An X-ray machine for large language models: TinyGPT Visualizer opens the black box
Tools for exploring the souls of large language models
TinyGPT Visualizer is an open-source tool that lets you peer inside a real, small-scale language model as it generates text. Built to make mechanistic interpretability hands-on, it shows the numeric matrices behind each thought—no surgery required. The project aims to help more people explore how LLMs work, from attention patterns to emergent behaviors, and even ask where consciousness might hide in the numbers.
Unlike the brains of humans, we can stop and look at the numbers as the thoughts are formed without killing the patient, and, unlike human brains, we can search those numbers for patterns and algorithms which explain the results these models emit as well-formed prose, code, music, and videos.