作者位次 #4/5ACL 2026Poster· Natural Language Processing
Step- GRPO : Internalizing Dynamic Early Exit for Efficient Reasoning
Benteng Chen · Weida Wang · Shufei Zhang · Mingbao Lin · Min Zhang
摘要
Large reasoning models that use long chain-of-thought excel at problem-solving yet waste compute on redundant checks. Curbing this overthinking is hard: training-time length penalties can cripple ability, while inference-time early-exit adds system overhead. To bridge this gap, we propose Step-GR...