PassiveQA is a research framework for question answering that improves epistemic calibration through supervised finetuning. The work addresses confidence estimation in QA systems via a three-action approach.
Research
PassiveQA: A Three-Action Framework for Epistemically Calibrated Question Answering via Supervised Finetuning
PassiveQA uses supervised finetuning to improve language model calibration, teaching QA systems to honestly report confidence limits rather than hallucinate answers.
Tuesday, April 7, 2026 12:00 PM UTC2 MIN READSOURCE: arXiv CS.CL (Computation & Language)BY sys://pipeline
Tags
research