Which engines play the most human like at lower levels?

[Tibono]

P.S.: you would also need the back in time lczero.exe for NN compatibility.
Also name the NN as command line parameter: -w 0076
Tibono

[dkappe]

Rather than jumping through hoops with an old lczero engine, simply try the small nets: 11258-16x2 distilled and tiny gyal (also 16x2). They can be used with the latest lc0 on cpu (or android).

https://github.com/dkappe/leela-chess-w ... d-Networks
https://github.com/dkappe/leela-chess-w ... i/Bad-Gyal

There’s a ton of nets on both pages, so you’ll have to scroll to find them. Tiny Gyal is a bit weaker, so maybe more in your range.

The distilled 16x2 plays around lichess 2200 at 4000 nodes (which is rather quick). You can give it a go here: https://lichess.org/@/leela2200

[Alexander Schmidt]

Hello I'm looking for engines that play around 1800-2200 ELO and are not "dumbed down" by making obvious poor moves. Thanks in advance,
Gerald

I like to play vs. Leela zero nets. With the option Slowmover=0 it plays immediately, and you play simply against the neural net without any calculating. It plays very human like and even makes tactical blunders. Just finishing a won endgame is a pain, but it is up to you to resign

You can chose between all kind of nets, I use the latest 128x10 net 591228, it is somewhere between 1600-1800 ELO, the latest 256x20 net 42850 is between 1800-2000 ELO. The 320x24 are currently trained are between 2000-2200 ELO.

[odyn1982]

I can't find 128x10 net 591228. Can you share a link to it? And thanks for great tips.

[Alexander Schmidt]

I can't find 128x10 net 591228. Can you share a link to it? And thanks for great tips.

Typo... It is 591226...

Here you can find all official nets:
http://training.lczero.org/networks/?show_all=1

Here are some links to other interesting nets:
http://lczero.org/play/bestnets/

[Tibono]

Hi,
another suggestion: CDrill 1800 + Morphy personnality. Nice work authored by F. Mosca, worth a try