How are you adjudicating the games? Both sides showing a huge eval. for some number moves?
I guess that's it.
Experimental Test: Leela Chess Zero v0.21.1 41715 vs Stockfish 240319 10m/40
Moderators: hgm, Rebel, chrisw
-
jp
- Posts: 1470
- Joined: Mon Apr 23, 2018 7:54 am
-
mwyoung
- Posts: 2727
- Joined: Wed May 12, 2010 10:00 pm
Re: Experimental Test: Leela Chess Zero v0.21.1 41715 vs Stockfish 240319 10m/40
It is set to auto in the GUI. Setting is Resign Late....
"The worst thing that can happen to a forum is a running wild attacking moderator(HGM) who is not corrected by the community." - Ed Schröder
But my words like silent raindrops fell. And echoed in the wells of silence.
But my words like silent raindrops fell. And echoed in the wells of silence.
-
jp
- Posts: 1470
- Joined: Mon Apr 23, 2018 7:54 am
Re: Experimental Test: Leela Chess Zero v0.21.1 41715 vs Stockfish 240319 10m/40
That's a boost for Lc, because it often doesn't convert winning endgames.
But playing to mate or a 6-man TB win would take much more time, and maybe that's not what you want to test anyway. (We already know Lc is bad at endgames, and maybe these tests are more about opening advantages, etc.)
But playing to mate or a 6-man TB win would take much more time, and maybe that's not what you want to test anyway. (We already know Lc is bad at endgames, and maybe these tests are more about opening advantages, etc.)
-
mwyoung
- Posts: 2727
- Joined: Wed May 12, 2010 10:00 pm
Re: Experimental Test: Leela Chess Zero v0.21.1 41715 vs Stockfish 240319 10m/40
You are incorrect. I only set this setting after I confirmed it did not affect the results. Check the tapes.....jp wrote: ↑Thu Mar 28, 2019 1:09 am That's a boost for Lc, because it often doesn't convert winning endgames.
But playing to mate or a 6-man TB win would take much more time, and maybe that's not what you want to test anyway. (We already know Lc is bad at endgames, and maybe these tests are more about opening advantages, etc.)
"The worst thing that can happen to a forum is a running wild attacking moderator(HGM) who is not corrected by the community." - Ed Schröder
But my words like silent raindrops fell. And echoed in the wells of silence.
But my words like silent raindrops fell. And echoed in the wells of silence.
-
jp
- Posts: 1470
- Joined: Mon Apr 23, 2018 7:54 am
Re: Experimental Test: Leela Chess Zero v0.21.1 41715 vs Stockfish 240319 10m/40
I should say that's a potential boost. Whether it affects results depends whether it gets those winning endgames it cannot convert or drawing endgames it cannot hold. e.g. It would have affected the TCEC results. The games are shown on the Lc blog. It failed to win a 7-man endgame where SF's eval was +150 or something. (But the time cost here may not be worth it anyway. As long as we know what the settings are for each test, it's fine.)