search

zc-alexfan / hold

[CVPR 2024✨Highlight] Official repository for HOLD, the first method that jointly reconstructs articulated hands and objects from monocular videos without assuming a pre-scanned object template and 3D hand-object training data.
https://zc-alexfan.github.io/hold
MIT License
271 stars 7 forks source link

Training on Mutiple GPUs #15

Closed eccv-workspace closed 2 months ago

eccv-workspace commented 2 months ago

Hi there, @zc-alexfan! Does HOLD support training on mutiple gpus? I've tried to modify train.py so that the trainer can use four gpus at the same time using strategy="ddp" in pl.Trainer() and specifying the devices beforehand, but there seems to be an error saying that the tensors are expected to be on the same device, and therefore I can't use more than one gpu. I'm wondering if there are any methods to solve this problem :D Thank you for your time!

zc-alexfan commented 2 months ago

Hi,

Sorry I didn't try multi-gpus. It should be feasible as the framework is in PL.

eccv-workspace commented 2 months ago

Thanks for the quick reply @zc-alexfan !