README.md

Gradient-based Adversarial Attacks against Text Transformers

0. Install Dependencies and Data Download:

1. Install HuggingFace dependences

conda install -c huggingface transformers
pip install datasets

2. (Optional) For attacks against DBPedia14, download from Kaggle and setup data directory to contain:

<data_dir>/dbpedia_csv/
       train.csv
       test.csv

1. Training finetuned models

Use the following training script to finetune a pre-trained transformer model from HuggingFace:

python text_classification.py --data_folder <data_dir> --dataset <dataset_name> --model <model_name> --finetune True

2. Attacking a finetuned model

To attack a finetuned model after running text_classification.py or from the TextAttack library:

python whitebox_attack.py --data_folder <data_dir> --dataset <dataset_name> --model <model_name> --finetune True --start_index 0 --num_samples 100 --gumbel_samples 100

This runs the GBDA on the first 100 samples from the test set.

2.1. Downloading GPT-2 trained on BERT tokenizer (optional)

To attack a BERT model, GBDA requires a casual language model trained on the BERT tokenizer. We provide a pretrained GPT-2 model for this purpose. Before the attack, please run the following script to download the model from the Amazon S3 bucket:

curl https://dl.fbaipublicfiles.com/text-adversarial-attack/transformer_wikitext-103.pth -o transformer_wikitext-103.pth

3. Evaluating transfer attack

After attacking a model, run the following script to query a target model from the optimized adversarial distribution:

python evaluate_adv_samples.py --data_folder <data_dir> --dataset <dataset_name> --surrogate_model <surrogate_model_name> --target_model <target_model_name> --finetune True --start_index 0 --num_samples 100 --end_index 100 --gumbel_samples 1000

Citation

Please cite [1] if you found the resources in this repository useful.

[1] C. Guo *, A. Sablayrolles *, Herve Jegou, Douwe Kiela. Gradient-based Adversarial Attacks against Text Transformers. EMNLP 2021.

@article{guo2021gradientbased,
  title={Gradient-based Adversarial Attacks against Text Transformers},
  author={Guo, Chuan and Sablayrolles, Alexandre and Jégou, Hervé and Kiela, Douwe},
  journal={arXiv preprint arXiv:2104.13733},
  year={2021}
}

Contributing

See the CONTRIBUTING file for how to help out.

License

This project is CC-BY-NC 4.0 licensed, as found in the LICENSE file.