Multi-GPU Training

[LcenArthas]

Hi, Have you tried to run training on multiple gpus?

[JKBox]

Hi, Have you tried to run training on multiple gpus?

Thanks to your reminder, I wrote the code with single gpu, I will change it to multiple gpus later.

[LcenArthas]

I tried, but failed TAT.....,but i found this： https://github.com/ultralytics/yolov3/pull/121 . I tried to fix the code,but failed. I hope it can help u :)

[LcenArthas]

by the way. i have fix the code follow by that url, and it can run in the multiple gpus, but it sooooo slow. So i think i have made the wrong code

[longxianlei]

os.environ["CUDA_VISIBLE_DEVICES"] = "4,5,6,7" if torch.cuda.device_count() > 1: model = nn.DataParallel(model, device_ids=[0, 1, 2, 3]) model.to(device).train() I have 8 GPUs. I set 4 of my device visiable. Then i use the model to parallel to these GPUs. but when i run the train.py. inter_area = torch.min(box1, box2).prod(2) RuntimeError: Expected object of type torch.cuda.FloatTensor but found type torch.FloatTensor for argument #2 'other' Is the code didn't support multi GPU training now.

[JKBox]

os.environ["CUDA_VISIBLE_DEVICES"] = "4,5,6,7" if torch.cuda.device_count() > 1: model = nn.DataParallel(model, device_ids=[0, 1, 2, 3]) model.to(device).train() I have 8 GPUs. I set 4 of my device visiable. Then i use the model to parallel to these GPUs. but when i run the train.py. inter_area = torch.min(box1, box2).prod(2) RuntimeError: Expected object of type torch.cuda.FloatTensor but found type torch.FloatTensor for argument #2 'other' Is the code didn't support multi GPU training now.

yes, the code only support single GPU training currently, I'll fix it later