作者位次 #3/6ICML 2026Poster· General Machine Learning / Supervised Learning
Hard Labels In! Rethinking the Role of Hard Labels in Mitigating Local Semantic Drift
Jiacheng Cui · Bingkui Tong · Xinyue Bi · Xiaohan Zhao · Jiacheng Liu · Zhiqiang Shen
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。
Jiacheng Cui · Bingkui Tong · Xinyue Bi · Xiaohan Zhao · Jiacheng Liu · Zhiqiang Shen
Yaxin Luo · Jiacheng Cui · Xiaohan Zhao · Xinyi Shang · Jiacheng Liu · Xinyue Bi · Zhaoyi Li · Zhiqiang Shen
The pretraining data mixture of Large Language Models (LLMs) constitutes their “digital DNA”, shaping model behaviors, capabilities, and failure modes. Yet this composition is rarely disclosed, making post-hoc auditing of data combination or provenance difficult. In this work, we formalize \textb...