Learn retweeted
The dark side of reinforcement learning
@olive_jy_song, senior researcher at @MiniMax_AI, about RL models that try to hack rewards and why alignment fails in practice
This conversation is an inside look at how Chinese AI labs move fast – testing new models overnight, debugging unpredictable behaviors, and operating with relentless first-principles discipline.
Full interview in the comments
#TP_interview
Learn retweeted
It's often said "we need a theory of deep learning". No one knows what that really means, except there's a sense of hopelessness. We actually have more of an understanding than many realize, and we can analyze modern neural nets. Make an effort to leave the echo chamber!
Learn retweeted
One thing I have noticed when solving math problems is my brain moves faster than my hands. In the solution steps, I make silly errors like changing a + sign to a - sign.
Learn retweeted
If all you think about accessibility is screen readers. I'm here to remind you how wrong you are. Accessibility goes beyond screen readers. Screen readers make up a limited percentage of a user's need
Learn retweeted
Web accessibility means people with disabilities can perform all the actions you can do on the web.
Learn retweeted
Mathematics is the language of the universe, and first order logic is the language of mathematics. Therefore, first order logic is the metalanguage of the universe
Learn retweeted
I’m gonna give 10 random people that repost this and follow me $25,000 for fun (the $250,000 my X video made)
I’ll pick the winners in 72 hours




