Leela reveals a beautifully simple positional chess secret vs Stockfish bypassing any complications

♚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
👕 Kingscrusher Store :teespring.com/...
♚Donating via Paypal: goo.gl/7HJcDq
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]

Replayable game with indented variations: www.chessworld.net/chessclubs/ltpgnviewer32/ltpgnboard.asp?GameID=5023947&v=gOS4jaFztqs

[@holahola-gp6vd]

Can you also show how Leela loses to stockfish, because it lost many games and we want to see its weak side too.

[@joseraulcapablanca8564]

Extremely simple and elegant, as you say Capablanca would be proud. Fine analysis too. Thanks KC keep up the good work

[@kingscrusher]

Cheers, K

[@roodborstkalf9664]

This game makes it clear to me that Stockfish's evaluation of pawn structures in the endgame can be much improved upon. It's clearly suboptimal in this game. It's a interesting challenge to human designers of traditional chess engines to figure out how to do this.

[@iamwishfordan]

she does the same thing fundamentally in Go, taking all of the positional key points in the opening and closing it out solidly. Its not uncommon for even top amateurs and some pros to feel like the game is completely lost by move 50. But for both Go and Chess I find her moves to be very entertaining and instructive. It seems positional play has the benefit of remaining flexible.

[@kingscrusher]

Great comment, K

[@paulbloemen7256]

In go, the main point of interest to me is the handling of initiative, whether one should answer a local threat or not (sente or gote). These AI-Zero machines teach us to not love the stones we played, leading to very complex swaps of groups, on an almost continuous basis. This, in return is shown in the modern (current to 1-2 year old) pro games by the very young and promising top stars from China and Korea. I surely would like to see more top Zero-Zero games, and more top pro - top Zero games to see the progress and develop the ideas behind those moves.

[@iamwishfordan]

@@paulbloemen7256 i agree. In GO, i tend to be flexible myself so I am intrigued at how they handle the middle game (although for me i cant quite get behind their fuseki/opening. Its already easier for me to be somewhat objective about stones and groups (im currently between 7-8d kgs atm). What interests me the most is the dynamic way they identify and solidify the key parts of the board, while seemingly avoiding risky exchanges, at least against humans.

[@marcusklaas4088]

KC - i enjoy all your leela videos without exception and this one in particular. very clean and instructive win. great commentary. thank you

[@mustafaunal1834]

Thank you so much KC. A detailed and wonderful analysis. This is a chess-art.

[@kingscrusher]

Cheers, K

[@RealCurrencies]

Very nice series, btw, Kingscrusher! You're definitely right, what Leela is showing is truly great chess, and much more advanced than the brute force engines, and also very much in line with what Alpha Zero was producing. It's truly beautiful to watch, no doubt about it.

[@tasdourian]

At what move did Stockfish realize it was losing by over half a pawn? When did it register being down a pawn? What I find surprising is that it couldn’t avoid the line by calculation, like it often does.

[@seasideman]

The long term strategy by Leela here was, indeed, most instructive. However, the route to success was strewn with many possible pitfalls and I doubt all but the very strongest human players could have navigated the dangerous waters to a win as Leela did. In particular, the pawn-pushing sequence after simplification is crucial and I know I would have messed it up!

[@vargas2022]

In a previous game I watched on your channel Stockfish greedily munching pawns. In this one it totally underestimates the endgame possibilities. I have to say, that makes Stockfish more human like than Leela. Funny enough, Leela is becoming an AI troll with godlike chess abilities, but definitely less human like.

[@noseonscent1935]

Whilst the idea may seem basic the execution is anything but. Thanks KC!

[@luciengrondin5802]

Wasn't the Greek gift working here? I guess not but it sure was tempting.

[@chessboy00]

I was also looking at Bxh2+ Greek giftishly

[@Attlanttizz]

Cheating, trolling, slow rolling... is there anything Leela can't do :D ?

[@kingscrusher]

Fun comment - Cheers, K

[@Attlanttizz]

@@paulwhite760 Seriously? Can't enjoy a chess game without spewing that retarded crap? Grow up and get a life already!

[@matthewviramontes3131]

I wanna see it go down in an actual tourney, not just in these side games.

[@paulbloemen7256]

Interesting indeed. At a few points, alternative White moves are indicated, giving an about equal position. Does this mean that in certain positions, Stockfish is losing it? At least to me, simple untalented but interested amateur, the winning position of Black came out of the blue, as it seems to have happened to the mighty Stockfish too, noticing its loss far too late.

[@kingscrusher]

Cheers, K

[@t6yz]

These videos are consistently interesting, sorry I don’t watch as frequently any more. You’re such a sound guy

[@JordanMetroidManiac]

End of era!

[@RealCurrencies]

It's not just a matter of 2 vs 3 (a majority, by the way, which only becomes reality after the white d and black e pawns are exchanged). Very important is that White's majority is cripled. If it was a wholesome 4 vs 3 majority, white would have had good play. Also, the Bishop is much stronger than the Knight. I agree, it's strange that Stockfish would call this 'equal'. It certainly isn't.

[@dannygjk]

Keep in mind that SF is penalizing black for 2 isolated pawns as compared to white having only 1 IP. Looking at that it makes sense that SF scores it as equal.

[@RealCurrencies]

They both have three pawn islands. It's true that especially h7 could prove a little shaky, but look at the bishop vs knight: while the Bishop can defend the pawn, while attacking the queenside, and supporting the pawns advance there, the white knight will prove lucky if he can just defend the Q-side.

[@RealCurrencies]

I think it's fair to say that most experienced players would prefer black always. Also with the rooks still on, it's a position that only black dreams of winning, while he risks nothing.

[@ot44eto]

Great game. May I ask, were can I find Leela 30? On their website there is 0.18 version?!

[@kingscrusher]

See www.lczero.org/

[@octane2344]

I liked the pawn play analysis

[@LJAreval0]

was looking forward to a greek gift but if cappy says so then it must be good anyway

[@kingscrusher]

Fun comment - Cheers, K

[@LJAreval0]

kingscrusher for second you must admit you did think of it as well right? would have loved a small analysis there as well

[@Socrates...]

I miss your blitz games

[@kingscrusher]

I still do banter blitz sessions for blitz. Cheers, K

[@Socrates...]

@@kingscrusher It's not the same because you don't play high rated players, do you?

[@chesslover2925]

Dear KC. Leela is the quenn of passed pawns. Poor stocfish 9 :(

[@holahola-gp6vd]

Stockfish 9 is stronger then Leela. Look at the database it won much more games then Leela.

[@sambowie4989]

This whole game was totally out of my league. Even with kc's great analysis. Oh man I hate it when that happens!! 😡

[@Christoff070]

Same. But that's fine, these are the most advanced computers that even Magnis couldn't beat!

[@eyeofhorus1301]

You've become quite obsessed with Leela huh? Not that I have a problem with that, I'm just saying.

[@paulbloemen7256]

Eye of Horus At least to me, rightfully so, not only on chess, but on go too. It's the nature of the technology that is fascinating to me. AI-Zero starts with zero knowledge about the subject except the rules, and builds up prowess by smart self learning based on the results. The fascinating part is: how far will it carry? If it would have remained at a sub human level, nobody would have been interested. But it surpassed human prowess fairly quickly, and it looks like it is surpassing all the human based AI systems fairly soon too. Not only is this interesting from the point of view of results, it far more is interesting because of the emerging ideas behind those result based moves, of which the AI-Zero machines are totally unaware. Which ideas were already known to us humans, which human ideas prove to be wrong, which new ideas emerge? So, GO LEELA, GO GO GO, become as strong as possible, but foremost, show us your truth about the game!

[@yamahantx7005]

Currently, with Leela running on my GTX 1080(CUDA) vs Stockfish 9 on my i5-4690k(overclocked), she's getting really close to parity. The million dollar question is WHEN will Leela surpass Stockfish for my hardware? I'll admit that the video card is more powerful than the CPU, but if stockfish wants to release a GPU version of their engine I will run it on the GPU.

[@eyeofhorus1301]

@@paulbloemen7256 I agree. I'm a lot more interested in Alpha Zero than Leela though. Did you hear? Demis Hassabis was on as a guest at the world championship a couple days ago and he implied Alpha Zero will be back in chess soon with many more released games. And he said Leela is pretty strong but not as strong as Alpha Zero.

[@paulbloemen7256]

Eye of Horus Hip hip hooray if Demis Hassabis does as he said, and hopefully this is also true for go! I hope some fundamental issues have been addressed the last year, like selecting a move not only based on its chance to win the game, but also on the winning margin with about equal winning chance, thereby buiding a buffer of points for when unpleasant surprises happen, which will be the case several times in every game. And how I would love to be able to use an analysis tool based on Zero technology that actually works on my iPad!