Class-conditional image synthesis on ImageNet 256x256 (val)
1.6FIDMomentum Guidance
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Momentum GuidanceCFG scale w=1.2, NFE=642026.02 | 1.6 | 254.6 | 0.81 | 0.62 | |
| Validation dataAcceptance Rate=1.02021.10 | 1.62 | 235 | — | — | |
| Momentum GuidanceCFG scale w=1.2, NFE=322026.02 | 1.71 | 250.89 | 0.81 | 0.62 | |
| Momentum GuidanceCFG scale w=1.4, NFE=162026.02 | 1.85 | 288.65 | 0.82 | 0.6 | |
| Classifier-Free GuidanceCFG scale w=1.2, NFE=642026.02 | 1.89 | 239.56 | 0.79 | 0.62 | |
| Momentum GuidanceCFG scale w=1.4, NFE=322026.02 | 1.9 | 300.71 | 0.84 | 0.59 | |
| Momentum GuidanceCFG scale w=1.4, NFE=642026.02 | 1.99 | 306.02 | 0.84 | 0.59 | |
| Momentum GuidanceCFG scale w=1.2, NFE=162026.02 | 2 | 238.29 | 0.8 | 0.61 | |
| Classifier-Free GuidanceCFG scale w=1.4, NFE=642026.02 | 2.03 | 301.29 | 0.84 | 0.58 | |
| Classifier-Free GuidanceCFG scale w=1.4, NFE=322026.02 | 2.04 | 293.03 | 0.84 | 0.58 | |
| Classifier-Free GuidanceCFG scale w=1.2, NFE=322026.02 | 2.2 | 230.71 | 0.79 | 0.62 | |
| DiT-XL/2†Model Category=Diffusion, Size=675M, Leverages pretrained ImageNet classifier=false2023.03 | 2.27 | 278.24 | — | — | |
| StyleGAN-XLModel Category=GAN, Size=166M, Leverages pretrained ImageNet classifier=false2023.03 | 2.32 | 297.62 | 0.82 | 0.61 | |
| Classifier-Free GuidanceCFG scale w=1.4, NFE=162026.02 | 2.38 | 275.06 | 0.83 | 0.57 | |
| Momentum GuidanceCFG scale w=1.6, NFE=162026.02 | 2.62 | 330.08 | 0.85 | 0.56 | |
| Momentum GuidanceCFG scale w=1.6, NFE=322026.02 | 2.89 | 342.27 | 0.85 | 0.56 | |
| ViT-VQGANAcceptance Rate=0.52021.10 | 3.04 | 227.4 | — | — | |
| Classifier-Free GuidanceCFG scale w=1.6, NFE=162026.02 | 3.13 | 325.55 | 0.86 | 0.52 | |
| Classifier-Free GuidanceCFG scale w=1.6, NFE=322026.02 | 3.17 | 340.88 | 0.87 | 0.54 | |
| Momentum GuidanceCFG scale w=1.6, NFE=642026.02 | 3.17 | 349.04 | 0.85 | 0.56 | |
| Momentum GuidanceCFG scale w=1.0, NFE=642026.02 | 3.26 | 179.66 | 0.74 | 0.67 | |
| Classifier-Free GuidanceCFG scale w=1.2, NFE=162026.02 | 3.26 | 212.83 | 0.78 | 0.6 | |
| Classifier-Free GuidanceCFG scale w=1.6, NFE=642026.02 | 3.34 | 348.87 | 0.87 | 0.54 | |
| GigaGANModel Category=GAN, Size=569M, Leverages pretrained ImageNet classifier=false2023.03 | 3.45 | 225.52 | 0.84 | 0.61 | |
| Momentum GuidanceCFG scale w=1.8, NFE=162026.02 | 3.49 | 353.77 | 0.85 | 0.54 | |
| Momentum GuidanceCFG scale w=1.0, NFE=322026.02 | 3.58 | 176.1 | 0.74 | 0.67 | |
| LDM-4-GModel Category=Diffusion, Size=400M, Leverages pretrained ImageNet classifier=true2023.03 | 3.6 | 247.67 | — | — | |
| RQ-TransformerModel Category=Transformer, Size=3.8B, Leverages pretrained ImageNet classifier=false2023.03 | 3.83 | 339.41 | 0.85 | 0.6 | |
| Momentum GuidanceCFG scale w=1.8, NFE=322026.02 | 3.96 | 370.72 | 0.86 | 0.53 | |
| ADM-G-UModel Category=Diffusion, Size=726M, Leverages pretrained ImageNet classifier=true2023.03 | 4.01 | 240.24 | 0.85 | 0.62 | |
| ViT-VQGANAcceptance Rate=1.02021.10 | 4.17 | 175.1 | — | — | |
| Momentum GuidanceCFG scale w=1.0, NFE=162026.02 | 4.46 | 165.85 | 0.73 | 0.66 | |
| ADM-GModel Category=Diffusion, Size=608M, Leverages pretrained ImageNet classifier=true2023.03 | 4.48 | 207.86 | 0.84 | 0.62 | |
| Classifier-Free GuidanceCFG scale w=1.8, NFE=162026.02 | 4.48 | 363.56 | 0.89 | 0.49 | |
| ADM-GAcceptance Rate=1.0, Guidance=1.0 guid.2021.10 | 4.59 | 186.7 | — | — | |
| Momentum GuidanceCFG scale w=1.8, NFE=642026.02 | 4.62 | 382.04 | 0.87 | 0.51 | |
| Momentum GuidanceCFG scale w=2.0, NFE=162026.02 | 4.62 | 382.79 | 0.87 | 0.51 | |
| Classifier-Free GuidanceCFG scale w=1.0, NFE=642026.02 | 4.75 | 163.5 | 0.72 | 0.67 | |
| Classifier-Free GuidanceCFG scale w=1.8, NFE=322026.02 | 4.76 | 377.39 | 0.89 | 0.49 | |
| CDMModel Category=Diffusion, Leverages pretrained ImageNet classifier=true2023.03 | 4.88 | 158.71 | — | — | |
| Classifier-Free GuidanceCFG scale w=1.8, NFE=642026.02 | 4.99 | 383.6 | 0.89 | 0.49 | |
| VQ-GANModel Category=Transformer, Size=1.4B, Leverages pretrained ImageNet classifier=false2023.03 | 5.2 | 314.61 | 0.81 | 0.57 | |
| Momentum GuidanceCFG scale w=2.0, NFE=322026.02 | 5.27 | 397.44 | 0.88 | 0.49 | |
| Mask-GITModel Category=Transformer, Size=227M, Leverages pretrained ImageNet classifier=false2023.03 | 5.4 | 216.38 | 0.87 | 0.6 | |
| Classifier-Free GuidanceCFG scale w=1.0, NFE=322026.02 | 5.57 | 156.1 | 0.72 | 0.67 | |
| Classifier-Free GuidanceCFG scale w=2.0, NFE=162026.02 | 5.94 | 392.5 | 0.9 | 0.45 | |
| Momentum GuidanceCFG scale w=2.0, NFE=642026.02 | 6.08 | 407.3 | 0.89 | 0.48 | |
| Classifier-Free GuidanceCFG scale w=2.0, NFE=322026.02 | 6.36 | 403.22 | 0.9 | 0.46 | |
| Classifier-Free GuidanceCFG scale w=2.0, NFE=642026.02 | 6.62 | 408.68 | 0.9 | 0.46 | |
| BigGAN-deepAcceptance Rate=1.02021.10 | 6.84 | 203.6 | — | — | |
| BigGAN-DeepModel Category=GAN, Size=112M, Leverages pretrained ImageNet classifier=false2023.03 | 6.95 | 224.46 | 0.89 | 0.38 | |
| VQGANAcceptance Rate=0.252021.10 | 7.35 | 188.6 | — | — | |
| BigGANAcceptance Rate=1.02021.10 | 7.53 | 168.6 | — | — | |
| LDM-8-GModel Category=Diffusion, Size=506M, Leverages pretrained ImageNet classifier=true2023.03 | 7.76 | 209.52 | — | — | |
| Classifier-Free GuidanceCFG scale w=1.0, NFE=162026.02 | 7.76 | 140.89 | 0.7 | 0.65 | |
| VQGANAcceptance Rate=0.52021.10 | 10.26 | 125.5 | — | — | |
| ADM-GAcceptance Rate=1.0, Guidance=no guid.2021.10 | 10.94 | 101 | — | — | |
| IDDPMAcceptance Rate=1.02021.10 | 12.3 | — | — | — | |
| VQGANAcceptance Rate=1.02021.10 | 17.04 | 70.6 | — | — | |
| VQVAE-2Acceptance Rate=1.02021.10 | 31 | 45 | — | — | |
| DCTransformerAcceptance Rate=1.02021.10 | 36.5 | — | — | — |