Fastai Docs about text

I was reading the documentation about text processing, and trying to run the toy example provided (IMDB).

The dataset is correctly downloaded, untarred, and shown as pandas dataframe.

But as I try to instantiate the databunch for the language model, it searches for train.csv in the same dir where it untarred the sample. But no train.csv is present:

data_lm = TextLMDataBunch.from_csv(path, 'texts.csv')


FileNotFoundError: File b'/home/poko/.fastai/data/imdb_sample/train.csv' does not exist

In that directory, only texts.csv does actually exist.

Note that I’m following the tutorial step by step here.

Should we split texts.csv manually?

You need the latest version of fastai :wink:

Heck, I upgraded just a few days ago.

But I realize you all are working continuously to improve the library, which is a commendable effort :slight_smile: