layerlens evaluate

layerlens evaluate — create and run an evaluation.

layerlens evaluate --model openai/gpt-4o --benchmark mmlu
layerlens evaluate --space my-space
layerlens evaluate --space my-space --fail-on-regression --tolerance 0.01

Options

Option	Description
--model	Model key
--benchmark	Benchmark key
--space	Saved space ID
--scorers	Comma-separated scorer IDs
--judges	Comma-separated judge IDs
--fail-on-regression	Exit non-zero on regression
--tolerance	Regression tolerance (default 0.01)

See also

• Tutorial 3: CI/CD gates