Search found 835 matches

by pedrox
Fri Jul 09, 2021 5:11 pm
Forum: Computer Chess Club: General Topics
Topic: Are hybrid approaches stronger than pure nnue?
Replies: 15
Views: 1533

Re: Are hybrid approaches stronger than pure nnue?

Shortly after nnue was merged into the official Stockfish project, one of the first patches to pass was to use Stockfish's handcrafted evaluation instead of nnue for positions that are above a certain evaluation (5.00 was the original threshold). An year has passed, and that patch has still not bee...
by pedrox
Sun May 23, 2021 9:10 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: NNUE scoring (egbb lib)
Replies: 14
Views: 5196

Re: NNUE scoring (egbb lib)

As said if you are going to use NNUE in a simple engine (TSCP type) that only uses the evaluation on the leaves of the tree, then it is not necessary to do any tuning or scaling of the evaluation. If the engine uses the evaluation on each node of the search for pruning (razoring, null move, static n...
by pedrox
Sat Apr 24, 2021 7:19 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: My test for DanaSah 8.8_x64_avx2 (NNUE)
Replies: 1
Views: 1275

Re: My test for DanaSah 8.8_x64_avx2 (NNUE)

Thanks to you for the test. The result is good. I saw the first game against Rybka and it looked very close, I didn't know how the final result would be. All the Elo in this new version corresponds to the evaluation, the search gives 0 points. Of course credit to NNUE. This also indicates that my cl...
by pedrox
Sat Apr 24, 2021 4:15 pm
Forum: Computer Chess Club: General Topics
Topic: New engine releases & news 2021
Replies: 533
Views: 194542

Re: DanaSah 8.8

If Ed would like to test the networks with SIM, I would be happy to see the results. stock2bd10.nnue versus SF12 dev0 dev1 dev2 dev3 dev4 dev5 RMS SIM 2063 1852 1543 1059 662 1044 63.34 53.87 stock2bd10.nnue versus SF13 dev0 dev1 dev2 dev3 dev4 dev5 RMS SIM 2048 1865 1480 1090 690 1054 62.95 53.79 ...
by pedrox
Sat Apr 24, 2021 1:56 pm
Forum: Computer Chess Club: General Topics
Topic: UCI_Elo 1500 engines and FIDE Elo
Replies: 51
Views: 21684

Re: UCI_Elo 1500 engines and FIDE Elo

Maia Lichess 1546 has a blitz Lichess rating of around 1546 . This engine is run with maia-1100.pb.gz net at node 1 on a gauntlet test below. The formula to get FIDE Elo from Lichess Blitz Elo is, FIDE rating = 65 + [0.87 x Lichess Blitz Rating] based from the regression in 2019 . Maia FIDE Elo = 6...
by pedrox
Sat Apr 24, 2021 12:10 pm
Forum: Computer Chess Club: General Topics
Topic: New engine releases & news 2021
Replies: 533
Views: 194542

Re: DanaSah 8.8

Does DanaSah 8.8 play chess960 under Cutechess GUI, either as UCI or WinBoard ? Currently trying it as XBoard at chess960 under Cutechess. Only 10 games played but I think it is probably fine. DanaSah can play FRC or Chess960 with either xboard or uci protocol. No user configuration is required. It...
by pedrox
Fri Apr 23, 2021 10:05 pm
Forum: Computer Chess Club: General Topics
Topic: UCI_Elo 1500 engines and FIDE Elo
Replies: 51
Views: 21684

Re: UCI_Elo 1500 engines and FIDE Elo

Maia Lichess 1546 has a blitz Lichess rating of around 1546 . This engine is run with maia-1100.pb.gz net at node 1 on a gauntlet test below. The formula to get FIDE Elo from Lichess Blitz Elo is, FIDE rating = 65 + [0.87 x Lichess Blitz Rating] based from the regression in 2019 . Maia FIDE Elo = 6...
by pedrox
Fri Apr 23, 2021 9:05 pm
Forum: Computer Chess Club: General Topics
Topic: Introducing Toga III chess engine
Replies: 55
Views: 30957

Re: Introducing Toga III chess engine

Hi Dieterich, I do not have better results here, do I use the wrong files? toga3_avx2.exe 21.978.765 18.04.2021 17:14 ---- 1 DanaSah 8.8DN1 x64 1CPU +24 +32/=43/-25 53.50% 53.5/100 2 Toga III 0.3NN x64 1CPU -24 +25/=43/-32 46.50% 46.5/100 Thank you for testing this version of DanaSah. I tried playi...
by pedrox
Fri Apr 23, 2021 8:37 pm
Forum: Computer Chess Club: General Topics
Topic: New engine releases & news 2021
Replies: 533
Views: 194542

Re: DanaSah 8.8

DanaSah 8.8 allows the use of a NNUE type neural network. It includes a network that has been created from another network created with CCRL 40/2 positions. More information in the readme.txt file. Although there are many changes, the version with the classic evaluation does not gain strength, abou...
by pedrox
Tue Apr 20, 2021 7:19 pm
Forum: Computer Chess Club: General Topics
Topic: New engine releases & news 2021
Replies: 533
Views: 194542

DanaSah 8.8

DanaSah 8.8 allows the use of a NNUE type neural network. It includes a network that has been created from another network created with CCRL 40/2 positions. More information in the readme.txt file. Although there are many changes, the version with the classic evaluation does not gain strength, about...