Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Grandmaster-level chess without search

github.com

Post reply on HN
Print view View original

141–150 of 171 posts

Re: Grandmaster-level chess without search

#141

Post by mewpmewp2 »

bob1029 wrote: post #136

https://arxiv.org/abs/2402.04494 > Board states s are encoded as FEN strings which we convert to fixed-length strings of 77 characters where the ASCII-code of each character is one token. A FEN string is a description of all pieces on the board, whose turn it is, the castling availability for both players, a potential en passant target, a half-move clock and a full-move counter. We essentially take any variable-lengt…

I don't know much about this space, but it seems like this could be solved by leaving a good amount of empty tokens that you would only start using when they arise. Or leave tokens which you can use together to combine anything for various edge cases. Because if you have all the characters as tokens you can combine them into anything.

Re: Grandmaster-level chess without search

#142

Post by mewpmewp2 »

sksxihve wrote: post #139

Isn't generating the training data by running stockfish on all the board positions for all the games just encoding the search tree into the transformer model? So increasing the number of parameters to the model would allow it to encode more of the search tree and give better performance, which doesn't seem all that interesting.

How could it be possible to encode a search tree like this though.

Re: Grandmaster-level chess without search

#143

Post by mewpmewp2 »

jvanderbot wrote: post #116
vanjajaja1 wrote: post #115

what i like about this is that it implies you can build heuristics good enough to make it to GM level. this is great because i find calculating moves a headache

The thing is the heuristic done by a huge network might be insanely complex and doing all kinds of calculations. it's just that it's one function call so we ignore all those calculations. It's not immediately obvious that deploying a transformer to solve for next best move means that a human mind can avoid difficult calculations and just play by gut. There's just too much wordplay going on with "heuristic"

Arguably intuition and gut feeling are also insanely complex systems doing all kinds of calculations.

Re: Grandmaster-level chess without search

#144

Post by osti »

stabbles wrote: post #100
osti wrote: post #31

https://lczero.org/blog/2024/02/how-well-do-lc0-networks-com... The best neural network chess engine's authors wrote about this deepminds publication.

LC0 hasn't been the best neural network chess engine since Stockfish added NNUE in 2020.

True, but since stockfish uses a way way smaller network, I still prefer to think of stockfish as the traditional engine.

Re: Grandmaster-level chess without search

#145

Post by adgjlsfhk1 »

7373737373 wrote: post #123
timerol wrote: post #117

Earlier quoted context omitted.

Note that the exact version of this technique is used in chess for the endgame, referred to as a tablebase. Chess is solved once there are 7 pieces on the board, in an 18.4TB database, described here: https://lichess.org/@/lichess/blog/7-piece-syzygy-tablebases...

Makes me wonder what % of games end with <=7 pieces

at a high level, almost all of them that aren't draws. if you have pieces in the right place, you should've be checkmateable. Without blunders, wins occur after a small advantage is followed by slightly favorable trades into a winning position

Re: Grandmaster-level chess without search

#146

Post by espadrine »

tzs wrote: post #20

OT: what's the state of the art in non-GM level computer chess? Say I want to play chess with an opponent that is at about the same skill level as me, or perhaps I want to play with an opponent about 100 rating points above me for training. Most engines let you dumb them down by cutting search depth, but that usually doesn't work well. Sure, you end up beating them about half the time if you cut the search down enoug…

KataGo has a special model weights release with human-like play at various Elo: https://github.com/lightvector/KataGo/blob/master/docs/Analy...

You can see in the release notes a few screenshot examples where a particular move changes likelihood as you get to higher-level play: https://github.com/lightvector/KataGo/releases/tag/v1.15.0

Re: Grandmaster-level chess without search

#147

Post by Out_of_Characte »

tzs wrote: post #122
gcr wrote: post #114

Earlier quoted context omitted.

Stockfish is the classic answer, though I’m not sure how well it’s graded. Someone must have a “Stockfish strength”-to-ELO mapping.

It's not getting an engine to play in the right rating range that is the problem. It's getting it to play like a human would play in that rating range. The average rating of tournament chess players in the US is around USCF 1550. I'm not sure what their FIDE rating would be. FIDE ratings are usually 50-100 points lower than USCF ratings but that's based on comparing people that have both ratings which for the most pa…

What I find fasinating is how bad human beings are at chess. Now that we have engines we're finally able to analise every game ever played and they show us everything in chess that we're blind to. Their ability to never blunder in 1 or 3 moves is admirable, and better than most players, and to say nothing about their ability to make you play out the longest possible chain before checkmate. What I found most insulting is when I played against the best bot I could beat, it gave up its rook for free.

Re: Grandmaster-level chess without search

#148

Post by jxy »

chvid wrote: post #78
chvid wrote: post #77

But the gigantic synthetic dataset that is used for training is created with plenty of traditional search. So it is all a bit silly but I guess cool none the less ...

If anything it demonstrates the limits of NN. A human brain can learn based on far fewer examples.

Nature's evolution algorithm took millions of years to find the architecture and the base model, which then takes decades to be fine tuned to be able to form this opinion.

Re: Grandmaster-level chess without search

#149

Post by tech_ken »

chvid wrote: post #106
amunozo wrote: post #90

Earlier quoted context omitted.

It's a knowledge distillation. You can then use this smaller, more efficient models instead of the larger one.

Or maybe it is just memorizing a very large number of games.

Seems more like a 'compression' of the large number of games, or even like an approximate 'index' of the database

Re: Grandmaster-level chess without search

#150

Post by timmg »

mewpmewp2 wrote: post #142
sksxihve wrote: post #139

Isn't generating the training data by running stockfish on all the board positions for all the games just encoding the search tree into the transformer model? So increasing the number of parameters to the model would allow it to encode more of the search tree and give better performance, which doesn't seem all that interesting.

How could it be possible to encode a search tree like this though.

Imagine you collected a billion unique, feasible board positions (all positions is intractable, but most possible positions are impractical) and the best nest move for each. That "best next move" is the result of a tree search.

Now use a transformer to "compress" that information into its model. It sounds like that is approximately what is going on here. Certainly, the model is likely to generalize some aspects of the data (just like LLMs do). But for the most part, the model encodes the information from the Stockfish evaluation.

(This is just my guess of what we are seeing.)

Post reply on HN