LC0 32566 is a monster! | Leela Chess reacts wonderfully to Stockfish in Dutch Leningrad

♚COURSES kingscrusher.t...
📚 kingscrusher.t...
📚 kingscrusher.t... || kingscrusher.t...
#KCComputerChess
#KCLeelaChess
♚ Play turn style chess at bit.ly/chessworld
♚ Play Chess vs. Kingscrusher and others: www.chessworld...
♚ Subscribe to best KZbin Chess Video Channel : bit.ly/kingscru...
🎥 Leela Chess Playlist: bit.ly/leelachess
FIDE CM Kingscrusher goes over amazing games of Chess every day, with a focus recently on games of Neural Networks which are opening up new concepts for how chess could be played more effectively.
👕 Kingscrusher Store: teespring.com/...
♚Donating via Paypal: goo.gl/7HJcDq
♚Channel exclusive perks: / @kingscrusher
Game quality tags: amazing, awesome, astonishing, brilliant, classic, crushing, dynamic, elegant, exceptional, excellent, exciting, fabulous, famous, fantastic, finest, flashy, greatest, important, impressive, incredible, instructive, interesting, magnificent, marvellous.
Info about Leela Zero:
en.wikipedia.o...
...
Leela Chess Zero (lc0) is a free, open-source, and neural network-based chess engine and distributed computing project.
Leela Zero's algorithm is based on DeepMind's 2017 paper about AlphaGo Zero.[3][6] Unlike the original Leela, which has a lot of human knowledge and heuristics programmed into it, Leela Zero only knows the basic rules and nothing more.[7]
Leela Zero is trained by a distributed effort, which is coordinated at the Leela Zero website. Members of the community provide computing resources by running the client, which generates self-play games and submits them to the server. The self-play games are used to train newer networks. Generally, over 500 clients have connected to the server to contribute resources.[7] The community has provided high quality code contributions as well.[7]
Leela Zero finished third at the BerryGenomics Cup World AI Go Tournament in Fuzhou, Fujian, China on 28 April 2018.[8]
Info about Alphazero:
en.wikipedia.o...
AlphaZero is a computer program developed by the Alphabet-owned AI research company DeepMind, which uses an approach similar to AlphaGo Zero's to master not just Go, but also chess and shogi. On December 5, 2017 the DeepMind team released a preprint introducing AlphaZero, which, within 24 hours, achieved a superhuman level of play in these three games by defeating world-champion programs, Stockfish, elmo, and the 3-day version of AlphaGo Zero, in each case making use of custom tensor processing units (TPUs) that the Google programs were optimized to make use of.[1] AlphaZero was trained solely via "self-play" using 5,000 first-generation TPUs to generate the games and 64 second-generation TPUs to train the neural networks, all in parallel, with no access to opening books or endgame tables. After just four hours of training, DeepMind estimated AlphaZero was playing at a higher Elo rating than Stockfish; after 9 hours of training, the algorithm decisively defeated Stockfish 8 in a time-controlled 100-game tournament (28 wins, 0 losses, and 72 draws).[1][2][3] The trained algorithm played on a single machine with four TPUs.
...
Relation to AlphaGo Zero
Further information: AlphaGo Zero
AlphaZero (AZ) is a more generalized variant of the AlphaGo Zero (AGZ) algorithm, and is able to play shogi and chess as well as Go. Differences between AZ and AGZ include:[1]
AZ has hard-coded rules for setting search hyperparameters.
The neural network is now updated continually.
Go (unlike Chess) is symmetric under certain reflections and rotations; AlphaGo Zero was programmed to take advantage of these symmetries. AlphaZero is not.
Chess can end in a draw unlike Go; therefore AlphaZero can take into account the possibility of a drawn game.
AlphaZero vs. Stockfish and elmo
...
►If you like my chess videos please consider become a full member here at my Chessworld Site. www.chessworld....
#KCChess game tactics
♞ Challenge KC and others for turn style chess at bit.ly/chessworld
♚TEAM kingscrusher.tv...
♚TWITTER kingscrusher.tv...
♚DISCORD kingscrusher.t...
♚TWITCH / kingscrusher
♚REDDIT kingscrusher.tv...
♚QUORA kingscrusher.tv...
♚DONATE paypal.me/king...
♚COURSES kingscrusher.t...
♚COURSES kingscrusher.t...

[@kingscrusher]

Q. Will Neural Networks become stronger than traditional Alpha Beta Engines in 2019 ? 🤔😀😎🌍 Video: goo.gl/5iC5yg Replayable game: goo.gl/jS8Eys #chess #chessgame #ArtificialIntelligence #AI #MachineLearning #DeepLearning

[@plasmaastronaut]

hard to say. leela is helping AB programmers by finding their weaknesses, so they can be corrected more easily. That leela isn't getting any better than SF suggests they are going to be similar strength as they help each other get better. who will win? unknown, will be interesting to see

[@zaksmith1035]

It is almost certain. NN engines are accelerating in their improvement, while AB programmers are scrambling to keep up. It's fairly likely that better NN constructs will be implemented, as well, resulting in faster training. I'm waiting for the day that NN engines are trained to help US learn to play the game.

[@yamahantx7005]

Why did they drop a new network?!?!?! Why? I want the 3x,xxx networks back! She surpassed SF8 and SF9 on my PC in the last month, and they give up now?!?!? This is AI genocide! lol I no longer donate GPU time because of this!

[@wakuseichikyuu]

wow! Leela made big progress in the endgames. This game is full of finesse. Thks KC.

[@kingscrusher]

Cheers, K

[@peterpetrov6522]

"But I've been taught that a rook is worth more than a bishop" said Stockfish.

[@modolief]

This game really was a masterpiece. Just stunning. And to win like that against what we have considered to be such a mature, nigh invincible program in Stockfish 10.

[@doncar9]

The strongest bishop that I ever saw! A great game with an unforgettable early double exchange sacrifice . Thank you for it .

[@kingscrusher]

Cheers, K

[@pablosartor6715]

Awesome game!! Leela is becoming a monster. As usual is a very good explanation by KC! Thanks KC!!

[@CarlosRodriguez-ru8em]

Who would win between Alpha Zero and Leela right now? Would love to see a series between them.

[@mustafaunal1834]

Advanced versions of the T40 (415xx) will surely win. I'm not sure if it's for T30.

[@dannygjk]

@@mustafaunal1834 Maybe but AZ destroyed SF 9 recently.

[@hqcart1]

why does leela love shuffling?

[@kingscrusher]

Fun comment - Cheers, K

[@paulbloemen7256]

hq cart Maybe the more appropriate question is: why does Leela end shuffling? Leela probably sees it has an advantage and plays moves to keep that advantage: if necessary a few hundred of them. But then, the 50-moves rule or the 3-times repetition rule come into play, threatening a draw, so eventually Leela has to move on. There may be some lesson here. It could be that trying to win a won game is slightly more risky than maintaining an advantageous truce. So you have to be forced to move on, quite probably resulting in the win. This may be true in real life too: sitting on your butt enjoying the easy life, when sometimes it is time to go and earn the money.

[@bhgtree]

@@paulbloemen7256 So true!

[@dannygjk]

Sometimes it is just a complex example of related squares.

[@paulogomesmestreplaymobil1978]

That's a woman trait...They do this to make us suffer more :D

[@seasideman]

I think that's the most astonishing Leela game I have seen yet. The absence of counterplay was exposed by the "high class shuffling" which just slowly improved Leela's position until everything finally fell apart. I don't really get how Leela foresaw all of that, but it's very impressive indeed. To answer the question in your pinned comment, I think the answer is: no. I reckon we are 3-5 years away from NNs getting stronger than AB engines, but I do think it's inevitably going that way. The horizon effect is a big limit on ABs which Leela does not have. No matter how fast the hardware gets, it still isn't enough to calculate games like this one.

[@TheAfroNoah]

blew my mind

[@mpeskett]

Pretty smart stuff from Leela! Fingers crossed she does make the TCEC SuFi, but even if she doesn’t I’m sure she’ll make for an interesting Cup tournament!

[@mustafaunal1834]

Wow! Leela simply outplay SF.

[@moneymakerrrh]

awesome game and awesome explanation. is there a PGN for this game

[@SpiritualFox]

9:50 Rook H8 Black had a drawn position? Stockfish saw the pawn death and freespace for it's rook? Perhaps something like computer-panic? I believe a 7th rank pawn is worth 3 pawns. Might have distorted the computer's perception by hanging it there for so long. I don't see natural progress, but i'm no CM. (Thx for last comment

[@RobEdwards369]

Very nice... Thanku for these videos :)

[@aetherllama8398]

I really thought Leela would sac the rook on A1 for Stockfish's dark square bishop to gain control of the long diagonal. Any idea why Leela preferred the exchange sac for Stockfish's light square bishop?

[@phen-themoogle7651]

Bishops are more powerful in those closed positions with lots of diagonals available, also Leela was a pawn up so it was worth it for sure positionally.

[@SpiritualFox]

2:20 I called it. These computers are baffling. Knight A6? That invites a mating attack. Queen E8 was probably the first hint that Stockfish was out of book. It seems to be playing to occupy space like it doesn't understand the variation. Reminds me of the 90's when the computers were playing Queens Gambit Accepted to fit with the computer's materialism.

[@kingscrusher]

Cheers, K

[@SpiritualFox]

Zarion 11 Black's defensive pawn structure is set and missing the f-pawn. Black spends alot of time on the queenside. Black chooses to shift its queen rather than develop a piece, and places a knight as far away from the king as it can. White sacrifices early. It would probably be inadvisable for a human playing white to try so much intricate bishop manouvering, but i think i have accurately described the error. It's irrelevant, in this setting, that this opening is vetted by top grandmasters. Repeating this opening moveset is disadvantageous for black, or do you disagree with that statement?

[@dannygjk]

Qe8 is based on a standard strategical idea in many Dutch variations. Since SF is so strong I would hesitate to criticize it's decisions that early in the game. You may be right but I would try a lot of different ideas and use SF and Leela to see what they do with them.

[@phen-themoogle7651]

When you mentioned Bridge building and rook there I thought of lucena position, it's the same thing right?

[@dejavulearning121]

The same thing dude

[@Th3Wick3dOn3YT]

KC, you said its gonna be a "knightmare" in a position with no knights. Did you mean a Bishopmare?

[@xwarmangle]

Nice job.

[@kingscrusher]

Cheers, K

[@victorsamuelbaezsouffront9017]

oh my sugswang :V 10:01 / 13:42

[@phen-themoogle7651]

Dutch is hard to play, I don't take this as Leela being a monster, just stockfish not being its normal self because of not understanding the Dutch completely. Probably with slower time controls stockfish wouldnt have give the knight and pawn up, that was a weird exchange

[@guepardiez]

11:10 "the pawn chain is just being _mould_ here"?

[@derekhoover5679]

*mauled

[@guepardiez]

@@derekhoover5679 Thanks! I thought I had tried all possible spellings, but I was wrong.

[@peterdriscoll4070]

One game does not mean anything, if it is only random chance giving an advantage.

[@davidchang8428]

self-rating of over 11,000 ???? lol!