List your company
ResearchFormerly WizWand

Text-to-Image Retrieval on SV-1k

98.6R@1TULIP
48.6861.6474.687.56Dec 4, 2024Feb 4, 2025Apr 8, 2025Jun 9, 2025Aug 11, 2025Oct 12, 2025Dec 14, 2025
Updated 8mo ago

Evaluation results

MethodLinks
TULIPData=WIT-400M + ShareGPT4V-1M2025.1298.6
Smart-CLIPData=WIT-400M + ShareGPT4V-1M2025.1298.1
FLAIRData=30M, Zero-shot=true, Vision Encoder=ViT-B/162024.1298
FLAIRData=DreamLIP 30M2025.1298
FLAIRData=15M, Zero-shot=true, Vision Encoder=ViT-B/162024.1296.7
FLAIRData=12M, Zero-shot=true, Vision Encoder=ViT-B/162024.1295.1
FG-CLIP (w/ hard negatives)Data=WIT-400M + LAION-1.6B + GRIT-12M + 40M reg. + 10M hard negatives2025.1294.9
β-CLIP (BCE)Data=WIT-400M + ShareGPT4V-1M, Objective=BCE2025.1294.4
β-CLIP (CE)Data=WIT-400M + ShareGPT4V-1M, Objective=CE2025.1293.7
Long-CLIPData=WIT-400M + ShareGPT4V-1M2025.1293.4
FLAIRData=3M, Zero-shot=true, Vision Encoder=ViT-B/162024.1289.7
FLAIRData=30M, Zero-shot=true, Vision Encoder=ViT-B/162024.1289.4
FineLIPData=WIT-400M + ShareGPT4V-1M2025.1289.3
OpenCLIPData=2B, Zero-shot=true, Vision Encoder=ViT-B/162024.1287.7
Long-CLIPData=400M, Zero-shot=true, Vision Encoder=ViT-B/162024.1287.4
LoTLIPData=100M, Zero-shot=true, Vision Encoder=ViT-B/162024.1286.8
FLAIRData=15M, Zero-shot=true, Vision Encoder=ViT-B/162024.1286.8
LoTLIPData=Mixed-100M2025.1286.8
EVA-CLIPData=Merged-2B2025.1285.5
ALIGNData=700M, Zero-shot=true, Vision Encoder=ViT-B/162024.1285.3
LoTLIPData=100M, Zero-shot=true, Vision Encoder=ViT-B/162024.1283.7
SigLIPData=10B, Zero-shot=true, Vision Encoder=ViT-B/162024.1283.4
FLAIRData=12M, Zero-shot=true, Vision Encoder=ViT-B/162024.1283.4
SigLIPData=WebLI-1B2025.1283.4
LiTData=100M, Zero-shot=true, Vision Encoder=ViT-B/162024.1280
CLIPData=WIT-400M2025.1279.6
Fine-CLIPData=CC2.5M + 10.4M reg.2025.1273.3
FLAIRData=3M, Zero-shot=true, Vision Encoder=ViT-B/162024.1270.6
OpenCLIPData=2B, Zero-shot=true, Vision Encoder=ViT-B/162024.1266.8
SigLIPData=10B, Zero-shot=true, Vision Encoder=ViT-B/162024.1263
ALIGNData=700M, Zero-shot=true, Vision Encoder=ViT-B/162024.1262.7
Long-CLIPData=400M, Zero-shot=true, Vision Encoder=ViT-B/162024.1262
LiTData=100M, Zero-shot=true, Vision Encoder=ViT-B/162024.1250.6