When All Rollouts Fail: Policy-Compatible Joint Decoding for Hard-Problem Reasoning
Reinforcement Learning with Verifiable Rewards (RLVR) derives its learning signal from binary rewards via on-policy exploration and becomes a powerful paradigm for improving the reasoning of large language models. Yet it faces a learning cliff on problems beyond model's current ability: when every rollout fails, the sa...
Ze-Hao Wang, Han-Zhu Chen, Run-Quan Gui et al.
· 1 citation