Lesson3 IMDB: data load query

In Lesson3 Imdb notebook, Jeremy saved the TextDataBunch for language model as a pickle file. i.e `data_lm.save(‘data_lm.pkl’)

Why didn’t he use feather format to save the file as it faster to load or are feather files not supported in this case?

I have problem like this, I can’t load imdb_sample.tgz file to my notebook
did you find a solution?