Oreilly – MLOps Engineering at Scale, Video Edition 2022-2

Published on: 2024-08-18 23:36:55

Categories: 28

Description

MLOps Engineering at Scale course, Video Edition. This course teaches you how to efficiently produce machine learning models using pre-built services on AWS and other cloud providers. You’ll learn how to quickly build flexible and scalable machine learning systems without getting involved in time-consuming operational tasks or incurring high physical hardware costs. Following a real example of taxi fare calculation, you will design an MLOps pipeline for a PyTorch model using AWS serverless capabilities.

What you will learn:

Data extraction, conversion and loading
Query data with SQL
Automatic differential comprehension in PyTorch
Deploy model training pipelines as a service endpoint
Pipeline life cycle monitoring and management
Measuring performance improvement

This course is suitable for people who:

They are familiar with Python, SQL and the basics of machine learning.
They seek to increase the speed and efficiency in the implementation of machine learning systems.
They want to take advantage of cloud computing to develop their models.

Course details

Publisher: Oreilly
Lecturer: Carl Osipov
Training level: beginner to advanced
Training duration: 8 hours and 5 minutes

Course headings

Part 1. Mastering the data set
Chapter 1. Introduction to serverless machine learning
Chapter 1. Challenges when designing a machine learning platform
Chapter 1. Public clouds for machine learning platforms
Chapter 1. What is serverless machine learning?
Chapter 1. Why serverless machine learning?
Chapter 1. Who is this book for?
Chapter 1. How does this book teach?
Chapter 1. When is this book not for you?
Chapter 1. Conclusions
Chapter 1. Summary
Chapter 2. Getting started with the data set
Chapter 2. Starting with object storage for the data set
Chapter 2. Discovering the schema for the data set
Chapter 2. Migrating to columnar storage for more efficient analytics
Chapter 2. Summary
Chapter 3. Exploring and preparing the data set
Chapter 3. Getting started with data quality
Chapter 3. Applying VACUUM to the DC taxi data
Chapter 3. Implementing VACUUM in a PySpark job
Chapter 3. Summary
Chapter 4. More exploratory data analysis and data preparation
Chapter 4. Summary
Part 2. PyTorch for serverless machine learning
Chapter 5. Introducing PyTorch: Tensor basics
Chapter 5. Getting started with PyTorch tensor creation operations
Chapter 5. Creating PyTorch tensors of pseudorandom and interval values
Chapter 5. PyTorch tensor operations and broadcasting
Chapter 5. PyTorch tensors vs. native Python lists
Chapter 5. Summary
Chapter 6. Core PyTorch: Autograd, optimizers, and utilities
Chapter 6. Linear regression using PyTorch automatic differentiation
Chapter 6. Transitioning to PyTorch optimizers for gradient descent
Chapter 6. Getting started with data set batches for gradient descent
Chapter 6. Data set batches with PyTorch Dataset and DataLoader
Chapter 6. Dataset and DataLoader classes for gradient descent with batches
Chapter 6. Summary
Chapter 7. Serverless machine learning at scale
Chapter 7. Using IterableDataset and ObjectStorageDataset
Chapter 7. Gradient descent with out-of-memory data sets
Chapter 7. Faster PyTorch tensor operations with GPUs
Chapter 7. Scaling up to use GPU cores
Chapter 7. Summary
Chapter 8. Scaling out with distributed training
Chapter 8. Parameter server approach to gradient accumulation
Chapter 8. Introducing logical ring-based gradient descent
Chapter 8. Understanding ring-based distributed gradient descent
Chapter 8. Phase 1: Reduce-scatter
Chapter 8. Phase 2: All-gather
Chapter 8. Summary
Part 3. Serverless machine learning pipeline
Chapter 9. Feature selection
Chapter 9. Feature selection case studies
Chapter 9. Feature selection using guiding principles
Chapter 9. Selecting features for the DC taxi data set
Chapter 9. Summary
Chapter 10. Adopting PyTorch Lightning
Chapter 10. Summary
Chapter 11. Hyperparameter optimization
Chapter 11. Neural network layers configuration as a hyperparameter
Chapter 11. Experimenting with the batch normalization hyperparameter
Chapter 11. Summary
Chapter 12. Machine learning pipeline
Chapter 12. Enabling PyTorch-distributed training support with Kaen
Chapter 12. Unit testing model training in a local Kaen container
Chapter 12. Hyperparameter optimization with Optuna
Chapter 12. Summary
Appendix A. Introduction to machine learning
Appendix A. Machine learning at first glance
Appendix A. Machine learning with structured data sets
Appendix A. Regression with structured data sets
Appendix A. Classification with structured data sets
Appendix A. Training a supervised machine learning model