InFACT trains LLMs with hierarchical informativeness rewards plus abstention, improving factual precision on QA benchmarks while largely preserving recall.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
InFact: Informativeness Alignment for Improved LLM Factuality
InFACT trains LLMs with hierarchical informativeness rewards plus abstention, improving factual precision on QA benchmarks while largely preserving recall.