Inconsistent GPU behavior when running FitDiT on AWS #41
Description
While running the FitDiT demo on an AWS GPU instance (24 GB VRAM), the model sometimes produces identical outputs or no visible garment changes, even though the run completes successfully.
I did nvidia-smi monitoring and found that it was using GPU. I tried running all the commands mentioned in the GPU , I have cuda 12.8 version install.
It ran one time and when I closes the server and run the gradio3_app.py again it never generates the 2d tried on output. Only giving the same image as input.
Would be great if anyone can help me out, it is urgent!!!!