ERR+：基于序列熵消解的 LLM 高效推理框架

AI 摘要

ERR+ 提出一种两阶段 RLVR 框架，基于 token 级熵降奖励（ERR）和鲁棒相对效率奖励（RRER）优化 LLM 推理。实验表明正确推理轨迹熵降更频繁，ERR+ 在五个数据集上提升准确率并缩短响应长度。

正文节选

ERR+: Sequential Entropy Resolution for Efficient and Decisive LLM Reasoning Abstract Large reasoning models achieve strong performance on complex tasks by generating extended chain-of-thought (CoT) traces via reinforcement learning with verifiable rewards (RLVR). While current RLVR methods have achieved strong results with correctness-based reward signals, they provide limited guidance on the quality of the reasoning process itself, leaving the internal reasoning structure largely unoptimized.

发布时间：2026-09-01 12:00

抓取时间：2026-09-01 12:13

来源机构：arXiv