Closed IceFtv closed 1 day ago
I encountered the same issue when I used the A800 for training. However, when I switched to the RTX 3090, the issue didn't occur anymore.
Hi,
This seems like a cuda-related bug of the distCUDA2
function. I would suggest simply switching to lower-end GPUs since our method is not GPU memory demanding.
Very Nice Work! But when I run: python train.py dataset=ps_female_3 option=iter30k pose_correction=none I got this problem:
I am using an A800 with 80GB GPU memory. Normally, this is sufficient. Could you tell me what's wrong? Thank you very much.