The Vals platform is designed to allow you to easily test LLM applications: copilots, RAG systems, and more. This documentation serves as a guide for the usage of both the web appplication as well as the CLI and SDK tools.

The goal of the platform is to create and run “Test Suites”. Each “Test Suite”, broadly, is meant to test your model’s performance in a certain aspect.

Each “Test Suite” is composed of multiple “Tests”. A “Test” has exactly one input, which is meant to represent an input you may receive from a user. For example, if you are testing a math copilot, an input may be “What is 3 * 2?”.

A test also has a set of “checks”. Each check is looking for a certain thing in the model’s answer. For example, if the input is “What is 3 * 2?”, a check may be “the output includes the number 6”.

The rest of the documentation further explains how to create test suites and running them against models. To get started, view our page on creating test suites