Unsupervised Label Refinement Improves Dataless Text Classification

This repository contains the code and data for the paper Unsupervised Label Refinement Improves Dataless Text Classification

repo structure

• dual-encoder: with cosine and L2 distances
• single-encoder
• data: comes with three training data files with 30k instances. The paper uses 300k instances, so this repo is just for demo purposes.

dependencies

• PyTorch 1.4.0
• transformers 2.6.0
• scipy 1.4.1
• numpy 1.19.1
• sklearn