作者位次 #5/6ICLR 2026Poster· Deep Learning / Algorithms
Reasoning Models Can be Accurately Pruned Via Chain-of-Thought Reconstruction
Ryan Lucas · Kayhan Behdin · Zhipeng Wang · Qingquan Song · shao tang · Rahul Mazumder
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。
Ryan Lucas · Kayhan Behdin · Zhipeng Wang · Qingquan Song · shao tang · Rahul Mazumder
XUANZHAO DONG · Wenhui Zhu · Xiwen Chen · Zhipeng Wang · Peijie Qiu · shao tang · Xin Li · Yalin Wang
Wei-Rui Chen · Vignesh Kothapalli · Ata Fatahibaarzi · Hejian Sang · shao tang · Qingquan Song · Zhipeng Wang · Muhammad Abdul-Mageed
Distilling the capabilities from a large reasoning model (LRM) to a smaller student model often involves training on substantial amounts of reasoning data. However, knowledge distillation (KD) over lengthy sequences with prompt (P), chain-of-thought (CoT), and answer (A) sections makes the proces...