From the course: Microsoft Azure AI Engineer Associate (AI-102) Cert Prep by Microsoft Press

Unlock this course with a free trial

Join today to access over 25,100 courses taught by industry experts.

Customize pre-trained models for unique use cases

Customize pre-trained models for unique use cases - Azure Tutorial

From the course: Microsoft Azure AI Engineer Associate (AI-102) Cert Prep by Microsoft Press

Start my 1-month free trial Buy for my team

Customize pre-trained models for unique use cases

“

- [Instructor] There are three main methods that Microsoft teaches us for making a GPT smarter. When we do a deployment in Azure AI service, that's taking a base model, and then what? Well, we can use prompt engineering for fast control. This is where we're not fine-tuning or training the GPT, we're just being smarter about our inputs. If you have the capacity to add a system message, that helps. Providing examples, I use Markdown for semantics and delimiters and that kind of stuff. We'll suggest prompt engineering for a business who's not yet advanced enough to embrace fine-tuning. Maybe they don't know about assistance yet. Use when cost and time has to stay minimized. There's really two ways to make a large language model smarter. Again, we're talking GPT in particular. To make it organically smarter, if you pardon that term, it's fine-tuning. This is where you're training the model on your terms. You think, "Tim, yeah, we've seen this with the ML or machine learning models…

Contents