Skip to content
Chaoran Huang

Docs home

Long-form learning notes on NLP, machine learning, and adjacent topics.

Welcome. These are long-form notes written in MDX. Use the sidebar to move between pages, or start with the series below.

Series

Part 1 — From One-Hot to BERT

How NLP learned to represent meaning — from one-hot vectors to contextual embeddings. (Full post.)

Part 2 — From Representation to Behavior

The roadmap from a pre-trained transformer to an aligned, reasoning assistant. (Outline; deep-dives in progress.)

Standalone posts

Bits, Math, Hashes, Streams

A systems refresher covering bit operations, modular arithmetic, hashing, and streams — written after an AI assistant generated a test with a magic number I could no longer explain.

External references

Next.js docs

Fumadocs

From One-Hot to BERT: How NLP Learned to Represent Meaning

Part 1 of a 2-part series. A historical, geometry-first walk through NLP representations: from one-hot vectors to contextual embeddings — the ideas that made LLMs possible. Each section is a hub for a future deep-dive page.

On this page

SeriesStandalone postsExternal references