BiLSTM in Pytorch for Multi-scale Sentiment Analysis of SemEval Twitter Data

This is my final project for Natural Language Processing, Spring 2018 at NYU.

This implements a BiLSTM in Pytorch, which has achieved considerable accuracy in the three-point scale sentiment analysis. For detailed network architecture, method and final result, please refer to the report in this repo.