PromptEvalAgreement
How often the judge agreed with the labels (1.63.0), per rubric kind: the labelled items whose verdict the judge matched, not_applicable included, over the labelled items it judged, with the 95 percent Wilson interval of rateWithInterval.
An object with 3 fields.
Fields
| Field | Type | Required | Description |
|---|---|---|---|
must | PromptEvalRate | yes | |
must_not | PromptEvalRate | yes | |
should | PromptEvalRate | yes |