Image Generation on MNIST 3/5 (test)
99.52AccuracyWGAN-GP + S2M sampling
Evaluation results
| Method | Links | ||
|---|---|---|---|
| WGAN-GP + S2M samplingsampling=S2M, MCMC iterations=100, backbone=WGAN-GP2022.06 | 99.52 | 0.88 | |
| GenPU2022.06 | 99.33 | 1.93 | |
| RumiLSGAN2022.06 | 77.06 | 13.2 | |
| CP-GAN2022.06 | 66.89 | 19.5 |