STstatsig
Evals

Send eval results

Changed on

Create a run on an existing eval from results scored outside Statsig. Each result carries its input, output, optional expected value and categories, and one or more named scores from 0 to 1.

post/console/v1/evals/send_results/{name}

Request

  • Base URL
    https://statsigapi.net
  • URL
    https://statsigapi.net/console/v1/evals/send_results/{name}
  • Auth
    API key in header STATSIG-API-KEY

Path parameters

namestring required

name

Request body

namestring nullable

The name of the eval run

parametersobject

Parameters

summaryScoresobject

Summary scores

primaryGraderstring

Score name shown as the primary result. Defaults to the first score.

criticalGradersstring[]

Score names treated as release gates: a mean of 0 fails the whole run.

Example request

{
  "name": "My Eval Run",
  "results": [
    {
      "input": "Foo",
      "expected": "Hi Foo",
      "output": "Hi Foo",
      "category": [
        "greeting"
      ],
      "scores": {
        "correctness": {
          "score": 1,
          "metadata": {
            "justification": "friendly"
          }
        },
        "tone": {
          "score": 1
        }
      }
    }
  ],
  "parameters": {
    "parameter_name": "parameter_value"
  },
  "summaryScores": {
    "summary_score_name": 100
  }
}

Response

Send eval results

messagestring required

A simple string explaining the result of the operation.

Changes