v1

latestOpenAPI 3.0.32026-07-223120123.1 KB

Monitoring

Publish inference

Publish an inference data point to an inference pipeline.

post/inference-pipelines/{inferencePipelineId}/data-stream

Path parameters

inferencePipelineIdstring uuid required

The inference pipeline id (a UUID).

Request body

rowsobject[] required

A list of inference data points with inputs and outputs

Example request

{
  "rows": [
    {
      "user_query": "what is the meaning of life?",
      "output": "42",
      "tokens": 7,
      "cost": 0.02,
      "timestamp": 1620000000
    }
  ],
  "config": {
    "numOfTokenColumnName": "num_tokens",
    "contextColumnName": "context",
    "costColumnName": "cost",
    "groundTruthColumnName": "ground_truth",
    "inferenceIdColumnName": "id",
    "inputVariableNames": [
      "user_query"
    ],
    "latencyColumnName": "latency",
    "outputColumnName": "output",
    "prompt": [
      {
        "role": "user",
        "content": "{{ user_query }}"
      }
    ],
    "questionColumnName": "question",
    "timestampColumnName": "timestamp",
    "userIdColumnName": "user_id",
    "sessionIdColumnName": "session_id"
  }
}

Response

Status OK.

successtrue required