Search found 222 matches

by Sergei S. Markoff
Mon May 13, 2019 5:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: One idea to solve chess?
Replies: 35
Views: 11423

Re: One idea to solve chess?

EGTB's are desirable because they are FAST. The classic Time vs Space tradeoff. NN's are not even close to optimal models of any domain. Not even close. Keep that in mind, that its not even close. Let me say it again, that it is not even close. Very far from optimal. So very far. Let me say it agai...
by Sergei S. Markoff
Sun May 12, 2019 2:06 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: One idea to solve chess?
Replies: 35
Views: 11423

Re: One idea to solve chess?

Let introduce me, I made a chess program in the 90s and organized one Spanish Computer Chess championship in Valencia, my city. Worked as an AI programmer in the videogame industry some years, and now on the internet sector. It looks, I've came the same idea :) See http://talkchess.com/forum3/viewt...
by Sergei S. Markoff
Sat May 11, 2019 4:52 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: The Plan-9 to finally solve chess :)
Replies: 0
Views: 1466

The Plan-9 to finally solve chess :)

An idea how to train NN to play chess w/o RL using only TBs. 1. Train some deep NN on 7-men tablebases. 2. Let's assume some trust men limit, initially L = 7. 3. Create empty minibatch for NN. 4. Gererate random L + 1 men position and perform N-ply search (optimal N value is disputable) val neg_valu...
by Sergei S. Markoff
Sat Apr 20, 2019 9:55 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Re-Pair compression questions
Replies: 11
Views: 4391

Re: Re-Pair compression questions

For the same reason i have some doubt that replacing Huffman with any of the ANS entropy encoders is a good idea. It would make the tables a bit smaller, but that is just one part of the story. (For DTZ tables probed only at the root it likely does make a lot of sense.) I think it's worth to try us...
by Sergei S. Markoff
Tue Dec 11, 2018 4:04 am
Forum: Computer Chess Club: General Topics
Topic: Advance chess
Replies: 1
Views: 553

Advance chess

Wikipedia suggests that chess engines are now stronger than centaur (human + engine) players (https://en.wikipedia.org/wiki/Advanced_Chess#cite_note-7). "Computer chess engine Zor won the freestyle 2017 Ultimate Challenge tournament. The best human plus computer came in 3rd place. In 2017, chess com...
by Sergei S. Markoff
Thu Mar 08, 2018 8:52 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Feed bayeselo with pure game results without PGN
Replies: 6
Views: 14310

Re: Feed bayeselo with pure game results without PGN

Thank you! I'm going to make a tool to feed results in a most compact form, for example: engine1: {engine2: {w: {100/100/101}, b: {95/100/105}}, engine3: {w: 101/100/100, b: {96:100:104}} engine2: ... etc So instead of multiple games or game headers you will have just results. In the case you're hav...
by Sergei S. Markoff
Thu Mar 08, 2018 8:47 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Feed bayeselo with pure game results without PGN
Replies: 6
Views: 14310

Re: Feed bayeselo with pure game results without PGN

Thank you! But I have a multiple versions, so I need to fit their ratings. My framework is based on genetical approach. The previous version was based on playing vs base version, but there are two problems: 1) if you will be able to use games of some sibligngs vs each other it will help you to save ...
by Sergei S. Markoff
Thu Mar 08, 2018 4:40 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Feed bayeselo with pure game results without PGN
Replies: 6
Views: 14310

Re: Feed bayeselo with pure game results without PGN

Why not feed it a minimal PGN file with the players and the result ? cutechess-cli should be able to produce minimal PGN. Because it's too slow to parse PGN with 10 mln games, even minimal, especially when you need to do it after every batch of games per CPU core. Moreover, it's hard to even store ...
by Sergei S. Markoff
Thu Mar 08, 2018 1:52 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Feed bayeselo with pure game results without PGN
Replies: 6
Views: 14310

Feed bayeselo with pure game results without PGN

Hello all! Do you have any ideas how to quickly feed bayeselo with game results array? In my tuning framework I'm going to play several millions of games per day, so storing them in PGN and then feeding them to elostat will be too slow. Of course, it's possible to tweak elostat source to do it, but ...
by Sergei S. Markoff
Sun Feb 04, 2018 9:26 pm
Forum: Computer Chess Club: General Topics
Topic: SmarThink 1.98 is out
Replies: 26
Views: 9029

Re: SmarThink 1.98 is out

May be it's because of some hashtable data from previous search? I've tried to compare on your position and it still looks the same: Transposition size is 524288 entries (13631488 bytes) Pawn hash size is 262144 entries (8912896 bytes) Aggressivity = 100 (50)% Cowardice = 100...