k1c / ACL_Workshop_Gender_Bias_NLP

ACL_Workshop_Gender_Bias_NLP

https://genderbiasnlp.talp.cat/

1. datasets:

   • GAP Coreference
   • Book Corpus
   • UMBC

2. dataloaders:

   • Gap Coreference Dataloader
   • Book Corpus Dataloader
   • UMBC Dataloader

3. experiments:

   • A1_filter: run filter on a small test set.
   • plot_tsne: run TSNE on Bert_feature_extraction.json produced by ./pytorch-pretrained-BERT/examples/extract_features.py
   • allennlp_models.py: Run [your_dataset].* through various AllenNLP pre-trained models. Currently supporting Coreference and SRL.

pip3 install allennlp

Run script with Coreference model:

python3 allennlp_models.py \
    https://s3-us-west-2.amazonaws.com/allennlp/models/coref-model-2018.02.05.tar.gz \
    [your_dataset].* --output-file [your_dataset]_coref.json

Run script with Semantic Role Labeling model:

python3 allennlp_models.py \
    https://s3-us-west-2.amazonaws.com/allennlp/models/srl-model-2018.02.27.tar.gz \
    [your_dataset].* --output-file [your_dataset]_srl.json

Extract features from BERT at this location ./pytorch-pretrained-BERT/examples/extract_features.py

python extract_features.py --input_file path_to_input_file --output_file output_file_name.json --bert_model bert-base-cased --layers -2