Yeah could use some more info. I added examples for the Python API. There’s some benchmarks buried in there. But it’d be nice to have a simple comparison in the readme to loading from a CSV or something like that.