Jobs API

A job is designed to execute a specific task on the Ray cluster. Jobs are created by the backend and are executed on the Ray cluster.

For example, a job may run inference with a model to generate ground truth for a dataset. Or, a job may evaluate the outputs of an inference job against the ground truth stored in the dataset.

Endpoints

POST /api/v1/jobs/inference/

Create Inference Job

Status Codes:
POST /api/v1/jobs/annotate/

Create Annotation Job

This uses a hardcoded model, that is, Lumigator’s opinion on what reference model should be used to generate annotations. See more: https://blog.mozilla.ai/lets-build-an-app-for-evaluating-llms/

Status Codes:
POST /api/v1/jobs/evaluator/

Create Evaluation Job

Status Codes:
GET /api/v1/jobs/

List Jobs

Retrieves job data from the Lumigator repository where Ray metadata is also available.

Results are a merged representation which form an augmented view of a β€˜job’.

NOTE: Lumigator repository data takes precedence over Ray metadata.

Query Parameters:

  • skip (integer)

  • limit (integer)

  • job_types (array)

Status Codes:
GET /api/v1/jobs/{job_id}

Get Job

Retrieves merged job data from the Lumigator repository and Ray for a valid UUID.

The result is a merged representation which forms an augmented view of a β€˜job’.

NOTE: Lumigator repository data takes precedence over Ray metadata.

Parameters:

  • job_id (string)

Status Codes:
GET /api/v1/jobs/{job_id}/logs

Get Job Logs

Parameters:

  • job_id (string)

Status Codes:
GET /api/v1/jobs/{job_id}/result

Get Job Result

Return job results metadata if available in the DB.

Parameters:

  • job_id (string)

Status Codes:
GET /api/v1/jobs/{job_id}/dataset

Get Job Dataset

Return the job-associated dataset if available in the DB.

Parameters:

  • job_id (string)

Status Codes:
GET /api/v1/jobs/{job_id}/result/download

Get Job Result Download

Return job results file URL for downloading.

Parameters:

  • job_id (string)

Status Codes: