The Leela Chess Zero project was first announced on TalkChess.com on January 9, 2018. Leela Chess Zero is a free, open-source, and neural network based chess engine. Leela Zero and AlphaGo Zero, Leela Chess Zero only knows the basic rules and nothing more.
According to Wikipedia, Leela Chess Zero quickly reached the GM level, surpassing the strength of the early releases of Rybka, Stockfish, and Komodo.
In April 2018, Leela Chess Zero became the first neural network engine to enter the Top Chess Engine Championship (TCEC), during season 12 in the lowest division, but did not perform well, scoring +1 -26 =1. But, it improved quickly.
In December 2018, Leela participated in season 14 of the TCEC and dominated divisions 3, 2, and 1, easily finishing first in all of
them. In the premier division, Stockfish dominated while Houdini,
Komodo and Leela competed for second place. It came down to a
final-round game where Leela needed to hold Stockfish to a draw with
black to finish second ahead of Komodo. This it successfully managed,
and therefore contested the superfinal against Stockfish. It narrowly
lost the superfinal against Stockfish with a 49.5-50.5 final score.
In February 2019, Leela scored its first major tournament win when it defeated Houdini in the final of the second TCEC cup where it did not lose a game. And, in April 2019, Leela won the Chess.com Computer Chess Championship: Blitz Bonanza.
Chessbase had a good article on it last year HERE. Installation instructions are also included, but you can also find them on Youtube.
You can play against it on lichess. On the right hand side, click on “Play with the machine” then pick your level, time and color.
Yesterday I played it a 10 minute game on level 1 and won quite easily, but today playing it 10 minute games on level 8 was a different story! I became the crushee. Anyway, it was fun to play against.
Here are the estimated Elo for each level.
Level 1 = 1350
Level 2 = 1420
Level 3 = 1500
Level 4 = 1600
Level 5 = 1700
Level 6 = 1900
Level 7 = 2200
Level 8 = 2700
[Event "Online"]
[Site "Lichess"]
[Date "2019.4.19"]
[Round "?"]
[White "Tartajubow"]
[Black "Leela (Level 8)"]
[Result "0-1"]
[WhiteELO "?"]
[BlackELO "?"]
%Created by Caissa's Web PGN Editor
{Urusov Gambit} 1. e4 e5 2. Bc4 Nf6 3. d4 {On level 1 the engine played the
inferior 3...Nxe4 4.dxe5 Nc6 which ia a blunder. 5.Bxf7+ followed by 6.Qd5+
gives white an overwhelming position.} 3... exd4 {Now 4.e5 looks inviting but
after the typical 4...d5 black has equalized.} 4. Nf3 Nxe4 {What theory there
is on the Urusov suggests that 4...Nc6 is best.} 5. Qxd4 Nf6 6. Nc3 {White has
a lead in development, but won't be able to maintain it. Also, with no Ps in
the center black will soon gain the upper hand there too.} 6... c6 7. O-O d5
8. Re1+ Be7 9. Bd3 O-O 10. Bg5 {Stockfish is evaluating this position at
almost dead equal after 10...h6 or 10...Na6 or 10...Re8. The more I look at
the position the more I like black's chances in the long run.} 10... a5
{Stockfish slapped this with two question marks in the auto-annotaion. The
reason is that after 11. Qh4 white has a considerable superiority because
11...h6 is pretty much forced and after 12.Bxh6+ Ne4 white has regained the P
and has the makings of a strong attack.} 11. Rad1 {Routine and played almost
instantly. I got two question marks from Stockfish for failing to play 11.Qh4.
I was going to play it, but just not yet. The position is evaluated at 0.00.
I still like black's long range chances.} 11... h5 {11...h6 would have been
met by 12.Bxh7 gxh6 13.Qe3} 12. Qh4 Bg4 {At this point I kind of ran out of
ideas. After my next move the position is still equal, but Stockfish prefers
13.Ne2 and 14.Nfd4.} 13. h3 Bxf3 14. gxf3 Nbd7 15. Ne2 {Transferring the N to
the K-side where it may find something to do. Vague, but it's a plan.} 15...
Ne5 {Most unpleasant. Besides threatening the fork on f3, at some point I am
going to lose the two Bs. Here I was beginning to feel a little pessimistic
even though Stockfish says black is better by only half a P.} 16. Kg2
{Clogging up the open g-file can't be good. 16.Nd4 and 17.f4 was better.}
16... Re8 {After a minute or so I couldn't decide where to put the N, on f4 or
g3. According to Stockfish there's not much difference.} 17. Nf4 Nfd7 18.
Bxe7 Qxe7 {A little more time was wasted on my next move. Either way I win
the h-Pawn, but should I trade Qs or not? I decided not too because there
seemed no point in simplification.} 19. Qxh5 Qf6 20. Qh7+ Kf8 21. Qf5 {I
decided simplification was unavoidable...my Q was starting to look a little
vulnerable. It turns out that 21.Bf5 was a little better, but black is
already getting the upper hand.} 21... Qh6 {This is interesting. Without
making any serious mistakes I have already achieved a position that is
lost...just about any engine should be able to win from this position.} 22.
Ne2 Nc5 {My next move loses quickly. Better was 23.f4, but even then, black
has the much better position.} 23. Ng3 g6 {Here I realized my Q doesn't have
any retreat squares.} 24. Qf6 Re6 {I have one rather pathetic shot.} 25. Nf5
{This might have scared a human, but not an engine.} 25... Qh7 26. Qh4 Qxh4
27. Nxh4 {At least I saved the Q.} 27... Nexd3 28. cxd3 Rxe1 29. Rxe1 Nxd3
30. Rb1 c5 {But I couldnt save the game...this ending is completely lost.}
0-1
No comments:
Post a Comment