Image Generation On Binarized Mnist
Métriques
nats
Résultats
Résultats de performance de divers modèles sur ce benchmark
Nom du modèle | nats | Paper Title | Repository |
|---|---|---|---|
| MADE 2hl (32 orders) | 86.64 | MADE: Masked Autoencoder for Distribution Estimation | - |
| CR-NVAE | 76.93 | Consistency Regularization for Variational Auto-Encoders | - |
| Efficient-VDVAE | 79.09 | Efficient-VDVAE: Less is more | - |
| Locally Masked PixelCNN (8 orders) | 77.58 | Locally Masked Convolution for Autoregressive Models | - |
| BFN | 77.87 | Bayesian Flow Networks | - |
| PixelCNN | 81.30 | Pixel Recurrent Neural Networks | - |
| EoNADE-5 2hl (128 orders) | 84.68 | Iterative Neural Autoregressive Distribution Estimator (NADE-k) | - |
| NADE | 88.33 | A Deep and Tractable Density Estimator | - |
| PixelRNN | 79.20 | Pixel Recurrent Neural Networks | - |
| EoNADE 2hl (128 orders) | 85.10 | Iterative Neural Autoregressive Distribution Estimator NADE-k |
0 of 10 row(s) selected.