PromptEvalCalibrationCreateRequest
Calibrates the suite's current judge (1.63.0): draws a sample of a completed run's verdicts to label, or reuses the labelled sample of an earlier calibration. Every issue is a detail of 422 validation_failed (PromptEvalCalibrationIssueCode).
An object with 3 fields.
Fields
| Field | Type | Required | Description |
|---|---|---|---|
from_calibration_id | PromptEvalCalibrationId | no | An earlier calibration of the suite whose sample is fully labelled, reused with its labels; the new calibration queues at once. |
run_id | PromptEvalRunId | no | A completed run of the suite to sample; exactly one of it and from_calibration_id. |
sample_size | integer | no | 50 when omitted (a target); fewer when the run holds fewer. |
Used by
The operations whose parameters, request body or responses name this shape directly.