One of multiple pull requests which you might find useful:
Labeler used to download and extract its own model (saved-models.zip), now a single script does both
Downloading and unzipping is done by shell utilities (curl and unzip) because that's way faster for bigger files than Python's requests + zipfile
Downloader creates .<model_filename>.downloaded for every model which has been downloaded and extracted successfully to avoid re-downloading and re-extracting it again
One of multiple pull requests which you might find useful:
saved-models.zip), now a single script does bothrequests+zipfile.<model_filename>.downloadedfor every model which has been downloaded and extracted successfully to avoid re-downloading and re-extracting it again