Search found 411 matches

by chrisw
Sat Mar 16, 2019 1:11 pm
Forum: Computer Chess Club: General Topics
Topic: Chess Fighter Neural Net
Replies: 12
Views: 1085

Re: Chess Fighter Neural Net

I have had a similar frustrating experience with reinforcement learning before i deemed it too big a project for one man. What input planes do you use besides piece placement? I use attack tables and piece counts as additional inputs and they seem to help accelerate learning. Do you use replay buff...
by chrisw
Mon Mar 11, 2019 7:44 pm
Forum: Computer Chess Club: General Topics
Topic: Allie & Stein
Replies: 22
Views: 2207

Re: Allie & Stein

Guenther, almost certainly the problem you are seeing is because of a mismatch between the cuda libraries I built with and the ones you have installed. I built the binary with cuda 10.0. You can look at the appveyor.yml script which has the exact cuda files to download and install. Tried several co...
by chrisw
Sat Mar 09, 2019 7:14 pm
Forum: Computer Chess Club: General Topics
Topic: Allie & Stein
Replies: 22
Views: 2207

Re: Allie & Stein

The cuda dlls are delivered with each LC0 download since several months at least. He will also need the appropriate qt5core.dll. But all of this is invain anyway, because he has no available cuda hardware (can be deducted from earlier posts of him). It depends. It used to be that one had to go thro...
by chrisw
Sat Mar 09, 2019 1:59 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: syzygy 345 download
Replies: 3
Views: 211

syzygy 345 download

does anybody have a link to a FUNCTIONAL download, compressed, for all the 345 men egtbs (I just want to start with those to get syzygy going) which

a) isn’t a torrent, and
b) hasn’t gone 404, which all the links I can find by google have done
by chrisw
Sat Mar 09, 2019 1:54 pm
Forum: Computer Chess Club: General Topics
Topic: Allie & Stein
Replies: 22
Views: 2207

Re: Allie & Stein

I am unfamiliar with QT, I get this error when I try it on Windows. A link is appreciated if someone has got it. https://i.imgur.com/idvSpOi.png I have been making an appveyor.yml script for building on windows, but it still doesn't quite work for reasons I don't understand. If I do roughly the sam...
by chrisw
Sat Mar 09, 2019 10:11 am
Forum: Computer Chess Club: General Topics
Topic: Allie & Stein
Replies: 22
Views: 2207

Re: Allie & Stein

I am unfamiliar with QT, I get this error when I try it on Windows. A link is appreciated if someone has got it. https://i.imgur.com/idvSpOi.png I have been making an appveyor.yml script for building on windows, but it still doesn't quite work for reasons I don't understand. If I do roughly the sam...
by chrisw
Tue Mar 05, 2019 6:52 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Training the trainer: how is it done for Stockfish?
Replies: 22
Views: 1238

Re: Training the trainer: how is it done for Stockfish?

On why AlphaZero (and hence LC0) opted for a Monte Carlo search... There are no strong chess engines which use alpha-beta pruning alone. All the strong engines use other techniques to do a huge amount of pruning. These other techniques are mostly chess-specific, so no good for DeepMind's purposes. ...
by chrisw
Tue Mar 05, 2019 4:08 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Training the trainer: how is it done for Stockfish?
Replies: 22
Views: 1238

Re: Training the trainer: how is it done for Stockfish?

Regarding: "Very simple. MCTS is the only viable approach given the 80000 nps they got using 4 TPUs." Well, A/B fully replaced any other scheme when computers where much slower than 80Knps, and MCTS never won above A/B in any period of chess engine history...so maybe not so simple :-) Regrading: "i...
by chrisw
Mon Mar 04, 2019 9:53 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Training the trainer: how is it done for Stockfish?
Replies: 22
Views: 1238

Re: Training the trainer: how is it done for Stockfish?

Regarding: "Very simple. MCTS is the only viable approach given the 80000 nps they got using 4 TPUs." Well, A/B fully replaced any other scheme when computers where much slower than 80Knps, and MCTS never won above A/B in any period of chess engine history...so maybe not so simple :-) Regrading: "i...