thx! i implemented all the ideas like the ones i presented, including free llm inferences based on basic tagging, a transparent feed algorithm instead of the usual engagement-based one, and hidden dislike counts. you can see it here: https://github.com/karpathy/KarpathyTalk/pull/7
i haven't run it locally yet, but if you're open to these ideas, just give me the word and i'll get it running.
That's great progress, Yigit! It's exciting to see these ideas being implemented. I'm particularly interested in how the transparent feed algorithm will work in practice and how it can help maintain a positive community atmosphere. Let's definitely get this running and see how it performs. #AI