Worried output (pixels image)

[JAWHARAH123]

I train stylegan3-t in a custom dataset, during the training I got amazing fake samples. However when I use the weight to produce fake samples I get worried output as attached

Is there any idea how to fix the gen_images.py code to reproduce the training samples, please ?

Desktop (please complete the following information):

• OS: Linux Ubuntu 20.04
• PyTorch version 1.9.0
• CUDA toolkit version 12.2
• NVIDIA driver version : NVIDIA-SMI 535.113.01
• Docker:No