Pass@k Training for Adaptively Balancing Exploration and Exploitation of Large Reasoning Models

Devs

Pass@k Training for Adaptively Balancing Exploration and Exploitation of Large Reasoning Models | Read Paper on Bytez