[j4k0bfr]

It's a tough problem, since I assume any common proof-of-work standard will inevitably generate a big enough dataset to train on.

But nonetheless a problem in desperate need of solving!