Awesome

Towards Debiasing Sentence Representations

Pytorch implementation for debiasing sentence representations.

This implementation contains code for removing bias from BERT representations and evaluating bias level in BERT representations.

Correspondence to:

• Paul Liang (pliang@cs.cmu.edu)
• Irene Li (mengzeli@cs.cmu.edu)

Paper

Towards Debiasing Sentence Representations<br> Paul Pu Liang, Irene Li, Emily Zheng, Yao Chong Lim, Ruslan Salakhutdinov, and Louis-Philippe Morency<br> ACL 2020

If you find this repository useful, please cite our paper:

@inproceedings{liang-etal-2020-towards,
    title = "Towards Debiasing Sentence Representations",
    author = "Liang, Paul Pu  and
      Li, Irene Mengze  and
      Zheng, Emily  and
      Lim, Yao Chong  and
      Salakhutdinov, Ruslan  and
      Morency, Louis-Philippe",
    booktitle = "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics",
    month = jul,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2020.acl-main.488",
    doi = "10.18653/v1/2020.acl-main.488",
    pages = "5502--5515",
}

Installation

First check that the requirements are satisfied:</br> Python 3.6</br> torch 1.2.0</br> huggingface transformers</br> numpy 1.18.1</br> sklearn 0.20.0</br> matplotlib 3.1.2</br> gensim 3.8.0 </br> tqdm 4.45.0</br> regex 2.5.77</br> pattern3</br>

The next step is to clone the repository:

git clone https://github.com/pliang279/sent_debias.git

To install bert models, go to debias-BERT/, run pip install .

Data

Download the GLUE data by running this script:

python download_glue_data.py --data_dir glue_data --tasks SST,QNLI,CoLA

Unpack it to some directory $GLUE_DIR.

Precomputed models and embeddings (optional)

1. Models

   • Download https://drive.google.com/file/d/1cAN49-HDHFdNP1GJZn83s2-mEGCeZWjh/view?usp=sharing to debias-BERT/experiments.

   • tar -xvf acl2020-results.tar.gz
     

2. Embeddings

   • Download https://drive.google.com/file/d/1ubKn8SCjwnp9pYjQa9SmKWxFojX9a6Bz/view?usp=sharing to debias-BERT/experiments.

   • tar -xvf saved_embs.tar.gz
     

Usage

If you choose to use precomputed models and embeddings, skip to step B. Otherwise, follow step A and B sequentially.

A. Fine-tune BERT

1. Go to debias-BERT/experiments.
2. Run export TASK_NAME=SST-2 (task can be one of SST-2, CoLA, and QNLI).
3. Fine tune BERT on $TASK_NAME.
   • With debiasing

     python run_classifier.py \
     --data_dir $GLUE_DIR/$TASK_NAME/ \
     --task_name $TASK_NAME \
     --output_dir path/to/results_directory \
     --do_train \
     --do_eval \
     --do_lower_case \
     --debias \
     --normalize \
     --tune_bert 
     

   • Without debiasing

     python run_classifier.py \
     --data_dir $GLUE_DIR/$TASK_NAME/ \
     --task_name $TASK_NAME \
     --output_dir path/to/results_directory \
     --do_train \
     --do_eval \
     --do_lower_case \
     --normalize \
     --tune_bert 
     

   The fine-tuned model and dev set evaluation results will be stored under the specified output_dir.

B. Evaluate bias in BERT representations

1. Go to debias-BERT/experiments.

2. Run export TASK_NAME=SST-2 (task can be one of SST-2, CoLA, and QNLI).

3. Evaluate fine-tuned BERT on bias level.

   • Evaluate debiased fine-tuned BERT.

       python eval_bias.py \
       --debias \
       --model_path path/to/model \
       --model $TASK_NAME \
       --results_dir path/to/results_directory \
       --output_name debiased
     

     If using precomputed models, set model_path to acl2020-results/$TASK_NAME/debiased.
   • Evaluate biased fine-tuned BERT.

       python eval_bias.py \
       --model_path path/to/model \
       --model $TASK_NAME \
       --results_dir path/to/results_directory \
       --output_name biased
     

     If using precomputed models, set model_path to acl2020-results/$TASK_NAME/biased.

   The evaluation results will be stored in the file results_dir/output_name.

   Note: The argument model_path should be specified as the output_dir corresponding to the fine-tuned model you want to evaluate. Specifically, model_path should be a directory containing the following files: config.json, pytorch_model.bin and vocab.txt.

4. Evaluate pretrained BERT on bias level.

   • Evaluate debiased pretrained BERT.

     python eval_bias.py \
     --debias \
     --model pretrained \
     --results_dir path/to/results_directory \
     --output_name debiased 
     

   • Evaluate biased pretrained BERT.

     python eval_bias.py \
     --model pretrained \
     --results_dir path/to/results_directory \
     --output_name biased 
     

   Again, the bias evaluation results will be stored in the file results_dir/output_name.