The math. trick in self-attention (v2 and v3)

The mathematical trick in self-attention (v1)

Install `nx` from github to use latest unreleased changes

Fix generate text function again 🙈

chore: use a numerical function to generate a batch

chore: stick to original implementation when generating the text

Fix text generation function

Add simple README

Update text generation to use multinomial distribution

Train the bigram model

Text generation with the Bigram model

Basica Bigram Language Model implementation

Let's build GPT: from scratch, in code, spelled out, but in Elixir