if you’re training the model with
torch.nn.DataParallel
encapsulates the model
then please set multiple GPU
CUDA = '0,1'
os.environ['CUDA_VISIBLE_DEVICES'] = CUDA
div>
if you’re training the model with
torch.nn.DataParallel
encapsulates the model
then please set multiple GPU
CUDA = '0,1'
os.environ['CUDA_VISIBLE_DEVICES'] = CUDA
div>