一作ICML 2026Poster· Reinforcement Learning / Everything Else
SWE-MiniSandbox: Container-Free Reinforcement Learning for Building Software Engineering Agents
Danlong Yuan · Wei Wu · Zhengren Wang · Xueliang Zhao · Huishuai Zhang · Dongyan Zhao
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。
Danlong Yuan · Wei Wu · Zhengren Wang · Xueliang Zhao · Huishuai Zhang · Dongyan Zhao
Danlong Yuan · Tian Xie · Shaohan Huang · Huishuai Zhang · Zhuocheng Gong · Chong Luo · Furu Wei · Dongyan Zhao
Long-reasoning models achieve strong accuracy on complex reasoning tasks, but their extended reasoning trajectories incur substantial memory and latency costs. Several existing shortening methods rely on additional supervision or multi-stage post-training, which primarily reduces inference length...