智能AI None 不要破坏你的特征：为什么 CenterLoss 会损害 OOD 检测，而多尺度 Mahalanobis 会获胜 21493v1 Announce Type: new Abstract: The ability to detect out-of-distribution (OOD) inputs is fundamental to safe deplo OOD that for 2026-05-23 Rahul D Ray
智能AI None 法学硕士分布外对齐失败的基准测试和改进监视器 arXiv:2605.21602v1 Announce Type: new Abstract: Many safety and alignment failures of large language models (LLMs) occur OOD that and 2026-05-23 Dylan Feng, Pragya Srivastava, Cassidy Laidlaw