一作ICLR 2026Poster· Reinforcement Learning / Everything Else
Strategic Scaling of Test-Time Compute: A Bandit Learning Approach
Bowen Zuo · Yinglun Zhu
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Bowen Zuo · Yinglun Zhu
Bowen Zuo · Dongruo Zhou · Yinglun Zhu
While scaling test-time compute can substantially improve model performance, existing approaches either rely on static compute allocation or sample from fixed generation distributions.In this work, we introduce a test-time compute allocation framework that jointly adapts where computation is spen...