Overview

The goal of this course is to provide the student with a tangible understanding and a real hands-on experience with generative AI applications deployed and optimized for IBM Power systems. This self-paced virtual course uses video lectures, review questions, and virtual lab machine exercises to provide the student with foundational knowledge and experience about the topics covered in the course. In the lecture, the student begins by learning the basics of generative AI, and the basic components of a generative AI application, then applies these concepts to real-world examples on Power where the student learns Power offerings for AI workloads, package management fundamentals in Python, Operating System deployment strategies on Power, and Power hardware use cases.

The lab exercises will start with viewing and manipulating files in a basic AI application running on IBM Power Red Hat Enterprise Linux. They will have hands-on experience identifying the components of that AI application and their connection to other components, and will add functionalities to the application that demonstrate more advanced AI application techniques like frameworks, prompt tuning, and conversation memory. Then, the students will go through the process of setting up a Python virtual environment, investigating Power hardware resources, and using the Hardware Management Console (HMC) to ensure memory and other resources are optimized for AI workloads. 

Audience

This course is open to all interested learners, regardless of their experience. Typical students may include customers, IBM technical personnel, Business Partner technical personnel, computer engineering students, IT consultants and architects.

Prerequisites

Objective

• Understand foundational genAI concepts and terms
• Familiarize with the larger genAI ecosystem for AI applications
• Distinguish between basic genAI math and hardware terms
• Understand IBM Power’s current offerings for AI workloads
• Set up package channels and repositories for AI libraries on Power
• Identify AI application components in an example implementation of AI inferencing on Power
• Describe Power-specific optimizations and recommendations for GenAI
• Refer to peripheral applications and services that help to develop AI apps on Power

Course Outline

• Unit 0: Introduction
• Video 0-1: Introduction
• Unit 1: Generative AI applications
• Video 1-1: Generative AI concepts
• Video 1-2: Inferencing
• Video 1-3: Code
• Unit 1 review questions
• Exercise 1: Generative AI applications
• EX01 Section 1: Working through a Jupyter Notebook of an AI application
• EX01 Section 2: Adding and managing advanced generative AI app features
• EX01 Section 3: Adding conversation memory to a generative AI application
• EX01 Section 4: Implementing an AI framework
• Unit 2: Math, hardware, and Power offerings for generative AI
• Video 2-1: Math and hardware terms
• Video 2-2: Power offerings for AI
• Unit 2 review questions
• Unit 3: Implementing generative AI on Power
• Video 3-1: Package management
• Video 3-2: Retrieval-augmented generation
• Video 3-3: Implementing genAI on Power
• Unit 3 review questions
• Unit 4: Deploying AI on Power
• Video 4-1: Performance considerations
• Video 4-2: AI deployment options
• Unit 4 review questions
• Exercise 2: Deploying AI on Power Red Hat Enterprise Linux (RHEL)
• EX02 Section 1: Viewing a virtual environment and installed packages
• EX02 Section 2: Viewing hardware resources
• EX02 Section 3: Creating a virtual environment and investigating HMC hardware resources