/transformertc

Transformer based Token Classification.

Primary LanguagePythonMIT LicenseMIT

TransformerTC

Documentation Status License

Introductions

Transformer based Token Classification.

Currently supports training and inference for Named Entity Recognition (NER) tasks using BERT models.

Examples

See examples/ directory.

Install

This is a python package with setupy.py.

Tip: It’s probably a good idea to install numpy, scikit-learn, and pytorch using your preferred installation packages (i.e. optimized for your CPU/GPU).

Example using virtualenv:

virtualenv venv
source venv/bin/activate
cd transformertc
pip install -r requirements.exact.txt
python setup.py install

Note: Two different requirements.txt files are provided:

  1. requirements.txt - just the required python packages;
  2. requirements.exact.txt - the pinned versions of the required packages.

If installing everything inside the virtualenv, there should be no problem getting the exact versions of the packages. However, if for some reason the environment is shared, using the pinned versions might be restrictive.