The Text Uncanny Valley: Non-Monotonic Performance Degradation in LLM Information Retrieval

概要

arXiv:2605.07186v1 Announce Type: cross Abstract: Existing Large Language Model (LLM) benchmarks primarily focus on syntactically correct inputs, leaving a significant gap in evaluation on imperfect text. In this work, we study how word-boundary corruption affects how LLMs detect targeted informati…