Feature request: untar_data URLs.ML_100K

[raybellwaves (Ray Bell)]

Just going through course-v3/nbs/dl1/lesson4-collab.ipynb at master · fastai/course-v3 · GitHub

Came across the first instance where I had to download data instead of running untar_data with the Movielens 100k data dataset.

Is there a way this can be read in on the command line without having to click and download the file?

Lesson 5 In-Class Discussion

Yes, I had to download the ml-100k folder to that location and correct the path such that the following path was put into the folder: new: path=Config.data_path()/'movie_lens_sample'/'ml-100k' old: path=Config.data_path()/'ml-100k' Then it showed up correctly for me.

I’ve noticed in same cases when running untar_data it downloads data from an S3 bucket.

Could we add a ‘Collaborative filtering’ section to https://course.fast.ai/datasets and put the data on AWS with reference to the original dataset of course.

I would be more than happy with this endeavour given a little guidance.

Some reference docs:
https://docs.fast.ai/datasets.html#untar_data
https://docs.fast.ai/datasets.html#download_data

[sgugger]

This is not an open-source dataset and its readme explicitly says not to redistribute. That’s why we can’t host it on our S3 bucket.