 |
+
"loss": 0.0001,
|
| 1942 |
+
"reward": 1.036210298538208,
|
| 1943 |
+
"reward_std": 0.37449972331523895,
|
| 1944 |
+
"rewards/correct_code_reward_func": 0.5416666716337204,
|
| 1945 |
+
"rewards/len_reward_func": 0.49454365670681,
|
| 1946 |
+
"step": 149
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"completion_length": 99.29166793823242,
|
| 1950 |
+
"epoch": 2.6371681415929205,
|
| 1951 |
+
"grad_norm": 63.75890524377285,
|
| 1952 |
+
"kl": 0.2841796875,
|
| 1953 |
+
"learning_rate": 1.6576126120299043e-08,
|
| 1954 |
+
"loss": 0.0003,
|
| 1955 |
+
"reward": 0.8559730350971222,
|
| 1956 |
+
"reward_std": 0.3255860209465027,
|
| 1957 |
+
"rewards/correct_code_reward_func": 0.3750000149011612,
|
| 1958 |
+
"rewards/len_reward_func": 0.4809730350971222,
|
| 1959 |
+
"step": 150
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"completion_length": 107.31250381469727,
|
| 1963 |
+
"epoch": 2.6548672566371683,
|
| 1964 |
+
"grad_norm": 36.89829026803691,
|
| 1965 |
+
"kl": 0.7239990234375,
|
| 1966 |
+
"learning_rate": 1.4894273886239206e-08,
|
| 1967 |
+
"loss": 0.0007,
|
| 1968 |
+
"reward": 0.6250000298023224,
|
| 1969 |
+
"reward_std": 0.3382536321878433,
|
| 1970 |
+
"rewards/correct_code_reward_func": 0.1666666716337204,
|
| 1971 |
+
"rewards/len_reward_func": 0.4583333432674408,
|
| 1972 |
+
"step": 151
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"completion_length": 64.35416793823242,
|
| 1976 |
+
"epoch": 2.672566371681416,
|
| 1977 |
+
"grad_norm": 2.096221752318607,
|
| 1978 |
+
"kl": 0.052490234375,
|
| 1979 |
+
"learning_rate": 1.3299753831787192e-08,
|
| 1980 |
+
"loss": 0.0001,
|
| 1981 |
+
"reward": 0.75,
|
| 1982 |
+
"reward_std": 0.38857419788837433,
|
| 1983 |
+
"rewards/correct_code_reward_func": 0.25,
|
| 1984 |
+
"rewards/len_reward_func": 0.5,
|
| 1985 |
+
"step": 152
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"completion_length": 96.64583587646484,
|
| 1989 |
+
"epoch": 2.6902654867256635,
|
| 1990 |
+
"grad_norm": 0.6573939843127445,
|
| 1991 |
+
"kl": 0.032470703125,
|
| 1992 |
+
"learning_rate": 1.1793158256050706e-08,
|
| 1993 |
+
"loss": 0.0,
|
| 1994 |
+
"reward": 0.7916666865348816,
|
| 1995 |
+
"reward_std": 0.3521219789981842,
|
| 1996 |
+
"rewards/correct_code_reward_func": 0.3125000074505806,
|
| 1997 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 1998 |
+
"step": 153
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"completion_length": 57.45833396911621,
|
| 2002 |
+
"epoch": 2.7079646017699117,
|
| 2003 |
+
"grad_norm": 2.8347099035822936,
|
| 2004 |
+
"kl": 0.051513671875,
|
| 2005 |
+
"learning_rate": 1.0375046797782867e-08,
|
| 2006 |
+
"loss": 0.0001,
|
| 2007 |
+
"reward": 0.8958333730697632,
|
| 2008 |
+
"reward_std": 0.204109326004982,
|
| 2009 |
+
"rewards/correct_code_reward_func": 0.395833358168602,
|
| 2010 |
+
"rewards/len_reward_func": 0.5,
|
| 2011 |
+
"step": 154
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"completion_length": 97.02083587646484,
|
| 2015 |
+
"epoch": 2.725663716814159,
|
| 2016 |
+
"grad_norm": 89.9892277373503,
|
| 2017 |
+
"kl": 0.531982421875,
|
| 2018 |
+
"learning_rate": 9.045946227499297e-09,
|
| 2019 |
+
"loss": 0.0005,
|
| 2020 |
+
"reward": 0.9904971122741699,
|
| 2021 |
+
"reward_std": 0.4233323186635971,
|
| 2022 |
+
"rewards/correct_code_reward_func": 0.5208333432674408,
|
| 2023 |
+
"rewards/len_reward_func": 0.46966375410556793,
|
| 2024 |
+
"step": 155
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"completion_length": 75.37500381469727,
|
| 2028 |
+
"epoch": 2.7433628318584073,
|
| 2029 |
+
"grad_norm": 41.09603274193404,
|
| 2030 |
+
"kl": 0.0634765625,
|
| 2031 |
+
"learning_rate": 7.806350251804483e-09,
|
| 2032 |
+
"loss": 0.0001,
|
| 2033 |
+
"reward": 0.8541666865348816,
|
| 2034 |
+
"reward_std": 0.3492931127548218,
|
| 2035 |
+
"rewards/correct_code_reward_func": 0.3541666716337204,
|
| 2036 |
+
"rewards/len_reward_func": 0.5,
|
| 2037 |
+
"step": 156
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"completion_length": 92.85417175292969,
|
| 2041 |
+
"epoch": 2.7610619469026547,
|
| 2042 |
+
"grad_norm": 1.148928120100101,
|
| 2043 |
+
"kl": 0.0374755859375,
|
| 2044 |
+
"learning_rate": 6.6567193299996996e-09,
|
| 2045 |
+
"loss": 0.0,
|
| 2046 |
+
"reward": 0.7610294222831726,
|
| 2047 |
+
"reward_std": 0.485231876373291,
|
| 2048 |
+
"rewards/correct_code_reward_func": 0.3125,
|
| 2049 |
+
"rewards/len_reward_func": 0.4485294073820114,
|
| 2050 |
+
"step": 157
|
| 2051 |
+
},
|
| 2052 |
+
{
|
| 2053 |
+
"completion_length": 98.41667175292969,
|
| 2054 |
+
"epoch": 2.7787610619469025,
|
| 2055 |
+
"grad_norm": 0.2382852510613484,
|
| 2056 |
+
"kl": 0.02642822265625,
|
| 2057 |
+
"learning_rate": 5.597480503041485e-09,
|
| 2058 |
+
"loss": 0.0,
|
| 2059 |
+
"reward": 0.5000000149011612,
|
| 2060 |
+
"reward_std": 0.1178511306643486,
|
| 2061 |
+
"rewards/correct_code_reward_func": 0.0416666679084301,
|
| 2062 |
+
"rewards/len_reward_func": 0.4583333432674408,
|
| 2063 |
+
"step": 158
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"completion_length": 77.20833587646484,
|
| 2067 |
+
"epoch": 2.7964601769911503,
|
| 2068 |
+
"grad_norm": 6.517708531460766,
|
| 2069 |
+
"kl": 0.0968017578125,
|
| 2070 |
+
"learning_rate": 4.629027234912986e-09,
|
| 2071 |
+
"loss": 0.0001,
|
| 2072 |
+
"reward": 0.9473089575767517,
|
| 2073 |
+
"reward_std": 0.46372126042842865,
|
| 2074 |
+
"rewards/correct_code_reward_func": 0.5208333432674408,
|
| 2075 |
+
"rewards/len_reward_func": 0.4264756143093109,
|
| 2076 |
+
"step": 159
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"completion_length": 80.27083396911621,
|
| 2080 |
+
"epoch": 2.814159292035398,
|
| 2081 |
+
"grad_norm": 3.3787178462046454,
|
| 2082 |
+
"kl": 0.0364990234375,
|
| 2083 |
+
"learning_rate": 3.751719266468584e-09,
|
| 2084 |
+
"loss": 0.0,
|
| 2085 |
+
"reward": 0.8541666865348816,
|
| 2086 |
+
"reward_std": 0.13607724383473396,
|
| 2087 |
+
"rewards/correct_code_reward_func": 0.3541666679084301,
|
| 2088 |
+
"rewards/len_reward_func": 0.5,
|
| 2089 |
+
"step": 160
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"completion_length": 77.83333587646484,
|
| 2093 |
+
"epoch": 2.831858407079646,
|
| 2094 |
+
"grad_norm": 1.0432210255226626,
|
| 2095 |
+
"kl": 0.04864501953125,
|
| 2096 |
+
"learning_rate": 2.9658824818044325e-09,
|
| 2097 |
+
"loss": 0.0,
|
| 2098 |
+
"reward": 0.5659293830394745,
|
| 2099 |
+
"reward_std": 0.21801644191145897,
|
| 2100 |
+
"rewards/correct_code_reward_func": 0.08333333395421505,
|
| 2101 |
+
"rewards/len_reward_func": 0.4825960248708725,
|
| 2102 |
+
"step": 161
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"completion_length": 68.77083587646484,
|
| 2106 |
+
"epoch": 2.849557522123894,
|
| 2107 |
+
"grad_norm": 4.445236186593027,
|
| 2108 |
+
"kl": 0.061279296875,
|
| 2109 |
+
"learning_rate": 2.271808787206092e-09,
|
| 2110 |
+
"loss": 0.0001,
|
| 2111 |
+
"reward": 1.0243507027626038,
|
| 2112 |
+
"reward_std": 0.4195573776960373,
|
| 2113 |
+
"rewards/correct_code_reward_func": 0.6041666865348816,
|
| 2114 |
+
"rewards/len_reward_func": 0.4201839864253998,
|
| 2115 |
+
"step": 162
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"completion_length": 86.83333587646484,
|
| 2119 |
+
"epoch": 2.8672566371681416,
|
| 2120 |
+
"grad_norm": 4.1412440697628705,
|
| 2121 |
+
"kl": 0.5286865234375,
|
| 2122 |
+
"learning_rate": 1.6697560027171543e-09,
|
| 2123 |
+
"loss": 0.0005,
|
| 2124 |
+
"reward": 0.6875000149011612,
|
| 2125 |
+
"reward_std": 0.23144195601344109,
|
| 2126 |
+
"rewards/correct_code_reward_func": 0.2083333432674408,
|
| 2127 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 2128 |
+
"step": 163
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"completion_length": 72.6875,
|
| 2132 |
+
"epoch": 2.8849557522123894,
|
| 2133 |
+
"grad_norm": 7.176170560573175,
|
| 2134 |
+
"kl": 0.04962158203125,
|
| 2135 |
+
"learning_rate": 1.1599477663696843e-09,
|
| 2136 |
+
"loss": 0.0,
|
| 2137 |
+
"reward": 0.7519232630729675,
|
| 2138 |
+
"reward_std": 0.26646677777171135,
|
| 2139 |
+
"rewards/correct_code_reward_func": 0.27083333395421505,
|
| 2140 |
+
"rewards/len_reward_func": 0.4810899347066879,
|
| 2141 |
+
"step": 164
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"completion_length": 87.64583587646484,
|
| 2145 |
+
"epoch": 2.9026548672566372,
|
| 2146 |
+
"grad_norm": 1.9523625673288243,
|
| 2147 |
+
"kl": 0.05029296875,
|
| 2148 |
+
"learning_rate": 7.425734511116998e-10,
|
| 2149 |
+
"loss": 0.0001,
|
| 2150 |
+
"reward": 0.6250000298023224,
|
| 2151 |
+
"reward_std": 0.22233545035123825,
|
| 2152 |
+
"rewards/correct_code_reward_func": 0.1250000037252903,
|
| 2153 |
+
"rewards/len_reward_func": 0.5,
|
| 2154 |
+
"step": 165
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"completion_length": 79.41666793823242,
|
| 2158 |
+
"epoch": 2.920353982300885,
|
| 2159 |
+
"grad_norm": 29.99743474150218,
|
| 2160 |
+
"kl": 0.346923828125,
|
| 2161 |
+
"learning_rate": 4.17788094463023e-10,
|
| 2162 |
+
"loss": 0.0003,
|
| 2163 |
+
"reward": 0.5346376597881317,
|
| 2164 |
+
"reward_std": 0.21488384157419205,
|
| 2165 |
+
"rewards/correct_code_reward_func": 0.06250000186264515,
|
| 2166 |
+
"rewards/len_reward_func": 0.4721376597881317,
|
| 2167 |
+
"step": 166
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"completion_length": 71.14583587646484,
|
| 2171 |
+
"epoch": 2.938053097345133,
|
| 2172 |
+
"grad_norm": 2.049279192654788,
|
| 2173 |
+
"kl": 0.037353515625,
|
| 2174 |
+
"learning_rate": 1.857123409250705e-10,
|
| 2175 |
+
"loss": 0.0,
|
| 2176 |
+
"reward": 0.8958333730697632,
|
| 2177 |
+
"reward_std": 0.24056155234575272,
|
| 2178 |
+
"rewards/correct_code_reward_func": 0.3958333432674408,
|
| 2179 |
+
"rewards/len_reward_func": 0.5,
|
| 2180 |
+
"step": 167
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"completion_length": 57.000003814697266,
|
| 2184 |
+
"epoch": 2.9557522123893807,
|
| 2185 |
+
"grad_norm": 11.198599869127536,
|
| 2186 |
+
"kl": 0.0416259765625,
|
| 2187 |
+
"learning_rate": 4.6432397166285e-11,
|
| 2188 |
+
"loss": 0.0,
|
| 2189 |
+
"reward": 0.9583333730697632,
|
| 2190 |
+
"reward_std": 0.29602527618408203,
|
| 2191 |
+
"rewards/correct_code_reward_func": 0.4791666716337204,
|
| 2192 |
+
"rewards/len_reward_func": 0.4791666716337204,
|
| 2193 |
+
"step": 168
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"completion_length": 50.625,
|
| 2197 |
+
"epoch": 2.9734513274336285,
|
| 2198 |
+
"grad_norm": 26.983307381671835,
|
| 2199 |
+
"kl": 0.271484375,
|
| 2200 |
+
"learning_rate": 0.0,
|
| 2201 |
+
"loss": 0.0003,
|
| 2202 |
+
"reward": 0.9375,
|
| 2203 |
+
"reward_std": 0.39486490190029144,
|
| 2204 |
+
"rewards/correct_code_reward_func": 0.4375,
|
| 2205 |
+
"rewards/len_reward_func": 0.5,
|
| 2206 |
+
"step": 169
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 2.9734513274336285,
|
| 2210 |
+
"step": 169,
|
| 2211 |
+
"total_flos": 0.0,
|
| 2212 |
+
"train_loss": 6.70051409522318e-05,
|
| 2213 |
+
"train_runtime": 7347.2678,
|
| 2214 |
+
"train_samples_per_second": 0.138,
|
| 2215 |
+
"train_steps_per_second": 0.023
|
| 2216 |
+
}
|
| 2217 |
+
],
|
| 2218 |
+
"logging_steps": 1,
|
| 2219 |
+
"max_steps": 169,
|
| 2220 |
+
"num_input_tokens_seen": 0,
|
| 2221 |
+
"num_train_epochs": 4,
|
| 2222 |
+
"save_steps": 50,
|
| 2223 |
+
"stateful_callbacks": {
|
| 2224 |
+
"TrainerControl": {
|
| 2225 |
+
"args": {
|
| 2226 |
+
"should_epoch_stop": false,
|
| 2227 |
+
"should_evaluate": false,
|
| 2228 |
+
"should_log": false,
|
| 2229 |
+
"should_save": true,
|
| 2230 |
+
"should_training_stop": true
|
| 2231 |
+
},
|
| 2232 |
+
"attributes": {}
|
| 2233 |
+
}
|
| 2234 |
+
},
|
| 2235 |
+
"total_flos": 0.0,
|
| 2236 |
+
"train_batch_size": 1,
|
| 2237 |
+
"trial_name": null,
|
| 2238 |
+
"trial_params": null
|
| 2239 |
+
}
|