Xin Zhang

Hey! Have you handled this problem?

Same problem with A100 -40G. I try to use Xformers to solve it. It works. But my question is why the authors can run it successfully without Xformers

@nitinmukesh @Borismartirosyan In fact, here is my version of xformers xformers == 0.0.22 torch ==1.13.1+cu116 unet_vton.enable_xformers_memory_efficient_attention() unet_garm.enable_xformers_memory_efficient_attention() By the way, torch >= 2.0 is recommended.

Hi! I have same problem with you. Do you follow the [SMPL & Rendering doc](https://github.com/fudan-generative-vision/champ/blob/master/docs/data_process.md) to produce your own motion datas? I guess it's the key. I'm trying to do...