一作ACL 2026Poster· Natural Language Processing
Think Better, Not Longer: Token-Level Marginal Utility for Efficient Reasoning in Large Reasoning Models
Jiawei Li · Yang Gao · Huashan Sun · Chong Feng
摘要
While Large Reasoning Models (LRMs) have demonstrated remarkable capabilities through explicit Chain-of-Thought (CoT) generation, they frequently suffer from “overthinking”. In this work, we bridge this gap by introducing Token-level Marginal Utility , which quantifies the per-token log-probabili...