Multitask Learning on 20 vision tasks (accuracy)
72.9AccuracyAE (Ours)
Evaluation results
| Method | Links | |
|---|---|---|
| AE (Ours)Backbone=ViT-L/14, Basis size (M)=50%, Merging Method=L&S2025.02 | 72.9 | |
| AE (Ours)Backbone=ViT-B/32, Basis size (M)=50%, Merging Method=L&S2025.02 | 62.8 | |
| PretrainedBackbone=ViT-B/162025.02 | 59.8 | |
| L&SBackbone=ViT-B/32, Basis size (M)=100%2025.02 | 59.8 |