Leela network questions

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
Guenther
Posts: 3347
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Full name: Guenther Simon
Contact:

Re: Leela network questions

Post by Guenther » Fri Apr 13, 2018 10:32 am

SzG wrote:I can see there are several networks for download, but Leela uses only one of them at a time.
What is the difference between the networks? Do they impact playing strength? Or playing style? Which is the one to use?

Another question: does Leela learn from playing in a tournament against various opponents?
1. Of course they impact playing strength, because all 'knowledge' is in the NN file

2. Since NN ID123 the network size was increased to add more blocks+filters (from 6*64 to 10*128) and there should be some bigger jump in strength coming soon.
There was also a change to affect variety of opening moves in selfplay, which should also result in more strength (in real play) to previously 'neglected' opening trees. (I guess this needs quite some time)

3. You should always use the latest one (which did not fail = green),
even if it appears to have less selfplay rating, because this does not correlate to strength in games vs. other opponents.

4. The NN learns only in training mode

5. If you want to start testing it now it is suggested to do some 'fulltuning' of your gpu to get the most out of it. This will increase the nps of LCZero,
otherwise it quickly creates a working tuning file in the first run which is not optimized, but works with your gpu.

e.g. (might take a while - this produces a file 'leelaz_opencl_tuning')

Code: Select all

lczero.exe --tune-only --full-tuner -w ID123
Last edited by Guenther on Fri Apr 13, 2018 10:40 am, edited 1 time in total.
https://rwbc-chess.de
Greg Strong@ovyron wrote: What I think is that you have a very, very, VERY strong desire to post, and very little self-control.
All the time. In every conversation. You must post. Something. It's like you just can't help yourself.

Guenther
Posts: 3347
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Full name: Guenther Simon
Contact:

Re: Leela network questions

Post by Guenther » Fri Apr 13, 2018 10:41 am

SzG wrote:Thanks Günther.
*added another point, while I was editing my post
https://rwbc-chess.de
Greg Strong@ovyron wrote: What I think is that you have a very, very, VERY strong desire to post, and very little self-control.
All the time. In every conversation. You must post. Something. It's like you just can't help yourself.

Werewolf
Posts: 1308
Joined: Thu Sep 18, 2008 8:24 pm

Re: Leela network questions

Post by Werewolf » Fri Apr 13, 2018 11:13 am

Guenther wrote:
SzG wrote:I can see there are several networks for download, but Leela uses only one of them at a time.
What is the difference between the networks? Do they impact playing strength? Or playing style? Which is the one to use?

Another question: does Leela learn from playing in a tournament against various opponents?
1. Of course they impact playing strength, because all 'knowledge' is in the NN file

2. Since NN ID123 the network size was increased to add more blocks+filters (from 6*64 to 10*128) and there should be some bigger jump in strength coming soon.
There was also a change to affect variety of opening moves in selfplay, which should also result in more strength (in real play) to previously 'neglected' opening trees. (I guess this needs quite some time)

3. You should always use the latest one (which did not fail = green),
even if it appears to have less selfplay rating, because this does not correlate to strength in games vs. other opponents.

4. The NN learns only in training mode

5. If you want to start testing it now it is suggested to do some 'fulltuning' of your gpu to get the most out of it. This will increase the nps of LCZero,
otherwise it quickly creates a working tuning file in the first run which is not optimized, but works with your gpu.

e.g. (might take a while - this produces a file 'leelaz_opencl_tuning')

Code: Select all

lczero.exe --tune-only --full-tuner -w ID123
Thanks, very helpful post

MonteCarlo
Posts: 111
Joined: Sun Dec 25, 2016 3:59 pm

Re: Leela network questions

Post by MonteCarlo » Fri Apr 13, 2018 3:00 pm

Just to be clear, the change to get more variety in openings (the temperature decay) is only for the match games used for gating (determining whether the new network passes or not); it is not used in self-play games.

There's already a high variety of openings in self-play because a temperature of 1 is used for the whole game.

In the matches, where temperature wasn't used, they were getting crazy things like half the games starting from the same position after 8 moves, which they were worried was making the match results less informative.

That's what the change is meant to address.

carldaman
Posts: 1985
Joined: Sat Jun 02, 2012 12:13 am

Re: Leela network questions

Post by carldaman » Fri Apr 13, 2018 9:31 pm

Guenther wrote:
SzG wrote:
5. If you want to start testing it now it is suggested to do some 'fulltuning' of your gpu to get the most out of it. This will increase the nps of LCZero,
otherwise it quickly creates a working tuning file in the first run which is not optimized, but works with your gpu.

e.g. (might take a while - this produces a file 'leelaz_opencl_tuning')

Code: Select all

lczero.exe --tune-only --full-tuner -w ID123
Hi Guenther,

Is full-tuning to be done for cpu's as well, or just gpu's?

Thanks,
CL

Jhoravi
Posts: 270
Joined: Wed May 08, 2013 4:49 am

Re: Leela network questions

Post by Jhoravi » Sat Apr 14, 2018 1:36 am

MonteCarlo wrote: In the matches, where temperature wasn't used, they were getting crazy things like half the games starting from the same position after 8 moves, which they were worried was making the match results less informative.
That's probably the reason why Google didn't publish all 100 games between Stockfish and Alpha0. Since no book to vary SF opening, it keeps on repeating the same loosing variation in the Queens Indian Defense as shown in their selected 10 game sample. It also explains why too many looses as black(27) than as white(3)

MonteCarlo
Posts: 111
Joined: Sun Dec 25, 2016 3:59 pm

Re: Leela network questions

Post by MonteCarlo » Sat Apr 14, 2018 2:30 am

Full tuning is only for the GPU version.

Cheers!

carldaman
Posts: 1985
Joined: Sat Jun 02, 2012 12:13 am

Re: Leela network questions

Post by carldaman » Sat Apr 14, 2018 8:26 pm

MonteCarlo wrote:Full tuning is only for the GPU version.

Cheers!
OK, good to know, thanks! :)

Post Reply