Skip to content

🪢 Open source agent evals & observability: Trace, evaluate, and improve LLM applications with one open platform.

319 B
Raw Blame
// Shared bounds for every sampling slider in evals v2 so the rule editor, the
// activation dialog, and the evaluator-saved dialog read the same. The floor is
// non-zero because a rule sampled at 0% would never evaluate anything.
export const SAMPLING_SLIDER_MIN = 0.0001;
export const SAMPLING_SLIDER_STEP = 0.0001;