Sentiment Analysis App (BERT) 🎭

A professional-grade sentiment analysis application utilizing a fine-tuned BERT (Bidirectional Encoder Representations from Transformers) model. This project features a deep learning training pipeline and a real-time web interface for sentiment prediction.

🚀 Project Overview

This repository contains a complete end-to-end NLP pipeline:

• Training: A Jupyter Notebook optimized for Google Colab/Local GPU using Hugging Face Transformers.
• Inference: A Streamlit web application that provides instant sentiment classification and confidence scores.
• Model: Fine-tuned bert-base-uncased capable of understanding complex linguistic nuances.

---

🛠️ Tech Stack

• Language: Python 3.14.3
• Deep Learning: PyTorch, Transformers, Datasets
• Web Interface: Streamlit
• Data Science: Pandas, NumPy, Scikit-Learn
• Environment: Virtualenv / Git LFS

---

📦 Project Structure

sentiment_app/
├── bert_sentiment_final/   # Local folder containing model weights
│   ├── model.safetensors   # The 438MB BERT weights
│   ├── tokenizer.json      # Tokenizer configuration
│   └── config.json         # Model architecture config
├── Sentiment_Analysis.ipynb # Training pipeline (Colab compatible)
├── sentiment_app.py        # Streamlit web application code
├── requirements.txt        # Dependency list for Python 3.14
└── README.md               # Documentation