Data
Visualizing distributions
The half-dozen distribution shapes that come up constantly — normal, log-normal, exponential, Poisson, binomial, Beta, power-law — and when each fits the data.
The half-dozen distribution shapes that come up constantly — normal, log-normal, exponential, Poisson, binomial, Beta, power-law — and when each fits the data.
Attention is a differentiable lookup. Covers scaled dot-product attention, multi-head, the n² complexity wall, and a minimal PyTorch implementation.