04:00
2026-05-26
arxiv.org
large-language-models
DRInQ: Evaluating Conversational Implicature with Controlled Context Variation
Researchers introduced DRinQ, a benchmark designed to evaluate how well large language models understand conversational implicature—meaning suggested rather than explicitly stated—by systematically va…