My favorite part of the bob dylan movie is how faithfully it portrays how easy it is to be a rock star.

Now that the election is over, all our work is still ahead of us. I'll be called naive. I'll be called foolish. But these guidelines are the best ways I know to do that work. news.fairforall.org/p/what-now-w...

Teach your llms to pass by reference. Huge time/cost/accuracy improvements

@jeremie.com inspired me to give o1-preview a crack at an old project, and it’s killing it. Made me realize that perhaps the best evals for new models is the boneyard of stuff you tried and failed to acvomplish over the last 6 months. You need to be failmaxxing to stay on top!

Story: I’ve been doing a lot of work trying to use LLMs to generate large codebases (well outside context length) and while I don’t have a silver bullet, I’ve developed a bit of a pattern language for how to approach different commonly recurring problems 1/n

There’s a trope in sci-fi of the utopian alien culture that buys its flourishing at the price of some horrifically draconian law. If you were writing such a story, what is the draconian law?