Density Estimation On Cifar 10
Métriques
NLL (bits/dim)
Résultats
Résultats de performance de divers modèles sur ce benchmark
Tableau comparatif
| Nom du modèle | NLL (bits/dim) |
|---|---|
| denoising-diffusion-probabilistic-models | 3.69 |
| masked-autoregressive-flow-for-density | - |
| ffjord-free-form-continuous-dynamics-for | 3.4 |
| flow-matching-for-generative-modeling | 2.99 |
| how-to-train-your-neural-ode | 3.38 |
| conditional-image-generation-with-pixelcnn | 3.03 |
| variational-diffusion-models | 2.65 |
| pixelcnn-improving-the-pixelcnn-with | 2.92 |
| diffusion-models-with-learned-adaptive-noise | 2.55 |
| multi-resolution-continuous-normalizing-flows | 3.54 |
| improved-techniques-for-maximum-likelihood | 2.42 |
| image-transformer | 2.90 |
| score-based-generative-modeling-through-1 | 2.99 |
| generative-modeling-with-bayesian-sample | 2.64 |