Human Ratings

What are Galileo human ratings?

Galileo allows users to create or rate runs based on human ratings offering inside of Galileo Evaluate. Human ratings show in the Feedback section inside of Galileo Evaluate to offer the capability to see these ratings side-by-side with the runs and customize them based on the goals of the human rating. They allow users to add their own rating to a given run. The human rating types offered include:

  • ๐Ÿ‘/ ๐Ÿ‘Ž

  • 1 - 5 โญs

  • Numerical ratings

  • Categorical ratings (self-defined categories)

  • Free-form text

Along with each rating, you can also allow users to provide a rationale. These ratings are aggregated against all of the runs in a project or run.

Human ratings are a great way to extend Galileo's generative AI evaluation platform to meet the needs of human evaluators, reviewers, business users, subject matter experts, data scientists, or developers. Because they are entirely customizable (through the Configure button) they can enable users to add their own feedback to a run. This is helpful in cases where metrics don't capture everything being evaluated, review of metrics is being done, or additional information is gathered during evaluation. For more information, visit the Evaluate with Human Feedback page.

Last updated