List your company

Image-Text Retrieval on Flickr30K (R@1, R@5, nDCG@10)

82.4Recall@1SaMer (w/ FT)
40.07251.06162.0573.039Jul 6, 2026

Evaluation results

MethodLinks
SaMer (w/ FT)Architecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Merging, Adapted=true2026.0782.496.390.9
EVA-CLIPArchitecture=Single-vector2026.0779.994.789.2
SigLIP2Architecture=Single-vector2026.0779.894.288.8
DFN-CLIPArchitecture=Single-vector2026.0779.694.989
SaMer (w/ FT)Architecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Merging, Adapted=true2026.0779.394.788.7
ColPaliArchitecture=Multi-vector, Backbone=ColPali, Compression=None2026.077792.986.9
OpenCLIPArchitecture=Single-vector2026.0776.593.587
GMEArchitecture=VLM-based2026.0774.691.885.2
H-PoolArchitecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Pooling2026.0773.791.484.7
ColQwen2Architecture=Multi-vector, Backbone=ColQwen2, Compression=None2026.0773.691.384.7
SaMer (w/o FT)Architecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Merging, Adapted=false2026.0773.591.184.5
H-PoolArchitecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Pooling2026.0771.890.883.6
SaMer (w/o FT)Architecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Merging, Adapted=false2026.0771.390.483.3
VisRAGArchitecture=VLM-based2026.0769.69082.2
SAPArchitecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Structural anchor pruning2026.0769.389.982.2
SAPArchitecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Structural anchor pruning2026.0768.388.680.8
HPCArchitecture=Multi-vector, Backbone=ColQwen2, Token Compression (K)=64, Compression Strategy=Attention-based pruning2026.0767.188.780.4
MetaCLIPArchitecture=Single-vector2026.0762.885.877.2
CLIPArchitecture=Single-vector2026.0758.282.973.7
HPCArchitecture=Multi-vector, Backbone=ColPali, Token Compression (K)=64, Compression Strategy=Attention-based pruning2026.075477.869
VLM2Vec-v2Architecture=VLM-based2026.0741.768.859