DataStories for Sentiment Analysis

Description

• This is the pytorch implementation of the first Model (Task A Model) of this paper https://aclweb.org/anthology/S17-2126
• I have implemented this network for my Data Mining Project

Emdeddings

• please download one of the following embeddings and put it inside the embeddings directory:
  • 50-d twittter word embeddings
  • 100-d twittter word embeddings
  • 200-d twittter word embeddings
  • 300-d twittter word embeddings

Software Requirements:

• Python 3
• pip
• Pytorch

To install other requirements

pip3 install -r requirements.txt

To Do

• Train the model and Show some statistics (on the original hyper-parameters)
• Add tensorboard graphs (Scalars of the training, development and testing metrics across epochs)
• Add option to continue from a checkpoint
• Add option to save checkpoint every iteration
• Provide links for the pre-trained models