When I run this code, something is wrong…
vgg = VGGNet()
if use_cuda:
# vgg.cuda()
vgg = torch.nn.DataParallel(vgg, [0, 1, 2])for step in range(config.total_step):
# Extract multiple(5) conv feature vectors target_features = vgg(target) # error
target is in GPU_0
How to solve this? Thanks~