README.md

---
title: README
emoji: 👀
colorFrom: gray
colorTo: indigo
sdk: static
pinned: false
---

🐙 OctoThinker is led by [GAIR](https://huggingface.co/GAIR)

🎯 Our Goal: To reshape the pre-training trajectory so models scale better under RL.


Check our [*technical report*](https://arxiv.org/abs/2506.20512v1) for more details!

![image/png](https://cdn-uploads.huggingface.co/production/uploads/62cbeb2d72dfd24b86bdf977/Ygt1ofrt-eXBQyUT6D-6H.png)