Closed summersunshine1 closed 2 years ago
Could you provide the line numbers?
In the model.py, it may be necessary for LSTM to do a mask for tokens with paddings. In the solve.py, it may be required for masking the loss of padded tokens.
Thanks for your suggestion. Please feel free to try it yourself and see how it works.
I have a question about padding. It seems that the training of the model misses mask operation for padding.