Reproducibility?

[Prasanna1991]

Hi,

Very good repo. I am also doing some experiments with beta-VAE and your work was helpful. However, I couldn't reproduce the shape as you are showing in one of the notebooks (Peiyao_Sheng_beta_VAE.ipynb). Did you reproduce that using 5000 training samples and in 260 epochs? Or some other configurations were used?

Thanks.

[simplespy]

Hi, I review my log and find that the reconstruction results using 5000 training samples is not good enough so I tried larger size of training set, the result shown in Figure 'reconstruction_260.png' uses train_set = 695810, test_set = 4975. I will revise my report to point it out, thanks.

[Prasanna1991]

Thanks for the info. Good repo though.