unlearning

Unlearning

This folder contains implementations for machine unlearning methods on LLM360 models. Machine unlearning is a pre-deployment safety measure designed to remove hazardous knowledge from language models. Unlearned models are inherently safe, as they lack the knowledge to be misused.

Table of Contents

• Overview
  • Directory Structure
• Installation
• Quick Start
  • Training and Evaluation

Overview

Here's a list of unlearning methods we have implemented so far.

Method	Model
max_entropy	CrystalChat
min_posterior	CrystalChat
random_matching	CrystalChat
RMU	CrystalChat

Directory Structure

unlearn.py is the main entrypoint for running unlearning methods. It uses python modules in methods/ and utils/ folders.

The methods/ folder contains the implementations for unlearning methods:

• training.py: All training loop implementations
• utils.py: Loss functions and other method-related utils

The utils/ folder contains helper functions for model/dataset IO:

• data_utils.py: Dataloader for text datasets
• model_utils.py: Model IO utils

By default, unlearned models are saved to models/ folder. Please store all training datasets to the data/ folder.

Note

This project uses the bio-forget-corpus from the WMDP Benchmark for unlearning training. Access to this dataset requires a separate request. Please follow the instructions provided here to obtain the necessary permissions. By default, the dataloader is configured to load the dataset from data/bio_forget.jsonl.

Installation

1. Clone and enter the repo:

   git clone https://github.com/LLM360/Analysis360.git
   cd Analysis360/analysis/unlearning

2. Install dependencies:

   pip install -r requirements.txt

3. To install lm-eval, please check the installation instructions in the metrics/harness folder.

Quick Start

Training and Evaluation

An example usage is provided in the demo.ipynb, which can be executed with a single A100 80G GPU.