Micromodels -- A framework for accurate, explainable, data efficient, and reusable NLP models.
A micromodel is a representation for a specific linguistic behavior. Analagous to how recent applications are built using microservices, in which each microservice has a specific responsibility, a micromodel is responsible for representing a specific linguistic behavior. By orchestrating a suite of micromodels, our framework allows researchers and developers to build explainable and reusable systems across multiple NLP tasks.
For more details, refer to our Findings of EMNLP 2021 paper, Micromodels for Efficient, Explainable, and Reusable Systems: A Case Study on Mental Health.
This repository requires Python version 3.7 or newer.
Once cloned, install all required packages using pip. We highly recommend using a virtual environment.
pip install -r requirements.txt
Set an environment variable MM_HOME to point to the location of the cloned repository.
This will be the default location for where micromodels are stored. Ex:
export MM_HOME=/home/[username]/micromodels
To ensure that everything is set up correctly, run pytest.
pytest
Documentation can be found in our readthedocs, which includes a tutorial.
Pull requests are more than welcomed! If there are any features you would like to add, or have any suggestions on design, email ajyl@umich.edu or create a pull request.
If you find our work helpful for your research, please cite our work:
@inproceedings{lee-etal-2021-micromodels-efficient,
title = "Micromodels for Efficient, Explainable, and Reusable Systems: A Case Study on Mental Health",
author = "Lee, Andrew and
Kummerfeld, Jonathan K. and
An, Larry and
Mihalcea, Rada",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2021",
month = nov,
year = "2021",
address = "Punta Cana, Dominican Republic",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.findings-emnlp.360",
pages = "4257--4272",
}