The Future of Facts: Tracing the Factual Generation-Verification Gap 事件

The Future of Facts: Tracing the Factual Generation-Verification Gap arXiv:2605.27564v1 Announce Type: new Abstract: Language models are becoming the default interface to factual knowledge, yet they often verify outputs more reliably than they generate them. This generation-verification gap (GV-gap) underlies many recent advances in self-improvement and reasoning, but its dynamics on factual knowledge specifically remain poorly understood. We focus on the training mechanisms underlying factual