LAB 3: Sentiment analysis using NLP models

Prerequisites

The project requires Python 3.

1 - Create a Virtual Environment (Optional)

You can use virtualenv but we recommend that you use conda. Download the appropriate Miniconda version for your system. Then follow the installation instructions.

2 - Install PyTorch

Follow the instructions from the PyTorch home page: https://pytorch.org/

3 - Install Requirements

pip install -r lab3/requirements.txt

4 - Download pre-trained Word Embeddings

In order to minimize the memory requirements you can use low dimensional word embeddings, such as the 50d Glove embeddings. However, if your computer has enough RAM you will get better results with higher dimensional embeddings.

• Glove 6B: Generic english word embeddings - 50d, 100d, 200d, & 300d vectors.

• Glove Twitter: Twitter specific word embeddings - 25d, 50d, 100d, & 200d vectors

• fastText: Generic english word embeddings - only 300d vectors.

The project expects the file(s) to be in the /embeddings folder.