一作ACL 2026Poster· Natural Language Processing
Less is More: Improving LLM Reasoning with Minimal Test-Time Intervention
Zhen Yang · Mingyang Zhang · Feng Chen · Ganggui Ding · Liang Hou · Xin Tao · Ying-Cong Chen
摘要
Recent progress in large language models (LLMs) has focused on test-time scaling to improve reasoning via increased inference computation, but often at the cost of efficiency. We revisit test-time behavior and uncover a simple yet underexplored phenomenon: reasoning uncertainty is highly localize...