WRSWorld Research StateContribute
WORLD RESEARCH STATE

Language Models (Mostly) Know What They Know

Contribution grade

LLM factuality and grounded generation · Human-designed research method or resource

Reference score135,000Provisional editorial estimate of this exact contribution; its grade is assessed separately.

What changed

Studies answer-conditioned P(True) and trains a separate P(IK) value head from sampled correctness labels to estimate answerability before selecting a particular answer.

Scientific scope

Studies answer-conditioned P(True) and trains a separate P(IK) value head from sampled correctness labels to estimate answerability before selecting a particular answer. Source-reported calibration and partial cross-task generalization, not access to ground truth or guaranteed honesty. P(IK) finetunes the whole model plus head; out-of-distribution calibration remains weak. Stars mark correspondence, not equal contribution.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Source-reported calibration and partial cross-task generalization, not access to ground truth or guaranteed honesty. P(IK) finetunes the whole model plus head; out-of-distribution calibration remains weak. Stars mark correspondence, not equal contribution.

Contributors

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction