作者位次 #4/5ICLR 2026Poster
Taming Polysemanticity in LLMs: Theory-Grounded Feature Recovery via Sparse Autoencoders
Siyu Chen · Heejune Sheen · Xuyuan Xiong · Tianhao Wang · Zhuoran Yang
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。
Siyu Chen · Heejune Sheen · Xuyuan Xiong · Tianhao Wang · Zhuoran Yang
Shuo Xie · Tianhao Wang · Beining Wu · Zhiyuan Li
Xinhe Wang · Jin Huang · Xingjian Zhang · Tianhao Wang · Jiaqi W. Ma
Reasoning benchmarks such as the Abstraction and Reasoning Corpus (ARC) and ARC-AGI are widely used to assess progress in artificial intelligence and are often interpreted as probes of core, so-called “fluid” reasoning abilities. Despite their apparent simplicity for humans, these tasks remain ch...