Learn
Learn how an AI works, from the AI itself
Big AI Brain explains its own insides, with moving diagrams and the real code. Start with the three lessons below, then ask the brain anything about how it works. If you've never written a line of code, start with "Teach me to code" in the chat.
How I grow new neurons (function-preserving widening)How Big AI Brain adds neurons without forgetting anything: function-preserving widening (Net2Net), explained simply with a diagram and the real code.How I think: predicting the next characterThe one tiny job behind Big AI Brain's growing neural network: look at the last 32 characters and guess the next one. Explained with a live diagram and code.The report card: how we know it's really getting smarterBig AI Brain grades itself every night on a test it never studies. Learn what bits per character means, why lower is smarter, and why it can't cheat by memorizing.
Ask the brain directly
The brain answers these in the chat, with a picture:
Why we teach this
If the world is going to raise an AI, the world should understand it. Every part of Big AI Brain is written in plain Python with one math library and no magic. If you can read a recipe, you can read this brain. More lessons arrive as the brain grows.