Image Generation on FFHQ 256 x 256
Metric: bits/dimension (lower is better)
Metric: bits/dimension (lower is better)
| # | Model↕ | bits/dimension▲ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | Efficient-VDVAE | 0.53 | No | Efficient-VDVAE: Less is more | 2022-03-25 | Code |
| 2 | Very Deep VAE | 0.61 | No | Very Deep VAEs Generalize Autoregressive Models ... | 2020-11-20 | Code |
| 3 | NVAE w/ flow | 0.69 | No | NVAE: A Deep Hierarchical Variational Autoencoder | 2020-07-08 | Code |