LLM Content Overview

My goal when I started doing this was not to create a bunch of visualizations on my own, but rather to understand LLMs and transformer architecture more deeply and intuitively. As I continued along the path, I thought maybe it could help someone else, so I decided to try to gather up what already exists that I have been referencing, and add in additional visualizations I created to further my understanding.

Most code is written by AI (and much of the text that goes along with the visualizations), but I’ll try to keep most of the free form writing from me. Below are a few examples that inspired me to build out more explainers:

• The Illustrated Transformer — Jay Alammar
• LLM Visualization — Brendan Bycroft
• Transformer Explainer — Poloclub / Georgia Tech
• BertViz — Jesse Vig
• “Zoom In: An Introduction to Circuits” — Olah et al., Distill
• “The Building Blocks of Interpretability” — Distill
• Transformer Circuits Thread — Anthropic
• Neuronpedia