Image Classification on Continual Learning Task Stream S
97.4Caltech AccuracyBETA
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| BETABackbone=SigLIP, Variant=B/16-224, Shot-setting=Full-shot2026.06 | 97.4 | 95.3 | 94.6 | 99.6 | 91.5 | 58.1 | 78.5 | 81.4 | 96.7 | 86.4 | 88 | |
| BETABackbone=SigLIP 2, Variant=B/16-224, Shot-setting=Full-shot2026.06 | 97.4 | 95.5 | 94.4 | 99.6 | 92.3 | 65.6 | 80 | 82.1 | 96.6 | 88 | 89.2 | |
| BETABackbone=SigLIP 2, Variant=So/14-224, Shot-setting=Full-shot2026.06 | 97.2 | 96.4 | 96.3 | 99.9 | 95.2 | 79.6 | 83.3 | 84.4 | 96.4 | 92.7 | 92.1 | |
| BETABackbone=CLIP, Variant=L/14-224, Shot-setting=Full-shot2026.06 | 96.8 | 95.5 | 90.9 | 99.2 | 93.1 | 63.2 | 80.8 | 79.4 | 97.2 | 90 | 88.6 | |
| BETABackbone=SigLIP, Variant=So/14-224, Shot-setting=Full-shot2026.06 | 96.4 | 96.4 | 95.9 | 99.8 | 94.7 | 71.7 | 82.9 | 83.5 | 97.1 | 90.8 | 90.9 | |
| BETABackbone=SigLIP 2, Variant=So/14-2242026.06 | 95.7 | 96.2 | 95.9 | 99.8 | 93.4 | 77.3 | 81.1 | 78.5 | 87 | 84.4 | 88.9 | |
| BETABackbone=CLIP, Variant=B/16-224, Shot-setting=Full-shot2026.06 | 95.5 | 94 | 87.1 | 98.8 | 89.8 | 53.2 | 77.2 | 75.9 | 95.9 | 86.4 | 85.4 | |
| BETABackbone=SigLIP, Variant=B/16-2242026.06 | 95.4 | 95 | 93.6 | 99.2 | 89.4 | 53.5 | 76.4 | 76.7 | 87.9 | 80.8 | 84.8 | |
| BETABackbone=SigLIP 2, Variant=B/16-2242026.06 | 95.2 | 95 | 94 | 99.4 | 90.2 | 60.3 | 78 | 76.8 | 87.9 | 83 | 86 | |
| BETABackbone=SigLIP, Variant=So/14-2242026.06 | 94.6 | 96.2 | 95.3 | 99.6 | 93 | 69 | 81 | 79 | 90.1 | 82.7 | 88.1 | |
| BETABackbone=CLIP, Variant=L/14-2242026.06 | 94 | 94.9 | 89.2 | 98.7 | 90.8 | 59.1 | 78.7 | 71.5 | 91.6 | 85.1 | 85.4 | |
| BETABackbone=CLIP, Variant=B/16-2242026.06 | 93.4 | 92.8 | 83.9 | 97.1 | 86.8 | 46.9 | 74.7 | 67.8 | 88.4 | 83.1 | 81.5 | |
| Zero-shotBackbone=SigLIP 2, Variant=B/16-224, Shot-setting=Full-shot2026.06 | 85.8 | 92.7 | 92.5 | 81.6 | 88.8 | 28 | 69.4 | 55.1 | 35.8 | 66.2 | 69.6 | |
| Zero-shotBackbone=SigLIP 2, Variant=B/16-2242026.06 | 85.8 | 92.7 | 92.5 | 81.6 | 88.8 | 28 | 69.4 | 55.1 | 35.8 | 66.2 | 69.6 | |
| Zero-shotBackbone=SigLIP, Variant=So/14-224, Shot-setting=Full-shot2026.06 | 85.4 | 95 | 87.5 | 89.5 | 92.6 | 50.1 | 72 | 56.4 | 49.4 | 77.6 | 75.5 | |
| Zero-shotBackbone=SigLIP, Variant=So/14-2242026.06 | 85.4 | 95 | 87.5 | 89.5 | 92.6 | 50.1 | 72 | 56.4 | 49.4 | 77.6 | 75.5 | |
| Zero-shotBackbone=SigLIP, Variant=B/16-224, Shot-setting=Full-shot2026.06 | 85.3 | 92.8 | 89 | 81.7 | 88 | 29.4 | 65.7 | 55.1 | 31.6 | 72.2 | 69.1 | |
| Zero-shotBackbone=SigLIP 2, Variant=So/14-224, Shot-setting=Full-shot2026.06 | 85.3 | 94.2 | 95 | 89.2 | 92.5 | 55 | 70.8 | 60.1 | 43.3 | 76 | 76.1 | |
| Zero-shotBackbone=SigLIP, Variant=B/16-2242026.06 | 85.3 | 92.8 | 89 | 81.7 | 88 | 29.4 | 65.7 | 55.1 | 31.6 | 72.2 | 69.1 | |
| Zero-shotBackbone=SigLIP 2, Variant=So/14-2242026.06 | 85.3 | 94.2 | 95 | 89.2 | 92.5 | 55 | 70.8 | 60.1 | 43.3 | 76 | 76.1 | |
| Zero-shotBackbone=CLIP, Variant=L/14-224, Shot-setting=Full-shot2026.06 | 75.7 | 93.3 | 76.8 | 79.5 | 90.3 | 32.5 | 64.4 | 41.8 | 61 | 73.3 | 68.9 | |
| Zero-shotBackbone=CLIP, Variant=L/14-2242026.06 | 75.7 | 93.3 | 76.8 | 79.5 | 90.3 | 32.5 | 64.4 | 41.8 | 61 | 73.3 | 68.9 | |
| Zero-shotBackbone=CLIP, Variant=B/16-224, Shot-setting=Full-shot2026.06 | 72.4 | 88.6 | 65.5 | 70.7 | 85.3 | 24.8 | 59.1 | 33.7 | 45.2 | 64.9 | 61 | |
| Zero-shotBackbone=CLIP, Variant=B/16-2242026.06 | 72.4 | 88.6 | 65.5 | 70.7 | 85.3 | 24.8 | 59.1 | 33.7 | 45.2 | 64.9 | 61 |