transformer-explainer

An interactive browser tool for learning how GPT-style Transformer models work. Enter text and watch a live GPT-2 model visualize how its components predict the next tokens.

Browse open issues

Pick a small issue you can reproduce, then fix it with your AI coding agent.

Share on XLicense: MIT

Overview

Transformer Explainer is an interactive browser-based learning tool for understanding Transformer models such as GPT. It runs a live GPT-2 model in the browser, so learners can enter their own text and see how the model predicts the next tokens. Visualizations show how internal components and operations work together during prediction. Use the live demo to explore, or run the project locally with Node.js and npm to open it in a browser.

Key features

  • Runs a live GPT-2 model in the browser
  • Accepts text entered by the learner
  • Visualizes model components and operations involved in next-token prediction

Best for

Learners who want to explore how GPT-style language models make predictions. It is useful for interactive study of Transformer internals using sample or self-written text.

Upstream
poloclub/transformer-explainer
Fork on GitHub
Guo-astro/transformer-explainer
Upstream stars
9.1k
Category
Learning, interviews and curated lists
License
MIT
Forked
2026-10-10
Sync status
In syncLast synced 2026-11-08