一作NeurIPS 2026Poster
Crafting Reversible SFT Behaviors in Large Language Models
Yuping Lin · Pengfei He · Yue XING · Yingqian Cui · Jiayuan Ding · Subhabrata Mukherjee · Hui Liu · Zhen Xiang
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。
Yuping Lin · Pengfei He · Yue XING · Yingqian Cui · Jiayuan Ding · Subhabrata Mukherjee · Hui Liu · Zhen Xiang
Zidi Xiong · Yuping Lin · Wenya Xie · Pengfei He · Zirui Liu · Jiliang Tang · Himabindu Lakkaraju · Zhen Xiang
Memory is a critical component in large language model (LLM)-based agents, enabling them to store and retrieve past executions to improve task performance over time. In this paper, we conduct an empirical study on how memory management choices impact the LLM agents’ behavior, especially their lon...
Yuping Lin · Zitao Li · Yue Xing · Pengfei He · Yingqian Cui · Yaliang Li · Bolin Ding · Jingren Zhou · Jiliang Tang
Recent studies have identified “retrieval heads” in Large Language Models (LLMs) responsible for extracting information from input contexts. However, prior works largely rely on static statistics aggregated across datasets, identifying heads that perform retrieval on average. This perspective ove...