Weird looking outputs with random artificats/patterns

[its-nmt05]

After training the model for around 1000 epochs, I am getting weird patterns/artifacts as output instead of meaningful images. I have not been able to trace out any particular reason for this issue yet. PS. Will update the issue if I find one.

Take a look at some of the outputs here 👇

With GroupNorm

with batchNorm

You can find all the code here: https://github.com/its-nmt05/DeepLLMs/tree/main/diffusion