LLMs as Zero-shot Graph Learners: Alignment of GNN Representations with LLM Token Embeddings

Oct-9-2025, 18:10:17 GMT–Neural Information Processing Systems 

GNN's representations into a fixed number of graph token embeddings without