Link Prediction on CDs
93.5AccuracyMario-8B
Evaluation results
| Method | Links | |
|---|---|---|
| Mario-8BModality=Text+Image, Training Protocol=Mix-Training2026.03 | 93.5 | |
| LLaGAModality=Text-only, Training Protocol=Mix-Training2026.03 | 90.32 | |
| MLaGAModel Type=Ours2025.06 | 90.2 | |
| MLaGAModality=Text+Image, Training Protocol=Mix-Training2026.03 | 86.53 | |
| LLaGA-NDModel Type=Graph LLM, Training Setting=Cross-Task2025.06 | 84.15 | |
| GraphPrompterModality=Text-only, Training Protocol=Mix-Training2026.03 | 83.2 | |
| LLaGA-NDModel Type=Graph LLM, Training Setting=Cross-Domain2025.06 | 82.69 | |
| LLaGA-NDModel Type=Graph LLM, Training Setting=Single Focus2025.06 | 81.36 | |
| Unigraph2Model Type=Multimodal Graph Model2025.06 | 81 | |
| Qwen2.5-VL-7BModel Type=MLLM2025.06 | 80.9 | |
| GraphPrompterModel Type=Graph LLM2025.06 | 79.83 | |
| MLPModel Type=GNN-based2025.06 | 78 | |
| LLaMA3-8BModality=Text-only, Training Protocol=Mix-Training2026.03 | 74.1 | |
| GCNModality=Text-only, Training Protocol=Mix-Training2026.03 | 72.93 | |
| GCNModel Type=GNN-based2025.06 | 71.07 | |
| GraphGPTModel Type=Graph LLM2025.06 | 68.95 | |
| GCNModality=Text+Image, Training Protocol=Mix-Training2026.03 | 68.6 | |
| SAGEModality=Text+Image, Training Protocol=Mix-Training2026.03 | 68.43 | |
| GraphSAGEModel Type=GNN-based2025.06 | 68.37 | |
| MGATModel Type=Multimodal Graph Model2025.06 | 67.66 | |
| GCNModality=Image-only, Training Protocol=Mix-Training2026.03 | 67.07 | |
| MMGCNModel Type=Multimodal Graph Model2025.06 | 65.22 | |
| SAGEModality=Image-only, Training Protocol=Mix-Training2026.03 | 62.93 | |
| SAGEModality=Text-only, Training Protocol=Mix-Training2026.03 | 62.03 | |
| LLaVA1.5-Next-7BModel Type=MLLM2025.06 | 51.8 |