Skip to main content
At its core, the Evaluate pillar allows you to assess the performance of your prompts by running evaluations on datasets after you defined a set of evaluators. In other words, evaluators will assess the performance of each prompt on different variables’ values. The values of the variables are stored in datasets. For this reason, when selecting any evaluator, you must link an existing dataset to it: Link datasets to evaluators in Adaline In Adaline, datasets are datastores of test cases. Use datasets to run tests on row-by-row bases in the Playground with a linked dataset, or on all rows by running evaluations.

​
Dynamic Columns

Datasets support dynamic columns that can fetch values from external APIs or execute other prompts at runtime. This transforms static datasets into dynamic data sources, enabling you to:
  • Fetch live data from external HTTP endpoints using API variables
  • Chain prompts together using prompt variables
  • Populate columns automatically instead of manually entering values
Learn more about dynamic columns. Learn more about datasets:

Setup for Evaluations

Create new datasets and set them up to run evaluations.

Dynamic Columns

Configure API and Prompt variables at the column level.

Bulk Import Data

Bulk import your data to create new datasets.