Big 🍎

2⁵

AGI’s Achilles heel.

Ripples of attention in liquid latent space

New paper: What happens when an LLM reasons? We created methods to interpret reasoning steps & their connections: resampling CoT, attention analysis, & suppressing attention We discover thought anchors: key steps shaping everything else. Check our tool & unpack CoT yourself 🧵

Token: the smallest representational unit of meaning that is composable with itself (self-referential).

This flow is so agentic that even my agents have agents

See

New programing language: ─ │ ┌ ┐ └ ┘ ├ ┤ ┬ ┴ ┼ █

Good afternoon

Concept: Learning hands

Edge of chaos

Here?

In Nate intelligence is not probabilistic, if anything, it could be quite the opposite: it is the ability to defeat the odds, that is pretty much what every living being does every second of its existence. Note that this fact does not contradicts that, when we look backwards, we

Good evening

Liquid attention