Search found 872 matches

by AlvaroBegue
Mon Oct 15, 2018 1:30 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 3 million games for training neural networks
Replies: 22
Views: 3627

Re: 3 million games for training neural networks

Can you tell me the purpose producing 3 miilion games from only 5953 unique opening lines of which 21% are exact doubles? Yeah, I was hoping Stockfish's behavior would be more random than that. :( If I do something like this again, I'll probably use UCT to grow an opening-book tree with much more v...
by AlvaroBegue
Fri Oct 12, 2018 7:16 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 3 million games for training neural networks
Replies: 22
Views: 3627

Re: 3 million games for training neural networks

If you do use games where White and Black are different engines (TCEC or CCRL), it would be interesting to inform the network of the Elo difference between the players (assuming Elo estimates are available). That input can then be used to implement something like contempt factor in a more principled...
by AlvaroBegue
Thu Oct 11, 2018 11:53 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 3 million games for training neural networks
Replies: 22
Views: 3627

Re: 3 million games for training neural networks

Hi, I had my 8-core Ryzen 7 computer spend about 2 months generating quick Stockfish-vs-Stockfish games. The result is 3 million games that I think could be used for training neural networks, or to tune evaluation functions. I have the games available in PGN format with comments that include the sc...
by AlvaroBegue
Thu Sep 20, 2018 7:02 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: My PHP Chess Move Generator is slow. Help!
Replies: 12
Views: 890

Re: My PHP Chess Move Generator is slow. Help!

People say C is even faster than C++, that's why I personally write in C. People say a lot of things, but C is not faster than C++. Anything you can write in C you can write in C++, and the performance will be the same. C++ has additional features and some of those might be slow, especially if misu...
by AlvaroBegue
Thu Sep 13, 2018 7:52 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: pg_key.c link
Replies: 5
Views: 351

Re: pg_key.c link

Try using quotes to make the whole FEN string into a single argument:

pg_key "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"
by AlvaroBegue
Wed Sep 12, 2018 7:43 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Can it be a "real" hash collision
Replies: 11
Views: 606

Re: Can it be a "real" hash collision

That looks like a bug. Something I used to do in my engine Ruy-López is clearing up the search state (hash, killers, history-heuristic tables...) after every move, so I could reproduce everything it did exactly (except for maybe time control). It may make you lose ~30 Elo, but it will let you find b...
by AlvaroBegue
Wed Jul 11, 2018 9:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: statistic magicians help required
Replies: 2
Views: 475

Re: statistic magicians help required

First of all, I think they report 1.96 standard deviations, not 2. But that's small potatoes. You haven't been precise enough in describing the two-tournaments scenario. If you do two tournaments between the same two engines and you obtain Elo estimates with means Elo1, Elo2 and corresponding standa...
by AlvaroBegue
Wed Jul 11, 2018 5:40 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Testing a set of engines
Replies: 3
Views: 622

Re: Testing a set of engines

If you look at the CCRL table that lists engines in decreasing order of strength, you'll see a column "LOS". That is something like one minus the p-value of the hypothesis that the engine in one line is actually stronger than the one below. Running more games allows you to make those numbers closer ...
by AlvaroBegue
Tue Jun 26, 2018 5:44 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Hoziron effect and extensions
Replies: 9
Views: 1148

Re: Hoziron effect and extensions

In my experience, it is important to be restrictive in the definition of recapture. My code currently does something like this: bool extend = false; // Extend checks if (!extend && in_check) extend = true; // Extend recaptures if (!extend && last_two_moves_had_the_same_destination_square() && last_t...