Image Classification on ImageNet PGD
3.8AccuracyPixMix
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| PixMixExternal Data-free=No, Backbone=ViT-Base2026.03 | 3.8 | — | — | — | |
| MoireMixExternal Data-free=Yes, Backbone=ViT-Base2026.03 | 3.1 | — | — | — | |
| RandAugExternal Data-free=Yes, Backbone=ViT-Base2026.03 | 1.9 | — | — | — | |
| CutoutExternal Data-free=Yes, Backbone=ViT-Base2026.03 | 1.8 | — | — | — | |
| IPMixExternal Data-free=No, Backbone=ViT-Base2026.03 | 1.8 | — | — | — | |
| BaselineExternal Data-free=Yes, Backbone=ViT-Base2026.03 | 1.3 | — | — | — | |
| MixUpExternal Data-free=Yes, Backbone=ViT-Base2026.03 | 1.3 | — | — | — | |
| CutMixExternal Data-free=Yes, Backbone=ViT-Base2026.03 | 1.2 | — | — | — | |
| AugMixExternal Data-free=Yes, Backbone=ViT-Base2026.03 | 1.1 | — | — | — | |
| DiffuseMixExternal Data-free=No, Backbone=ViT-Base2026.03 | 1 | — | — | — | |
| EllipticalBackbone=ViT-Ti/16, Training Recipe=DeiT, Attention mechanism=Elliptical2026.03 | — | 46.3 | 80.05 | 2.231 | |
| Elliptical-QUESTBackbone=ViT-Ti/16, Training Recipe=DeiT, Attention mechanism=Elliptical-QUEST2026.03 | — | 47.25 | 80.61 | 2.211 | |
| QUESTBackbone=ViT-Ti/16, Training Recipe=DeiT, Attention mechanism=QUEST2026.03 | — | 45.26 | 79.33 | 2.448 | |
| StandardBackbone=ViT-Ti/16, Training Recipe=DeiT, Attention mechanism=Standard2026.03 | — | 43.65 | 78.18 | 2.503 | |
| Swin-ACMoE-BaseBackbone=Swin-Base2025.02 | — | 46.28 | 80.24 | — | |
| Swin-ACMoE-Top 1Top-k routing=12025.02 | — | 40.66 | 73.46 | — | |
| Swin-ACMoE-Top 2Top-k routing=22025.02 | — | 43.74 | 78.55 | — | |
| Swin-BaseBackbone=Swin-Base2025.02 | — | 44.61 | 79.2 | — | |
| Swin-Top 1Top-k routing=12025.02 | — | 39.69 | 74.59 | — | |
| Swin-Top 2Top-k routing=22025.02 | — | 40.85 | 75.51 | — |