wrhn

Recent submissions

Comment · 2 years ago
I tried writing an AlphaZero clone to play Chess on my home PC (I only had an RTX 3070) and I failed for essentially the same reason as they mentioned: iteration time was too slow and you couldn’t tell if the model was getting any better at all after weeks of training.

I thought I’d work on it further and maybe write some blog or put some dev vlogs on YouTube but never got around to doing it. Might do it some day.

Till then I’ll just post some links to my Github if anyone wants to check out what I was doing:

1. https://github.com/krkartikay/AlphaZero-proto

2. https://github.com/krkartikay/AlphaZeroFinal

3. https://github.com/krkartikay/mcts-chess

Comment · 3 years ago
Pessimistic take: after the two stages you mentioned, in the third stage they will carry on BSing as usual (only more efficiently with LLMs now) and there will be no one to point out that the emperor is naked.
Comment · 3 years ago
That is not the first rule in the book. Although yes Jordan Peterson might say something like that.

Just to get the facts clear: the first rule in his book states “Stand straight with your shoulders back.” which he discusses both in its literal (biological) meaning and the metaphorical meaning.

Comment · 3 years ago
Video summary: https://www.youtube.com/watch?v=BrjAt-wvEXI

I wonder how long it is until someone finally figures out how to combine MCTS + RL (like AlphaZero) with LLMs and it's game over for humans. I truly think that has a pretty real chance of creating AGI.

Comment · 3 years ago
I had seriously thought the title was a reference to GPT 4.

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton