Call Me When Necessary: LLMs can Efficiently and Faithfully Reason over Structured Environments

arXiv:2403.08593v1 Announce Type: new
Abstract: Large Language Models (LLMs) have shown potential in reasoning over structured environments, e.g., knowledge graph and table. Such tasks typically require multi-hop reasoning, i.e., match natural language utterance with instances in the environment. Previous methods leverage LLMs to incrementally build a reasoning path, where the LLMs either invoke tools or pick up schemas by step-by-step interacting with the environment. We propose Reasoning-Path-Editing (Readi), a novel framework where LLMs can efficiently and faithfully reason over structured …

abstract arxiv build call cs.ai cs.cl environment environments graph instances knowledge knowledge graph language language models large language large language models llms match natural natural language reason reasoning table tasks the environment type