debajyotidasgupta/ais-eval-results-uqheads
Updated • 3
EU-Halt — Evaluation Datasets
Per-probe AUROC + paired-bootstrap CIs. v3-mahafixed is the canonical headline grid; others are baseline head-to-heads.
Per-probe AUROC + paired-bootstrap CIs. v3-mahafixed is the canonical headline grid; others are baseline head-to-heads.