2025 ICML ICML 2025

GIVE: Structured Reasoning of Large Language Models with Knowledge Graph Inspired Veracity Extrapolation