[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • You may highlight syntax and preserve whitespace by using [code] tags.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


Janitor applications are now open. Apply here!


[Advertise on 4chan]


File: GXHVoRNXcAAa5fH.jpg (97 KB, 1024x1024)
97 KB JPG
I'm still freaked out about stockfish, forget LLMs, I mean shouldn't the halting problem or some other law cockblock machines from being better than the data they train on.
>>
Chess is completely logical and can theoretically be solved with stupid algorithms like alpha-beta pruning. The problem is the phase space is fuckhuge. Checkers was actually solved, you may want to look into that. Computers are perfect at Checkers and impossible to beat.
>>
Stockfish is not some classifier that is trained on a dataset mapping board positions to a best move. It uses a search algorithm. So logically it can find a best move without having seen a best move before.

All machine learning is artificial intelligence, but not all artificial intelligence uses machine learning.
>>
>>108881472
stockfish uses machine learning
https://official-stockfish.github.io/docs/nnue-pytorch-wiki/docs/nnue.html
>>
>>108881005
No human could ever beat stockfish either even if it was all super GMs given 1000 years per move
>>
>>108881488
that's a fairly recent addition to SF. it was already insanely good before AI slop was slapped on top of it.
>>
you can have the computer play itself millions or billions of times or even more, with small variations in how it plays. the computer records how it plays when it wins the most often. this is called "synthetic data" and it eventually reaches superhuman level and it is recorded and saved how it plays when it reaches superhuman level
>>
>>108881511
No, but no matter how good Stockfish is, it can always be reliably beaten by a better Stockfish.
>>
>>108881562
Stockfish used its' own evaluation to train. It did not mercilessly steal data from competing engines, like Chessbase did.
>>
>>108881698
True. Chessbase stole stockfish's code, including NNUE, and claimed it to be its own invention (the only proprietary part was the NN weights - not NNUE itself). This resulted in a lawsuit because it violated GPL or whatever FOSS license Stockfish had, and rightfully so.
>>
>>108881783
It was GPL violation. They did reach a settlement with FSF Europe. Chessbase could still use stockfish provided that any modifications they did to it would have to be released in accordance with the GPL.
>>
>>108881472
That means Chess is a shit game.
>>
>>108882740
i mean it is. at higher levels it's a memorization contest, boring as fuck to watch or play
>>
>>108880824
Chess is a complete information game. It can also be perfectly simulated, so you have infinite perfect data for it.
There is no infinite data for text or videos or sound, however, and our best attempts to generate them are diffusions and LLMs. There's the limit.



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.