Quick Start Guide to Large Language Models: Strategies and Best Practices for ChatGPT, Embeddings, Fine-Tuning, and Multimodal AI, 2nd Edition

5

Optimizing LLMs with Customized Fine-Tuning

Introduction

So far, we’ve almost exclusively used LLMs, both open- and closed-source, just as they are off the shelf. We were relying on the power of the Transformer’s attention mechanisms and their speed of computation to perform some pretty complex problems with relative ease. As you can probably guess, that isn’t always enough.

In Chapter 2, I showcased the power of updating LLMs with custom data to increase accuracy in information retrieval. But that’s just the tip of the iceberg. In this chapter, we will dive deeper into the world of fine-tuning LLMs to unlock their full potential. Fine-tuning updates off-the-shelf models—specifically, the values of their parameters—and empowers them to achieve ...