My code run well under torch 0.2, however, the following error occurs when I use torch 0.3.1, why?

Traceback (most recent call last):
File "../train.py", line 149, in
loss.backward()
File "/home/jxliu/.local/lib/python3.6/site-packages/torch/autograd/variable.py", line 167, in backward
torch.autograd.backward(self, gradient, retain_graph, create_graph, retain_variables)
File "/home/jxliu/.local/lib/python3.6/site-packages/torch/autograd/init.py", line 99, in backward
variables, grad_variables, retain_graph)
RuntimeError: dimension specified as 0 but tensor has no dimensions