Skip to main content

a wrapper for the huggingface transformer libraries

Project description


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.


To install using pip, run

pip install smaberta

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

git clone

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

pip install -r requirements.txt
python 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_model(training_sample, training_labels)

For further details, see Tutorial.ipynb in the (examples)[] directory.


Code for this project was adapted from version 0.6 of

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

Megan Brown contributed to documentation and publication.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

smaberta-0.0.2.tar.gz (12.3 kB view hashes)

Uploaded Source

Built Distribution

smaberta-0.0.2-py3-none-any.whl (12.1 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page