Multi-task Regression on MovieLens (test)
3,679LossDSelect-k
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DSelect-kGating Setting=Per-example, alpha=0.92021.06 | 3,679 | 1.1 | |
| Shared Bottomalpha=0.92021.06 | 3,687 | — | |
| Gumbel SoftmaxGating Setting=Static, alpha=0.92021.06 | 3,688 | 3.6 | |
| DSelect-kGating Setting=Static, alpha=0.92021.06 | 3,690 | 1.3 | |
| Top-kGating Setting=Static, alpha=0.92021.06 | 3,693 | 2 | |
| Top-kGating Setting=Per-example, alpha=0.92021.06 | 3,741 | 2 | |
| DSelect-kGating Setting=Static, alpha=0.52021.06 | 3,804 | 1.5 | |
| Top-kGating Setting=Static, alpha=0.52021.06 | 3,818 | 2 | |
| DSelect-kGating Setting=Per-example, alpha=0.52021.06 | 3,823 | 1.2 | |
| Top-kGating Setting=Per-example, alpha=0.52021.06 | 3,841 | 2 | |
| Softmax MoEalpha=0.92021.06 | 3,847 | 8 | |
| Shared Bottomalpha=0.52021.06 | 3,868 | — | |
| Gumbel SoftmaxGating Setting=Static, alpha=0.52021.06 | 3,898 | 2.6 | |
| Softmax MoEalpha=0.52021.06 | 3,960 | 8 | |
| DSelect-kGating Setting=Per-example, alpha=0.12021.06 | 4,006 | 1.5 | |
| Top-kGating Setting=Static, alpha=0.12021.06 | 4,012 | 2 | |
| DSelect-kGating Setting=Static, alpha=0.12021.06 | 4,015 | 2.7 | |
| Top-kGating Setting=Per-example, alpha=0.12021.06 | 4,027 | 2 | |
| Shared Bottomalpha=0.12021.06 | 4,037 | — | |
| Softmax MoEalpha=0.12021.06 | 4,090 | 8 | |
| Gumbel SoftmaxGating Setting=Static, alpha=0.12021.06 | 4,171 | 2.7 |