e
math
What the incentives and game theory suggest about how the AI race will play out
Why repeating any sequence of moves on a Rubik's cube always returns it to its starting state
Innovation comes from combining breadth and depth of knowledge, not from genius in a vacuum
A linear map can approximate any function if you encode and decode in high enough dimensions
The intuition behind L'Hôpital's rule
Constrained optimization with Lagrange multipliers
Computing gradients through a neural network
K-means minimizes a function by alternating two local improvements
At its core, machine learning is just finding the minimum of a function
Exponentially many nearly orthogonal vectors fit in n dimensions
A simple operation with many connections
Why comparison sorts are limited to O(n log n): an information theory view
“Building a tiny x86-64 ELF that prints “Hello, World!””