Qijia Shen

Thank you very much for your reply. However the GPU memory is still unbalanced after adding these two lines at here https://github.com/NVlabs/eg3d/blob/493076071bc400be3289dd6725ef1eef0a552a9d/eg3d/training/training_loop.py#L129  Am I adding to the correct place?

I was also using 8 24G GPU, and I could set the total batch size to 64 without having memory issue. The dataset I used was shapenet cars.

I just reproduced your error on another machine. It turns out that nvcc isn't correctly installed on that machine. Not sure if that'll be the same cause for your problem.

However, when I set batch-gpu=8, gpus=8, batch=64, the GPU memory consumption reduced. It's so weird, I'm wondering if someone might know any clue about this? 

你好！想问一下你有没有试过使用stylegan2ada的stylegan xl呀？似乎默认并不支持stylegan2

Hi，我不知道你们的例子是怎么样的呀？我比较感兴趣stylegan2 能不能用于stylegan_xl训练