AI RESEARCH

Exploring Pass-Rate Reward in Reinforcement Learning for Code Generation

arXiv CS.LG

ArXi:2605.02944v1 Announce Type: new Reinforcement learning (RL) from unit-test feedback has become a standard post-