About

• Efficiently analyze sentiments in entity-level text using small Bert (part of the BERT family) created by Google.
• This project employs advanced natural language processing techniques to predict sentiment behind entities mentioned in textual data.

Key Features

• Entity-Level Analysis: Focus on sentiments associated with specific entities within the text.
• BERT Pre-Trained model: Pre-trained small Bert for this project.

Tech Stack

• Python.
• Tensorflow.
• Pandas.
• Kaggle.
• BERT.

Results

• Achieved an accuracy score of 83.68% on testing datasets.

Important

• Uncomment lines that have "!pip install" in them.
• Make sure you have GPUs or TPUs on your machine before training the model.
• If you dont have any, you can use either Colabs or Kaggle to have access to GPUs or TPUs.