ImageNet imagenet-4 Leaderboard
Auto-discovered from papers reporting ImageNet (Recall). Β· Metric: Recall (higher is better) Β· π’ Updated 12 min ago
| # | Model | Recall | Paper |
|---|---|---|---|
| 1 | + GAN FT (Full MIMFlow) | 0.71 | link |
| 2 | RAE** | 0.67 | link |
| 3 | All | 0.66 | link |
| 4 | D+CLIP | 0.66 | link |
| 5 | DINO | 0.66 | link |
| 6 | K=128 | 0.66 | link |
| 7 | + Learnable Tokens, - GAN Loss | 0.66 | link |
| 8 | + Masking (0.4β0.6) | 0.66 | link |
| 9 | MIM weight=1 | 0.66 | link |
| 10 | REPA-E** | 0.66 | link |
| 11 | Ο=0.2 | 0.66 | link |
| 12 | Ο=0.3 | 0.66 | link |
| 13 | DDT** | 0.65 | link |
| 14 | K=64 | 0.65 | link |
| 15 | MIM weight=5 | 0.64 | link |
| 16 | REPA** | 0.64 | link |
| 17 | Baseline (e2e, SD-VAE, 256tok, GAN) | 0.63 | link |
| 18 | eMIGM-H | 0.63 | link |
| 19 | + Aux Loss (DINO+CLIP) | 0.62 | link |
| 20 | MAR-H | 0.62 | link |
| 21 | Ο=0.5 | 0.62 | link |
| 22 | eMIGM-L | 0.60 | link |
| 23 | MAR-L | 0.60 | link |
| 24 | PixelFlow | 0.60 | link |
| 25 | C+HOG | 0.59 | link |
| 26 | eMIGM-B | 0.57 | link |
| 27 | K=192 | 0.57 | link |
| 28 | MAR-B | 0.57 | link |
| 29 | MIM weight=10 | 0.55 | link |
| 30 | eMIGM-S | 0.54 | link |
| 31 | eMIGM-XS | 0.51 | link |