PABI

This is the code repository for the ArXiv paper Foreshadowing the Benefits of Incidental Supervision. If you use this code for your work, please cite

@article{he2020foreshadowing,
  title={Foreshadowing the Benefits of Incidental Supervision},
  author={He, Hangfeng and Zhang, Mingyuan and Ning, Qiang and Roth, Dan},
  journal={arXiv preprint arXiv:2006.05500},
  year={2020}
}

Installing Dependencies

Use virtual environment tools (e.g miniconda) to install packages and run experiments
python>=3.6
pip install -r requirements.txt

Code Organization

The code is organized as follows:

• bpp.py (CWBPP algorithm for learning with various inductive signals)
• run_ner.py (BERT for NER)
• run_squad.py (BERT for QA)

Reproducing experiments

To reproduce the experiments for learning with various inductive signals:

sh run_experiments.sh

To reproduce the experiments for cross-domain signals:

sh run_xdomain_ner_experiments.sh
sh run_xdomain_qa_experiments.sh