作者位次 #6/11ACL 2026Findings· Findings of ACL
AD a PT : Token-Level Decoupling for Efficient Large Reasoning Models
tingyun li · Zishang Jiang · Jinyi Han · Xinyi Wang · Sihang Jiang · Han Xia · Zhaoqian Dai · Ma Shuguang · F. Richard Yu · Jiaqing Liang · Yanghua Xiao
摘要
Large reasoning models rely on long chain-of-thought to achieve strong performance, but applying such reasoning uniformly incurs high computational cost. Existing efficiency-oriented methods attempt to shorten or mix reasoning strategies, yet often degrade reasoning capability. We identify the ro...