And the pseudocode algorithm for quick reference. 15/n

But this is the gist of the magic. And it results in reported 38x convergence speedup compared to MCTS & impressive benchmark gains. 14/n

Quick primer for non-wizards about the post-MCTS LLM reasoning future:

How will LLMs learn to reason efficiently?

No math in this thread, ~simple words only! Let's go through the "Process Reinforcement through IMplicit REwards" (PRIME) method. 1/n

curvy-check-498.notion.site/Process-Rein...

Building your institutional knowledge (painfully gained from rolling the product out for 1000s of customers, with all the real-world painful mess and nitty-gritty) is an endurance sport, but also something ChatGPT will not replace in a second. 8/n

Can we build something lasting in a world where technology is easily reproducible?

How much time do we have before Artificial General Intelligence (AGI) is here, and how much time do we have afterwards? 2/n

The product strategy of most SaaS isn't ready for AGI.

A disconnect is widening between how product builders think, and how Artificial Intelligence researchers project short-term future.

But I believe it's a solvable problem (in any world where a solution exists). 1/n🧵

Lowkey the most emotional moment in TV history?

Damn, rewatching after 10y hits as hard as ever.

I now realize Haibane Renmei is not just my most favorite show ever, but also the core of my moral philosophy.

What can RBMK teach us about building AGI?

THEY HAVE PLAYED US FOR ABSOLUTE FOOLS

Claude knows how to use vim! And... you can do anything from vim.

Embedding Claude in vim means you get a fully agentic AI pair programmer in <1000 LoC. (Or a generic chatbot, or Linux co-admin.)

Unlike CLIs, you can review Claude's changes, and keep the chat history.

github.com/pasky/claude...