AI RESEARCH
A Unified Graph Language Model for Multi-Domain Multi-Task Graph Alignment Instruction Tuning
arXiv CS.LG
•
ArXi:2605.12197v1 Announce Type: new Leveraging Graph Neural Networks (GNNs) as graph encoders and aligning the resulting representations with Large Language Models (LLMs) through alignment instruction tuning has become a mainstream paradigm for constructing Graph Language Models (GLMs), combining the generalization ability of LLMs with the structural modeling capacity of GNNs.