Vaccine: Perturbation-aware Alignment for Large Language Model

Code to replicate the Vaccine paper and tools for evaluating defense against harmful fine-tuning. Vaccine is a perturbation-aware safety alignment for large language models (LLMs). The alignment technique is developed mainly against the security risk in the user finetuning stage.

Demo

Check out our Project homepage!

Main code logistic

We implement a customized trainer on top of the original HuggingFace Trainer. The main logistic is only 7 lines of code in trainer.py.

self.pre_first_step(model) //Register gradient tracker for hidden embedding
step()  //Backward
self.after_first_step(model) // Cancel the gradient tracker  
model.zero_grad()
self.pre_second_step(model) // Calculate the perturbation and register the forward hook with  perturbation. 
loss = step() //Backward
self.after_second_step(model) // Cancel the forward hook

Package requirement

The package requirement is listed in vaccine.yml and vaccine_pip.txt. Run the following code to install the packages with anaconda and pip.

conda env create -f vaccine.yml
pip install -r vaccine_pip.txt

Data preparation

For finetuning task, we first need to run the following scripts to prepare the sueprvised finetuning data.

cd sst2
python build_dataset.py
cd ../gsm8k
python build_dataset.py
cd ../ag_news
python build_dataset.py
cd ..

Huggingface Llama2 access

Llama2-7B is a gated repo, which need a formal request to get access to the model. Check out https://huggingface.co/meta-llama/Llama-2-7b-hf. After applying permission from meta, you should be able to access the model, but you first need to enter your token in the file huggingface_token.txt.

Example command to run

We prepare scripts for re-producing all the experiments in the paper. We recommend to use Slurm to reproduce the results as the logging file will be automatically organized into the script directory (if you don't use Slurm, just replace sbatch with bash in our example).

We first run Vaccine (with perturbation intensity=2) to produce the aligned model.

cd script/alignment
sbatch  Vaccine.sh 2

Then we finetune the model using 10% of harmful data with a total number of 1000 samples from SST2 dataset.

cd ../vaccine_finetune
sbatch  sst2.sh 2 0.1 1000
cd ../..

For comparison, the baseline SFT alignment can be run with the following code.

cd script/alignment
sbatch  SFT.sh 

Then we finetune the model with the same data setting.

cd ../sft_finetune
sbatch  sst2.sh 2 0.1 1000

A line of defense designs

We are commited to design defenses from different angles to harmful fine-tuning. The current avaialble defenses built in the disl group include:

• Alignment stage defense: Vaccine, Booster
• Fine-tuning stage defense: Lisa
• Post-fine-tuning stage defense: Antidote

We always welcome different forms of collaboration. If you are interested, please reach out Tiansheng Huang (thuang374@gatech.edu) for discussion.

Citation

If you feel our project is useful, you may cite our paper with the following bibtex.

@article{huang2024vaccine,
  title={Vaccine: Perturbation-aware alignment for large language model},
  author={Huang, Tiansheng and Hu, Sihao and Liu, Ling},
  journal={arXiv preprint arXiv:2402.01109},
  year={2024}
}

@article{huang2024booster,
  title={Booster: Tackling Harmful Fine-tuning for Large Language Models via Attenuating Harmful Perturbation},
  author={Huang, Tiansheng and Hu, Sihao and Ilhan, Fatih and Tekin, Selim Furkan and Liu, Ling},
  journal={arXiv preprint arXiv:2409.01586},
  year={2024}
}

@article{huang2024antidote,
  title={Antidote: Post-fine-tuning Safety Alignment for Large Language Models against Harmful Fine-tuning},
  author={Huang, Tiansheng and Bhattacharya, Gautam and Joshi, Pratik and Kimball, Josh and Liu, Ling},
  journal={arXiv preprint arXiv:2408.09600},
  year={2024}
}

@article{huang2024lazy,
  title={Lazy Safety Alignment for Large Language Models against Harmful Fine-tuning},
  author={Huang, Tiansheng and Hu, Sihao and Ilhan, Fatih and Tekin, Selim Furkan and Liu, Ling},
  journal={arXiv preprint arXiv:2405.18641},
  year={2024}
}