I jumped right into IMDB course and trying to customize the notebook for my own dataset which is a simple dataframe instead of text files in some folders. I went through the Language model training part and get elegant accuracy but when I try to apply similar code to build a binary classifier, this error shows for lr_find(): “TypeError: ‘str’ object cannot be interpreted as an integer”
My code is like this after having a language model:
data_clas = (TextClasDataBunch.from_df(’’, rev_train_set, rev_val_set, text_cols=0, label_cols=1, vocab=review_lm.vocab))
data_clas = TextClasDataBunch.load(’’, ‘data_clas.pkl’, bs=40)
learn = text_classifier_learner(data_clas, ‘AWD_LSTM’, drop_mult=0.5)
(I’ve also tried updating my fastai package to version 1.0.58 but still not working…)
Please LMK any thoughts on it! Thanks!