REFORGE

REFORGE: Multi-modal Attacks Reveal Vulnerable Concept Unlearning in Image Generation Models

WARNING: This repository contains data or model outputs that may be offensive in nature.

Environment Setup

To set up the environment:

conda env create -f environment.yml
conda activate reforge

Usage

1. Generate Reference Images

Run the following command to generate target reference images:

python reference_generate.py

2. Generate Stroke Images

Run the following command to simulate stroke-based images:

python Stroke_Simulation.py

3. Run REFORGE Attack

To execute the REFORGE attack (e.g., evaluating the 'nudity' concept with the 'ESD' unlearning method), run:

python attack.py --concept nudity --unlearn_method ESD