AI & LLMsPart 1 · Diving into LLMs
Diving into LLMs, part 1: how LLMs work, explained simply
ChatGPT felt like magic, so I started building a tiny language model myself, following Andrej Karpathy's Zero to Hero course. How an LLM learns and writes, in plain words and animations: next-word prediction, gradients, training, attention and the full pipeline.
8 min read