RuntimeError: grad can be implicitly created only for scalar outputs

[gsx0015]

When I use a single GPU, it works fine.
But with model = torch.nn.DataParallel(model, device_ids=[0, 1]), I get an error.
How to solve this problem?

---

i try "loss.backward(torch.Tensor([1, 1])) or loss.sum().backward().", but not work.

[shaoxiongji]

Hi, I have the same error. Have you solved this problem?

[Flova]

Multiple GPUs are not supported at the moment. Thread for multi GPU training #520