This repo is written with Transformers library (PyTorch), packed with Pytorch-lightning.
DemoSite - Folder consists of HTML, CSS and JS files for demo webpage
paraphrase_data - Folder consists of combination of QQP data and PAWS data for questions and sentences paraphrase generation training.
flaskAPI.py - Using Flask to serve the model for the web application
ParaphraseDataLoader.py - Data module for loading data into model training.
T5FineTuner.py - Model definition.
test.py - File to run test with pretrained weights.
train.py - File to train the model.