Transformer Explainer

Use it when

• You are teaching or learning how GPT-style transformers turn text into next-token predictions and static architecture diagrams are not clicking
• You want to probe attention and prediction behavior on your own example text without setting up a Python interpretability stack

What it solves

Not the fit when

• Inspecting models other than the bundled in-browser GPT-2
• Research-grade interpretability experiments or activation patching
• Serving or benchmarking LLMs
• production model serving
• interpretability analysis of your own custom model checkpoints
• training or fine-tuning models

Install

git clone https://github.com/poloclub/transformer-explainer.git && cd transformer-explainer && npm install && npm run dev (requires Node.js v20+ and NPM v10+), or use the hosted demo with no install

Invoke

Open http://localhost:5173 (or https://poloclub.github.io/transformer-explainer), enter your own text, and observe in real time how the in-browser GPT-2 model's internal components produce next-token predictions

Alternatives

No reviewed alternatives recorded yet.