leela-zero 0.17-1build2 (armhf binary) in ubuntu hirsute

 A Go program with no human provided knowledge. Using MCTS (but without Monte
 Carlo playouts) and a deep residual convolutional neural network stack.
 .
 This is a fairly faithful reimplementation of the system described in the
 Alpha Go Zero paper "Mastering the Game of Go without Human Knowledge". For
 all intents and purposes, it is an open source AlphaGo Zero.
 .
 https://deepmind.com/documents/119/agz_unformatted_nature.pdf
 .
 No network weights are in this repository. If you manage to obtain the AlphaGo
 Zero weights, this program will be about as strong, provided you also obtain a
 few Tensor Processing Units. Lacking those TPUs, the author recommends a top
 of the line GPU - it's not exactly the same, but the result would still be an
 engine that is far stronger than the top humans.
 .
 Recomputing the AlphaGo Zero weights will take about 1700 years on commodity
 hardware. Upstream is running a public, distributed effort to repeat this
 work. Working together, and especially when starting on a smaller scale, it
 will take less than 1700 years to get a good network (which you can feed into
 this program, suddenly making it strong). To help with this effort, run the
 leelaz-autogtp binary provided in this package. The best-known network weights
 file is at http://zero.sjeng.org/best-network