Image Generation On Celeba 256X256
Métriques
bpd
Résultats
Résultats de performance de divers modèles sur ce benchmark
Tableau comparatif
| Nom du modèle | bpd |
|---|---|
| generating-high-fidelity-images-with-subscale | 0.61 |
| score-based-generative-modeling-in-latent | 0.70 |
| styleswin-transformer-based-gan-for-high-1 | - |
| ncp-vae-variational-autoencoders-with-noise-1 | - |
| macow-masked-convolutional-generative-flow | 0.95 |
| efficient-vdvae-less-is-more | 0.51 |
| improved-transformer-for-high-resolution-gans | - |
| adversarial-latent-autoencoders | - |
| glow-generative-flow-with-invertible-1x1 | 1.03 |
| residual-flows-for-invertible-generative | 0.992 |
| locally-masked-convolution-for-autoregressive | 0.74 |
| generative-latent-flow-a-framework-for-non | - |
| latent-space-factorisation-and-manipulation | - |
| taming-transformers-for-high-resolution-image | - |
| augmented-normalizing-flows-bridging-the-gap | 0.72 |
| nvae-a-deep-hierarchical-variational | 0.70 |
| macow-masked-convolutional-generative-flow | 0.67 |