Hi there I am going to build NLP from scratch for my language using fastai. I have train.txt and val_data.txt files separately. I followed the NLP chapter of the book but there are some problems.
- First error is splitting train and valid data.
- Second is that there is no metric values in the training process.
And it shows that my generator is empty:
One thing we used AWD_LSTM in the learner. Does it affect the model in other languages?
Anyone can help?