Supervised uncertainty probes for LLMs show poor robustness under distribution shift, with middle-layer representations and multi-token aggregation proving more reliable than final-layer or single-token features.
InProceedings of the 63rd Annual Meet- ing of the Association for Computational Linguistics (V olume 1: Long Papers), pages 6089–6104, Vienna, Austria
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Hidden Failures in Robustness: Why Supervised Uncertainty Quantification Needs Better Evaluation
Supervised uncertainty probes for LLMs show poor robustness under distribution shift, with middle-layer representations and multi-token aggregation proving more reliable than final-layer or single-token features.