作者位次 #6/9ICLR 2026Poster· Reinforcement Learning / Everything Else
SWE-RM: Execution-free Feedback for Software Engineering Agents
KaShun SHUM · Binyuan Hui · Jiawei Chen · Lei Zhang · X. W. · Jiaxi Yang · Yuzhen Huang · Junyang Lin · Junxian He
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。
KaShun SHUM · Binyuan Hui · Jiawei Chen · Lei Zhang · X. W. · Jiaxi Yang · Yuzhen Huang · Junyang Lin · Junxian He
Jiajun Zhang · Zeyu Cui · Jiaxi Yang · Lei Zhang · Yuheng Jing · Zeyao Ma · Tianyi Bai · Zilei Wang · Qiang Liu · Liang Wang · Binyuan Hui · Junyang Lin
The dominant Fill-in-the-Middle (FIM) paradigm for code completion is constrained by its rigid inability to correct contextual errors and reliance on unaligned, insecure Base models. While Chat LLMs offer safety and Agentic workflows provide flexibility, they suffer from performance degradation a...