File size: 2,301 Bytes
d6ea141 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 |
---
license: llama3.2
datasets:
- OctoThinker/MegaMath-Web-Pro-Max
- LLM360/MegaMath
language:
- en
base_model:
- meta-llama/Llama-3.2-1B
pipeline_tag: text-generation
---
# [OctoThinker: Mid-training Incentivizes Reinforcement Learning Scaling](https://arxiv.org/abs/2506.20512)
## OctoThinker-1B-Short-Zero
The OctoThinker family is built on carefully studied mid-training insights, starting from the Llama-3 family, to create a reinforcement learning–friendly base language model.
OctoThinker-1B-Short-Zero is trained using the R1-Zero-style reinforcement learning technique, starting from OctoThinker-1B-Short-Base without any supervised fine-tuning (SFT).
### Training Recipe for OctoThinker-1B-Short-Base
<div style="display: flex; justify-content: left; gap: 20px;">
<img src="https://cdn-uploads.huggingface.co/production/uploads/62cbeb2d72dfd24b86bdf977/nQEwjHnW88XIThK8jHAA3.png" alt="Data Pipeline" style="width:90%;">
</div>
### Evaluation Results of OctoThinker-1B-Base Series
Note that we adopt the few-shot prompting evaluation for these base language models.
<div style="display: flex; justify-content: left; gap: 20px;">
<img src="https://cdn-uploads.huggingface.co/production/uploads/62cbeb2d72dfd24b86bdf977/3n1cnG81wLjjPwzyMyQ-o.png" alt="Data Pipeline" style="width:80%;">
</div>
### RL Training Dynamics of OctoThinker-1B-Zero Series
<div style="display: flex; justify-content: left; gap: 20px;">
<img src="https://cdn-uploads.huggingface.co/production/uploads/62cbeb2d72dfd24b86bdf977/xoiYwO-fsOCdPTKsASJNK.png" alt="Data Pipeline" style="width:80%;">
</div>
### More about OctoThinker
<div style="display: flex; justify-content: left; gap: 20px;">
<img src="https://cdn-uploads.huggingface.co/production/uploads/62cbeb2d72dfd24b86bdf977/bn85CEB_DW6azJ7KJp11Q.png" alt="Data Pipeline" style="width:100%;">
</div>
## Citation
Check out our [paper](https://arxiv.org/abs/2506.20512) for more details. If you use our models, and datasets or find our work useful, please cite
```
@article{wang2025octothinker,
title={OctoThinker: Mid-training Incentivizes Reinforcement Learning Scaling},
author={Wang, Zengzhi and Zhou, Fan and Li, Xuefeng and Liu, Pengfei},
year={2025},
journal={arXiv preprint arXiv:2506.20512},
note={Preprint}
}
```
|