For text classifier, I am loading the data from a CSV file post-training language model.
data = (TextList.from_csv('/content/models/processed', 'cleanedData.csv','text', vocab=data_lm.vocab) .split_by_rand_pct(0.2) .label_from_df(cols=0) .databunch(bs = bs))
It creates the desired data bunch but when I try to do lr_find() I get #na# in my valid_loss. Whereas using the same command with same file for language modelling works. The only thing different is vocab and label_from_df.
learn = text_classifier_learner(data, AWD_LSTM, drop_mult=0.5) learn.load_encoder('/content/models/results/newsEncoder-train-epoch-11') learn.lr_find()
I have tried giving start and end range for lr_find() but still no result.