OneBench
Provable Limits and Certified Deferral for Verbalized Uncertainty in Small Language Models | OneBench: AI Insights