nvae icon indicating copy to clipboard operation
nvae copied to clipboard

An unofficial toy implementation for NVAE 《A Deep Hierarchical Variational Autoencoder》

Results 8 nvae issues
Sort by recently updated
recently updated
newest added

Hi~ Do you implement the resblock which is mentioned in the paper?

Hi, amazing implementation! Very simple to use. Is there any code available for variable control experiments? Best,

Hi! Nice work! I want to modify and apply this model to 1-D data. Do you think it feasible to input 1-D data and apply 1-D convolution in NVAE? Can...

Hi, thank you for the implementation, I am finding it very helpful! I was wondering if there is anyway to obtain latent embeddings using this architecture for visualisation with t-SNE...

Hi, Very nice implementation, thank you. I'd be very much interested in using it with discretized mix logistic distribution. I note you seem to have implemented this function in nvae/loss.py....

When tyring to run the code, nvae/vae_celeba.py throws import error robust_loss_pytorch. The file is missing.

Hi! Very Impressive work, thanks for sharing! I have a question regarding the hierarchical KL loss. As in the original paper, the hierarchical kl loss is stated as: ∑_L [KL(**q(z_l...

Hello! I am curious about the experiment settings. For 64x64 colour images, on which GPU and how many of them did you use? And how long did it trained to...