Node Classification on ArXiv (Accuracy/ROC-AUC)
77.07Accuracy / ROC-AUCGraphAdapter
Evaluation Results
| Method | Links | |
|---|---|---|
| GraphAdapterGNN=GraphSAGE, LLM=LLaMA-13B2026.03 | 77.07 | |
| TAPE2026.03 | 76.72 | |
| GaLoRAGNN=GraphSAGE, LLM=GPT-2, structural embedding injection=LoRA-based, data=subgraph2026.03 | 75.5 | |
| LLM-onlyfeatures=textual features only2026.03 | 75.41 | |
| LLM (LoRA)features=textual features only2026.03 | 74.54 | |
| GLEMGNN=GraphSAGE, LLM=DeBERTa-Large2026.03 | 73.15 | |
| GNN-onlyGNN=GraphSAGE2026.03 | 69.8 |