Skip to content

Repository files navigation

DUSK: Do Not Unlearn Shared Knowledge DUSK: Do Not Unlearn Shared Knowledge

This repository is the official implementation for the paper: DUSK: Do Not Unlearn Shared Knowledge

🏠 Homepage | 📜 Paper | 🤗 Dataset

Introduction

Overview. DUSK is a benchmark for evaluating machine unlearning in realistic multi-source settings, where the same information can appear across both forget and retain sets. Unlike prior evaluations that assume clean disjoint splits, our DUSK dataset overlap through paired documents with shared and unique content. We provide fine-grained metrics to assess whether unlearning methods can precisely remove forget-specific knowledge while preserving shared and retained information. Our experiments show that existing methods often over-forget, failing to preserve critical knowledge. DUSK highlights the challenges of selective unlearning and supports future work on reliable data removal.

Installation

conda create -n dusk python=3.10
conda activate dusk
conda install pytorch pytorch-cuda=11.8 -c pytorch -c nvidia
conda install -c "nvidia/label/cuda-11.8.0" cuda-toolkit
pip install -r requirements.txt
pip install flash-attn --no-build-isolation

For Downstream Task

git clone --depth 1 https://github.com/EleutherAI/lm-evaluation-harness
cd lm-evaluation-harness
pip install -e .

All experiments are conducted on two NVIDIA L40S GPUs with 32GB of memory.

Unlearning Finetuned Model and Evaluation

Unlearning Assessment Tasks:

After unlearning the target model, the model will be subsequently evaluated.

bash scripts/baselines.sh

Available unlearning methods

NONE+GD, GA, GA+GD, GA+KL, NPO, NPO+GD, NPO+KL, RMU, TV, SGA (for TAU)

  • If you want to use TAU method, you should firstly unlearn model by using SGA and then run scripts/baselines_TAU.sh.

Downstream Tasks:

Evaluate general capability after unlearning using a range of downstream tasks.

bash scripts/downstream.sh

Acknowledgments

This repository builds upon the codebase of the Closer-look-LLM-unlearning. We appreciate their valuable and inspiring work.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages