HG-CRC enforces simultaneous per-group risk guarantees in LLM selective prediction via Bonferroni correction over hierarchy nodes and a leaf-first threshold policy, reaching 0% empirical violations and WGER=0 on ARC Challenge for two models.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Hierarchical Group-Conditional Conformal Risk Control for Selective Prediction in Language Models
HG-CRC enforces simultaneous per-group risk guarantees in LLM selective prediction via Bonferroni correction over hierarchy nodes and a leaf-first threshold policy, reaching 0% empirical violations and WGER=0 on ARC Challenge for two models.