Sentiment Analysis API with Quantized Model

This repository contains a simple Flask API that performs sentiment analysis on text input using a quantized DistilBERT model. Quantization helps reduce the model size and inference time, making it more efficient for deployment.

Structure

• app.py: Main Flask application file. Defines the API endpoint and handles requests.
• models/sentiment_model.py: Handles model loading, quantization, and inference.

└── 📁 models
    └── 📄 sentiment_model.py
📄 app.py
📄 requirements.txt 

Setup and Usage

1. Install dependencies:

   pip install -r requirements.txt

   (Make sure you have a requirements.txt file with the necessary packages, or create one with: pip freeze > requirements.txt)

2. Run the Flask API:

   flask run 

   (Or python app.py, depending on your setup)

3. Send requests to the API:

   You can use tools like curl or Postman to send POST requests to the /predict endpoint.

   Example using curl:

   curl -X POST \
     -H "Content-Type: application/json" \
     -d '{"text": "This is a great movie! I highly recommend it."}' \
     http://127.0.0.1:5000/predict

   Example using Python's requests library:

   import requests
   
   api_url = 'http://127.0.0.1:5000/predict'
   text = "This is a great movie! I highly recommend it."
   data = {'text': text}
   
   response = requests.post(api_url, json=data)
   
   if response.status_code == 200:
     result = response.json()
     print(result)
   else:
     print(f"Request error: {response.status_code}")

   The API will respond with a JSON object containing:

   {
     "text": "This is a great movie! I highly recommend it.",
     "sentiment": "POSITIVE",
     "score": 0.9998656630516052
   }

Notes

• The API currently uses a pre-trained and quantized DistilBERT model. You can replace this with a different model by modifying the model_name and loading the appropriate model file in models/sentiment_model.py.
• This is a basic example, and you might need to adjust it based on your specific needs and deployment environment. Consider adding error handling, authentication, and other features for a more robust application.