Class-conditional generation on ImageNet 512x512 (test)
1.13FIDRAE
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| RAE#Params=839M2026.05 | 1.13 | 259.6 | — | — | — | |
| EDM2#Params=1.5B2026.05 | 1.25 | — | — | — | — | |
| DDT#Params=675M2026.05 | 1.28 | 305.1 | — | — | — | |
| OURS, FLOP HEAVYPARAMS=399M, NFE=3202026.03 | 1.45 | 310 | 74 | 63 | 3.91 | |
| SiD22026.05 | 1.5 | — | — | — | — | |
| SiD2, FLOP HEAVYPARAMS=397M, NFE=5122026.03 | 1.53 | 306.2 | 74 | 63 | 3.98 | |
| SphereAR-LType=AR, Order=raster, #Params=481M, #Epochs=4002025.09 | 1.67 | 292.5 | — | — | — | |
| FrequencyBooster-H/16Epochs=400, Params=1.1B, NFE=100 × 22026.05 | 1.69 | 332.6 | — | — | — | |
| xAR#Params=608M2026.05 | 1.7 | 281.5 | — | — | — | |
| MAR-LType=Mask., Order=random, #Params=481M, #Epochs=8002025.09 | 1.73 | 279.9 | — | — | — | |
| MAR#Params=481M2026.05 | 1.73 | 279.9 | — | — | — | |
| JIT-GPARAMS=2B2026.03 | 1.78 | 306.8 | — | — | — | |
| OURS, SMALLPARAMS=399M, NFE=2562026.03 | 1.79 | 306.1 | 73 | 64 | 4.39 | |
| MAGVIT-v2Type=Mask., Order=random, #Params=307M, #Epochs=2702025.09 | 1.91 | 324.3 | — | — | — | |
| MAGVIT-v2#Params=307M2026.05 | 1.91 | 324.3 | — | — | — | |
| JiT-H/16Epochs=600, Params=965M, NFE=100 × 22026.05 | 1.94 | 309.1 | — | — | — | |
| SRC-Flow#Params=1.4B2026.05 | 2.07 | 305.7 | — | — | — | |
| REPA#Params=675M2026.05 | 2.08 | 274.6 | — | — | — | |
| SiD2, SMALLPARAMS=397M, NFE=5122026.03 | 2.19 | 295.3 | 72 | 63 | 4.3 | |
| DeCo-XL/16Epochs=340, Params=682M, NFE=100 × 22026.05 | 2.22 | 290 | 80 | 60 | — | |
| DiP-XL/32Params=631M, NFE=100 × 22026.05 | 2.31 | 291.7 | 84 | 58 | — | |
| StyleGAN-XL2026.05 | 2.41 | 267.8 | — | — | — | |
| VAR-d36-sType=N.S., Order=-, #Params=2.3B, #Epochs=3502025.09 | 2.62 | 303.2 | — | — | — | |
| SiT-XLEpochs=600, Params=675M, NFE=250 × 22026.05 | 2.62 | 252.2 | 84 | 57 | — | |
| SiT#Params=674M2026.05 | 2.62 | 252.2 | — | — | — | |
| VAR#Params=2.3B2026.05 | 2.63 | 303.2 | — | — | — | |
| DiffiT2026.05 | 2.67 | 252.1 | — | — | — | |
| SimFlow#Params=1.4B2026.05 | 2.74 | 304.9 | — | — | — | |
| PIXNERD-XL/16PARAMS=700M, NFE=1002026.03 | 2.84 | 245.6 | 80 | 59 | 5.95 | |
| PixNerd-XLEpochs=340, Params=700M, NFE=100 × 22026.05 | 2.84 | 245.6 | 80 | 59 | — | |
| STARFlow#Params=1.4B2026.05 | 3 | — | — | — | — | |
| DiT-XL/2Type=Diff., Order=-, #Params=675M, #Epochs=4002025.09 | 3.04 | 240.8 | — | — | — | |
| DiT-XLEpochs=600, Params=675M, NFE=250 × 22026.05 | 3.04 | 240.8 | 84 | 54 | — | |
| DiT#Params=674M2026.05 | 3.04 | 240.8 | — | — | — | |
| Simple DiffusionEpochs=800, Params=2.0B, NFE=250 × 22026.05 | 3.54 | 205 | — | — | — | |
| ADM#Params=731M2026.05 | 3.85 | 221.7 | — | — | — | |
| MaskGITType=Mask., Order=random, #Params=227M, #Epochs=3002025.09 | 4.46 | 342 | — | — | — | |
| FCDM-XLTraining iterations=1M, FLOPs (G)=257.7, Throughput (it/s)=129.62026.03 | 7.46 | 133.6 | 79 | 61 | — | |
| ADM-UEpochs=400, Params=554M, NFE=2502026.05 | 7.72 | 172.7 | 87 | 53 | — | |
| DiCo-XLTraining iterations=1.3M, FLOPs (G)=349.8, Throughput (it/s)=82.02026.03 | 8.1 | 132.9 | 78 | 62 | — | |
| BigGAN-deep#Params=158M2026.05 | 8.43 | 177.9 | — | — | — | |
| FCDM-XLTraining iterations=400K, FLOPs (G)=257.7, Throughput (it/s)=129.62026.03 | 10.23 | 108.7 | 79 | 60 | — | |
| DiT-XL/2Training iterations=3M, FLOPs (G)=524.7, Throughput (it/s)=18.62026.03 | 12.03 | 105.25 | 75 | 64 | — | |
| DiC-HTraining iterations=400K, FLOPs (G)=817.2, Throughput (it/s)=68.62026.03 | 12.89 | 101.8 | — | — | — | |
| DiT-XL/2Training iterations=1.3M, FLOPs (G)=524.7, Throughput (it/s)=18.62026.03 | 13.78 | — | — | — | — | |
| DiC-XLTraining iterations=400K, FLOPs (G)=464.3, Throughput (it/s)=124.22026.03 | 15.32 | 93.6 | — | — | — | |
| DiG-XL/2Training iterations=400K2026.03 | 17.36 | 69.4 | 75 | 63 | — | |
| DiT-XL/2Training iterations=400K, FLOPs (G)=524.7, Throughput (it/s)=18.62026.03 | 20.94 | 66.3 | 74 | 58 | — |