Zero-shot Image Classification on DataComp
63.47DataComp ScoreTuneCLIP
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| TuneCLIPBase Model=SigLIP ViT-B/162026.01 | 63.47 | 9.27 | 74.16 | — | |
| TuneCLIPBase Model=SigLIP ViT-B/162026.01 | 63.47 | — | — | — | |
| BaselineBase Model=SigLIP ViT-B/162026.01 | 62.32 | — | — | — | |
| TuneCLIPBase Model=OpenAI ViT-B/162026.01 | 58.62 | 8.62 | 68.96 | — | |
| TuneCLIPBase Model=OpenAI ViT-B/162026.01 | 58.62 | — | — | — | |
| TuneCLIPBase Model=LAION ViT-B/322026.01 | 57.22 | 4.32 | 36.24 | — | |
| TuneCLIPBase Model=LAION ViT-B/322026.01 | 57.22 | — | — | — | |
| BaselineBase Model=LAION ViT-B/322026.01 | 56.94 | — | — | — | |
| OpenCLIPBase Model=LAION ViT-B/322026.01 | 56.75 | 3.1 | 24.8 | — | |
| OpenCLIPBase Model=LAION ViT-B/322026.01 | 56.75 | — | — | — | |
| BaselineBase Model=OpenAI ViT-B/162026.01 | 56.26 | — | — | — | |
| FastCLIPBase Model=LAION ViT-B/322026.01 | 55.89 | 2.33 | 18.64 | — | |
| FastCLIPBase Model=LAION ViT-B/322026.01 | 55.89 | — | — | — | |
| OpenCLIPBase Model=OpenAI ViT-B/162026.01 | 55.11 | 5.46 | 43.68 | — | |
| OpenCLIPBase Model=OpenAI ViT-B/162026.01 | 55.11 | — | — | — | |
| TuneCLIPBase Model=OpenAI ViT-B/322026.01 | 54.34 | 5.05 | 40.4 | — | |
| TuneCLIPBase Model=OpenAI ViT-B/322026.01 | 54.34 | — | — | — | |
| FastCLIPBase Model=OpenAI ViT-B/162026.01 | 53.53 | 4.21 | 33.68 | — | |
| FastCLIPBase Model=OpenAI ViT-B/162026.01 | 53.53 | — | — | — | |
| BaselineBase Model=OpenAI ViT-B/322026.01 | 52.45 | — | — | — | |
| OpenCLIPBase Model=OpenAI ViT-B/322026.01 | 51.02 | 2.66 | 21.28 | — | |
| OpenCLIPBase Model=OpenAI ViT-B/322026.01 | 51.02 | — | — | — | |
| FastCLIPBase Model=OpenAI ViT-B/322026.01 | 49.78 | 2.22 | 17.76 | — | |
| FastCLIPBase Model=OpenAI ViT-B/322026.01 | 49.78 | — | — | — | |
| OpenCLIPBase Model=SigLIP ViT-B/162026.01 | 48.1 | 7.55 | 60.4 | — | |
| OpenCLIPBase Model=SigLIP ViT-B/162026.01 | 48.1 | — | — | — | |
| FastCLIPBase Model=SigLIP ViT-B/162026.01 | 45.8 | 4.28 | 34.24 | — | |
| FastCLIPBase Model=SigLIP ViT-B/162026.01 | 45.8 | — | — | — | |
| BaselineBackbone=ViT-H/14-quickgelu2026.01 | — | — | — | 69.61 | |
| TuneCLIPBackbone=ViT-H/14-quickgelu2026.01 | — | — | — | 69.23 |