This is still pretty rough ... HONEY-NN release

[MikeB]

Recommended for advanced users only, preferably those who can build their own.

https://github.com/MichaelB7/Stockfish/releases/tag/NN

There are some exe's posted for those with Ryzen processors and the like - AVX2 instruction set required for the NN engines.

Only Weakfish is currently able to play at weaker levels - although the UCI options will indicate otherwise.

If you are looking for a build, let me know. I might be able to get to it on the weekend.

[JohnS]

Great work Mike, thanks for the presents.

[maxdeg]

+3
Thanks Mike.
You have mail !

[Paloma]

Recommended for advanced users only, preferably those who can build their own.

https://github.com/MichaelB7/Stockfish/releases/tag/NN

There are some exe's posted for those with Ryzen processors and the like - AVX2 instruction set required for the NN engines.

Only Weakfish is currently able to play at weaker levels - although the UCI options will indicate otherwise.

If you are looking for a build, let me know. I might be able to get to it on the weekend.

Do I need a GPU (RTX2070) for Honey-NN or is a CPU all I need?

And will there not be an XI-R3.1?

[zullil]

Recommended for advanced users only, preferably those who can build their own.

https://github.com/MichaelB7/Stockfish/releases/tag/NN

There are some exe's posted for those with Ryzen processors and the like - AVX2 instruction set required for the NN engines.

Only Weakfish is currently able to play at weaker levels - although the UCI options will indicate otherwise.

If you are looking for a build, let me know. I might be able to get to it on the weekend.

Do I need a GPU (RTX2070) for Honey-NN or is a CPU all I need?

And will there not be an XI-R3.1?

You should only need a CPU for this.

[Paloma]

Thanks Louis

[peter]

If you are looking for a build, let me know. I might be able to get to it on the weekend.

Thanks! Looking forward to weekend and a popcnt- compile.

[Raphexon]

Together with first Sergio net this is honest to god the best puzzle solver I've ever seen.
It literally finds everything, some extremely difficult ones take a few billion nodes. But that's no worry, it finds it all.

[nnnnnnnn]

What do you mean by "puzzle"? Study? Direct-mate?

And why would a neural net be good at these? I would have thought that nets would be weaker than pure engines when the solution was purely tactical, as is the case for most compositions, particularly when the winning moves are often ipso facto improbable - underpromotions, tempo losses, and so on.

[Raphexon]

What do you mean by "puzzle"? Study? Direct-mate?

And why would a neural net be good at these? I would have thought that nets would be weaker than pure engines when the solution was purely tactical, as is the case for most compositions, particularly when the winning moves are often ipso facto improbable - underpromotions, tempo losses, and so on.

It finds this sequence for example.

https://en.chessbase.com/post/solution- ... able-study

Normal SF never did.