Anand K Subramanian

Hi, I am glad that you find this code useful. Yes, the numerical issues occur mostly due to the features of the two models having different scales. I usually tackle...

Appreciate the quick response! I tried the `thumbnail` method, but it still resizes by the aspect ratio. 

Another related question - How do I set the interpolation method for resizing the image? According to [docs](https://www.libvips.org/API/current/libvips-resample.html#vips-thumbnail), the default seems to be LANCZOS3. My goal here is to get...

As you said in the title, it is the logarithm of the variance. If the variance is too small (< 1) then `log_var` can be negative.

The FactorVAE as implemented here has this issue. I believe it is because of the way PyTorch Lightning calls the `loss_function` as the `self.D_Z_reserve` variable is updated by the model...

How does your config file look like? If you are training on a single CPU, then simply set the `gpu` field in the config file empty or none.

Try increading the MMD loss weight. Also, play around with different kernels - RBF and IMQ

I have the same issue but even after downloading from your link, I get an error from the `_check_integrity()` function saying that `Dataset not found or corrupted. You can use...

I did set `download=False` after downloading the files manually and checked their size as well. I figured the problem was with the [checkintegrity function](https://github.com/pytorch/vision/blob/main/torchvision/datasets/celeba.py#L132) where it returns `False`. So, I...

Hi, thanks for pointing this out. I had the same problem and fixed it by generating a random string for the image namees during each run.