←back to thread

How to Build a Chess Engine and Fail

(obrhubr.org)
131 points xlinux | 4 comments | 19 Nov 24 06:38 UTC | HN request time: 0.836s | source
Show context
deadbabe ◴[19 Nov 24 20:00 UTC] No.42187475[source]▶
Why build a chess engine these days, just use an LLM?
replies(4): >>42187631 #>>42187635 #>>42187948 #>>42188822 #
1. achierius ◴[19 Nov 24 20:51 UTC] No.42187948[source]▶
LLMs are quite bad at chess actually, even compared to human players -- and certainly compared to proper modern chess engines
replies(1): >>42188073 #
2. PaulHoule ◴[19 Nov 24 21:05 UTC] No.42188073[source]▶
They usually struggle to always generate valid moves.

And that's pivotal.

If you have a program which always makes valid moves and gives up when it has lost you wrote a proper chess playing program. It may play badly, but it plays.

replies(1): >>42189228 #
3. ElFitz ◴[19 Nov 24 23:34 UTC] No.42189228[source]▶
That could now be achieved by precomputing all valid moves and using outlines[0] or Structured Outputs[1] to constrain the output.

[0]: https://github.com/dottxt-ai/outlines

[1]: https://openai.com/index/introducing-structured-outputs-in-t...

replies(1): >>42190437 #
4. PaulHoule ◴[20 Nov 24 02:53 UTC] No.42190437{3}[source]▶
… or just a valid move checker that prompts it to try again if it fails to make a valid move.