first scape there is gorgeous, I have some of the same plants in my shrimp tank!

I like the posts, it's just that if I'm going to read them I click on one and scroll to the top. Thought I'd say something instead of just unfollowing, but I'll stop reading if it's that or spam

would you mind only reposting the parent tweet of a given thread? twenty out-of-order posts in a row is a bit obstrusive

saw a million bakeries when I visited there, such a strange thing to say

I did this and I was wiped the next day but I think it's fine

can you tack into the wind like on a boat?

This is excellent advice. It solved my computer problems instantly

that bee needs to finish the job

People have a knee-jerk response to religion but it's such a huge part of our intellectual tradition! You lose so much context without engaging with it. Cool paper, thanks for posting

deadlifts are lifesaving if you work at a desk

that's what's up

I agree with you re: bailouts but I do think there's some potential risk of contagion with stuff like data center securitization. I don't know how large of a market data center mortgage backed securities are, but they're definitely on some institutional balance sheets

shrimp proof

Right, and I believe that "AI animated movie" is being done with a lot of human animators as well! I expect to see a lot more stuff like this, especially as the Open AI IPO gets closer

You can imagine the hysterical response on places like LinkedIn.

it was marketed as is the LLM had created the hypothesis de novo. If you read the paper you'll see that the actual experimental design was done by humans and the model used high quality perturbation data to score a list of small molecules based on their predicted effect.

www.biorxiv.org/content/10.1...
I read this paper recently because it was being touted as an LLM generating a hypothesis and unsurprisingly found that to be untrue. The experiment was using ML to predict responses to drugs and they got a candidate that validated experimentally, but (cont.)