Blog
2026-10-11
The Math Behind AIAn interactive introduction to the math behind modern AI: vectors, dot products, matrices, derivatives, the chain rule, and probability, each shown as numbers, geometry, and code.
2026-09-27
Steering the Latent SpaceAn interactive explanation of how to control generative models through their latent spaces: concept directions, interpolation on the Gaussian shell, disentangled edits, diffusion guidance, activation steering in language models, and learning directions from human preferences, each shown as geometry and as code.
2026-09-13
Inside the TransformerAn interactive explanation of attention and the Transformer, from a single dot product to a small model in PyTorch.
2026-08-30
Inside the AutoencoderAn interactive explanation of autoencoders, variational autoencoders, and sparse autoencoders, from a single weighted sum to learned features.
2026-08-16
Inside the Graph Neural NetworkAn interactive explanation of graph neural networks, from a single averaged neighborhood to GCNs, graph attention, over-smoothing, and the limits of message passing, each shown as a graph, as geometry, and as code.