Description
colah's blog is the personal site of Christopher Olah, who works on reverse engineering artificial neural networks into human understandable algorithms and is a co-founder of Anthropic and of the journal Distill. The index groups his essays by theme: neural networks in general, recurrent networks, convolutional networks and visualizing neural networks, with pieces such as Understanding LSTM Networks, Calculus on Computational Graphs: Backpropagation, Understanding Convolutions and Neural Networks, Manifolds, and Topology. It also links his later interpretability work, including the Circuits thread on Distill and Transformer Circuits. A strong starting point for developers who want intuition for how networks compute, not just how to train them.
