Practical Deep Learning at Scale with MLflow

Track, store, versionize, and deploy deep learning models and model pipelines for different use cases

Key Features

• Focus on deep learning models and frameworks that solve practical problems
• Explore detailed example applications using deep learning models and pipeline management
• Learn to train, test, and tune pipelines at scale and implement model explainability

Book Description

Deep learning models used for solving practical problems can be developed, managed, shipped, and refined with their code and data using MLflow.

This book teaches you how to build reproducible ML pipelines and run them in a local or cloud environment. With the help of easy-to-use frameworks provided by MLflow, you'll bridge the gap between diverse environments. You'll find out how to switch between different frameworks, track your code, create versions of a pipeline, and model along with its parameters and metrics using MLflow's tracking and registry APIs and Delta Lake. Next, you'll customize the model pipeline to save and load the model to decouple offline model experimentation and production, all while keeping the model's behavior consistent with different data sets managed by Delta Lake. The book also helps you to recognize the patterns of different inference pipelines and shows you how to implement them. Finally, you'll be able to figure out how to choose the right framework for the right model serving scenarios and enjoy the best model serving experience in a scalable and cost-effective way.

By the end of this deep learning book, you'll be able to build and deploy your own deep learning applications using MLflow and deep learning model libraries, addressing the key pain points encountered in the deep learning model development life cycle.

What you will learn

• Track deep learning models in different environments using MLflow
• Build, deploy, and run deep learning model pipelines
• Understand the key challenges in the deep learning model development life cycle
• Manage the data dependencies and data versioning associated with deep learning models
• Scale up deep learning model pipeline training, testing, tuning, and deployment
• Ship practical NLP and image deep learning solutions from experimentation to production

Who This Book Is For

This book is for machine learning practitioners including data scientists, data engineers, and ML engineers and scientists interested in building scalable deep learning models using MLflow. Basic understanding of data science and machine learning is necessary to understand the concepts presented in this book.

Table of Contents

1. Overview of Deep Learning Life Cycle and MLOps Challenges
2. Getting Started with MLflow for Deep Learning
3. Tracking Models, Parameters, and Metrics
4. Tracking Code and Data Versioning
5. Write Once, Run Anywhere Using MLflow MLproject
6. Running Hyperparameter Tuning at Scale
7. Multi-Step Deep Learning Pipeline
8. Deploying DL Inference Pipeline at Scale
9. Fundamentals of Deep Learning Explainability
10. Implementing Model Explainability with MLflow SHAP at Scale