Image-Text Retrieval on DocVQA
53.2R@1ColQwen2
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ColQwen2Architecture=Multi-vector, Backbone=ColQwen2, Compression=None2026.07 | 53.2 | 67.9 | 63 | |
| ColPaliArchitecture=Multi-vector, Backbone=ColPali, Compression=None2026.07 | 51 | 66.5 | 60.7 | |
| H-PoolArchitecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Pooling2026.07 | 50.1 | 64.8 | 59.3 | |
| SaMer (w/ FT)Architecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Merging, Adapted=true2026.07 | 48.8 | 62.5 | 58.1 | |
| SaMer (w/o FT)Architecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Merging, Adapted=false2026.07 | 47.2 | 63.9 | 57.3 | |
| SaMer (w/ FT)Architecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Merging, Adapted=true2026.07 | 45.7 | 60.5 | 55.6 | |
| SAPArchitecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Structural anchor pruning2026.07 | 45.5 | 59.4 | 54.7 | |
| H-PoolArchitecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Pooling2026.07 | 45 | 61.9 | 55.5 | |
| SaMer (w/o FT)Architecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Merging, Adapted=false2026.07 | 44.6 | 62.1 | 55.6 | |
| GMEArchitecture=VLM-based2026.07 | 41.5 | 60.3 | 53.3 | |
| SAPArchitecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Structural anchor pruning2026.07 | 40.4 | 55.2 | 50.2 | |
| HPCArchitecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Attention-based pruning2026.07 | 39.9 | 55.2 | 49.6 | |
| VisRAGArchitecture=VLM-based2026.07 | 34.8 | 51.2 | 45.6 | |
| HPCArchitecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Attention-based pruning2026.07 | 26.4 | 39.7 | 34.9 | |
| SigLIP2Architecture=Single-vector2026.07 | 15.5 | 29.1 | 24.3 | |
| OpenCLIPArchitecture=Single-vector2026.07 | 15.3 | 26.8 | 23.1 | |
| EVA-CLIPArchitecture=Single-vector2026.07 | 12.9 | 25.1 | 19.8 | |
| DFN-CLIPArchitecture=Single-vector2026.07 | 12.4 | 27.7 | 21.8 | |
| CLIPArchitecture=Single-vector2026.07 | 6.7 | 13.8 | 11.8 | |
| MetaCLIPArchitecture=Single-vector2026.07 | 5.4 | 14.4 | 11.5 | |
| VLM2Vec-v2Architecture=VLM-based2026.07 | 4 | 10.4 | 8.5 |