New framework enables LLMs to explore graph neighborhoods for better reasoning

A new framework called Call Neighbours Yourself (CNY) has been proposed for large language models (LLMs) to reason over text-attributed graphs. Unlike existing methods that use a fixed set of neighbors, CNY allows LLMs to proactively explore graph neighborhoods by learning when to expand candidate neighbors for additional evidence. This approach addresses the challenge of delayed credit in neighbor exploration through destination-conditioned on-policy self-distillation, which converts changes in action preference into training signals. Experiments show that CNY outperforms fixed-context baselines on standard graph reasoning benchmarks and demonstrates transferability to unseen graphs and different tasks. AI

IMPACT This framework could improve LLM performance on tasks requiring reasoning over complex, interconnected data.

RANK_REASON The cluster contains a research paper detailing a new framework for LLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• Call Neighbours Yourself
• CatalyzeX Code Finder for Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• Influence Flower
• large language models
• ScienceCast
• text-attributed graphs

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →