Our full dataset uses around 20GB of storage and is distributed as a multi-part zip archive (13 files, each smaller than 2GB). We provide a python script that downloads and assembles the archive for your convenience. The parts can also be downloaded individually below.