PMLDL-1

Assignment Overview

This repository contains the solution for an assignment on the course of Practical Machine Learning and Deep Learning (PMLDL). The task involves deploying a machine learning model and creating a simple automated MLOps pipeline. The main task is to deploy a model in an API and create a web application that interacts with the API. The model API accepts requests from the web application and sends back responses.

Model

For this solution, I use the distil-bert model from HuggingFace for Named Entity Recognition (NER). NER is a personally interesting field of research. I chose it for this assignment to get more practice for the course project. The possible entity types are:

• MISC: Miscellaneous entity
• PER: Person’s name
• ORG: Organization
• LOC: Location

Repository Structure

PMLDL-1/
├── services/
│   ├── backend/
│   │   ├── src/
│   │   │   └── main.py
│   │   ├── Dockerfile
│   │   ├── pyproject.toml
│   │   └── oetry.lock
│   ├── frontend/
│   │   ├── src/
│   │   │   └── main.py
│   │   ├── Dockerfile
│   │   ├── pyproject.toml
│   │   └── poetry.lock
├── docker-compose.yml
├── .dockerignore
├── .gitignore
└── README.md

Prerequisites

• Docker
• Docker Compose
• Poetry (for dependency management)

Setup

1. Clone the repository:

   git clone https://github.com/your-username/PMLDL-1.git
   cd PMLDL-1

2. Build and run the Docker containers:

   docker-compose up --build

3. Access the web application:

   Open your web browser and navigate to http://localhost:8501 to access the Streamlit frontend.

Backend

The backend is built using FastAPI and runs the distil-bert model for NER. It includes:

• Model Caching: To improve performance by caching the model.
• Optimized Docker Configuration: To ensure efficient resource usage.
• Health Checks: To monitor the status of the API.

Frontend

The frontend is built using Streamlit and includes:

• Input Fields: For users to input text.
• Prediction Area: To display the annotated entities.

For color annotation, I used the st-annotated-text library.

Technologies Used

• Docker: For containerization.
• FastAPI: For building the backend API.
• Streamlit: For building the frontend web application.
• HuggingFace Transformers: For the distil-bert model.
• st-annotated-text: For color annotation in the frontend.
• Poetry: For dependency management.

Acknowledgments

• Thanks to the creators of the st-annotated-text library for providing a simple way to annotate text in Streamlit.
• Thanks to the HuggingFace team for the distil-bert model and the Transformers library.