Hands-On Large Language Models

Chapter 6. Prompt Engineering

In the first chapters of this book, we took our first steps into the world of large language models (LLMs). We delved into various applications, such as supervised and unsupervised classification, employing models that focus on representing text, like BERT and its derivatives.

As we progressed, we used models trained primarily for text generation, models that are often referred to as generative pre-trained transformers (GPT). These models have the remarkable ability to generate text in response to prompts from the user. Through prompt engineering, we can design these prompts in a way that enhances the quality of the generated text.

In this chapter, we will explore these generative models in more detail and dive into the realm of prompt engineering, reasoning with generative models, verification, and even evaluating their output.

Using Text Generation Models

Before we start with the fundamentals of prompt engineering, it is essential to explore the basics of utilizing a text generation model. How do we select the model to use? Do we use a proprietary or open source model? How can we control the generated output? These questions will serve as our stepping stones into using text generation models.

Choosing a Text Generation Model

Choosing a text generation model starts with choosing between proprietary models or open source models. Although proprietary models are generally more performant, we focus in this book more on open source models as they offer ...