PyTorch-Tutorial/tutorial-contents/403_RNN_regressor.py contains a weird line of code
BlackTentacle opened this issue · comments
I tried some RNN regression learning based on the code in the "PyTorch-Tutorial/tutorial-contents/403_RNN_regressor.py" file, which did not work for me at all.
According to an accepted answer on stack-overflow (https://stackoverflow.com/questions/52857213/recurrent-network-rnn-wont-learn-a-very-simple-function-plots-shown-in-the-q?noredirect=1#comment92916825_52857213), it turns out that the line
h_state = h_state.data
of the tutorial file does not do what it is supposed to be doing. It does not actually break the connection to the previous iteration as specified in the commment. Rather the line
h_state = None
should do the trick (according to the accepted answer).
Indeed, I guess the comment is for the case whereby elements in batches are unrelated, like the MNIST example, however the code is correct here for continuous data.
Is the bug related to pytorch's version? The pytorch '0.4.1' runs fine.
when I use torchvision.datasets.MNIST, why occurs TypeError: init() missing 1 required positional argument: 'total'