Cross-modal Retrieval Suite MS-COCO & Flickr30K
79.4Average ScoreTuringViT-24L
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TuringViT-24LResolution=dyn., Pretraining data=0.85B, Evaluation Protocol=Zero-shot2026.06 | 79.4 | 59.4 | 76 | 86 | 96.3 | |
| TuringViT-18LResolution=dyn., Pretraining data=0.85B, Evaluation Protocol=Zero-shot2026.06 | 78.9 | 58.6 | 75.9 | 85.7 | 95.3 | |
| SigLIP2-LResolution=384, Pretraining data=10B, Evaluation Protocol=Zero-shot2026.06 | 76.7 | 55.3 | 71.4 | 85 | 95.2 | |
| MobileCLIP2-LResolution=224, Pretraining data=1.9B, Evaluation Protocol=Zero-shot2026.06 | 72.5 | 51.6 | 69 | 77.2 | 92 |