Multi-Task Learning on QM9
4.36Mean Relative Error (MR)APT-Fair
Evaluation results
| Method | Links | ||
|---|---|---|---|
| APT-Fair2026.04 | 4.36 | 64.9 | |
| FairGrad2026.04 | 5.82 | 57.9 | |
| Nash-MTL2026.04 | 6 | 62 | |
| APT-Muon2026.04 | 6 | 72.9 | |
| APT-FAMO2026.04 | 6.64 | 55.5 | |
| COST2026.04 | 6.82 | 58.3 | |
| SI2026.04 | 7.36 | 77.8 | |
| FAMO2026.04 | 7.91 | 58.5 | |
| UW2026.04 | 9.09 | 108 | |
| APT-CG2026.04 | 9.73 | 108.9 | |
| IMTL-G2026.04 | 10 | 77.2 | |
| PCGrad2026.04 | 10.73 | 125.7 | |
| CAGrad2026.04 | 11.45 | 112.8 | |
| MGDA2026.04 | 12.09 | 120.5 | |
| DWA2026.04 | 12.27 | 175.3 | |
| LS2026.04 | 12.45 | 177.6 | |
| RLW2026.04 | 14.27 | 203.8 |