KL divergence of attention heads from uniform distribution predicts LLM answer correctness across datasets and model families.
and Salakhutdinov, Ruslan and Manning, Christopher D
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
-
Detecting Hallucinations in Large Language Models via Internal Attention Divergence Signals
KL divergence of attention heads from uniform distribution predicts LLM answer correctness across datasets and model families.