Formalizes calibration for label ranking via a hierarchy of notions with implication proofs and reports empirical miscalibration in models plus correlation with but distinction from accuracy in RLHF.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Calibrated Preference Learning: The Case of Label Ranking
Formalizes calibration for label ranking via a hierarchy of notions with implication proofs and reports empirical miscalibration in models plus correlation with but distinction from accuracy in RLHF.