ShashChess

[Werewolf]

As expected bigger net and other changes make ShashChess worse solver (High Tal). Just one example. 110 positions HTC test. ShashChess 35.1 solved 93 and used 26 minutes. Version 36 solved 84 and used 36 minutes. Also questionable "hide current move" patch included.

What does this mean for practical game play? Will it play worse chess?

It's tactically weaker but positionally and strategically stronger (Stockfish 17 +46 elo).

Laughable.

Yes.

[chessica]

Well, just let the numbers do the talking.

[Rebel]

Well, just let the numbers do the talking.

Do the math yourself, start a match, and Shashchess should score 56.5% (=46 elo) against SF17.

[chessica]

Yes, you can do that, but I don't do that. So many other testers are already
doing that and they report very diligently.

[Rebel]

Yes, you can do that, but I don't do that. So many other testers are already
doing that and they report very diligently.

I don't have to do the test to know, but will be very diligently debunk the +46 nonsense.

Code: Select all

No. Name             Win Draw Loss Unf.  Score Games       %
------------------------------------------------------------
  1 SF17            +165 =764  -71   *0  547.0  1000   54.7%
  2 ShashChess-35.3  +71 =764 -165   *0  453.0  1000   45.3%

It looks more it is the other way around, isn't it?

[chessica]

Yes, that's just how it is... There's nothing you can do about it. The numbers are clear and describe exactly what?

[Werewolf]

Yes, that's just how it is... There's nothing you can do about it. The numbers are clear and describe exactly what?

That your claim was false.

[chessica]

Yes, that's just how it is... There's nothing you can do about it. The numbers are clear and describe exactly what?

That your claim was false.

Oh what? The "Hai" said it, not me. See here: viewtopic.php?p=968855#p968855

[Werewolf]

Yes, that's just how it is... There's nothing you can do about it. The numbers are clear and describe exactly what?

That your claim was false.

Oh what? The "Hai" said it, not me. See here: viewtopic.php?p=968855#p968855

Sorry. Hai's claim was false.

[supernova]

https://github.com/amchess/ShashChess/releases/tag/36
https://github.com/amchess/Alexander/releases/tag/2.0
https://github.com/amchess/BrainLearn/releases/tag/29

It may be premature to draw definitive conclusions at this stage, as 100 games may not provide a sufficiently large sample size. However, it appears that Version 36 has shown a decline in performance compared to Version 35.3. In my initial testing at a 2-minute 3-second increment time control, Version 36 has achieved less favorable results against a Stockfish-clone engine known as Vanilla 15.

ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Alexander 1.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : 100 : 37+ : 63= : 0- : 68.5%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Beast 15 6t1024h : 100 : 0+ : 99= : 1- : 49.5%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Chess-System-Tal 2.00.v21-E1019 6t1024h : 100 : 14+ : 86= : 0- : 57.0%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Clover 6.2 6t1024h : 100 : 21+ : 79= : 0- : 60.5%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Clover 7.1 6t1024h : 100 : 15+ : 84= : 1- : 57.0%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Coiled 1.2 6t1024h : 100 : 44+ : 56= : 0- : 72.0%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Lc0.cuda 0.31.1 611246.384 6t1024h : 100 : 18+ : 82= : 0- : 59.0%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Lc0.cuda 0.31.1 715893.256 6t1024h : 100 : 47+ : 53= : 0- : 73.5%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Lc0.onnx-dml 0.31.1 715893.256 6t1024h : 100 : 56+ : 44= : 0- : 78.0%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : LittleBeast 15 6t1024h : 100 : 0+ : 100= : 0- : 50.0%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Vafra 14.12.1 6t1024h : 100 : 24+ : 76= : 0- : 62.0%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Vafra 14.12.2 6t1024h : 100 : 10+ : 90= : 0- : 55.0%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Vanilla 15 6t1024h : 100 : 2+ : 97= : 1- : 50.5%

ShashChess 36 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Vanilla 15 6t1024h : 100 : 4+ : 81= : 15- : 44.5%