https://github.com/di37/imdb-reviews-sentiment-analysis-using-transformers

This repository contains a complete implementation of a sentiment analysis model using Transformers and TensorFlow. The project classifies IMDB movie reviews as either positive or negative by training Transformer model on the IMDB dataset.
https://github.com/di37/imdb-reviews-sentiment-analysis-using-transformers

keras machine-learning natural-language-processing nlp python sentiment-analysis tensorflow text-classification transformers

Last synced: 13 days ago
JSON representation

This repository contains a complete implementation of a sentiment analysis model using Transformers and TensorFlow. The project classifies IMDB movie reviews as either positive or negative by training Transformer model on the IMDB dataset.

• Host: GitHub
• URL: https://github.com/di37/imdb-reviews-sentiment-analysis-using-transformers
• Owner: di37
• Created: 2024-04-07T01:42:04.000Z (9 months ago)
• Default Branch: main
• Last Pushed: 2024-12-11T22:45:54.000Z (about 1 month ago)
• Last Synced: 2024-12-11T23:26:37.672Z (about 1 month ago)
• Topics: keras, machine-learning, natural-language-processing, nlp, python, sentiment-analysis, tensorflow, text-classification, transformers
• Language: Jupyter Notebook
• Homepage:
• Size: 1.95 KB
• Stars: 2
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Training an IMDB Movie Sentiment Analyzer Using TensorFlow and Transformers

This project demonstrates how to use Transformer-based models for sentiment analysis of movie reviews from the IMDB dataset. Transformers, a cutting-edge architecture for NLP tasks, provide contextual understanding of text, enabling highly accurate sentiment classification. TensorFlow and Keras libraries were used to build, train, and evaluate the model.

## Environment

The whole experiment was done in Colab notebook. 

## Machine Learning

The pre-processing and feature engineering is almost the same except padding was done. The whole experiment was done in `Full_Assignment_Transformers_IMDB_Sentiment_Analysis.ipynb` from Data Ingestion till Prediction Pipeline.

## Results

Evaluation using test dataset as follows:

235/235 - 11s - loss: 0.2903 - accuracy: 0.8815 - 11s/epoch - 47ms/step

loss: 0.290

accuracy: 0.881