Multi-label Classification on ImageNet ReaL
89.3Top-1 Accuracy+Multi-label FT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| +Multi-label FTModel=ViT-large2026.03 | 89.3 | 91.1 | |
| Multi-label E2EModel=ViT-large2026.03 | 89.3 | 91.2 | |
| +Multi-label FTModel=ViT-base2026.03 | 88.9 | 90.6 | |
| +IN1k-Mul FTBackbone=ViT-B/16, Resolution=224, Pre-training=ImageNet-21K (MIIL), Supervision=Multi-label (Mul), Training Protocol=Fine-tuning2026.03 | 88.9 | 91 | |
| Multi-label E2EModel=ViT-base2026.03 | 88.8 | 90.7 | |
| IN1k-Mul E2EBackbone=ViT-B/16, Resolution=224, Supervision=Multi-label (Mul), Training Protocol=End-to-end2026.03 | 88.8 | 90.7 | |
| +IN1k-Sig FTBackbone=ViT-B/16, Resolution=224, Pre-training=ImageNet-21K (MIIL), Supervision=Single-label (Sig), Training Protocol=Fine-tuning2026.03 | 88.7 | 90.9 | |
| Single-label E2EModel=ViT-large2026.03 | 88.6 | 90.8 | |
| Single-label E2EModel=ViT-base2026.03 | 88.2 | 90.6 | |
| Multi-label E2EModel=ViT-small2026.03 | 88.1 | 90.3 | |
| +Multi-label FTModel=ViT-small2026.03 | 87.7 | 90.1 | |
| Single-label E2EModel=ViT-small2026.03 | 87 | 89 | |
| Multi-label E2EModel=ResNet-1012026.03 | 86.7 | 89 | |
| +Multi-label FTModel=ResNet-1012026.03 | 86.6 | 89 | |
| Multi-label E2EModel=ResNet-502026.03 | 85.6 | 88.2 | |
| +Multi-label FTModel=ResNet-502026.03 | 85.4 | 88.2 | |
| Single-label E2EModel=ResNet-1012026.03 | 84.7 | 87.1 | |
| Single-label E2EModel=ResNet-502026.03 | 84.1 | 87 |