Search found 3576 matches

by Daniel Shawul
Wed Jan 13, 2021 8:32 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: From Esoteric to Transcendental Chess Programming?
Replies: 5
Views: 377

Re: From Esoteric to Transcendental Chess Programming?

Heyho, I do not follow SF development, but I get here and there a breadcrumb, for example: "is LVA as in MVV-LVA useless ?" http://talkchess.com/forum3/viewtopic.php?t=70918 "...Lazy SMP feeds on chaos..." http://talkchess.com/forum3/viewtopic.php?f=7&t=72684#p824068 So I ponder if we left the para...
by Daniel Shawul
Sun Jan 03, 2021 2:48 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: YBW engines past and present?
Replies: 15
Views: 1218

Re: YBW engines past and present?

ABDADA was comparable to YBW even in the old days too, just wasn't given as much attention. From what I recall, the author was bad-mouthed when the paper came out claiming a YBW algorithm done all through the hashtable. Lazy SMP is certainly lazier but it is not better than ABDADA which is still the...
by Daniel Shawul
Sat Dec 26, 2020 10:52 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: question about gensfen
Replies: 9
Views: 485

Re: question about gensfen

I am looking at: https://github.com/nodchip/Stockfish/blob/master/src/learn/gensfen.cpp The question is - it looks like by default it occasionally inserts a random move into the played games to increase variety. But then when the game is ended, it assigns the same result to all the generated FENs f...
by Daniel Shawul
Sat Dec 26, 2020 10:39 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: question about gensfen
Replies: 9
Views: 485

Re: question about gensfen

You need some randomness for variety and this is often controlled by the "temperature" parameter for MCTS-NN reinforcement learning. I wouldn't worry about the final score being assigned to all positions as long as the randomness (temperature) is limited to a reasonable value. I use a similar appro...
by Daniel Shawul
Sat Dec 26, 2020 2:31 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: question about gensfen
Replies: 9
Views: 485

Re: question about gensfen

You need some randomness for variety and this is often controlled by the "temperature" parameter for MCTS-NN reinforcement learning. I wouldn't worry about the final score being assigned to all positions as long as the randomness (temperature) is limited to a reasonable value. I use a similar approa...
by Daniel Shawul
Fri Dec 25, 2020 9:37 pm
Forum: Computer Chess Club: General Topics
Topic: Tornado News
Replies: 52
Views: 4767

Re: Tornado News

At the moment i am very pissed off :-( There are some trolls around that complains everything what programmers are doing, even they are spend many years at own work and let the community enjoy for free ! Hi Engin, Your anger is very understandable. I have another implementation that is not GPL but ...
by Daniel Shawul
Fri Dec 25, 2020 6:52 pm
Forum: Computer Chess Club: General Topics
Topic: Tornado News
Replies: 52
Views: 4767

Re: Tornado News

At the moment i am very pissed off :-( There are some trolls around that complains everything what programmers are doing, even they are spend many years at own work and let the community enjoy for free ! Hi Engin, Your anger is very understandable. I have another implementation that is not GPL but ...
by Daniel Shawul
Thu Nov 26, 2020 3:01 pm
Forum: Computer Chess Club: General Topics
Topic: Tony Thomas has passed away ....
Replies: 11
Views: 3204

Re: Tony Thomas has passed away ....

Rest in Peace Tony!
by Daniel Shawul
Thu Nov 19, 2020 3:55 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: You've trained a brilliant NN(UE) King-Piece Network. Now what?
Replies: 22
Views: 4104

Re: You've trained a brilliant NN(UE) King-Piece Network. Now what?

for floating point, it should be possible to rely on the compiler #pragma float_control(precise, off, push) float dot_product(int count, const float *a, const float *b) { float result = 0; for (int i=0; i<count; i++) result += a[i] * b[i]; return result; } #pragma float_control(pop) clang -O3 -marc...
by Daniel Shawul
Tue Nov 17, 2020 5:23 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Pytorch NNUE training
Replies: 61
Views: 11455

Re: Pytorch NNUE training

Yes, this comes from using the GPU (hooray pytorch!). As well, Sopel implemented a super fast C++ data parser, which feeds the inputs to pytorch as sparse tensors, which was a very large speedup (since the inputs to the first layer are super sparse). What was the batch size for this? That makes a b...