MUSE uses Jensen-Shannon divergence to pick a coherent subset of LLM predictions and averages them, improving binary-prediction calibration in several tasks.
InProceedings of the 2024 Conference on Empiri- cal Methods in Natural Language Processing, pages 21635–21645
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Simple Yet Effective: An Information-Theoretic Approach to Multi-LLM Uncertainty Quantification
MUSE uses Jensen-Shannon divergence to pick a coherent subset of LLM predictions and averages them, improving binary-prediction calibration in several tasks.