Read run results

Every dataset row of a run with what the target predicted, plus one entry per evaluator per row. Runs older than the status cache need experimentSlug as well, since a run id is only unique within its experiment.

GET /api/v1/experiments/runs/{runId}/results

Read run results

Python

import requests

url = "https://app.langwatch.ai/api/v1/experiments/runs/{runId}/results"

headers = {"X-Auth-Token": "<api-key>"}

response = requests.get(url, headers=headers)

print(response.text)

200

{
  "experimentId": "<string>",
  "runId": "<string>",
  "projectId": "<string>",
  "dataset": [
    {
      "index": 123,
      "entry": {},
      "targetId": "<string>",
      "predicted": {},
      "cost": 123,
      "duration": 123,
      "error": "<string>",
      "domainError": {
        "code": "<string>",
        "kind": "<string>",
        "message": "<string>",
        "meta": {},
        "httpStatus": 123,
        "fault": "<string>",
        "traceId": "<string>",
        "spanId": "<string>",
        "traceUrl": "<string>",
        "retryable": true,
        "tips": [
          "<string>"
        ],
        "docsUrl": "<string>",
        "reasons": [
          "<unknown>"
        ]
      },
      "traceId": "<string>"
    }
  ],
  "evaluations": [
    {
      "evaluator": "<string>",
      "status": "processed",
      "index": 123,
      "name": "<string>",
      "targetId": "<string>",
      "score": 123,
      "label": "<string>",
      "passed": true,
      "details": "<string>",
      "cost": 123,
      "duration": 123,
      "inputs": {}
    }
  ],
  "timestamps": {
    "createdAt": 123,
    "updatedAt": 123,
    "finishedAt": 123,
    "stoppedAt": 123
  },
  "completeness": {
    "complete": true,
    "dataset": {
      "received": 4503599627370495,
      "expected": 4503599627370495
    },
    "evaluations": {
      "received": 4503599627370495,
      "expected": 4503599627370495
    }
  },
  "workflowVersionId": "<string>",
  "progress": 123,
  "total": 123,
  "targets": [
    {
      "id": "<string>",
      "name": "<string>",
      "type": "<string>",
      "promptId": "<string>",
      "promptVersion": 123,
      "agentId": "<string>",
      "evaluatorId": "<string>",
      "model": "<string>",
      "metadata": {}
    }
  ]
}

Authorizations

X-Auth-Token · string · header · required

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.

Path Parameters

runId · string · required

Minimum string length: 1

Query Parameters

experimentSlug · string

Owning experiment. Required once the run has aged out of the status cache.

Response

200

application/json

Success

experimentId · string · required

runId · string · required

projectId · string · required

dataset · object[] · required

One row per dataset entry, with what the target predicted

Show child attributes

evaluations · object[] · required

One row per evaluator per dataset entry

Show child attributes

timestamps · object · required

Show child attributes

completeness · object · required

What is stored against what the run reported. Results are stored after they are reported, so a read can hold part of a run: complete is false until the run has ended and every reported row and evaluation is stored. expected is null when the run reported no counts

Show child attributes

workflowVersionId · string | null

progress · number | null

total · number | null

targets · object[] | null

Resolves the targetId each dataset row and evaluation carries

Show child attributes