Claude 3 models are 🐐

The talk demonstrates Claude 3’s 1‑million‑token context, its instruction-following ability, and how a swarm of such models can tackle complex workflow problems through extensive reasoning.

Overview

Claude 3 models feature approx 1M context window with really good instruction following.
In our exploration of these models, we can see early signs of using a swarm of these models to solve a bigger workflow problem and how well they are able to reason on 1M context window.