分类：.to(device)

2021
07-12

pytorch中.to(device) 和.cuda()的区别说明

原理.to(device)可以指定CPU或者GPUdevice=torch.device("cuda:0"iftorch.cuda.is_available()else"cpu")#单GPU或者CPUmodel.to(device)#如果是多GPUiftorch.cuda.device_count()>1:model=nn.DataParallel(model，device_ids=[0,1,2])model.to(device).cuda()只能指定GPU#指定某个GPUos.environ['CUDA_VISIBLE_DEVICE']='1'model.cuda()#如果是多GPUos.environment['CUDA_VISIBLE_DEVICES']='0,1,2,3'device_i... 继续阅读 >