adamlin120 / Zero-shot_Classification_of_News_Title

Zero-shot Classification of News Title - Final Project for IRTM NTU 2018

Zero-shot Classification of News Title

Getting Started

Prerequisites

GPU enabled environment is required.

numpy
PyTorch >= 1.0
pytorch_pretrained_bert

tqdm
redis
flask

Running the tests

User can get the prediction result by sending request to server.

Send Request

python3 send_request [--title TITLE] [--tags TAGS [TAGS ...]] [--api_url API_URL]

Deployment With

• Flask - The python web framework used
• Redis - In-memory cache

Authors

• YenTing Lin - GitHub

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Acknowledgments

• pytorch_pretrained_bert is cloned from https://github.com/huggingface/pytorch-pretrained-BERT
• Zero-shot classification for text is inspired by https://arxiv.org/abs/1712.05972