SMaBERTa

This repository contains the code for SMaBERTa, a wrapper for the huggingface transformer libraries. It was developed by Zhanna Terechshenko and Vishakh Padmakumar through research at the Center for Social Media and Politics at NYU.

Setup

To install using pip, run

pip install smaberta

To install from the source, first download the repository by running

git clone https://github.com/SMAPPNYU/SMaBERTa.git

Then, install the dependencies for this repo and setup by running

cd SMaBERTa
pip install -r requirements.txt
python setup.py install

Using the package

Basic use:

from smaberta import TransformerModel

epochs = 3
lr = 4e-6

training_sample = ['Today is a great day', 'Today is a terrible day']
training_labels = [1, 0]

model = TransformerModel('roberta', 'roberta-base', num_labels=25, reprocess_input_data=True, 
                         num_train_epochs=epochs, learning_rate=lr, output_dir='./saved_model/', 
                         overwrite_output_dir=True, fp16=False)

model.train(training_sample, training_labels)

For further details, see Tutorial.ipynb in the examples directory.

Acknowledgements

Code for this project was adapted from version 0.6 of https://github.com/ThilinaRajapakse/simpletransformers

Vishakh Padmakumar and Zhanna Terechshenko contributed to the software writing, implementation, and testing.

Megan Brown contributed to documentation and publication.

If you use this software in your research please cite it as:

@misc{padmakumar_terechshenko,
  author       = {Vishakh Padmakumar and Zhanna Terechshenko},
  title        = {SMAPPNYU/SMaBERTa},
  month        = dec,
  year         = 2020,
  doi          = {10.5281/zenodo.5090728},
  url          = {https://doi.org/10.5281/zenodo.5090728}
}