Visual Document Retrieval on InfoVQA
78.6NDCG@5ReAlign
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ReAlignRetriever Type=Multi-modal, Backbone=Qwen2026.04 | 78.6 | 80.3 | — | — | |
| ReAlignRetriever Type=Multi-modal, Backbone=Phi3V2026.04 | 76.9 | 78.6 | — | — | |
| VDocRetrieverRetriever Type=Multi-modal2026.04 | 72.7 | 74.9 | — | — | |
| NV-EmbedRetriever Type=Text-based2026.04 | 70.3 | 72.5 | — | — | |
| E5-MistralRetriever Type=Text-based2026.04 | 68.5 | 70.7 | — | — | |
| DSERetriever Type=Multi-modal2026.04 | 65.9 | 67.8 | — | — | |
| ColPaliRetriever Type=Multi-modal2026.04 | 62 | 64.1 | — | — | |
| BGERetriever Type=Text-based2026.04 | 59.4 | 61.6 | — | — | |
| VLM2VecRetriever Type=Multi-modal2026.04 | 46.8 | 50.1 | — | — | |
| SigLIP 2Retriever Type=Multi-modal2026.04 | 41.6 | 44.8 | — | — | |
| BM25Retriever Type=Text-based2026.04 | 39.9 | 42.8 | — | — | |
| E5-VRetriever Type=Multi-modal2026.04 | 38.2 | 40.6 | — | — | |
| CLIPRetriever Type=Multi-modal2026.04 | 36.1 | 38.9 | — | — | |
| BGE-LargeModality=Text-Based2026.05 | — | — | 87.98 | 70.86 | |
| BM25Modality=Text-Based2026.05 | — | — | 83.19 | 69.03 | |
| ColPaliModality=Visual-Based2026.05 | — | — | 94.82 | 81.92 | |
| DSEModality=Visual-Based2026.05 | — | — | 97.07 | 84.96 | |
| DSEModality=Hybrid2026.05 | — | — | 96.58 | 85.39 | |
| GTRModality=Text-Based2026.05 | — | — | 84.95 | 67.82 | |
| MiniCPMModality=Text-Based2026.05 | — | — | 94.67 | 82.34 | |
| NV-EmbedModality=Text-Based2026.05 | — | — | 91.84 | 77.05 | |
| SigLIPModality=Visual-Based2026.05 | — | — | 68.08 | 47.4 | |
| UnveilModality=Visual-Textual2026.05 | — | — | 97.26 | 86.19 | |
| UnveilModality=Visual-Only2026.05 | — | — | 97.61 | 86.39 | |
| VisRAGModality=Visual-Based2026.05 | — | — | 96.33 | 85.53 | |
| VisRAGModality=Hybrid2026.05 | — | — | 96.63 | 85.7 |