Ten thousand books, six million ratings
This initial release contains a ZIP archive of all CSV files in the dataset, as well as ZIP archives for individual files (ratings, books, etc.).
There is also a HDF5 file with integer data: ratings, to_read and book_tags.