AI News

Treating an LM Ensemble as a Jury: EM-based, Label-Free Reliability Estimation and a Log‑Odds Fusion Rule

ABC2 min read
AI Evaluation