1 paper
Zihang Li, Rui Zhou, Yingcheng Shi +8
When a large language model under reinforcement learning commits a wrong reasoning step early in a trajectory, standard algorithms force it to keep generating until the maximum hor…