Image Generation on CelebA
2.71FIDF-PNDM
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| F-PNDMsteps=200, variance schedule=linear2022.02 | 2.71 | — | — | — | — | — | — | |
| F-PNDMsteps=250, variance schedule=linear2022.02 | 2.71 | — | — | — | — | — | — | |
| F-PNDMsteps=125, variance schedule=linear2022.02 | 2.75 | — | — | — | — | — | — | |
| F-PNDMsteps=500, variance schedule=linear2022.02 | 2.77 | — | — | — | — | — | — | |
| F-PNDMsteps=100, variance schedule=linear2022.02 | 2.81 | — | — | — | — | — | — | |
| F-PNDMsteps=1000, variance schedule=linear2022.02 | 2.86 | — | — | — | — | — | — | |
| S-PNDMsteps=1000, variance schedule=linear2022.02 | 2.99 | — | — | — | — | — | — | |
| S-PNDMsteps=500, variance schedule=linear2022.02 | 3.01 | — | — | — | — | — | — | |
| S-PNDMsteps=250, variance schedule=linear2022.02 | 3.19 | — | — | — | — | — | — | |
| S-PNDMsteps=200, variance schedule=linear2022.02 | 3.3 | — | — | — | — | — | — | |
| F-PNDMsteps=50, variance schedule=linear2022.02 | 3.34 | — | — | — | — | — | — | |
| DDIM*steps=1000, variance schedule=linear2022.02 | 3.41 | — | — | — | — | — | — | |
| DDIMsteps=1000, variance schedule=linear2022.02 | 3.51 | — | — | — | — | — | — | |
| F-PNDMsteps=40, variance schedule=linear2022.02 | 3.67 | — | — | — | — | — | — | |
| S-PNDMsteps=125, variance schedule=linear2022.02 | 3.72 | — | — | — | — | — | — | |
| DDIM*steps=500, variance schedule=linear2022.02 | 3.75 | — | — | — | — | — | — | |
| S-PNDMsteps=100, variance schedule=linear2022.02 | 4.03 | — | — | — | — | — | — | |
| FONsteps=1000, variance schedule=linear2022.02 | 4.17 | — | — | — | — | — | — | |
| DDIM*steps=250, variance schedule=linear2022.02 | 4.44 | — | — | — | — | — | — | |
| FONsteps=500, variance schedule=linear2022.02 | 4.49 | — | — | — | — | — | — | |
| F-PNDMsteps=25, variance schedule=linear2022.02 | 4.75 | — | — | — | — | — | — | |
| DDIM*steps=200, variance schedule=linear2022.02 | 4.78 | — | — | — | — | — | — | |
| FONsteps=250, variance schedule=linear2022.02 | 5.14 | — | — | — | — | — | — | |
| FONsteps=200, variance schedule=linear2022.02 | 5.45 | — | — | — | — | — | — | |
| F-PNDMsteps=20, variance schedule=linear2022.02 | 5.51 | — | — | — | — | — | — | |
| S-PNDMsteps=50, variance schedule=linear2022.02 | 5.69 | — | — | — | — | — | — | |
| COCO-GANFramework=GAN2020.11 | 5.7 | — | — | — | — | — | — | |
| DDIM*steps=125, variance schedule=linear2022.02 | 5.74 | — | — | — | — | — | — | |
| Dia-HingeGAN2021.02 | 5.98 | — | — | — | — | — | — | |
| FONsteps=125, variance schedule=linear2022.02 | 6.28 | — | — | — | — | — | — | |
| DO-SGAN/CBackbone=SGAN, Strategy=Continual Learning2021.02 | 6.3 | — | — | — | — | — | — | |
| DO-SGAN/PBackbone=SGAN, Strategy=Pruning2021.02 | 6.32 | — | — | — | — | — | — | |
| DDIM*steps=100, variance schedule=linear2022.02 | 6.36 | — | — | — | — | — | — | |
| S-PNDMsteps=40, variance schedule=linear2022.02 | 6.5 | — | — | — | — | — | — | |
| DDIMsteps=100, variance schedule=linear2022.02 | 6.53 | — | — | — | — | — | — | |
| HingeGAN2021.02 | 6.66 | — | — | — | — | — | — | |
| FONsteps=100, variance schedule=linear2022.02 | 6.7 | — | — | — | — | — | — | |
| DO-SNGAN/CBackbone=SNGAN, Strategy=Continual Learning2021.02 | 6.74 | — | — | — | — | — | — | |
| DO-SNGAN/PBackbone=SNGAN, Strategy=Pruning2021.02 | 6.92 | — | — | — | — | — | — | |
| SGAN2021.02 | 6.98 | — | — | — | — | — | — | |
| DO-DCGAN/PBackbone=DCGAN, Strategy=Pruning2021.02 | 7.11 | — | — | — | — | — | — | |
| DO-DCGAN/CBackbone=DCGAN, Strategy=Continual Learning2021.02 | 7.16 | — | — | — | — | — | — | |
| ProGANFramework=GAN2020.11 | 7.3 | — | — | — | — | — | — | |
| SNGAN2021.02 | 7.62 | — | — | — | — | — | — | |
| F-PNDMsteps=10, variance schedule=linear2022.02 | 7.71 | — | — | — | — | — | — | |
| FONsteps=50, variance schedule=linear2022.02 | 8.13 | — | — | — | — | — | — | |
| S-PNDMsteps=25, variance schedule=linear2022.02 | 8.42 | — | — | — | — | — | — | |
| FONsteps=40, variance schedule=linear2022.02 | 8.89 | — | — | — | — | — | — | |
| DDIM*steps=50, variance schedule=linear2022.02 | 8.95 | — | — | — | — | — | — | |
| DDIMsteps=50, variance schedule=linear2022.02 | 9.17 | — | — | — | — | — | — | |
| S-PNDMsteps=20, variance schedule=linear2022.02 | 9.45 | — | — | — | — | — | — | |
| DDIM*steps=40, variance schedule=linear2022.02 | 9.99 | — | — | — | — | — | — | |
| FONsteps=25, variance schedule=linear2022.02 | 10.6 | — | — | — | — | — | — | |
| DCGAN2021.02 | 10.92 | — | — | — | — | — | — | |
| F-PNDMsteps=5, variance schedule=linear2022.02 | 11.3 | — | — | — | — | — | — | |
| FONsteps=20, variance schedule=linear2022.02 | 11.6 | — | — | — | — | — | — | |
| S-PNDMsteps=10, variance schedule=linear2022.02 | 12.2 | — | — | — | — | — | — | |
| DDIM*steps=25, variance schedule=linear2022.02 | 12.3 | — | — | — | — | — | — | |
| TTURarchitecture=DCGAN, learning rate (b, a)=1e-5, 5e-4, updates=225k2017.06 | 12.5 | — | — | — | — | — | — | |
| DDIM*steps=20, variance schedule=linear2022.02 | 13.4 | — | — | — | — | — | — | |
| DDIMsteps=25, variance schedule=linear2022.02 | 13.7 | — | — | — | — | — | — | |
| DuelGANreported_in_original_paper=false2021.01 | 13.95 | — | — | — | — | — | — | |
| DC-VAEFramework=VAE, Generation Mode=Reconstruction2020.11 | 14.3 | — | — | — | — | — | — | |
| DRAGANreported_in_original_paper=false2021.01 | 14.57 | — | — | — | — | — | — | |
| S-PNDMsteps=5, variance schedule=linear2022.02 | 15.2 | — | — | — | — | — | — | |
| WGANreported_in_original_paper=true2021.01 | 15.23 | — | — | — | — | — | — | |
| LSGANreported_in_original_paper=false2021.01 | 15.35 | — | — | — | — | — | — | |
| FONsteps=10, variance schedule=linear2022.02 | 16 | — | — | — | — | — | — | |
| DDIM*steps=10, variance schedule=linear2022.02 | 16.9 | — | — | — | — | — | — | |
| DDIMsteps=10, variance schedule=linear2022.02 | 17.3 | — | — | — | — | — | — | |
| D2GANreported_in_original_paper=false2021.01 | 17.3 | — | — | — | — | — | — | |
| DCGANreported_in_original_paper=false2021.01 | 17.38 | — | — | — | — | — | — | |
| DC-VAEFramework=VAE, Generation Mode=Sampling2020.11 | 19.9 | — | — | — | — | — | — | |
| Original one time-scale update rulearchitecture=DCGAN, learning rate (b=a)=5e-4, updates=70k2017.06 | 21.4 | — | — | — | — | — | — | |
| Dist-GANExperimental Setup=[20]2018.03 | 23.7 | — | — | — | — | — | — | |
| Dist-GANreported_in_original_paper=true2021.01 | 23.7 | — | — | — | — | — | — | |
| DDIM*steps=5, variance schedule=linear2022.02 | 24.4 | — | — | — | — | — | — | |
| OriginalModel=SAGAN, MACs=23.45M2021.10 | 24.87 | — | — | — | — | — | — | |
| GCCModel=SAGAN, MACs=15.45M, Compression Ratio=34.12%2021.10 | 25.21 | — | — | — | — | — | — | |
| WGANGPExperimental Setup=[20]2018.03 | 26.8 | — | — | — | — | — | — | |
| VAEGANExperimental Setup=[20]2018.03 | 27.5 | — | — | — | — | — | — | |
| PresGANFramework=GAN, Resolution=64x642020.11 | 29.1 | — | — | — | — | — | — | |
| WGAN GPCategory=adversarial training2020.07 | 30 | — | — | — | — | — | — | |
| GAN2021.06 | 32.85 | — | — | — | — | — | — | |
| MicroBatchGANreported_in_original_paper=true2021.01 | 34.5 | — | — | — | — | — | — | |
| PruneModel=SAGAN, MACs=15.45M, Compression Ratio=34.12%2021.10 | 36.6 | — | — | — | — | — | — | |
| SIGCategory=iterative, T=12020.07 | 37.3 | — | — | — | — | — | — | |
| BEGANExperimental Setup=[20]2018.03 | 38.1 | — | — | — | — | — | — | |
| WGANCategory=adversarial training2020.07 | 41.3 | — | — | — | — | — | — | |
| WAE-GANFramework=VAE, Resolution=64x642020.11 | 42 | — | — | — | — | — | — | |
| two-stage VAECategory=AE based2020.07 | 44.4 | — | — | — | — | — | — | |
| ms-DRAECategory=AE based2020.07 | 46 | — | — | — | — | — | — | |
| VEE-GANFramework=VAE, Resolution=128x1282020.11 | 46.2 | — | — | — | — | — | — | |
| Best default GANCategory=adversarial training2020.07 | 48 | — | — | — | — | — | — | |
| SWAECategory=AE based, Reference=Wu et al., 20192020.07 | 48.9 | — | — | — | — | — | — | |
| PAECategory=AE based2020.07 | 49.2 | — | — | — | — | — | — | |
| CWAECategory=AE based2020.07 | 49.7 | — | — | — | — | — | — | |
| Wasserstein Iterative NetworksGeneration Function=Average map over maps to input measures, Resolution=64x642022.01 | 52.85 | — | — | — | — | — | — | |
| LSGANExperimental Setup=[20]2018.03 | 53.6 | — | — | — | — | — | — | |
| LSGANFramework=GAN2020.11 | 53.9 | — | — | — | — | — | — |