Hacker News Reader

Comment by anton-107 | Hacker News Reader

Would be interesting to see AlphaZero in this contest, but I don't think it is open source

[1] How AlphaZero Completely CRUSHED Stockfish https://www.youtube.com/watch?v=8dT6CR9_6l4

Leela Chess Zero is the open source successor of Alpha Zero, and has been the second strongest engine (except a few times where it bested stockfish) for the last few years.

[1] https://lczero.org/

That was against an old version of Stockfish. Both Stockfish and Leela are much stronger now than AlphaZero.

...because they adopted neural network enhancements inspired by AlphaZero, or so I hear.

That’s correct, but Stockfish uses neutral nets differently than AlphaZero.

And the reason for that is because they haven't been actively developing AlphaZero anymore. It's not worth it to them to spend their limited resources doing so when others in the Chess community are perfectly willing to. DeepMind has meanwhile instead been spending their efforts on solving protein folding, which I think we can all agree is a better use of their limited resources.

DeepMind has limited resources? I'll remember that and smile next time I'm running my experiments on the same machine as another two PhD students.

The limited resources I'm referring to is people. It's a relatively small organization, and focusing people on Chess comes at an opportunity cost of not focusing on something else.

Just because your situation is even more limited doesn't mean that they don't also have limitations. That's just gatekeeping.

Sorry, I don't understand. What is gatekeeping?

I appreciate that nobody has unlimited resoureces but, for example, the latest DeepMind paper posted on HN (on AlphaCode) had a couple dozen authors. By comparison I do all my work myself, with the help of my advisor and so do his other PhD students. And all for a PhD student's stipend, when DeepMind researchers are paid at least at post-doc levels, presumably. There's no comparison in the resources that I and DeepMind can throw at a problem.

I mean, please, give me DeepMind's limitations. I'd be so happy!

It is. Leela Chess 0 (LC0) won in seasons 15 and 17 in 2019 and 2020.

LC0 is the open source implementation of Alpha Zero’s Monte Carlo tree search + neural nets for static evaluation approach.

Perhaps someone submitted LeelaZero at some point?

Yeah Leela even one once in 2019. Since stockfish moved to a hybrid approach (conventional evaluation + neural network) it has been unstoppable though.

Reply on news.ycombinator.com