SURE-RAG aggregates pair-level claim-evidence relations into interpretable signals for selective RAG answering, reaching 0.9075 Macro-F1 on HotpotQA-RAG v3 while providing auditability and reducing unsafe answers by 37% at 30% coverage.
FActScore: Fine-grained atomic evaluation of factual precision in long form text generation,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
SURE-RAG: Sufficiency and Uncertainty-Aware Evidence Verification for Selective Retrieval-Augmented Generation
SURE-RAG aggregates pair-level claim-evidence relations into interpretable signals for selective RAG answering, reaching 0.9075 Macro-F1 on HotpotQA-RAG v3 while providing auditability and reducing unsafe answers by 37% at 30% coverage.