loss.backward() one of the variables needed for gradient computation has been modified by an inplace operation error

Question

I can\'t figure out the error, I went exactly according to the pytorch tutorials but loss.backward() is giving an error.

 def forward(self, x):
        hidden_inp