3 ms·
Checkout Alpha Zero chess which has no other information other than the chess rules. No opening books or repetoires. It sometimes beats Stockfish like a python
by systemvoltage 4y ago
Checkout Alpha Zero chess which has no other information other than the chess rules. No opening books or repetoires. It sometimes beats Stockfish like a python strangling and squeeing a prey. It makes moves that no one understands.
> In a 1000-game match, AlphaZero won with a score of 155 wins, 6 losses, and 839 draws. DeepMind also played a series of games using the TCEC opening positions; AlphaZero also won convincingly. Stockfish needed 10-to-1 time odds to match AlphaZero.
https://en.wikipedia.org/wiki/AlphaZero https://en.wikipedia.org/wiki/AlphaZero
It is savage.
- awuji 4y agoIt played Stockfish 8, which was already a year old when they played, and the match was under arguably unbalanced circumstances. The current version of Stockfish is Stockfish 15, and although it hasn't played AlphaZero (because AlphaZero doesn't compete), it has played clones of AlphaZero such as Leela Chess Zero in TCEC which it won against consistently for the past 5 seasons. AlphaZero beating Stockfish was more a propaganda attempt from Google to appear as a world leader in AI then an actually attempt to find the superior engine.
- systemvoltage 4y agoStockfish is definitely strong but the way LCZ or AZ beats its opponents is totally different and foreign to humans. Stockfish has a database and a whole bunch of openings prebaked into it. Here are the actual stats: https://en.wikipedia.org/wiki/Top_Chess_Engine_Championship#Statistics https://en.wikipedia.org/wiki/Top_Chess_Engine_Championship#... Leela Chess Zero is definitely not as strong as Stockfish 14 but still ranked #4. That's not nothingburger. > AlphaZero beating Stockfish was more a propaganda attempt from Google to appear as a world leader in AI then an actually attempt to find the superior engine. Wtf?
- srvmshr 4y agoAgree. I feel awful that Deepmind honchos take a round table meeting once in a while, where they pick up a board or online game and totally demolish the top player just to prove how good ML algorithms are.