Score your AI responses with custom Arize judges.
Builds Arize judges and scoring tasks to assess model responses in recorded spans or experiment runs.
When to use it
Use for requests to create or improve evaluators, score spans or experiments, configure column mappings, or set up continuous evaluation.
Give it an Arize evaluation request; it creates evaluators and scoring tasks, asking for choices when needed.
Requires the ax CLI, version 0.14.0 or higher; troubleshooting instructions install or upgrade it if necessary.
Requires Arize access through a configured profile with an AI integration for the judge model.
This skill
Arize profile
Saves or replaces profile settings (irreversible)
Arize
Starts hosted evaluation jobs
Saves judges and evaluation tasks