Multi-task Image Classification on TA-8 (test)
94.3AccuracyFine-tuned
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Fine-tunedBackbone=CLIP ViT-L/142025.12 | 94.3 | 100 | |
| IndividualBackbone=ViT-L-142025.08 | 93.7 | — | |
| EMR-MergingBackbone=ViT-L-142025.08 | 93 | 99.6 | |
| MTLBackbone=ViT-L-142025.08 | 92.6 | — | |
| MTLBackbone=CLIP ViT-L/142025.12 | 92.3 | 97.9 | |
| WUDI-MergingBackbone=ViT-L-142025.08 | 91.7 | 97.7 | |
| SAMergingBackbone=CLIP ViT-L/14, k=162025.12 | 91.2 | 96.7 | |
| TSVMBackbone=ViT-L-142025.08 | 90.5 | 96.3 | |
| Iso-CTSBackbone=ViT-L-142025.08 | 90.4 | 96.4 | |
| Fine-tunedBackbone=CLIP ViT-B/322025.12 | 90.3 | 100 | |
| IndividualBackbone=ViT-B-322025.08 | 89.9 | — | |
| MTLBackbone=CLIP ViT-B/322025.12 | 88.5 | 98 | |
| EMR-MergingBackbone=ViT-B-322025.08 | 88.5 | 98.4 | |
| MTLBackbone=ViT-B-322025.08 | 87.8 | — | |
| ProDistillBackbone=CLIP ViT-L/14, k=162025.12 | 87.2 | 92.5 | |
| WUDI-MergingBackbone=ViT-B-322025.08 | 85.5 | 93.9 | |
| Consensus TABackbone=ViT-L-142025.08 | 85.3 | 90.7 | |
| TIESBackbone=ViT-L-142025.08 | 85 | 91.9 | |
| Task arithmeticBackbone=ViT-L-142025.08 | 84 | 89.3 | |
| SAMergingBackbone=CLIP ViT-B/32, k=162025.12 | 83.8 | 92.8 | |
| TSVMBackbone=ViT-B-322025.08 | 83.3 | 92.4 | |
| ProDistillBackbone=CLIP ViT-B/32, k=162025.12 | 81.1 | 89.8 | |
| Iso-CTSBackbone=ViT-B-322025.08 | 81 | 89.7 | |
| Consensus TABackbone=ViT-B-322025.08 | 74.8 | 82.8 | |
| TIESBackbone=ViT-B-322025.08 | 74 | 82 | |
| Task arithmeticBackbone=ViT-B-322025.08 | 70.4 | 78 | |
| Pre-trainedBackbone=ViT-L-142025.08 | 65.1 | — | |
| Pre-trainedBackbone=ViT-B-322025.08 | 47.3 | — |