Class-Conditional Image Generation on ImageNet-256 (val)
1.57Inception ScoreOpen-MAGVIT2-L
Evaluation Results
| Method | Links | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Open-MAGVIT2-LNFE=12026.07 | 1.57 | 1.39 | 4.87 | 2.93 | 6.33 | 4.94 | 6.59 | 5.33 | 5.95 | 4.5 | 4.66 | 7.05 | 2.7 | 1.41 | 4.3 | 4.03 | |
| DiT-XL/2NFE=12026.07 | 1.38 | 0.99 | 4.12 | 2.51 | 5.5 | 4.72 | 8.92 | 4.91 | 5.98 | 3.5 | 3.82 | 6.01 | 2.35 | 1.03 | 3.98 | 3.39 | |
| SiT-XL/2NFE=12026.07 | 1.29 | 1.12 | 4.53 | 2.68 | 6.24 | 4.75 | 9.94 | 5.21 | 6.36 | 3.87 | 4.09 | 6.43 | 2.32 | 0.91 | 4.27 | 3.63 | |
| iRDMNFE=1, external representation encoder in training=true2026.07 | 1.27 | 0.98 | 1.35 | 0.83 | 1.3 | 1.02 | 1.11 | 1.9 | 1.22 | 1.56 | 1.55 | 1.44 | 1.32 | 1.36 | 1.3 | 1.54 | |
| pMF-H (base)NFE=1, Model Scale=base2026.07 | 1.25 | 0.88 | 3.91 | 3.15 | 6.43 | 3.93 | 6.62 | 4.88 | 6.69 | 4 | 4.25 | 6.87 | 2.63 | 1.71 | 4.09 | 3.63 | |
| VAR-d30NFE=12026.07 | 1.08 | 1.12 | 4.18 | 3.18 | 5.71 | 4.51 | 6.37 | 5.19 | 6.52 | 3.77 | 3.88 | 6.28 | 2.63 | 0.88 | 3.95 | 3.51 | |
| JiT-HNFE=12026.07 | 1.07 | 1.46 | 3.73 | 2.78 | 5.52 | 6.17 | 5.34 | 4.84 | 6.59 | 3.85 | 3.75 | 6.11 | 2.81 | 1.13 | 3.94 | 3.39 | |
| Validation baseline2026.07 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | |
| MAR-HNFE=12026.07 | 1 | 1.04 | 4.16 | 2.39 | 5.25 | 4.34 | 9.05 | 4.86 | 6.21 | 3.23 | 4.01 | 6.04 | 2.2 | 0.48 | 3.87 | 3.18 | |
| Drifting-Lexternal representation encoder in training=true2026.07 | 0.97 | 1.61 | 6.12 | 3.2 | 8.84 | 5.83 | 18.2 | 7.12 | 7.89 | 5.31 | 5.7 | 8.45 | 2.69 | 1.07 | 5.93 | 4.91 | |
| iMF-XLNFE=12026.07 | 0.96 | 1.22 | 5.07 | 2.96 | 6.89 | 5.04 | 15.2 | 6.01 | 7.26 | 4.19 | 4.75 | 7.06 | 2.62 | 1.03 | 5.02 | 4.08 | |
| LightningDiT-XLNFE=1, external representation encoder in training=true2026.07 | 0.89 | 0.9 | 3.25 | 2.04 | 4.44 | 3.76 | 4.9 | 3.92 | 5.22 | 2.8 | 3.29 | 5.18 | 1.99 | 0.78 | 3.1 | 2.69 | |
| MDTv2-XL/2NFE=12026.07 | 0.86 | 0.98 | 3.76 | 2.44 | 5.53 | 4.71 | 9.78 | 4.94 | 6.33 | 3.08 | 3.59 | 5.48 | 2.3 | 0.79 | 3.9 | 3.14 | |
| SiT-XL/2+REPANFE=1, external representation encoder in training=true2026.07 | 0.85 | 1.01 | 3.63 | 2.35 | 5.13 | 4.33 | 8.15 | 4.58 | 5.99 | 3.02 | 3.34 | 5.29 | 2.12 | 0.72 | 3.61 | 2.99 | |
| DDT-XL/2NFE=1, external representation encoder in training=true2026.07 | 0.83 | 0.96 | 3.74 | 2.36 | 5.29 | 4.6 | 9.07 | 4.68 | 6.18 | 3.06 | 3.49 | 5.44 | 2.15 | 0.97 | 3.77 | 3.11 | |
| REG-XLNFE=1, external representation encoder in training=true2026.07 | 0.79 | 0.91 | 3.13 | 2.03 | 4.68 | 3.92 | 7.09 | 4.02 | 5.74 | 2.6 | 2.88 | 4.65 | 1.83 | 0.71 | 3.21 | 2.62 | |
| RAE-XLNFE=1, external representation encoder in training=true2026.07 | 0.75 | 1.3 | 2.38 | 2.11 | 2.74 | 3.52 | 2.76 | 2.8 | 4.51 | 2.39 | 2.31 | 3.88 | 1.51 | 1.13 | 2.43 | 2.18 | |
| REPA-E SiT-XL/1NFE=1, external representation encoder in training=true2026.07 | 0.75 | 1 | 2.79 | 1.86 | 3.41 | 2.83 | 3.3 | 3.07 | 3.89 | 2.04 | 2.41 | 4.13 | 1.47 | 0.66 | 2.4 | 2.14 | |
| pMF-H (FD-SIM)NFE=1, external representation encoder in training=true2026.07 | 0.67 | 0.67 | 1.81 | 0.6 | 1.86 | 2.69 | 2.33 | 2.63 | 4.76 | 2.14 | 2.31 | 3.68 | 1.24 | 1.35 | 2.05 | 1.98 |