Read news on ARC-AGI-1 with our app.
Read more in the app
I trained a small transformer in 1.5hrs and it beats many LLMs
A tiny recursive reasoning model achieves 45% on ARC-AGI-1 and 8% on ARC-AGI-2
