Skip to content

Latest commit

 

History

History
19 lines (10 loc) · 900 Bytes

README.md

File metadata and controls

19 lines (10 loc) · 900 Bytes
Raw

Mechanistic Interpretability of Binary and Ternary Networks

This repository contains the source code used to recreate the figures for the paper "Mechanistic Interpretability of Binary and Ternary Networks".

Description

The code in this repository was used to generate the figures in the paper "Mechanistic Interpretability of Binary and Ternary Networks". The paper discusses the interpretability of binary and ternary networks from a mechanistic perspective.

Instructions

To recreate the figures, follow these steps:

  1. Clone this repository to your local machine.
  2. Navigate to the root directory of the repository.
  3. There is one notebook that generates figures for the binary transformer and another one for the ternary transformer.

Please note that you may need to install certain Python packages to run the scripts. These packages are listed in the requirements.txt file.