Predictable Failure in Multi-Hop Retrieval: Score-Distributional Confidence Scoring and Abstention
RegimeAbstain uses structural retrieval features to abstain on predictable multi-hop failures, cutting confident wrong answers.
1 stories, newest first · one card per story
RegimeAbstain uses structural retrieval features to abstain on predictable multi-hop failures, cutting confident wrong answers.
Today 1.6K readers · 2.5K views · 779 reads · all time 189K views