Multi-Task Learning on Cityscapes
2.5MRNash-MTL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Nash-MTLCost=O(md)2026.03 | 2.5 | 6.82 | |
| MoCoCost=O(md)2026.03 | 3.75 | 9.9 | |
| MARIGOLDCost=O(d)2026.03 | 4.5 | 8.62 | |
| IMTL-GCost=O(md)2026.03 | 5 | 11.1 | |
| SDMGradCost=O(md)2026.03 | 5.25 | 7.79 | |
| CAGradCost=O(md)2026.03 | 7 | 11.64 | |
| FAMOCost=O(d)2026.03 | 7 | 8.13 | |
| GradDropCost=O(md)2026.03 | 7.5 | 23.73 | |
| FAMO (rerun)Cost=O(d)2026.03 | 7.75 | 9.92 | |
| PCGradCost=O(md)2026.03 | 8.5 | 18.29 | |
| MGDACost=O(md)2026.03 | 9.5 | 44.14 | |
| MoDoCost=O(md)2026.03 | 9.75 | 18.89 |