Full Stack Deep Learning with Python
Introduction
Full Stack Deep Learning
Understanding and expertise in all components and stages of building and deploying deep learning systems - getting deep learning systems from prototype to production.
Steps in Full Stack Deep Learning
- Planning and project setup
- Data collection and labeling
- Model training and debugging
- Deploying, testing, and maintenance
MLOps
Machine Learning Operations (MLOps) is a set of practices and tools to streamline and automate the end-to-end machine learning lifecycle. This is a part of the full stack deep learning.
Components and Concepts of MLOps
- Version Control
- Continuous Integration
- Continuous Delivery
- Model Packaging
- Monitoring and Logging
- Model Versioning
MLflow
- An open-source platform for managing the machine learning lifecycle.
- It is designed to simplify and streamline the end-to-end process of developing, training, and deploying machine learning models.
- It does two things well:
- Model Training and Hyperparameter tuning
- Model Serving
- Model Tracking
Components and Features of MLflow
- Tracking experiments and model runs.
- Package machine learning code into projects
- Package and share machine learning models
- Model versioning and management
- Model deployment - locally and on the cloud
Full stack deep learning
Project Planning and Setup
- Define project goals --> Choose metrics. --> Evaluate baselines --> Set up codebase
- Find the impact and feasibility of the project.
Data collection and labeling
- Strategy --> Ingest --> Labeling
- Define the data source, quality, quantity, and frequency of data.
Model Training and Debugging
- Here, the typical model training and hyperparameter tuning cycle take place.
- In this phase, we will also implement the MLOps workflow.
Deploying, testing, and maintenance
- Pilot in production --> Testing --> Deployment --> Monitoring
- Here also we do MLOps.
Abandoned the course....
