From the course: Building in Azure AI Studio
Join today to access over 23,200 courses taught by industry experts.
Evaluate your AI solution
From the course: Building in Azure AI Studio
Evaluate your AI solution
- [Instructor] After developing your AI solution, you need to evaluate its performance. A typical evaluation involves some key elements. We need to test datasets as the input. We need to evaluate the changes in variance, like flows, prompts, system message, your data, models, and model parameters. We also need to run evaluations against variance metrics, like relevance, coherence, and the groundedness. Because there are so many factors in an evaluation, a simple test like entering some prompts in the chat is insufficient and time consuming. We need to use an evaluation tool to help us. In Azure AI Studio, we can use the tool to run two types of evaluations. Manual evaluations for manually reviewing the results after generating AI responses with your test datasets. Metric evaluations for evaluating the performance of your AI application using industry standard metrics. Now, let's do a quick demo. Here's the Chat playground in my Azure AI project. I already set up a system message. Now,…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.