Image Classification on CIFAR-100 (test) (Last-10 vs Best-checkpoint Accuracy)
32.72Last-10 AccuracyM-CovJac
Evaluation results
| Method | Links | ||
|---|---|---|---|
| M-CovJack=1280k, L=6, iters=100k2026.05 | 32.72 | 32.88 | |
| Gumbel-STk=1280k, L=6, iters=100k2026.05 | 31.65 | 31.88 | |
| Soft-Mixk=1280k, L=6, iters=100k2026.05 | 30.56 | 30.87 | |
| M-CovJack=256k, L=6, iters=100k2026.05 | 28.37 | 28.91 | |
| Soft-Mixk=256k, L=6, iters=100k2026.05 | 27.92 | 28.21 | |
| Gumbel-STk=256k, L=6, iters=100k2026.05 | 27.72 | 28.08 | |
| M-STEk=1280k, L=6, iters=100k2026.05 | 27.7 | 29.77 | |
| M-STEk=256k, L=6, iters=100k2026.05 | 24.02 | 25.02 |