Making an NLP Dataset

So I did the lesson 3/4 IMDB tutorial successfully, but I’m wondering how to make my own dataset for NLP. I have a CSV file in the same format that the text.csv (in IMDB_SAMPLE) is. I saw that the full IMDB data file is separated into three different folders, train, test, and unsupported, each one containing a single movie review. So how can I take my CSV file, and turn it into the format that IMDB is? Thanks in advance!