Everything, Everywhere, All at Once: Is Mechanistic Interpretability Identifiable?

Maxime Méloux1, Silviu Maniu1, François Portet1, Maxime Peyrard1

1 Université Grenoble Alpes, CNRS, Grenoble INP, LIG, 38000 Grenoble, France

Article thumbnail
Article and poster ArXiv Preprint thumbnail
ArXiv
Pre-print GitHub Repo thumbnail
GitHub
Repo Demo Notebook XOR thumbnail
Demo Notebook:
XOR Demo Notebook MNIST thumbnail
Demo Notebook:
MNIST