Search found 65 matches

by Branko Radovanovic
Sat Nov 17, 2018 4:09 pm
Forum: Computer Chess Club: General Topics
Topic: what happened to scorpio NN in TCEC?
Replies: 33
Views: 7600

Re: what happened to scorpio NN in TCEC?

Meanwhile, the entire Div4 has been restarted, without Scorpio. The chat message is: hi everyone i am sorry to inform you Entrance Division 4 had to be restarted due to too many inconsistencies in book orders, probably as a consequence of the many crashes among other things. this seemed to be the on...
by Branko Radovanovic
Tue Oct 23, 2018 6:06 pm
Forum: Computer Chess Club: General Topics
Topic: What was the top Chess engine before it was Stockfish, Komodo or Houdini?
Replies: 58
Views: 13598

Re: What was the top Chess engine before it was Stockfish, Komodo or Houdini?

The last time the strongest chess engine was not Stockfish, Komodo, or Houdini was in December 2010, right before Houdini 1.5 was released. At that time, Rybka 4 was #1. The last archived version of CCRL 40/40 with Rybka as #1: https://web.archive.org/web/20101121055257/http://computerchess.org.uk/c...
by Branko Radovanovic
Wed Sep 26, 2018 12:09 pm
Forum: Computer Chess Club: General Topics
Topic: Leela doesn't obey well the Elo curve set by regular engines
Replies: 9
Views: 7788

Re: Leela doesn't obey well the Elo curve set by regular engines

This is very interesting and may at first appear difficult to explain. However, consider an engine that randomly crashes in e.g. 20% of the games. Elo-wise, this would create the same phenomenon: such an engine would always lose at least 20% of the games, even against the weakest of opponents, and i...
by Branko Radovanovic
Wed Aug 29, 2018 10:57 am
Forum: Computer Chess Club: General Topics
Topic: Komodo MCTS
Replies: 4
Views: 2273

Re: Komodo MCTS

My feelings too, and Komodo MCTS is my #1 wish for TCEC S14. In fact, I'd really like to see a non-NN MCTS engine entering. Apart from Komodo, there's only Scorpio, but it's not strong enough even for Div4.
by Branko Radovanovic
Tue Aug 14, 2018 11:38 pm
Forum: Computer Chess Club: General Topics
Topic: Strange Lc0 TCEC performance
Replies: 132
Views: 21202

Re: Strange Lc0 TCEC performance

Both Lc0 (testnet and Deus) in both Div 4 and DIv 3 perform consistently below expectations (although people cheered their sore promotion from Div 4). "Too few games" in some total of 80+ games (yes, different nets whatever, matters less) and under-performance of some 200 Elo points is a marginal a...
by Branko Radovanovic
Tue Aug 14, 2018 10:26 pm
Forum: Computer Chess Club: General Topics
Topic: Strange Lc0 TCEC performance
Replies: 132
Views: 21202

Re: Strange Lc0 TCEC performance

Not unprecedented, happens to A-B engines too. In S12, Ethereal got 23.5 points in Div4 (counting Scorpio's games, but still!) - that was outrageous , something like mid-DivP performance if I'm not mistaken. Yet, as many will remember, Div3 was another story entirely, in which Ethereal was eclipsed ...
by Branko Radovanovic
Sun Aug 12, 2018 10:21 am
Forum: Computer Chess Club: General Topics
Topic: Copyright and Machine Learning IP
Replies: 102
Views: 12050

Re: Copyright and Machine Learning IP

Just some brief remarks - I could write 5 pages on this issue. Creativity is not an absolute requirement for copyright; see https://en.wikipedia.org/wiki/Sweat_of_the_brow. NN weights are intellectual property, but they are not source code. The GPL covers source code only. With the rise of NN, this ...
by Branko Radovanovic
Fri Aug 10, 2018 6:34 pm
Forum: Computer Chess Club: General Topics
Topic: Leela is more Chess-domain specific than regular Chess engines
Replies: 34
Views: 9222

Re: Leela is more Chess-domain specific than regular Chess engines

This is why deep residual neural networks perform better. They combine many convolution filters. They require better hardware to run on though. Yes - the problem with Leela is that it relies on an NN architecture which was used primarily for image recognition (where it performed brilliantly) and wa...
by Branko Radovanovic
Fri Aug 10, 2018 2:35 pm
Forum: Computer Chess Club: General Topics
Topic: Leela is more Chess-domain specific than regular Chess engines
Replies: 34
Views: 9222

Re: Leela is more Chess-domain specific than regular Chess engines

Also, your conclusion seems to confirm my feeling that Leela would suffer greatly in Chess960. Obviously, its opening-phase advantage over other engines would be nullified, but it's probably much more than that. If you train it to Chess960 too, it'd work fine. If I write a classical engine and make...
by Branko Radovanovic
Thu Aug 09, 2018 9:46 pm
Forum: Computer Chess Club: General Topics
Topic: Leela is more Chess-domain specific than regular Chess engines
Replies: 34
Views: 9222

Re: Leela is more Chess-domain specific than regular Chess engines

All in all, it is true than 3x3 filters do have some difficulty with heavy pieces in Chess, but another conclusion would be that Leela is more Chess-domain specialized than traditional Chess engines, which perform better in variants deviating seriously from Chess. An excellent find and - what is mo...