Another day looking at lots of data.

RT @willccbb: huge day in terms of general optimism vibes.

Join millions who have switched to Grok.

RT @tenderizzation:

When you run into a dependency issue working in a open-source library and wonder how it was solved by other maintainers and see them sneaking in a unittest.skip lol.

This will be a fun weekend.

I'm only friends with people who can type "Qwen3/Qwen3-4B" in under 3 seconds.

My after-work routine @PrimeIntellect

Claude Code stuffing the hole that is the TRL documentation.

Self-learning will be Q-learning with a smart solution for long horizon biases.

curl -LsSf | sh.

AI ecosystem still feels way too chaotic.

It’s crazy that people choose to be outraged about neutral events that have literally no effect on their lives. Imagine redirecting that energy toward something that actually matters to you.

Claude really has those 10.000 hours under his belt.

Kinda like my code tbh.

Even though so much is happening, it still feels like a loop somehow.

What we thought would happen: AGI, post-work society, self-driving everything…. What actually happened: „You can save 2 cents when you speed up your inputs to Whisper 🤯“.

Why is the music AI field so underwhelming?.

RT @seohong_park: Q-learning is not yet scalable. I wrote a blog post about my thoughts on scalable RL algorithms.….

Tech bros saying „objectively“ are anything but objective.

Nothing makes me rethink learning more than Claude spitting out these better-than-textbook artifacts when I ask it to explain something deeply.