Probably! But maybe less than hoped, as a ton of that 50g...
@cldellow @datasetteproj Probably! But maybe less than hoped, as a ton of that 50g is small integers (i.e. ids) or DB indices, which I doubt compress super-well. Maybe 15gb of it is tokenizable text, which would squeeze down a lot. I'll put it all out there for folks to play with though.
➛