AI
How a diffusion model draws pictures out of noise, and how a network memorises the answers before it learns the rule.
3 plates
Plates 3
- Plate 43 More Features Than Neurons AIGoing further A single neuron in a neural network often stands for several unrelated things. A model with just two neurons can arrange five features into a pentagon and keep them all. Open →
- Plate 44 Pictures Out of Noise AIGoing further Many AI image generators start from random static and remove it, one step at a time. A small network does the same in your browser, and a cloud of dots slowly condenses into a knot. Open →
- Plate 46 Learning by Heart, Then Learning the Rule AIGoing further A small neural network learns to add hours on a 24-hour clock. It gets its practice questions right almost at once, by memorising them, and only long afterwards answers questions it has never seen. Inside, it has built clocks. Open →