overfitting.io
Optimizers
How Optimizers Work
From vanilla gradient descent to Adam — the intuition behind SGD, Momentum, RMSProp, and AdamW, with interactive demos at every step.
Learning
Interactive deep dives into ML concepts. Read the intuition, play with the demos.
Foundations
How Neural Networks Bend Space
A neural network warps the space your data lives in until the answer becomes obvious. The geometric intuition, with interactive demos.
Foundations
Activation Functions
Without nonlinearity, depth is useless. From sigmoid to GELU, how each activation shapes learning and why the field keeps inventing new ones.
Backpropagation
How Backpropagation Works
Walk through the chain rule step by step, from a simple computation graph to a full neural network pass.
Attention
The Attention Mechanism
How queries, keys and values interact — and why scaling by √d matters more than it sounds.