PixelVAE

PixelVAE copied to clipboard

hi, very nice and helpful tflib. I wonder that could you give conv3d.py & deconv3d.py? Thanks a lot !

gxwangupc

I am trying to reproduce the reported results for the LSUN bedrooms dataset Can you provide a script the preprocessing the dataset or the .hdf5 file used for the training?

GaParmar

satyenrajpal

Should the placeholder `all_images` be of `float32` type?

3

Hi, I tried to run the code with imagenet dataset and I land into this following error ``` TypeError: Expected int32 passed to parameter 'y' of op 'Minimum', got 50.0...

Sentient07

one-level network for 64x64 image size datasets

11

Any chance you can add this?

jcpeterson

Is there a way to make image generation faster by caching activations or avoiding using feed_dict?

jcpeterson

Hi, How do you ensure the mask is applied on every iteration, given that it's not a theano tensor? Thanks https://github.com/igul222/PixelVAE/blob/d174aaf440a324f93e3398693163096102504e22/lib/ops/conv2d.py#L104

alexmansbridge

any way to do reconstructions?

1

Do you already have code to do reconstructions? I think it would be enough to have code to get image encodings for images given a trained model, since there's already...

jcpeterson

Hi, I am trying to trace the code and understand the implementation details. I saw the the [hyperparameter alpha_iters](https://github.com/igul222/PixelVAE/blob/d174aaf440a324f93e3398693163096102504e22/pixelvae.py#L813-L814) and the [coefficients of KL divergence terms](https://github.com/igul222/PixelVAE/blob/d174aaf440a324f93e3398693163096102504e22/pixelvae.py#L835-L836), but I have no...

KJLiu