This repository contains a simple example of how to use BERT for sentiment analysis. The dataset used for this example is the IMDB dataset. The dataset contains 50,000 movie reviews from IMDB, labeled by sentiment (positive/negative). For illustrative purposes, we ran the model once with preprocessing and once without preprocessing. Predictively the model performs better without preprocessing. These notebooks were run on Google Colab. The code is written in Python 3.8.10. The model was trained on a Tesla T4 GPU.
- Haguar Tarek
- Manar Amgad
- Mariem Mostafa
- Nada Elwazane