76 lines
2.2 KiB
Markdown
76 lines
2.2 KiB
Markdown
---
|
||
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}
|
||
}
|
||
```
|