Text-to-Image Retrieval on M5Product (test)
38.1Recall@1VLM2Vec + AGCL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| VLM2Vec + AGCLModel Architecture Category=LLM-based Baselines, Enhanced Component=AGCL2026.04 | 38.1 | 64.5 | 75 | |
| VLM2VecModel Architecture Category=LLM-based Baselines2026.04 | 36.4 | 63 | 73.6 | |
| LLM2CLIP + AGCLModel Architecture Category=LLM-based Baselines, Enhanced Component=AGCL2026.04 | 29.5 | 55.9 | 67.5 | |
| LLM2CLIPModel Architecture Category=LLM-based Baselines2026.04 | 27.7 | 54.3 | 66.2 | |
| BLIPModel Architecture Category=BERT-based Baselines2026.04 | 23.5 | 46.2 | 57.5 | |
| ALBEFModel Architecture Category=BERT-based Baselines2026.04 | 20.2 | 43.2 | 54.9 | |
| CLIPModel Architecture Category=BERT-based Baselines2026.04 | 19.5 | 40.9 | 52.1 | |
| FG-CLIPModel Architecture Category=BERT-based Baselines2026.04 | 19.4 | 40.2 | 51.5 | |
| Region-CLIPModel Architecture Category=BERT-based Baselines2026.04 | 19.2 | 40.3 | 51.7 | |
| VL-CLIPModel Architecture Category=BERT-based Baselines2026.04 | 19.2 | 39.9 | 50.5 |