ImageNet imagenet-3 Leaderboard
Auto-discovered from papers reporting ImageNet (Precision). Β· Metric: Precision (higher is better) Β· π’ Updated 8 min ago
| # | Model | Precision | Paper |
|---|---|---|---|
| 1 | MAR-B | 0.82 | link |
| 2 | eMIGM-B | 0.81 | link |
| 3 | MAR-H | 0.81 | link |
| 4 | MAR-L | 0.81 | link |
| 5 | PixelFlow | 0.81 | link |
| 6 | eMIGM-H | 0.80 | link |
| 7 | eMIGM-L | 0.80 | link |
| 8 | eMIGM-S | 0.80 | link |
| 9 | eMIGM-XS | 0.80 | link |
| 10 | DDT** | 0.79 | link |
| 11 | REPA** | 0.79 | link |
| 12 | REPA-E** | 0.79 | link |
| 13 | RAE** | 0.78 | link |
| 14 | + GAN FT (Full MIMFlow) | 0.77 | link |
| 15 | K=64 | 0.71 | link |
| 16 | All | 0.70 | link |
| 17 | D+CLIP | 0.70 | link |
| 18 | DINO | 0.70 | link |
| 19 | K=128 | 0.70 | link |
| 20 | MIM weight=1 | 0.70 | link |
| 21 | Ο=0.3 | 0.70 | link |
| 22 | MIM weight=5 | 0.69 | link |
| 23 | Baseline (e2e, SD-VAE, 256tok, GAN) | 0.68 | link |
| 24 | Ο=0.2 | 0.66 | link |
| 25 | + Aux Loss (DINO+CLIP) | 0.65 | link |
| 26 | Ο=0.5 | 0.65 | link |
| 27 | + Masking (0.4β0.6) | 0.64 | link |
| 28 | C+HOG | 0.63 | link |
| 29 | K=192 | 0.57 | link |
| 30 | + Learnable Tokens, - GAN Loss | 0.52 | link |
| 31 | MIM weight=10 | 0.46 | link |