What's wrong with my train.py?

I run perprocess.py and then run train.py. When it run at Epoch 1, 150/ 157,it stop.
I didn’t change the code.


Could you copy the error log as text (not as image) so that we can see it completely?

raise ValueError("lengths array has to be sorted in decreasing order")

ValueError: lengths array has to be sorted in decreasing order

I guess it is because of valid_iter. There is a bug there. I mentioned it here. It seems not fixed.

use pytorch text v0.1.1 instead of pytorch text v0.2
source code: https://github.com/pytorch/text/archive/v0.1.1.tar.gz

it works for me.