Multi-task Learning on DUNE (Normalized Performance)
90.9Normalized PerformanceAdaMerging+TAP
Evaluation results
| Method | Links | |
|---|---|---|
| AdaMerging+TAPBase Model=DINOv2, Selection mechanism=TAP2026.04 | 90.9 | |
| TSVBase Model=DINOv22026.04 | 89.3 | |
| ConsensusBase Model=DINOv22026.04 | 89.2 | |
| TA+TAPBase Model=DINOv2, Selection mechanism=TAP2026.04 | 88.9 | |
| TIESBase Model=DINOv22026.04 | 88.7 | |
| DINOv2 (Pretrained)Status=Pretrained2026.04 | 87.1 | |
| STAR+TAPBase Model=DINOv2, Selection mechanism=TAP2026.04 | 86.3 | |
| DUNE†Type=Distilled baseline2026.04 | 84.7 |