DASE is an adaptive stopping rule for LLM ensembles that creates large accuracy gaps between early-commit and fallback answers on GPQA and AIME benchmarks.
and Shadlen, Michael N
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Adaptive Consensus in LLM Ensembles via Sequential Evidence Accumulation: Automatic Budget Identification and Calibrated Commit Signals
DASE is an adaptive stopping rule for LLM ensembles that creates large accuracy gaps between early-commit and fallback answers on GPQA and AIME benchmarks.