> Judging by the increase in accuracy from 16K images to 60K images there is something of a diminishing rate of return here. This is a well-known thing in mach

> Judging by the increase in accuracy from 16K images to 60K images there is something of a diminishing rate of return here. This is a well-known thing in machine learning. As far as I know all ML algorithms have this property, where the learning curve is basically logarithmic and each additive increase in performance requires multiplicatively more training material.