MLPublishMixin

MLPublishMixin#

class MLPublishMixin#

Bases: object

Methods Summary

get_ml_published_model(project_id, publish_id)

Returns the details of the a published model.

get_ml_published_models(project_id[, ...])

Returns all the publishing model for a given project.

ml_publish_model(project_id, published_as, ...)

Publish new ML Model.

modify_published_model(project_id, ...)

Update ML Published model

unpublish_ml_model(project_id, publish_id)

Delete ML Model

Methods Documentation

get_ml_published_model(project_id, publish_id, include_pipeline_workflows=False)#

Returns the details of the a published model.

Parameters:

  • project_id – Id of the Squirro project.

  • publish_id – id of the published model.

  • include_pipeline_workflows – Flag to include the information about pipeline workflows that are using the published model with the response.

get_ml_published_models(project_id, include_pipeline_workflows=False)#

Returns all the publishing model for a given project.

Parameters:

  • project_id – Id of the Squirro project.

  • include_pipeline_workflows – Flag to include the information about pipeline workflows that are using the published models with the response.

ml_publish_model(project_id, published_as, description, model_id=None, model_name=None, gt_id=None, gt_name=None, gt_version=None, template_id=None, template_name=None, workflow_id=None, workflow_config=None, workflow_name=None, tagging_level=None, labels=None, location=None, global_id=None, orig_project_id=None, external_model=False)#

Publish new ML Model.

Parameters:

  • project_id – Id of the Squirro project.

  • published_as – Name of the published model.

  • description – Text to be shown describing the model.

  • model_id – Id of the model to be published.

  • model_name – Name of ais model.

  • gt_id – Id of the groundtruth used for the model.

  • gt_name – Name of the groundtruth used for the model.

  • gt_version – Version of the groundtruth used for the model.

  • template_id – Id of the template used for the model.

  • template_name – Name of the template used for the model.

  • workflow_id – Id of the workflow of the model to be published.

  • workflow_config – Workflow config of the model to be published.

  • workflow_name – Workflow name of the model to be published.

  • tagging_level – Tagging level of the model to be published.

  • labels – Labels of the model to be published.

  • location – Location of the model to be published.

  • global_id – Global_id of the model to be published.

  • orig_project_id – Id of the project where it was publish originally.

  • external_model – Flag to determine if the published model is from external.

This function can be used in three cases of publishing a model:

Case 1: Publish an ais model: saves an ais model in the publish database,

fetches and save the heritage information (model, gt) while creation. Use the param ´external_model´=False and the model-id of the ais model you want to publish.

Case 2: Publish a model from external directly to the ingestion pipeline by providing the ml workflow.

Use ´external_model´=True. Populate ´workflow_config´ and ´workflow_name´ along with the mandatory params.

Case 3: Import a model published on another cluster. Use external_model=True,

and provide the heritage information along with its heritage information (´model_id´, ´model_name´, ´gt_id´, ´gt_name´, ´gt_version´, ´template_id´, ´template_name´, ´tagging_level´, ´labels´, ´location´, ´global_id´, ´orig_project_id´)

modify_published_model(project_id, publish_id, published_as, description)#

Update ML Published model

Parameters:

  • project_id – Id of the Squirro project.

  • publish_id – id of the published model.

  • published_as – Name of the published model.

  • description – Text to be shown describing the model.

unpublish_ml_model(project_id, publish_id)#

Delete ML Model

Parameters:

  • project_id – Id of the Squirro project.

  • publish_id – Id of the published Model.

Contents