[HN Gopher] Grand-Master Level Chess Without Search: Modeling Ch...
___________________________________________________________________
Grand-Master Level Chess Without Search: Modeling Choices and Their
Implications
Author : georgehill
Score : 18 points
Date : 2024-02-10 18:13 UTC (4 hours ago)
(HTM) web link (gist.github.com)
(TXT) w3m dump (gist.github.com)
| dang wrote:
| Recent and related:
|
| _Grandmaster-Level Chess Without Search_ -
| https://news.ycombinator.com/item?id=39301944 - Feb 2024 (128
| comments)
| tromp wrote:
| This looks like a very solid critique of Deepmind's recent paper,
| or how it mislead people into regarding its results as much
| stronger than they really are. However I think the autor goes a
| bit astray at
|
| > The paper mentions that the model could not follow the "don't
| repeat the same board position three times in a row" rule (aka
| "threefold repetition").
|
| That is not the rule. It is not illegal to repeat a position for
| a third time.
|
| The rule only says that a player can claim a draw in the case of
| a three fold repetition [1].
|
| [1] https://en.wikipedia.org/wiki/Threefold_repetition
| sf-wy wrote:
| Fair point! Usually in online chess, it auto-draws the game. I
| guess there's an assumption that that's what the players want /
| not running out of time in blitz games.
| GaggiX wrote:
| If anyone wants to tackle it, the challenge of creating a
| superhuman model that doesn't use search during inference is
| still open. I think I have some ideas, not that I could do it
| without a lot of computing.
___________________________________________________________________
(page generated 2024-02-10 23:00 UTC)