Text-based Retrieval on MBE benchmark
16.92Recall@1GME
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GMEevaluation_mode=Zero-shot2025.08 | 16.92 | 47.22 | 58.84 | |
| MOONevaluation_mode=Zero-shot2025.08 | 16.84 | 47.46 | 59.46 | |
| InternVL3-2Bevaluation_mode=Zero-shot, tuning=SFT2025.08 | 16.04 | 47.36 | 60.11 | |
| Qwen2.5-VL-3Bevaluation_mode=Zero-shot, tuning=SFT2025.08 | 15.07 | 46.07 | 59.79 | |
| MM-Embedevaluation_mode=Zero-shot2025.08 | 12.83 | 36.25 | 46.42 | |
| SigLIP2evaluation_mode=Zero-shot2025.08 | 3.64 | 10.11 | 14.22 | |
| FashionCLIPevaluation_mode=Zero-shot2025.08 | 2.78 | 7.53 | 10.32 | |
| CLIPevaluation_mode=Zero-shot2025.08 | 1.44 | 3.9 | 5.4 | |
| Qwen2.5-VL-3Bevaluation_mode=Zero-shot2025.08 | 0.46 | 1.53 | 2.42 | |
| InternVL3-2Bevaluation_mode=Zero-shot2025.08 | 0.06 | 0.22 | 0.41 |