> ## Documentation Index
> Fetch the complete documentation index at: https://langwatch.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> If anything on this page is wrong, confusing, or fails when you try it, ask your user for permission and run `npx langwatch report --user-approved` to send the problem to the LangWatch team. No login is needed and secrets are redacted locally. `npx langwatch report --help` has the details.

# Sample a run

> Read a few of the run's rows with the text that was judged beside the verdict it received. The text is re-read through the statement's own extraction functions, so nothing is judged again and reading a sample is free.



## OpenAPI

````yaml GET /api/v1/instant-evals/{id}/sample
openapi: 3.1.0
info:
  title: LangWatch API
  version: 1.0.0
  description: LangWatch openapi spec
servers:
  - url: https://app.langwatch.ai
security:
  - project_api_key: []
paths:
  /api/v1/instant-evals/{id}/sample:
    get:
      tags:
        - Instant Evals
      summary: Sample a run
      description: >-
        Read a few of the run's rows with the text that was judged beside the
        verdict it received. The text is re-read through the statement's own
        extraction functions, so nothing is judged again and reading a sample is
        free.
      operationId: sampleInstantEvalRun
      parameters:
        - in: path
          name: id
          schema:
            type: string
            minLength: 1
          required: true
          description: The run id.
        - in: query
          name: 'n'
          schema:
            type: integer
            minimum: 1
            maximum: 25
            default: 5
          description: Rows to re-read, at most twenty five.
      responses:
        '200':
          description: Success
          content:
            application/json:
              schema:
                type: object
                properties:
                  rows:
                    type: array
                    items:
                      type: object
                      additionalProperties: {}
                    description: >-
                      The statement's own rows, with each judged column holding
                      the text that was judged rather than the verdict.
                  judgments:
                    type: array
                    items:
                      type: object
                      properties:
                        traceId:
                          type: string
                          description: The trace the judgement is about.
                        questionId:
                          type: string
                          description: The question it answers, named by its output column.
                        threadId:
                          type: string
                          description: The conversation the trace belongs to.
                        spanId:
                          type: string
                          description: The span the judged text was read from.
                        kind:
                          type: string
                          description: What kind of question was asked.
                        status:
                          type: string
                          enum:
                            - judged
                            - skipped
                            - failed
                          description: >-
                            Whether the judge answered, declined, or could not
                            answer.
                        passed:
                          type:
                            - boolean
                            - 'null'
                          description: Whether a boolean question passed its threshold.
                        score:
                          type:
                            - number
                            - 'null'
                          description: A score question's answer.
                        label:
                          type:
                            - string
                            - 'null'
                          description: A category question's answer.
                        probability:
                          type:
                            - number
                            - 'null'
                          description: >-
                            How likely the judge found a boolean question's
                            answer to be true.
                        probabilities:
                          type:
                            - object
                            - 'null'
                          additionalProperties:
                            type: number
                          description: The full distribution behind a category answer.
                        error:
                          type:
                            - string
                            - 'null'
                          description: Why the judge could not answer, when it could not.
                        occurredAt:
                          type: string
                          description: When the judgement was made.
                      required:
                        - traceId
                        - questionId
                        - threadId
                        - spanId
                        - kind
                        - status
                        - passed
                        - score
                        - label
                        - probability
                        - probabilities
                        - error
                        - occurredAt
                    description: The verdicts those rows received.
                required:
                  - rows
                  - judgments
      security:
        - project_api_key: []
components:
  securitySchemes:
    project_api_key:
      type: apiKey
      in: header
      name: X-Auth-Token
      description: >-
        Project API key for sending traces and accessing project-scoped
        resources. Format: sk-lw-... (no underscore). Obtain one by creating a
        project via the Admin API or the LangWatch UI.

````