Models flip between correct and incorrect answers on over 23% of questions under meaning-preserving paraphrases, so single-prompt accuracy overstates reliable knowledge.
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
-
Same Question, Different Answers: Evaluating LLM Reliability Beyond Accuracy
Models flip between correct and incorrect answers on over 23% of questions under meaning-preserving paraphrases, so single-prompt accuracy overstates reliable knowledge.