Leela reacts beautifully to Stockfish's outrageous materialism in Nimzo Indian

♚COURSES kingscrusher.t...
📚 kingscrusher.t...
📚 kingscrusher.t... || kingscrusher.t...
#KCComputerChess
#KCLeelaChess
♚ Play turn style chess at bit.ly/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.
[Event "CCC3: Rapid Redux Stage 3 (30|5)"]
[Site "CCC"]
[Date "2019.01.17"]
[Round "1"]
[White "Lc0"]
[Black "Stockfish"]
[Result "1-0"]
[ECO "E21"]
[WhiteElo "-"]
[BlackElo "-"]
[MastersGameID "5024145"]
👕 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...

[@luezma]

Leela's moves are so "obvious" and natural, it's amazing

[@julioandresgomez3201]

Yes, I think among humans only a beginner would capture the g2 pawn. It´s an interesting way to play to win though. It´s true that it opens the file towards black king, but it also weakens the white king´s position that now will not castle king side and black can try to open up lines on its own with c5, d5, e5. Dynamic balance. If you play, say, knight bd7 it´s like you are not up to the challenge, you are playing to not lose.

[@JoaoTito_1]

Not really... Bxg2 was a reasonable move. It's a pawn, it ruins white pawn structure on the kingside that takes away short castle for the white king. But it is sharp.

[@mustafaunal1834]

Leela and her trolling endgames! :) :)

[@jonathanbush6197]

She's quite the sadist isn't she. I wish they would fix that. Alpha Go Zero does much the same thing, making ridiculous endgame moves which still win by the barest margin

[@neutralrobot]

@@jonathanbush6197 I think the Alpha Zero algorithm optimizes for wins, but doesn't necessarily optimize for fewest moves. In a position where the chances of losing are basically zero, like in these K+R vs K endgames, there's probably no incentive built into the reinforcement learning algorithm to improve technique.

[@StefanReich]

Maybe they don't program her to win quickly? The goal function might only be based on whether it's a win. She just wasted time in the end game.

[@modolief]

I wonder if they could bake something into the learning routine that would favor shorter winning paths without compromising winning percentages. Perhaps when Leela's estimated chance to win is above a certain number x, say x = 90%, then more points would be added for moves that lead to shorter game playouts. Maybe x would start very high (say x=99.9%) when the learning process is just beginning (think early versions of Leela), in order not to limit Leela's creativity, but get progressively lower (x=90% or x=80%) as time goes by and learning reaches it's plateau values (e.g. most recent versions of Leela--though we don't really know where that plateau is, I think).

[@StefanReich]

@@modolief Yes, exactly

[@kingscrusher]

Q. Will neural networks like Leela be used to rewrite Chess Opening books ? 🤔😀😎🌍 Video: goo.gl/AsJRiH Replayable game: goo.gl/Z7hstV #chess #chessgame #ArtificialIntelligence #AI #MachineLearning #DeepLearning

[@mjgayle52]

The way LCZ plays for open lines reminds me a bit of a strategy in space invaders. - but then i am ancient - lol

[@user-lo9tl5fx6o]

Great commentary sir is it Stockfish 10 vs Leela plz give answer bcoz L0 is playing now like Alpha zero but in different manner ie In LZ Styles !!!

[@Askhat08]

It is SF 11 dev.

[@seasideman]

As far as I can see, Leela's dominance here is enabled by the pawn grab which opened the g-file. This is a rather fascinating moment in which, I think, most human players would not have been as greedy as Stockfish was (even I wouldn't have grabbed that pawn). The horizon effect prevents Stockfish from being able to analyse deeply enough to see that this is a major positional blunder. Although, really, a brutish engine like Stockfish doesn't have any conception of "positional".

[@windowslogo3577]

>greedy

[@SoldierOfTwilight]

0:13 "Let's see what happened." Just a wild guess, but maybe Leela reacted beautifully to Stockfish's outrageous materialism in Nimzo Indian? :)

[@piyushsharma5938]

A moral lesson taught by Leela. ..... one shouldn't be greedy. ....

[@modolief]

Dear KC: You know, when I came up in chess, there were books like _Bobby Fischer Teaches Chess_ (got me my first win), Fred Reinfeld's _1001 Brilliant Ways to Checkmate_ and various game collection and strategy books, such as _Paul Keres Chess Master Class_ by by IAkov I. Neishtadt and Kenneth P. Neat (which was significantly above my head). But in this modern era of KZbin videos, game collections must be done by video instead: chess is such a visual / geometric game that only the top 0.1% of chess players/enthusiasts can keep a game in their head when viewed of the written page, so that the video format is orders of magnitude better. The viewer gets a much better flow, can pause and rewind when needed, and the content is information packed without being too abstract. Then, when supplemented by annotated PGN, video playlists along various themes, and perhaps some choice static content (i.e. book or book-like materials -- html pages or pdf files, for example), there can be an amazing learning package. What I'm getting to is: I think you are basically becoming the Fred Reinfeld of the 2010s and 2020s ... really good video exposition, themes developed over time (thorn pawn, pieces in Siberia, and classical themes like central control, flexible moves, like one of the rook moves in this video) ... just really high quality stuff. Please keep it coming, I think you're creating a great legacy 😀 !!

[@jimbo92107]

Dance little king, dance! Leela made the black king do a little twirl in the middle!

[@dannygjk]

Dance little -Jean- king... XD

[@martinet1985]

almost 100.000 subscriptions!

[@dannygjk]

If a traditional engine doesn't find a concrete line which demonstrates that grabbing material is bad they will grab the material. That was true 30 years ago and it is still true today because traditional engines have rudimentary positional skill. I could add caveats to that but I won't bother unless someone pulls off one glove, uses it to slap my virtual face then says, "I challenge you to a duel.".

[@julioandresgomez3201]

www.chessgames.com/perl/chessgame?gid=1937947 www.chessgames.com/perl/chessgame?gid=1938021 Curiously enough, there´s a couple of games in which A 0 like its little "sister" plays bishop d3 and knight d2 offering the g2 pawn, then Stockfish plays the natural retreat bishop b7 instead of the awkward-looking bishop c6 (what?).

[@dannygjk]

Stockfish you will reap what you sow. If you sow little you will reap little.

[@dannygjk]

Let us not mince words. Just word it, "...Stockfish's outrageous greed..."

[@kevinbyrne4538]

13:19 -- The mating (?) sequence was a bit ... unconventional.

[@LeelooMinai]

I was a bit worried there ("what if this goes on for the next 1000 moves...?")

[@dannygjk]

Yes Leela filet the brute.

[@feelzthedon994]

LMAO 😂😂 😂😂 😂. Leela is funny as hell with her trolling of Stockfish. I can tell Leela and her team don't much care for Stockfish.

[@romankusnir45958]

Do you have any idea about why Leela is trolling?

[@amjadabbas5952]

Stockfish loves pawns...I use the same Idea to beat many players I give them a pawn the beat them... hhhhh

[@ChessExpert-fq8hb]

☺☺☺

[@kingscrusher]

Cheers, K

[@kamboonseng]

And this version of Stockfish is ....?

[@Askhat08]

SF 11 dev.

[@dannygjk]

LOL sorry to burst your cherry! (not)

[@Indiancommander]

Which ID is this?

[@kingscrusher]

Leela ID here is 32425

[@bhgtree]

@@Straight_Talk I'm not KC, but maybe I can help. The ID is the number of the networks as they are trained and uploaded on lczero.org ( the 30xxx and 40xxx are the current runs and higher id means more recent, but may not be stronger, network).