Laminar

Laminar Evaluations API

The lower-level LaminarClient.evals surface for wiring evaluations into an existing pipeline: create an evaluation, pre-register each datapoint so a row is visible in the UI before the executor runs, then update that datapoint with a trace id, executor output, and scores. Because update can be called any time after create, the executor and the scorer can run in different processes, which supports async LLM-judge scoring and pure backfills of scored rows.

Documentation GitHub

Documentation

📖

Documentation

https://laminar.sh/docs/evaluations/manual-evaluation

📖

APIReference

https://laminar.sh/docs/sdk/client

API entry from apis.yml

aid: laminar:laminar-evaluations-api
name: Laminar Evaluations API
humanURL: https://laminar.sh/docs/evaluations/manual-evaluation
baseURL: https://api.lmnr.ai
tags:
- Evaluations
- Datasets
- Scoring
description: 'The lower-level LaminarClient.evals surface for wiring evaluations into an existing pipeline:
  create an evaluation, pre-register each datapoint so a row is visible in the UI before the executor
  runs, then update that datapoint with a trace id, executor output, and scores. Because update can be
  called any time after create, the executor and the scorer can run in different processes, which supports
  async LLM-judge scoring and pure backfills of scored rows.'
properties:
- url: https://laminar.sh/docs/evaluations/manual-evaluation
  type: Documentation
- url: https://laminar.sh/docs/sdk/client
  type: APIReference