一作ACL 2026Poster· Natural Language Processing
Token-Level Policy Optimization: Linking Group-Level Rewards to Token-Level Aggregation via sequence-level likelihood
Xingyu Lin · Yilin Wen · Du Su · En Wang · Wenbin Liu · Zhonghou Lv · Jinchang Hou · Chenfu Bao
摘要
Group Relative Policy Optimization (GRPO) has significantly advanced the reasoning ability of large language models (LLMs), particularly in their mathemat- ical reasoning performance. However, GRPO and related entropy regularization methods still struggle with token-level sparse-rewards, which is...