Search found 3216 matches

by Daniel Shawul
Tue May 14, 2019 2:18 pm
Forum: Computer Chess Club: General Topics
Topic: Wb2uci adaptor
Replies: 8
Views: 372

Re: Wb2uci adaptor

You didn't say (and maybe the CCC team didn't say) what the issue was. I use cutechess-cli with Winboard protocol all the time, and have had no issues. You can use the -debug flag to cutechess-cli to get all engine output/input logged, which can be helpful in finding engine-specific issues. --Jon I...
by Daniel Shawul
Tue May 14, 2019 1:33 pm
Forum: Computer Chess Club: General Topics
Topic: Wb2uci adaptor
Replies: 8
Views: 372

Re: Wb2uci adaptor

You didn't say (and maybe the CCC team didn't say) what the issue was. I use cutechess-cli with Winboard protocol all the time, and have had no issues. You can use the -debug flag to cutechess-cli to get all engine output/input logged, which can be helpful in finding engine-specific issues. --Jon I...
by Daniel Shawul
Tue May 14, 2019 12:46 pm
Forum: Computer Chess Club: General Topics
Topic: Wb2uci adaptor
Replies: 8
Views: 372

Re: Wb2uci adaptor

there may be UCI-only issues. If the pv or something like that can not be parsed by the GUI properly, one can just NOT do that and be done with it I recall some engines in TCEC used to not display pvs, nps properly etc, but they still played because all that mattered is that the move they made be p...
by Daniel Shawul
Tue May 14, 2019 11:48 am
Forum: Computer Chess Club: General Topics
Topic: Wb2uci adaptor
Replies: 8
Views: 372

Wb2uci adaptor

I have a problem getting in to ccc tournament because my engine is winboard apparently. I don’t really understand why a GUI will be ok with only uci engines when the tournament is played via cutechess-cli anyway. If the GUI can parse cutechess-cli output I think everything should be fine e.g tcec wo...
by Daniel Shawul
Fri May 10, 2019 1:57 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: catastrophic forgetting
Replies: 5
Views: 488

Re: catastrophic forgetting

A less complex problem is to train one neural network for the same game but on different board sizes. This has been done for Go which is played on board sizes of 9x9, 13x13, and 19x19. This paper https://arxiv.org/abs/1902.10565 uses an input plane mask to indicate on-board locations. They go furthe...
by Daniel Shawul
Thu May 09, 2019 4:16 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: catastrophic forgetting
Replies: 5
Views: 488

catastrophic forgetting

I am trying to train neural networks for my chess variant playing program Nebiyu. It supports > 10 chess variants and can also play Go/Hex/Reversi/Amazons/Checkers. Focusing only on the chess variants, if I set out to train a separate network for each variant, it will be cumbersome. So I am thinking...
by Daniel Shawul
Sat May 04, 2019 4:10 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Question to Remi about CrazyZero
Replies: 5
Views: 690

Re: Question to Remi about CrazyZero

I was tinkering with making nebiyu work with neural nets past couple of days and i think i've succeeded now -- though i still have to train nets for the games it supports. I had to move all the chess specific logic out of the neural network trainer and prober code. Egbbdll has effectively become a g...
by Daniel Shawul
Mon Apr 29, 2019 10:40 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Training using 1 playout instead of 800
Replies: 12
Views: 818

Re: Training using 1 playout instead of 800

What's the value of alpha parameter in the dirichlet noise? Do you use a softmax temperature of 0 after 30 half moves? I'm no expert but it seems like you could easily wipe out a winning signal from a position in chess with just a tiny % of random moves. alpha=0.3 and beta=1.0. After 30 moves, the ...
by Daniel Shawul
Sun Apr 28, 2019 12:05 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Wouldn't it be nice if C++ GPU
Replies: 24
Views: 1091

Re: Wouldn't it be nice if C++ GPU

In my experience, there is very little overhead for transferring data from the CPU to the GPU. But that may be because I am doing everything with cuDNN in C++, and build a half-precision host tensor directly on the CPU for transfer. Sending two half-size batches takes almost the same time as one fu...
by Daniel Shawul
Sun Apr 28, 2019 11:56 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Training using 1 playout instead of 800
Replies: 12
Views: 818

Re: Training using 1 playout instead of 800

Oh, in fact what you do is policy gradient, but your baseline is surprising. As trulses wrote, using a draw as a baseline looks like a better approach. I don't think using draw as baseline matters. The 'correct' baseline is to use the current position's evaluation as baseline to calculate Advantage...