Class conditional image generation on ImageNet
2.21FIDSODA
Evaluation Results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SODASampling=DDPM, Ns=1252026.03 | 2.21 | 272.01 | — | 82 | 58 | — | 4.72 | — | — | — | — | 1.778 | 76.7 | 1.55 | |
| DiT-XL/2Sampling=DDPM, Steps=2502026.03 | 2.23 | 275.65 | — | 82 | 58 | — | 4.57 | — | — | — | — | 2.508 | 118.68 | 1 | |
| DiT-XL/2Sampling=DDIM, Steps=502026.03 | 2.25 | 239.97 | — | 80 | 59 | — | 4.33 | — | — | — | — | 0.533 | 23.74 | 1 | |
| DDIM-50 stepsBackbone=DiT-XL/2, Sampling steps=50, MACs (T)=11.44, Speedup=1.00×2026.04 | 2.26 | — | — | 80 | 60 | — | 4.29 | — | — | — | — | — | — | — | |
| Full-ComputeNFC=50, FLOPs (T)=23.74, Latency (s)=7.612026.02 | 2.32 | 241.25 | — | — | — | — | 4.32 | — | — | — | — | — | — | — | |
| ResilPhaseN=3, O=4, Speed=2.78×, Backbone=DiT-XL/22026.06 | 2.347 | 233.57 | — | — | — | — | 4.672 | — | — | — | — | — | — | — | |
| DDIMSampling steps=50, Speed=1.00×, Backbone=DiT-XL/22026.06 | 2.367 | 236.06 | — | — | — | — | 4.387 | — | — | — | — | — | — | — | |
| SODASampling=DDIM, Ns=232026.03 | 2.39 | 240.75 | — | 81 | 58 | — | 4.55 | — | — | — | — | 0.314 | 12.21 | 1.94 | |
| DDIM-40 stepsBackbone=DiT-XL/2, Sampling steps=40, MACs (T)=9.14, Speedup=1.24×2026.04 | 2.39 | — | — | 80 | 59 | — | 4.28 | — | — | — | — | — | — | — | |
| SODASampling=DDPM, Ns=722026.03 | 2.47 | 262.3 | — | 81 | 59 | — | 5.09 | — | — | — | — | 1.07 | 43.42 | 2.73 | |
| RFCm=2, NFC=14.01, FLOPs (T)=6.67, Latency (s)=2.862026.02 | 2.52 | 231 | — | — | — | — | 4.6 | — | — | 0.3 | 1.33 | — | — | — | |
| TaylorSeerN=4, m=2, NFC=14, FLOPs (T)=6.66, Latency (s)=2.842026.02 | 2.55 | 232.26 | — | — | — | — | 5.3 | — | — | 0.44 | 2.17 | — | — | — | |
| ToCaSampling=DDPM2026.03 | 2.58 | 256.26 | — | 80 | 59 | — | 5.74 | — | — | — | — | 1.211 | 43.22 | 2.75 | |
| DuCaSampling=DDPM2026.03 | 2.59 | 256.36 | — | 80 | 59 | — | 5.68 | — | — | — | — | 1.17 | 43.42 | 2.73 | |
| SpeCaτ0=0.1, β=0.5, Speed=2.65×, Backbone=DiT-XL/22026.06 | 2.633 | 231.32 | — | — | — | — | 4.848 | — | — | — | — | — | — | — | |
| TaylorSeerN=5, m=2, NFC=11, FLOPs (T)=5.24, Latency (s)=2.522026.02 | 2.7 | 228.76 | — | — | — | — | 5.45 | — | — | 0.56 | 2.74 | — | — | — | |
| RFCm=2, NFC=11.01, FLOPs (T)=5.24, Latency (s)=2.522026.02 | 2.71 | 227.72 | — | — | — | — | 4.8 | — | — | 0.48 | 2.14 | — | — | — | |
| RFCm=2, NFC=10.04, FLOPs (T)=4.78, Latency (s)=2.462026.02 | 2.75 | 226.93 | — | — | — | — | 5.02 | — | — | 0.54 | 2.43 | — | — | — | |
| SODASampling=DDIM, Ns=182026.03 | 2.75 | 235.65 | — | 80 | 58 | — | 4.56 | — | — | — | — | 0.263 | 9.55 | 2.49 | |
| FORASampling=DDPM2026.03 | 2.8 | — | — | 80 | 59 | — | 6.13 | — | — | — | — | — | 43.36 | 2.74 | |
| DiffSparseBackbone=DiT-XL/2, Sampling steps=50, MACs (T)=4.97, Speedup=2.07×2026.04 | 2.81 | — | — | 80 | 59 | — | 4.61 | — | — | — | — | — | — | — | |
| ResilPhaseN=5, O=3, Speed=4.41×, Backbone=DiT-XL/22026.06 | 2.832 | 219.19 | — | — | — | — | 5.026 | — | — | — | — | — | — | — | |
| SODASampling=DDIM, Ns=162026.03 | 2.88 | 234.91 | — | 80 | 58 | — | 4.71 | — | — | — | — | 0.218 | 8.43 | 2.82 | |
| DDIM-25 stepsBackbone=DiT-XL/2, Sampling steps=25, MACs (T)=5.73, Speedup=1.96×2026.04 | 3.01 | — | — | 79 | 58 | — | 4.6 | — | — | — | — | — | — | — | |
| ToCaN=4, NFC=17, FLOPs (T)=8.73, Latency (s)=3.642026.02 | 3.03 | 229.01 | — | — | — | — | 5.02 | — | — | 0.86 | 3.56 | — | — | — | |
| DuCaBackbone=DiT-XL/2, Sampling steps=50, MACs (T)=4.94, Speedup=2.10×2026.04 | 3.04 | — | — | 79 | 57 | — | 4.7 | — | — | — | — | — | — | — | |
| ToCaSampling=DDIM2026.03 | 3.05 | — | — | 79 | 57 | — | 4.7 | — | — | — | — | 0.289 | 10.23 | 2.32 | |
| DuCaSampling=DDIM2026.03 | 3.05 | 233.21 | — | 80 | 58 | — | 4.66 | — | — | — | — | 0.276 | 9.58 | 2.48 | |
| ToCaBackbone=DiT-XL/2, Sampling steps=50, MACs (T)=4.97, Speedup=2.09×2026.04 | 3.05 | — | — | 79 | 57 | — | 4.7 | — | — | — | — | — | — | — | |
| TaylorSeerN=6, m=2, NFC=10, FLOPs (T)=4.76, Latency (s)=2.272026.02 | 3.1 | 222.85 | — | — | — | — | 6.47 | — | — | 1 | 4.63 | — | — | — | |
| RFCm=2, NFC=8.02, FLOPs (T)=3.83, Latency (s)=2.152026.02 | 3.12 | 219.2 | — | — | — | — | 5.07 | — | — | 0.81 | 3.1 | — | — | — | |
| ClusCaN=6, K=8, O=4, Speed=2.61×, Backbone=DiT-XL/22026.06 | 3.251 | 217.6 | — | — | — | — | 5.207 | — | — | — | — | — | — | — | |
| DuCaN=4, NFC=17, FLOPs (T)=7.66, Latency (s)=2.942026.02 | 3.39 | 224.03 | — | — | — | — | 4.95 | — | — | 1.19 | 4.22 | — | — | — | |
| RFCm=2, NFC=7.04, FLOPs (T)=3.37, Latency (s)=1.992026.02 | 3.4 | 215.39 | — | — | — | — | 5.21 | — | — | 1.03 | 3.66 | — | — | — | |
| TaylorSeerN=7, m=2, NFC=8, FLOPs (T)=3.82, Latency (s)=2.022026.02 | 3.46 | 217.18 | — | — | — | — | 6.97 | — | — | 1.3 | 5.61 | — | — | — | |
| DDIM-20 stepsBackbone=DiT-XL/2, Sampling steps=20, MACs (T)=4.58, Speedup=2.42×2026.04 | 3.48 | — | — | 79 | 56 | — | 4.64 | — | — | — | — | — | — | — | |
| ToCaN=5, NFC=14, FLOPs (T)=7.43, Latency (s)=3.352026.02 | 3.53 | 224.07 | — | — | — | — | 5.36 | — | — | 1.28 | 4.5 | — | — | — | |
| ToCaN=6, NFC=13, FLOPs (T)=7.01, Latency (s)=3.272026.02 | 3.54 | 226.32 | — | — | — | — | 5.54 | — | — | 1.44 | 5.18 | — | — | — | |
| DiT-XL/2Sampling Steps=20, Backbone=DiT-XL/22026.03 | 3.56 | 221.27 | — | 78 | 58 | — | — | — | — | — | — | 1.71 | 9.49 | — | |
| LDM-4-GNparams=400M, Sampling steps=250 steps, Classifier-free guidance (c.f.g.)=true, Guidance scale (s)=1.52021.12 | 3.6 | — | — | 87 | 48 | — | — | — | — | — | — | — | — | — | |
| DDIMSampling steps=20, Speed=2.50×, Backbone=DiT-XL/22026.06 | 3.858 | 216.76 | — | — | — | — | 5.201 | — | — | — | — | — | — | — | |
| FORABackbone=DiT-XL/2, Sampling steps=50, MACs (T)=4.13, Speedup=2.12×2026.04 | 3.88 | — | — | 79 | 56 | — | 6.74 | — | — | — | — | — | — | — | |
| ToCaN=7, NFC=14, FLOPs (T)=6.25, Latency (s)=2.92026.02 | 4.04 | 214.25 | — | — | — | — | 5.72 | — | — | 1.62 | 4.72 | — | — | — | |
| GENIEAFS=false, Denoising=true, NFEs=252022.10 | 4.13 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ViT-VQGANModel Type=Autoregressive, #Params=2.2B2023.05 | 4.17 | 175.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| GENIEAFS=true, Denoising=true, NFEs=252022.10 | 4.27 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FORAN=4, NFC=14, FLOPs (T)=6.66, Latency (s)=2.372026.02 | 4.33 | 215.33 | — | — | — | — | 7.38 | — | — | 1.9 | 6.32 | — | — | — | |
| GENIEAFS=false, Denoising=true, NFEs=202022.10 | 4.48 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ADM-Gclassifier guidance=1.02021.11 | 4.59 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ADM-GNparams=608M, Sampling steps=250 DDIM steps2021.12 | 4.59 | 186.7 | — | 82 | 52 | — | — | — | — | — | — | — | — | — | |
| GENIEAFS=true, Denoising=true, NFEs=202022.10 | 4.68 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FORAN=4, Speed=2.60×, Backbone=DiT-XL/22026.06 | 4.77 | 210.98 | — | — | — | — | 7.591 | — | — | — | — | — | — | — | |
| TaylorSeerN=8, O=3, Speed=2.14×, Backbone=DiT-XL/22026.06 | 4.807 | 197.22 | — | — | — | — | 7.088 | — | — | — | — | — | — | — | |
| TaylorSeerN=9, m=2, NFC=7, FLOPs (T)=3.35, Latency (s)=1.892026.02 | 4.9 | 198.46 | — | — | — | — | 7.92 | — | — | 2.33 | 7.35 | — | — | — | |
| GENIEAFS=false, Denoising=false, NFEs=252022.10 | 5 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DDIMAFS=false, Denoising=true, NFEs=252022.10 | 5.19 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GENIEAFS=false, Denoising=true, NFEs=152022.10 | 5.23 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GENIEAFS=true, Denoising=false, NFEs=252022.10 | 5.27 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VQ-Diffusionacceptance rate=0.052021.11 | 5.32 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DDIMAFS=true, Denoising=true, NFEs=252022.10 | 5.35 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GENIEAFS=false, Denoising=false, NFEs=202022.10 | 5.36 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GENIEAFS=true, Denoising=true, NFEs=152022.10 | 5.36 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GENIEAFS=true, Denoising=false, NFEs=202022.10 | 5.73 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DDIMAFS=false, Denoising=true, NFEs=202022.10 | 5.83 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VQGANacceptance rate=0.052021.11 | 5.88 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DDIMAFS=true, Denoising=true, NFEs=202022.10 | 5.93 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DuCaN=5, NFC=14, FLOPs (T)=6.32, Latency (s)=2.572026.02 | 6.02 | 196.96 | — | — | — | — | 6.71 | — | — | 4.19 | 8.18 | — | — | — | |
| StackformerModel Type=Autoregressive, #Params=651M2023.05 | 6.04 | 172.6 | — | — | — | — | — | — | — | — | — | — | — | — | |
| FORAN=5, NFC=11, FLOPs (T)=5.24, Latency (s)=2.032026.02 | 6.12 | 196.04 | — | — | — | — | 10.51 | — | — | 3.57 | 10.19 | — | — | — | |
| GENIEAFS=false, Denoising=false, NFEs=152022.10 | 6.13 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ToCaN=5, R=93%, Speed=2.67×, Backbone=DiT-XL/22026.06 | 6.321 | 196.06 | — | — | — | — | 7.027 | — | — | — | — | — | — | — | |
| DuCaN=6, NFC=13, FLOPs (T)=5.86, Latency (s)=2.532026.02 | 6.37 | 187.74 | — | — | — | — | 6.74 | — | — | 4.51 | 7.86 | — | — | — | |
| GENIEAFS=true, Denoising=false, NFEs=152022.10 | 6.59 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DDIMAFS=false, Denoising=false, NFEs=252022.10 | 6.64 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CR-BigGAN*Noise truncation=false, Consistency regularization=true2019.10 | 6.66 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BigGAN-deep2021.11 | 6.84 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SpeCaτ0=1.5, β=0.5, Speed=4.36×, Backbone=DiT-XL/22026.06 | 6.866 | 171.68 | — | — | — | — | 8.25 | — | — | — | — | — | — | — | |
| DDIMAFS=true, Denoising=false, NFEs=252022.10 | 6.92 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TC-PadéCache Type=Prediction-based, Mode=fast, Backbone=DiT-XL/22026.03 | 6.93 | 185.12 | — | 72 | 54 | — | — | — | — | — | — | 1.17 | 5.79 | — | |
| S-PNDMAFS=false, Denoising=true, NFEs=252022.10 | 6.94 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| S-PNDMAFS=true, Denoising=true, NFEs=252022.10 | 6.94 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DDIM-GMM-ORTHO-VUBSteps=10, Guidance Scale=2.52023.11 | 6.94 | 207.85 | — | — | — | — | — | — | — | — | — | — | — | — | |
| BigGan-deepNparams=340M2021.12 | 6.95 | — | — | 87 | 28 | — | — | — | — | — | — | — | — | — | |
| DDIMAFS=true, Denoising=true, NFEs=152022.10 | 7.14 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DDIMAFS=false, Denoising=true, NFEs=152022.10 | 7.17 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| S-PNDMAFS=false, Denoising=false, NFEs=252022.10 | 7.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| S-PNDMAFS=true, Denoising=false, NFEs=252022.10 | 7.22 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GENIEAFS=true, Denoising=true, NFEs=102022.10 | 7.41 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ImageBARTacceptance rate=0.052021.11 | 7.44 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| S-PNDMAFS=true, Denoising=true, NFEs=202022.10 | 7.52 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BigGAN2021.11 | 7.53 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RQ-TransformerModel Type=Autoregressive, #Params=3.8B2023.05 | 7.55 | 134 | — | — | — | — | — | — | — | — | — | — | — | — | |
| DDIMAFS=false, Denoising=false, NFEs=202022.10 | 7.57 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GENIEAFS=false, Denoising=true, NFEs=102022.10 | 7.59 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| S-PNDMAFS=false, Denoising=true, NFEs=202022.10 | 7.59 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F-PNDMAFS=true, Denoising=false, NFEs=252022.10 | 7.74 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BigGAN*Noise truncation=false, Implementation=Kurach et al. (2019)2019.10 | 7.75 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DuCaN=7, NFC=14, FLOPs (T)=4.96, Latency (s)=2.242026.02 | 7.76 | 181.5 | — | — | — | — | 7.77 | — | — | 5.45 | 8.8 | — | — | — | |
| TaylorSeerCache Type=Prediction-based, N (Cache interval)=3, O (Order of Taylor expansion)=2, Backbone=DiT-XL/22026.03 | 7.84 | 175.99 | — | 71 | 53 | — | — | — | — | — | — | 1.19 | 3.8 | — | |
| DDIMAFS=true, Denoising=false, NFEs=202022.10 | 7.86 | — | — | — | — | — | — | — | — | — | — | — | — | — |