Visual Document Retrieval on SlideVQA
82.5NDCG@5ReAlign
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ReAlignRetriever Type=Multi-modal, Backbone=Qwen2026.04 | 82.5 | 83.9 | |
| ColPaliRetriever Type=Multi-modal2026.04 | 79 | 80.6 | |
| ReAlignRetriever Type=Multi-modal, Backbone=Phi3V2026.04 | 77.5 | 79.5 | |
| VDocRetrieverRetriever Type=Multi-modal2026.04 | 77.2 | 78.8 | |
| NV-EmbedRetriever Type=Text-based2026.04 | 76.4 | 78.5 | |
| E5-MistralRetriever Type=Text-based2026.04 | 75.7 | 77.4 | |
| DSERetriever Type=Multi-modal2026.04 | 66.8 | 69.1 | |
| BGERetriever Type=Text-based2026.04 | 62.4 | 64.6 | |
| E5-VRetriever Type=Multi-modal2026.04 | 59 | 62 | |
| BM25Retriever Type=Text-based2026.04 | 49.8 | 52.1 | |
| VLM2VecRetriever Type=Multi-modal2026.04 | 44.7 | 48.1 | |
| SigLIP 2Retriever Type=Multi-modal2026.04 | 40.9 | 43.6 | |
| CLIPRetriever Type=Multi-modal2026.04 | 32.2 | 35 |