Compositional Reasoning on BIVLC (test)
61.3Group ScoreCLIP+TF_Local
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CLIP+TF_LocalTraining data=TROHN-Img2026.04 | 61.3 | 39 | |
| FSC-CLIPTraining data=COCO†2026.04 | 46.5 | 19.2 | |
| CLIP+TF_LocalTraining data=COCO2026.04 | 45.7 | 24 | |
| NegCLIPTraining data=COCO†2026.04 | 44.9 | 18.7 | |
| X-VLMTraining data=COCO+Others2026.04 | 40.9 | 12.3 | |
| FineCLIPTraining data=COCO†2026.04 | 39.4 | 8.6 | |
| CLIPTraining data=Pretrained2026.04 | 36.8 | 13.7 | |
| TripletCLIPTraining data=CC3M/12M2026.04 | 35.2 | 9.8 |