Search found 206 matches

by Tony P.
Fri Oct 02, 2020 5:19 pm
Forum: Computer Chess Club: General Topics
Topic: Busted French line?
Replies: 13
Views: 2919

Re: Busted French line?

The line hasn't been explored deeply on chessdb.cn yet :( You can submit moves for analysis there and see what SF says at depth 22 (and also make a better use of transpositions than Homefish would). So far, the score is estimated to be between +50 and +70 (centipawns?).
by Tony P.
Fri Oct 02, 2020 4:37 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Dangerous Positions
Replies: 9
Views: 1941

Re: Dangerous Positions

I wouldn't try to make a static eval training set deliberately quiet either, as that would be an unnatural move in the other direction :D A training set for a quiescence (/ eval uncertainty) model would need dangerous positions because the model's very purpose would be to discriminate between them a...
by Tony P.
Fri Oct 02, 2020 4:32 pm
Forum: Computer Chess Club: General Topics
Topic: Is Allie a 'Leela net player'?
Replies: 57
Views: 8457

Re: Is Allie a 'Leela net player'?

Realistically, any NNUE engine authors could take some old sergio net, put it through a short SL phase that doesn't change much its strength but change all the floats making up the weights so the source can't be traced, claim it as his own achievement, and get a net that's a few elo short of SF's o...
by Tony P.
Fri Oct 02, 2020 3:38 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Dangerous Positions
Replies: 9
Views: 1941

Re: Dangerous Positions

Sorry, I'm not qualified to advise on the centipawn cutoff. On this forum, I'm filling the niche vacated by Lyudmil Tsvetkov: I have very controversial views on how a chess engine should operate in general, not backed by code yet :mrgreen: In particular, I'm not a fan of incorporating tactical insig...
by Tony P.
Fri Oct 02, 2020 2:32 pm
Forum: Computer Chess Club: General Topics
Topic: A Crossroad in Computer Chess; Or Desperate Flailing for Relevance
Replies: 152
Views: 21156

Re: A Crossroad in Computer Chess; Or Desperate Flailing for Relevance

That's a good suggestion. The catch is that Andrew will then need to pursue something very different. It will be hard to build a business around Ethereal unless Andrew manages to undo all the other contributors' commits and end up with a well-working engine that doesn't break others' GPL rights and ...
by Tony P.
Fri Oct 02, 2020 6:48 am
Forum: Computer Chess Club: General Topics
Topic: A Crossroad in Computer Chess; Or Desperate Flailing for Relevance
Replies: 152
Views: 21156

Re: A Crossroad in Computer Chess; Or Desperate Flailing for Relevance

Andrew, as an ex-esportsman, you know very well that the meta is a rapidly moving target even while the rules are left intact (the latter is, btw, not a given in chess after the latest Alpha0 paper). CC is no different :P It's very competitive. Fortunately, it's just a game too. 60 hrs a week are in...
by Tony P.
Fri Oct 02, 2020 6:33 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Dangerous Positions
Replies: 9
Views: 1941

Re: Dangerous Positions

My 2c :P It doesn't seem efficient to teach a static centipawn/WDL estimator to detect double-edged positions like #1. It would make more sense to me to train a QS controller to extend in them and not extend in truly quiet ones*. To this end, it needs the attack tables, the king safety info and poss...
by Tony P.
Thu Oct 01, 2020 10:32 pm
Forum: Computer Chess Club: General Topics
Topic: Ethereal Pawn-King NN
Replies: 38
Views: 9586

Re: Ethereal Pawn-King NN

I've had a possibly stupid thought that it could make sense to count P & K moves during QS; if <=m P/K positions have changed between the QS start and the leaf where static eval is called, then update the net incrementally; if >m, update fully. (The optimal m is probably 1, but I'd test bigger value...
by Tony P.
Thu Oct 01, 2020 8:51 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: How to calculate piece weights with logistic regression?
Replies: 12
Views: 2887

Re: How to calculate piece weights with logistic regression?

Did you see Andrew Grant's article? It's the go-to for all things chess logistic regression :D As a bonus, his code examples are in C. Why would you need Python at all?

The PDF hosting is down now, let us know if it fails to go back up soon.
by Tony P.
Thu Oct 01, 2020 8:04 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Re-activation Programmer Code of Honor
Replies: 57
Views: 11255

Re: Re-activation Programmer Code of Honor

I personally don't consider finding datatables in an engine as reverse engineering. Its something you can teach a 5 year old to do with a text editor. Actually, there's a non-chess commercial binary where I tend to look up string constants with Notepad++ as a bit of business intelligence. Its core ...