Search found 150 matches

by dkappe
Sat Apr 20, 2019 8:55 pm
Forum: Computer Chess Club: General Topics
Topic: 41800-boost-7000 net released
Replies: 5
Views: 1720

41800-boost-7000 net released

About even with 41800, but plays more aggressively, less "trolly."

https://github.com/dkappe/leela-chess-w ... t-Networks
by dkappe
Wed Apr 17, 2019 9:10 pm
Forum: Computer Chess Club: General Topics
Topic: 32930 Boost network available
Replies: 12
Views: 3943

Re: 32930 Boost network available

After some additional testing, the boost net at 7000 steps is about the same head 2 head and vs sf10 at 1+1. The style of play is a bit different, however. Would appreciate feedback. https://github.com/dkappe/leela-chess-w ... t-Networks
by dkappe
Tue Apr 09, 2019 6:57 pm
Forum: Computer Chess Club: General Topics
Topic: 32930 Boost network available
Replies: 12
Views: 3943

Re: 32930 Boost network available

I applied this to my own 256x20-se net called Maddex and ran 2+2 for 800 games and got +30 elo. Right now I’m running tests for 500 to 4500 step nets. 32930 may need more steps. TBD.
by dkappe
Tue Apr 09, 2019 3:02 am
Forum: Computer Chess Club: General Topics
Topic: 32930 Boost network available
Replies: 12
Views: 3943

32930 Boost network available

https://github.com/dkappe/leela-chess-weights/wiki/Boost-Networks Based on my hypothesis on what was the critical element in the success of Antifish, this is 32930 sharpened with CCRL games at a low learning rate and a very few steps. Seems to give around 30 elo extra than the original net. I'll be ...
by dkappe
Sun Apr 07, 2019 5:44 pm
Forum: Computer Chess Club: General Topics
Topic: LC0-Leela0 for Android Devices...?
Replies: 25
Views: 3269

Re: LC0-Leela0 for Android Devices...?

I run a lichess bot on a raspberry pi 3. I use a 48x5 network, but it’s worth testing 80x7 and 64x6. 112x9 works well on a cpu, but may be too slow on a pi.
by dkappe
Sat Apr 06, 2019 3:29 pm
Forum: Computer Chess Club: General Topics
Topic: Yet another game analysis method
Replies: 34
Views: 3133

Re: Yet another game analysis method

From the recent US Women’s Championship (I upped the decisive score to 7.0): [Event "US Championship (Women)"] [Site "St Louis, MO USA"] [Date "2019.03.22"] [Round "3.1"] [White "Irina Krush"] [Black "Jennifer R Yu"] [Result "0-1"] [EventDate "2019.03.20"] [ECO "B13"] [WhiteElo "2451"] [BlackElo "22...
by dkappe
Sat Apr 06, 2019 3:22 pm
Forum: Computer Chess Club: General Topics
Topic: 32930-112x9-se and ender-112x9-se available
Replies: 11
Views: 2002

Re: 32930-112x9-se and ender-112x9-se available

# PLAYER : RATING ERROR POINTS PLAYED (%) CFS(%) W D L D(%) 1 11258-112x9-se : 0 15 132.5 250 53.0 92 62 141 47 56.4 2 50782 : -21 15 117.5 250 47.0 --- 47 141 62 56.4 White advantage = 40.99 +/- 14.96 Draw rate (equal opponents) = 57.55 % +/- 2.99 2+2 At low nodes, there’s an inflection point that...
by dkappe
Sat Apr 06, 2019 3:15 pm
Forum: Computer Chess Club: General Topics
Topic: Yet another game analysis method
Replies: 34
Views: 3133

Re: Yet another game analysis method

Ferdy wrote:
Sat Apr 06, 2019 3:01 pm
I see there is engine 2 and first engine will be used when material is still high :).
As per your formula.
by dkappe
Sat Apr 06, 2019 1:33 pm
Forum: Computer Chess Club: General Topics
Topic: Yet another game analysis method
Replies: 34
Views: 3133

Re: Yet another game analysis method

I took a whack at the combined leela and sf here https://github.com/dkappe/chess-artist The file is a separate new-artist.py along with an example.sh shell script. There is a point where the first and second engines are called on the same position, which probably isn’t what you want, but so far so g...
by dkappe
Fri Apr 05, 2019 10:00 pm
Forum: Computer Chess Club: General Topics
Topic: stockfish 10 vs. Mephisto III S Glasgow
Replies: 40
Views: 3398

Re: stockfish 10 vs. Mephisto III S Glasgow

You are at the low end of nodes, so you have a fair amount of noise. I don’t follow the logic on why you picked 1000 nodes. The advances in ab engines is more around search and forward pruning. Sf10 has a trade off in smarts vs speed. As an exercise, see where Mephisto and sf perform about equal ove...