Lucas Caccia comments

Results 28 comments of


                                            Lucas Caccia

Hyperparams for results reported in the paper

Hi, thanks for the fast response, and for your insight on PTB! is it also possible to get LM paramaters in the standard setting ?

Hyperparams for results reported in the paper

great! thanks again

Convolution Backward Pass with kernel_size > 1

Great news! Would you mind explaining how to implement handling of border cases ? If possible, could you share code ? Best, Lucas

Convolution Backward Pass with kernel_size > 1

ah smart! Thanks a lot for your solution :)

Replicating the bits / dim results in the original VQ-VAE paper

great! let me know what you end up getting :)

Replicating the bits / dim results in the original VQ-VAE paper

great! it's a very complete list, pretty cool :D

Replicating the bits / dim results in the original VQ-VAE paper

The best I got for VQ-VAE on CIFAR-10 was 5.00 BPD (using a unigram prior). This is still far from the results in the VQ-VAE paper. Hopefully you can manage...

[Question] What is PixelSnail? How to Train it?

would you mind sharing samples ? Just to get an idea of what to expect

[Question] What is PixelSnail? How to Train it?

that's pretty good! thanks for sharing :)

Will there be a torch implementation of it?

I did a pytorch implementation for what it's worth https://github.com/pclucas14/pixel-cnn