Multi Class Tweet Sentiment Classification

This repository presents two approaches to classify tweets into 5 classes based on their sentiment:

• Extremely Negative
• Negative
• Neutral
• Positive
• Extremely Positive

It is recommended to run the code in a GPU backend with ~15Gb VRAM.

Dataset

The dataset is available on kaggle

Approaches Used:

• Transformer model
• Bi-directional Encoder Representations (BERT)

Please refer to the report for further information on the preprocessing procedures used and results obtained.