Alphazero reveals unique trump card coordination with beautiful piece sac vs Stockfish - Game 16

♚COURSES kingscrusher.t...
📚 kingscrusher.t...
📚 kingscrusher.t... || kingscrusher.t...
#KCComputerChess
#KCAlphaZero
♚ 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...
Alphazero reveals unique trump card coordination with beautiful piece sac vs Stockfish - Game 16
👕 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...

[@chrisdarakjian6351]

I was a bit irritated that I couldnt figure out which side was stockfish and which side was alpha zero until 2:20 when you mentioned alpha pushing the pawn. Then I realized that alpha is always on the bottom (because the bottom was black rather than white). Why not squeeze the names somewhere on the board?

[@Maharani1991]

Agreed!

[@jesussavior6383]

www.godlikeproductions.com/forum1/message3965195/pg23#71808126 22

[@pnachtwey]

One thing I learned is that it makes little difference how deep the search is if searching for the wrong thing within reason. I think A0 uses a different way for scoring piece values from what we are taught. chess programs usually use a progressive breadth first search so the shouldn’t be blind spots. there is a quiescent search at the end points that is not a full width search. I bet this is the second place significant improvements have been made. The quiescent search is where there could be blind spots but these would be very deep. I am still amazed at how A0 can be 3 pawns down and still think it is OK. It is almost as if there is a algorithm for what squares are attacked but also a second algorithm for what squares can be attacked it a piece was moved. I have seen A0 sacrifice friendly pieces that are in the way or to lure enemy pieces out of the way but first the indirect attacks must be setup first.

[@kingscrusher]

Replayable game with indented variations: www.chessworld.net/chessclubs/ltpgnviewer32/ltpgnboard.asp?GameID=5024024&v=KmkXQKvsZYc

[@joseraulcapablanca8564]

A fascinating new idea and a commentary,which gives us chess, entertainment, and philosophical advice. Thanks KC keep up the good work

[@peterpetrov6522]

KC's analysis is truly awe inspiring! I have a feeling that even the super GMs can learn a thing or two!

[@kingscrusher]

Cheers, K

[@MrCiprian2179]

Thorn pawns will be the new era of chess novelties.

[@mustafaunal1834]

Absolutely. We are humans do not fully understand the value of thorn pawn. At least we didn't understand until now...

[@rogerrothman]

I’m fascinated by the way you use the word “celebrate” (as in: “This pin can be celebrated by pushing the g pawn...” I think it’s a great way of using that word. Did you coin that usage or did you hear it somewhere? I’ve searched online for similar usages of the word but haven’t found it.

[@reynaldovelasco5129]

I agree.

[@kingscrusher]

I think i tried to work it out myself from a dramatic game example once of a player called Stanton years back. I was wondering why he didn't exploit a pin immediately and wanted to try and conceptualise for myself what the other approach was of making use of the pin, and thought "celebrate" seemed a cool idea. I am not conscious of reading this term from any books or videos - but possibly it did slip into my conscious from somewhere previously to that. Cheers, K

[@rogerrothman]

kingscrusher: really interesting. I ask in part because I’m an academic (modern art) and am working on a book in which the idea of celebration is central. Your use of the term is very interesting to me and your explanation (in your reply to me) of why you realized that other words don’t fit is even more so. I think you’re right that the use to which you’ve put the word is one that isn’t met by other English words. I’m going to think more about this. Thanks!

[@rogerrothman]

... and i think it deserves to become part of the regular lexicon of chess analysis. I hope it does!

[@vargas2022]

Long live perpetual check. Kudos to Stockfish. We indeed have to try to keep stock of our soft spots. Otherwise, this happens. Opponent's tactical defenses creep in to save the day.

[@mustafaunal1834]

Good work KC. Thank you.

[@-ace-52]

Beautiful game and a great, accurate analysis. Wouldn't it be interesting if TCEC or CCCC let Leela play crucial positions from the A0 vs SF match in their bonus games? They have sufficient graphic cards.

[@Maharani1991]

Love the idea! :)

[@pnachtwey]

This almost makes me want to start writing chess programs again. I am amazed at how deep the computers can search now. I am still not convinced A0 uses NN for pattern recognition but it is obvious that it doesn't evaluate pieces in the normal 1,3,3,5,9 manner. It is if A0 is constantly asking itself what each piece or pawn is doing for it and if a friend pawn is in the way it gets sacrificed I think I need a new computer with a GTX 1080

[@afterthesmash]

This makes no sense at all. Pattern recognition/classification is the _only_ thing a neural network can do. You can train it to put a rook into class 5 and a queen into class 9 and a queen and a rook into class 14 (etc etc) but all you're doing deep down is classifying your way into the additive integers.

[@afterthesmash]

Another thing, Deep Blue from 1997 was searching 200 million positions per second, whereas Stockfish (on typical hardware) is searching more like 60 million positions per second (twenty years later, despite vastly faster silicon). The algorithmic improvements lie mainly is searching the good moves more narrowly and the bad moves less broadly. If the search is overly narrow, the program winds up with blind spots. If you've got blind spots, and your adversaries discover your blind spots, you'll never win under tournament conditions. Kasparov excelled at discovering and exploiting blind spots. So IBM went full blunderbuss with the overly broad search and the program mainly displayed a grinding, but uninspired style-with here and there a spot of clairvoyance, whenever the grinding caught scent of a stairway to heaven. Kasparov had no fear of Deep Blue's grinding style, but its odd forays into clairvoyance terrified him to the marrow. He had only those two mental categories: confidence and terror. So when Deep Blue famously blundered and made a random move due to a hardware or programming error, Kasparov reasoned that the move couldn't possibly be the result of grinding, therefore it has to be clairvoyance, and he just couldn't see the clairvoyance on the chess board (with good reason: there wasn't any). After ten minutes of staring into his ultimate abyss, he was never the same man again.

[@dannygjk]

@@afterthesmash You are projecting higher level abstract thinking onto AZ. A form of intuition does *emerge* from the nn approach but AZ's moves are still partially a function of a tree search. Also you can look at the (trained)nn itself as a sophisticated heuristic.

[@travisheck5979]

The FRIENDS!!!!!! :)

[@kingscrusher]

Fun comment - Cheers, K

[@sofbouss7891]

You rejoice whenever there's a thorn pawn in a game

[@kingscrusher]

Cheers, K

[@sandercoven2554]

When are you going to change the back round picture? Always watch your videos.

[@smashu2]

You should also check Alpha0 draw game with white also I found alpha missed the win in the 250+ move nimzo around move 20 alpha play Rb4 instead of the obvious Ra4 its very weird after alpha move it lead to a position SF dev giv + 2.28 to white but it is a fortress while after Ra4 the game would have been over very fast..

[@smashu2]

I also think many draw Alpha is white it possible to improve on white play after alpha achieve a clear edge Alpha sometimes slip in the tactic or endings.

[@Maharani1991]

+

[@kevin27966]

Can you post PGN (or link to allow for exploration of positions)?

[@jamesl939]

Fawn pawn??

[@amazimi10]

Stockfish, you coward!!

[@xyzct]

Soon AI-led wars will leave 3 billion dead on each side, forcing a draw. But in the meantime ... great fun, KC :-)

[@kingscrusher]

Fun comment - Cheers, K

[@Norpan506]

This might sound boring, but all games in the future will end in a draw :)

[@u.v.s.5583]

Only if the other side can counterfawn a pawn for every pawn you fawn, because otherwise you win by fawn.

[@mapifisher]

No fear of that if I'm still playing. I have lose-all-your-pieces and run-out-of-time skills honed to avoid draws.

[@mal2ksc]

I do think we're getting to that point, but remember that this was a TCEC style opening, and those are a distinct way to avoid draw death.