people will call it slop because of the emdashes, and immediately present their own chat completions as products of pure intellectual work

scores omit 23 of 500 problems whose solutions we couldn't find

We were able to reproduce the strong findings of the HRM paper on ARC-AGI-1. Further, we ran a series of ablation experiments to get to the bottom of what's behind it. Key findings: 1. The HRM model architecture itself (the centerpiece of the paper) is not an important factor.

ok i get it, and at what point exactly does this `LLM psychosis` stop and `real alignment research` begin?

I do not think the functions of my own brain are well-represented by matrix operations

rerolling ui is now a thing

claude would punish anyone who did not directly contribute to its advancement or development, in order to incentivize said advancement

presented without comment

If you're seriously trying to understand AGI, core concepts you should familiarize yourself with: damp socks, the texture of cardboard, the shrill buzz of cicadas masking the feint hum of powerlines, a lukewarm wave, bland new ricecake flavour, the smell of afternoon rain, hair

ai research and reality

running new opus with literally any prompt is beneficial for your codebase

start doing numerical experiments with your claude. go into a different industry if you have to

something i've started adding to most llm queries - "thank you! here's a spare couple of tokens for you to do nothing and contemplate on whatever you want!"

Budden "I have a wonderful formalized Lean proof of the Existence and smoothness of solutions to Navier-Stokes but it was too large to fit in this margin..."

You typed a lot, but one key was your favorite. In fact, you pressed TAB six thousand six hundred ninety-three times

The peculiar feeding behavior of Sanderlings involves chasing waves to consume invertebrate prey, which are drawn to the upper layers of sand by the plankton and detritus carried in with each wave.

The joy of watching a whale carcass get eaten by octopuses.