ViT-B-16+240 (LAION400M)

Reported on 4 benchmarks across 1 task · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision4 results

Image RetrievalonCREPE (Compositional REPresentation Evaluation)
Recall@1 (HN-Atom + HN-Comp, SC)· 2022-12-13
37.32
best: 39.44 (ViT-L-14 (LAION400M))
CREPE: Can Vision-Language Foundation Models Reason Compositionally?arXiv:2212.07796
Image RetrievalonCREPE (Compositional REPresentation Evaluation)
Recall@1 (HN-Atom + HN-Comp, UC)· 2022-12-13
32.26
best: 33.81 (ViT-L-14 (LAION400M))
CREPE: Can Vision-Language Foundation Models Reason Compositionally?arXiv:2212.07796
Image RetrievalonCREPE (Compositional REPresentation Evaluation)
Recall@1 (HN-Atom, UC)· 2022-12-13
46.53
best: 47.86 (ViT-L-14 (LAION400M))
CREPE: Can Vision-Language Foundation Models Reason Compositionally?arXiv:2212.07796
Image RetrievalonCREPE (Compositional REPresentation Evaluation)
Recall@1 (HN-Comp, UC)· 2022-12-13
60.19
best: 92.6 (RN-50 (MosaiCLIP, CC-12M))
CREPE: Can Vision-Language Foundation Models Reason Compositionally?arXiv:2212.07796