Getting it running was as easy as `git clone && docker run` and it actually worked, which I appreciate given how involved other ML projects tend to be.