Image Classification on ImageNet-C Gaussian Noise
65.2Top-1 AccuracyRandAR
Evaluation results
| Method | Links | |
|---|---|---|
| RandARModel size=XL/16, Model type=Autoregressive2026.03 | 65.2 | |
| RandAR XL/162026.03 | 65.2 | |
| ViTModel size=L/16, Model type=Discriminative2026.03 | 62 | |
| RandARModel size=L/16, Model type=Autoregressive2026.03 | 58.9 | |
| RandAR L/162026.03 | 58.9 | |
| DATModel size=L/16, Model type=Joint model2026.03 | 56.9 | |
| DiTModel size=XL/16, Model type=Diffusion2026.03 | 50.5 | |
| DINOv2Model size=XL/16, Model type=Discriminative2026.03 | 50 | |
| DINOv2Model size=L/16, Model type=Discriminative2026.03 | 49.7 | |
| RandAR rasterModel size=L/16, Model type=Autoregressive2026.03 | 47.6 | |
| DiTModel size=L/16, Model type=Diffusion2026.03 | 46.7 | |
| ResNet-1512026.03 | 44.3 | |
| SiT + REPAModel size=XL/16, Model type=Diffusion2026.03 | 41.6 | |
| ResNet-1012026.03 | 39.1 | |
| LlamaGenModel size=XL/16, Model type=Autoregressive2026.03 | 35.9 | |
| LlamaGenModel size=L/16, Model type=Autoregressive2026.03 | 35.8 | |
| SiTModel size=XL/16, Model type=Diffusion2026.03 | 31.1 | |
| ResNet-342026.03 | 30.1 | |
| ResNet-502026.03 | 30 | |
| ResNet-182026.03 | 19.5 | |
| A-VARCModel size=L/16, Model type=Autoregressive2026.03 | 15.5 | |
| VARModel size=XL/16, Model type=Autoregressive2026.03 | 14.6 | |
| VARModel size=L/16, Model type=Autoregressive2026.03 | 13.7 | |
| A-VARCModel size=XL/16, Model type=Autoregressive2026.03 | 7.7 |