Generative Artificial Intelligence (GenAI) is a rapidly developing area of machine learning, with application across business, government, and academia. In this course, you will learn about different types of GenAI and see examples of how SAS can enhance your efforts to make the most of these techniques.
This course will be released several lessons at a time until all lessons are available. We expect that each lesson can be completed in about an hour, and you can work at your own pace to complete the material. As we release new lessons, you might lose progress through the material that you have completed, so please make a note of where you are leaving off before exiting the course.
Learn How To
Explain what generative AI is and how it fits into the broader AI landscape. Describe several types of GenAI systems. Name some of the key challenges and opportunities in making a trustworthy AI system. Generate synthetic data with Synthetic Minority Oversampling Technique (SMOTE) and Generative Adversarial Networks (GANs). Generate meaningful text using Large Language Models (LLMs).Classify text for LLMs using Bidirectional Encoder Representations from Transformers (BERT). Improve the accuracy and relevance of LLM output using Retrieval Augmented Generation (RAG). Who Should Attend
Learners who want to know more about the techniques that comprise GenAI and how to make use of them with SAS
Prerequisites
Before taking this course, you should have some background in statistics and machine learning using SAS. You can gain this knowledge by taking the following courses:;
Machine Learning Using SAS Viya Statistics You Need to Know for Machine Learning SAS Products Covered
SAS Viya;SAS Machine Learning
Course Outline
Defining Generative AI: Unraveling the Concept
Generative AI. The AI landscape. Types of machine learning. Generative models. Generative AI and LLMs.Examples of generative AI technologies. The AI challenge: balancing risk and reward. Generative AI: key takeaways.Types of Generative AI SystemsGenerative AI in action. The input-output framework in generative AI. Generative AI systems. Other types of generative AI systems. Overview of Trustworthy AI and the Analytics Life CycleResponsible innovation.AI and analytics life cycle.Trustworthy AI: how can SAS help? Data chain of custody.Transformers and Large Language Models (LLMs)Language models.Sequence models.Attention-based models.Transformer overview.Embeddings.Positional encoding.From transformers to LLMs.GPT-1 and BERT.What is GPT?What is BERT?Comparing GPT-1 and BERT.Timelines.Synthetic Minority Oversampling Technique (SMOTE) in SASSeparate sampling techniques. What is SMOTE? Demo: understanding SMOTE by using the smoteSample action on a small data set. Demo: Using the smoteSample action to create new cases via SMOTE. SMOTE summary. Generative Adversarial Networks (GANs) in SASWhat are generative adversarial networks (GANs)? Training networks. GANs: advantages and challenges. Applications of GANs. Variants of GANs.Model architecture. Demo: augmenting Data. Synthetic data generation assessment.Comparing GANs and SMOTE.Classifying Text Using BERTWhat is BERT?What is BERT used for?Word embeddings.Multi-head attention.Encoding.Multi-head attention details.BERT Pre-training.BERT fine-tuning.SAS Viya VERT text classifier. Demo: Using the BERT text classifier CAS action.Retrieval Augmented Generation (RAG)Large language models.Retrieval-Augmented Generation (RAG).Supporting documents.RAG pipeline.Demo: Implementing a RAG pipeline, Part 1.Demo: Implementing a RAG pipeline, Part 1.Demo: Implementing a RAG pipeline, Part 2.Demo: Implementing a RAG pipeline, Part 3.Demo: Implementing a RAG pipeline, Part 4.(COMING SOON) Generating SAS Code Automatically with Simple Prompts (COMING SOON) Generating Synthetic Data with No Code/Low Code