Benchmarking and Improving Monitors for Out-Of-Distribution Alignment Failure in LLMs 文章

ArXiv CS.AI2026-05-26NEWSen作者: Dylan Feng, Pragya Srivastava, Anca Dragan, Cassidy Laidlaw

大语言模型

查看原文 →

Benchmarking and Improving Monitors for Out-Of-Distribution Alignment Failure in LLMs · 相关技术

相关技术

TTE ODE PLA LLM language model safety alignment Verse UAR Straight-Through Estimator SAFE Prompt Parts-of-Speech (POS) tags LMM large language models ICT HIS For FFI Effort Metric Attention ARG ANN