Next-token prediction in JavaScript — build fast language and diffusion models.

The course notes about Stanford CS224n Natural Language Processing with Deep Learning Winter 2019 (using PyTorch)

Dissecting Transformers via attention visualization

N-grams to build an autocomplet

An early version of a system that credits creators based on the similarity of their content to an LLM response. Giving back to creators is the only way for fair, sustainable AI economies that lead to true growth.

ELLE - Estonian language learning and analysis environment for learners, educators and linguists