Visual Document Retrieval on ArXivQA
76.2NDCG@5ReAlign
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ReAlignRetriever Type=Multi-modal, Backbone=Qwen2026.04 | 76.2 | 77.3 | |
| ReAlignRetriever Type=Multi-modal, Backbone=Phi3V2026.04 | 70.3 | 71.8 | |
| VDocRetrieverRetriever Type=Multi-modal2026.04 | 69.6 | 70.8 | |
| DSERetriever Type=Multi-modal2026.04 | 62.7 | 64 | |
| NV-EmbedRetriever Type=Text-based2026.04 | 42.6 | 44.1 | |
| E5-MistralRetriever Type=Text-based2026.04 | 42.1 | 43.3 | |
| E5-VRetriever Type=Multi-modal2026.04 | 40.9 | 42.9 | |
| VLM2VecRetriever Type=Multi-modal2026.04 | 39.5 | 42 | |
| SigLIP 2Retriever Type=Multi-modal2026.04 | 37.2 | 39.3 | |
| BM25Retriever Type=Text-based2026.04 | 33.6 | 34.9 | |
| BGERetriever Type=Text-based2026.04 | 32.4 | 33.2 | |
| CLIPRetriever Type=Multi-modal2026.04 | 22.6 | 23.5 |