matrice.projects module#

Module for interacting with backend API to manage projects.

class matrice.projects.Projects(session, project_name=None, project_id=None)[source]#

Bases: object

A class for handling project-related operations using the backend API.

session#

The session object used for API interactions.

Type:

Session

account_number#

The account number associated with the session.

Type:

str

project_name#

The name of the project.

Type:

str

project_id#

The ID of the project (initialized in the constructor).

Type:

str

project_input#

The input type for the project (initialized in the constructor).

Type:

str

output_type#

The output type for the project (initialized in the constructor).

Type:

str

Parameters:

  • session (Session) – The session object used for API interactions.

  • project_name (str) – The name of the project.

__init__(session, project_name=None, project_id=None)[source]#

Initialize a Projects object with project details.

Parameters:

  • session (Session) – The session object used for API interactions.

  • project_name (str) – The name of the project.

change_status(enable=True)[source]#

Enables or disable a project. It is set to enable by default.

Parameters:

type (str) – The type of action to perform: “enable” or “disable”.

Returns:

A tuple containing: - A success message if the project is enabled or disabled successfully. - An error message if the action fails.

Return type:

tuple

Example

>>> success_message, error = project.change_status(enable=True)
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(success_message)
create_annotation(project_type, ann_title, dataset_id, dataset_version, labels, only_unlabeled, is_ML_assisted, labellers, reviewers, guidelines)[source]#

Create a new annotation for a dataset.

Parameters:

  • project_type (str) – The type of the project for which the annotation is being created.

  • ann_title (str) – The title of the annotation.

  • dataset_id (str) – The ID of the dataset to annotate.

  • dataset_version (str) – The version of the dataset.

  • labels (list) – The list of labels for the annotation.

  • only_unlabeled (bool) – Whether to annotate only unlabeled data.

  • is_ML_assisted (bool) – Whether the annotation is ML-assisted.

  • labellers (list) – The list of labellers for the annotation.

  • reviewers (list) – The list of reviewers for the annotation.

  • guidelines (str) – The guidelines for the annotation.

Returns:

A tuple containing: - An Annotation object for the created annotation. - An Action object related to the annotation creation process.

Return type:

tuple

Example

>>> annotation, action = project.create_annotation("object_detection", "MyAnnotation", "dataset123", "v1.0", ["label1", "label2"], True, False, [{"email": "user-email", "name": "username", "percentageWork": '100'}],[{"email": "user-email", "name": "username", "percentageWork": '100'}], "Follow these guidelines")
>>> if action:
>>>     print(f"Annotation created: {annotation}")
>>> else:
>>>     print(f"Error: {annotation}")
create_deployment(deployment_name, model_id, gpu_required=True, auto_scale=False, auto_shutdown=True, shutdown_threshold=5, compute_alias='', model_type='trained', runtime_framework='Pytorch')[source]#

Create a deployment for a model.

Parameters:

  • deployment_name (str) – The name of the deployment.

  • model_id (str) – The ID of the model to be deployed.

  • gpu_required (bool, optional) – Flag to indicate if GPU is required for the deployment (default is True).

  • auto_scale (bool, optional) – Flag to indicate if auto-scaling is enabled (default is False).

  • auto_shutdown (bool, optional) – Flag to indicate if auto-shutdown is enabled (default is True).

  • shutdown_threshold (int, optional) – The threshold for auto-shutdown (default is 5).

  • compute_alias (str, optional) – The alias for the compute (default is an empty string).

Returns:

A tuple containing: - A Deployment object for the created deployment. - An Action object related to the deployment process.

Return type:

tuple

Example

>>> deployment, action = project.create_deployment("Deployment1", "model123", auto_scale=True)
>>> if action:
>>>     print(f"Deployment created: {deployment}")
>>> else:
>>>     print(f"Error: {deployment}")
create_experiment(name, dataset_id, target_run_time, dataset_version, primary_metric, matrice_compute=True, models_trained=[], performance_trade_off=-1)[source]#

Create a new experiment for model training.

Parameters:

  • name (str) – The name of the experiment.

  • dataset_id (str) – The ID of the dataset to be used in the experiment.

  • target_run_time (str) – The target runtime for the experiment.

  • dataset_version (str) – The version of the dataset.

  • primary_metric (str) – The primary metric to evaluate the experiment.

  • matrice_compute (bool, optional) – Flag to indicate whether to use matrix compute (default is True).

  • models_trained (list, optional) – List of models that have been trained in the experiment (default is an empty list).

  • performance_trade_off (float, optional) – The performance trade-off for the experiment (default is -1).

Returns:

A tuple containing: - An Experiment object for the created experiment. - An error message if the response indicates an error, or None if successful. - A status message describing the result of the operation.

Return type:

tuple

Example

>>> experiment, error, message = project.create_experiment("Experiment1", "dataset123", "runtimeA", "v1.0", "accuracy")
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Experiment created: {experiment}")
create_model_export(model_train_id, export_formats, model_config, is_gpu_required=False)[source]#

Add export configurations to a trained model.

Parameters:

  • model_train_id (str) – The ID of the trained model.

  • export_formats (list) – The list of formats to export the model.

  • model_config (dict) – The configuration settings for the model export.

  • is_gpu_required (bool, optional) – Flag to indicate if GPU is required for the export (default is False).

Returns:

A tuple containing: - An InferenceOptimization object related to the model export. - An Action object related to the export process.

Return type:

tuple

Example

>>> inference_opt, action = project.add_model_export("model123", ["format1", "format2"], {"configKey": "configValue"}, is_gpu_required=True)
>>> if action:
>>>     print(f"Model export added: {inference_opt}")
>>> else:
>>>     print(f"Error: {inference_opt}")
delete()[source]#

Delete a project by project ID.

Returns:

A tuple containing: - A success message if the project is deleted successfully. - An error message if the deletion fails.

Return type:

tuple

Example

>>> success_message, error = project.delete()
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(success_message)
get_actions_logs(action_id)[source]#

Fetch action logs for a specific action.

Parameters:

action_id (str) – The ID of the action for which logs are to be fetched.

Returns:

A tuple containing: - The action logs if the request is successful. - An error message if the request fails.

Return type:

tuple

Example

>>> logs, error = project.get_actions_logs_for_action("action123")
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Action logs: {logs}")
get_annotation(dataset_id=None, annotation_id=None, annotation_name='')[source]#

Get an Annotation instance.

Parameters:

  • dataset_id (str, optional) – The ID of the dataset associated with the annotation.

  • annotation_id (str, optional) – The ID of the annotation.

  • annotation_name (str, optional) – The name of the annotation.

Returns:

An Annotation instance with the specified dataset ID, annotation ID, and/or name.

Return type:

Annotation

Example

>>> annotation = project.get_annotation(annotation_id="annotation123")
>>> print(annotation)
get_annotations_status_summary()[source]#

Get the annotations status summary for the project.

Returns:

An ordered dictionary with annotations status and their counts.

Return type:

OrderedDict

Example

>>> annotations_status = project.get_annotations_status_summary()
>>> print(annotations_status)
get_dataset(dataset_id=None, dataset_name='')[source]#

Get a Dataset instance.

Parameters:

  • dataset_id (str, optional) – The ID of the dataset.

  • dataset_name (str, optional) – The name of the dataset.

Returns:

A Dataset instance with the specified ID and/or name.

Return type:

Dataset

Example

>>> dataset = project.get_dataset(dataset_id="dataset123")
>>> print(dataset)
get_dataset_status_summary()[source]#

Get the dataset status summary for the project.

Returns:

An ordered dictionary with dataset status and their counts.

Return type:

OrderedDict

Example

>>> dataset_status = project.get_dataset_status_summary()
>>> print(dataset_status)
get_deployment(deployment_id=None, deployment_name='')[source]#

Get a Deployment instance.

Parameters:

  • deployment_id (str, optional) – The ID of the deployment.

  • deployment_name (str, optional) – The name of the deployment.

Returns:

A Deployment instance with the specified ID and/or name.

Return type:

Deployment

Example

>>> deployment = project.get_deployment(deployment_id="deployment123")
>>> print(deployment)
get_deployment_status_summary()[source]#

Get the deployment status summary for the project.

Returns:

An ordered dictionary with deployment status and their counts.

Return type:

OrderedDict

Example

>>> deployment_status = project.get_deployment_status_summary()
>>> print(deployment_status)
get_experiment(experiment_id=None, experiment_name='')[source]#

Get an Experiment instance.

Parameters:

  • experiment_id (str, optional) – The ID of the experiment.

  • experiment_name (str, optional) – The name of the experiment.

Returns:

An Experiment instance with the specified ID and/or name.

Return type:

Experiment

Example

>>> experiment = project.get_experiment(experiment_id="experiment123")
>>> print(experiment)
get_exported_model(model_export_id=None, model_export_name='')[source]#

Get an InferenceOptimization instance.

Parameters:

  • model_export_id (str, optional) – The ID of the model export.

  • model_export_name (str, optional) – The name of the model export.

Returns:

An InferenceOptimization instance with the specified ID and/or name.

Return type:

InferenceOptimization

Example

>>> inference_optimization = project.get_inference_optimization(model_export_id="export123")
>>> print(inference_optimization)
get_exported_models()[source]#

Fetch all model exports for the project.

Returns:

A tuple containing: - The model export data if the request is successful. - An error message if the request fails.

Return type:

tuple

Example

>>> model_exports, error = project.get_model_exports()
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Model exports: {model_exports}")
get_latest_action_record(service_id)[source]#

Fetch the latest action logs for a specific service ID.

Parameters:

service_id (str) – The ID of the service for which to fetch the latest action logs.

Returns:

A tuple containing: - The response dictionary from the API. - An error message if the response indicates an error, or None if successful. - A status message describing the result of the operation.

Return type:

tuple

Example

>>> result, error, message = project.get_latest_action_record("service123")
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Status: {message}")
get_model(model_id=None, model_name='')[source]#

Get a Model instance.

Parameters:

  • model_id (str, optional) – The ID of the model.

  • model_name (str, optional) – The name of the model.

Returns:

A Model instance with the specified ID and/or name.

Return type:

Model

Example

>>> model = project.get_model(model_id="model123")
>>> print(model)
get_model_export_status_summary()[source]#

Get the model export status summary for the project.

Returns:

An ordered dictionary with model export status and their counts.

Return type:

OrderedDict

Example

>>> model_export_status = project.get_model_export_status_summary()
>>> print(model_export_status)
get_model_status_summary()[source]#

Get the model status summary for the project.

Returns:

An ordered dictionary with model status and their counts.

Return type:

OrderedDict

Example

>>> model_status = project.get_model_status_summary()
>>> print(model_status)
get_service_action_logs(service_id, service_name)[source]#

Fetch action logs for a specific service.

Parameters:

  • service_id (str) – The ID of the service for which to fetch action logs.

  • service_name (str) – The name of the service for which to fetch action logs.

Returns:

A tuple containing: - The response dictionary from the API. - An error message if the response indicates an error, or None if successful. - A status message describing the result of the operation.

Return type:

tuple

Example

>>> result, error, message = project.get_service_action_logs("service123")
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Status: {message}")
import_cloud_dataset(dataset_name, source, source_url, cloud_provider, dataset_type, input_type='image', dataset_description='', version_description='', credential_alias='', bucket_alias='', compute_alias='', source_credential_alias='', bucket_alias_service_provider='auto')[source]#

Upload a cloud dataset.

Parameters:

  • dataset_name (str) – The name of the dataset.

  • source (str) – The source of the dataset.

  • source_url (str) – The URL of the source.

  • cloud_provider (str) – The cloud provider for the dataset.

  • dataset_type (str) – The type of the dataset.

  • input_type (str, optional) – The input type for the dataset (default is “image”).

  • dataset_description (str, optional) – A description of the dataset (default is an empty string).

  • version_description (str, optional) – A description of the dataset version (default is an empty string).

  • credential_alias (str, optional) – The credential alias for accessing the dataset (default is an empty string).

  • bucket_alias (str, optional) – The bucket alias for the dataset (default is an empty string).

  • compute_alias (str, optional) – The compute alias (default is an empty string).

  • source_credential_alias (str, optional) – The source credential alias (default is an empty string).

  • bucket_alias_service_provider (str, optional) – The bucket alias service provider (default is “auto”).

Returns:

A tuple containing: - A Dataset object for the created dataset. - An Action object related to the dataset upload process.

Return type:

tuple

Example

>>> dataset, action = project.upload_cloud_dataset("MyCloudDataset", "cloud_source", "http://source.url", "AWS", "image")
>>> if action:
>>>     print(f"Dataset uploaded: {dataset}")
>>> else:
>>>     print(f"Error: {dataset}")
import_local_dataset(dataset_name, file_path, dataset_type, source='lu', dataset_description='', version_description='', input_type='image', credential_alias='', bucket_alias='', compute_alias='', source_credential_alias='', bucket_alias_service_provider='auto')[source]#

Upload a local dataset.

Parameters:

  • dataset_name (str) – The name of the dataset.

  • file_path (str) – The path to the local file.

  • dataset_type (str) – The type of the dataset.

  • source (str, optional) – The source of the dataset (default is “lu”).

  • dataset_description (str, optional) – A description of the dataset (default is an empty string).

  • version_description (str, optional) – A description of the dataset version (default is an empty string).

  • input_type (str, optional) – The input type for the dataset (default is “image”).

  • credential_alias (str, optional) – The credential alias for accessing the dataset (default is an empty string).

  • bucket_alias (str, optional) – The bucket alias for the dataset (default is an empty string).

  • compute_alias (str, optional) – The compute alias (default is an empty string).

  • source_credential_alias (str, optional) – The source credential alias (default is an empty string).

  • bucket_alias_service_provider (str, optional) – The bucket alias service provider (default is “auto”).

Returns:

A tuple containing: - A Dataset object for the created dataset. - An Action object related to the dataset upload process.

Return type:

tuple

Example

>>> dataset, action = project.upload_local_dataset("MyLocalDataset", "path/to/data.csv", "image")
>>> if action:
>>>     print(f"Dataset uploaded: {dataset}")
>>> else:
>>>     print(f"Error: {dataset}")
invite_user_to_project(email, permissions)[source]#

Invite a user to the current project with specific permissions.

This function sends an invitation to a user, identified by their email address, to join the specified project. The user will be assigned the provided permissions for different project services.

Parameters:

  • email (str) – The email address of the user to invite.

  • permissions (dict) – A dictionary specifying the permissions for various project services.

Returns:

A tuple containing three elements: - API response (dict): The raw response from the API. - error_message (str or None): Error message if an error occurred, None otherwise. - status_message (str): A status message indicating success or failure.

Return type:

tuple

Example

>>> email = "ashray.gupta@matrice.ai"
>>> permissions = {
...     'datasetsService': {
...         'read': True,
...         'write': True,
...         'admin': True
...     },
...     'annotationService': {
...         'read': True,
...         'write': False,
...         'admin': False
...     },
...     'modelsService': {
...         'read': True,
...         'write': False,
...         'admin': False
...     },
...     'inferenceService': {
...         'read': True,
...         'write': False,
...         'admin': False
...     },
...     'deploymentService': {
...         'read': True,
...         'write': True,
...         'admin': False
...     },
...     'byomService': {
...         'read': True,
...         'write': False,
...         'admin': False
...     }
... }
>>> resp, err, msg = project.invite_user_to_project(email, permissions)
>>> if err:
...     print(f"Error: {err}")
... else:
...     print("User invited successfully")
list_annotations(page_size=10, page_number=0)[source]#

List all annotations in the project.

Returns:

A tuple containing: - A list of annotations if the request is successful. - An error message if the request fails.

Return type:

tuple

Example

>>> annotations, error = project.list_annotations()
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Annotations: {annotations}")
list_collaborators()[source]#

List all collaborators associated with the current project along with the permissions.

This function retrieves a list of all collaborators for the specified project ID.

Returns:

A tuple containing three elements: - API response (dict): The raw response from the API. - error_message (str or None): Error message if an error occurred, None otherwise. - status_message (str): A status message indicating success or failure.

Return type:

tuple

Example

>>> resp, err, msg =project.list_collaborators()
>>> if err:
>>>     print(f"Error: {err}")
>>> else:
>>>     print(f"Collaborators : {resp}")
list_datasets(status='total', page_size=10, page_number=0)[source]#

List all datasets in the project.

Returns:

A tuple containing: - A list of datasets if the request is successful. - An error message if the request fails.

Return type:

tuple

Example

>>> datasets, error = project.list_datasets()
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Datasets: {datasets}")
list_deployments(page_size=10, page_number=0)[source]#

List all deployments inside the project.

Returns:

A tuple containing: - A list of deployments if the request is successful. - An error message if the request fails.

Return type:

tuple

Example

>>> deployments, error = project.list_deployments()
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Deployments: {deployments}")
list_drift_monitorings(page_size=10, page_number=0)[source]#

Fetch a list of all drift monitorings.

Returns:

A tuple containing three elements: - API response (dict): The raw response from the API. - error_message (str or None): Error message if an error occurred, None otherwise. - status_message (str): A status message indicating success or failure.

Return type:

tuple

Example

>>> resp, err, msg = projects.list_drift_monitorings()
>>> if err:
>>>     print(f"Error: {err}")
>>> else:
>>>     print(f"Drift Monitoring detail : {resp}")
list_experiments(page_size=10, page_number=0)[source]#

List all experiments in the project.

Returns:

A tuple containing: - A list of experiments if the request is successful. - An error message if the request fails.

Return type:

tuple

list_exported_models(page_size=10, page_number=0)[source]#

List all exported models in the project.

Returns:

A tuple containing: - A list of exported models if the request is successful. - An error message if the request fails.

Return type:

tuple

Example

>>> exported_models, error = project.list_exported_models()
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Exported models: {exported_models}")
list_model_train(page_size=10, page_number=0)[source]#

List model training sessions in the project with pagination.

Returns:

A tuple containing: - A paginated list of model training sessions if the request is successful. - An error message if the request fails.

Return type:

tuple

Example

>>> model_train_sessions, error = project.list_model_train()
>>> if error:
>>>     print(f"Error: {error}")
>>> else:
>>>     print(f"Model training sessions: {model_train_sessions}")
refresh()[source]#

Refresh the instance by reinstantiating it with the previous values.

update_permissions(collaborator_id, permissions)[source]#

Update the permissions for a collaborator in the current project.

This function updates the permissions for a specified collaborator in the current project.

Parameters:

  • collaborator_id (str) – The ID of the collaborator whose permissions are to be updated.

  • permissions (list) – A list containing the updated permissions for various project services.

Returns:

A tuple containing three elements: - API response (dict): The raw response from the API. - error_message (str or None): Error message if an error occurred, None otherwise. - status_message (str): A status message indicating success or failure.

Return type:

tuple

Example

>>> collaborator_id = "12345"
>>> permissions = [
...     "v1.0",
...     True,  # isProjectAdmin
...     {"read": True, "write": True, "admin": False},  # datasetsService
...     {"read": True, "write": False, "admin": False},  # modelsService
...     {"read": True, "write": False, "admin": False},  # annotationService
...     {"read": True, "write": False, "admin": False},  # byomService
...     {"read": True, "write": True, "admin": False},  # deploymentService
...     {"read": True, "write": False, "admin": False},  # inferenceService
... ]
>>> resp, err, msg = project.update_permissions(collaborator_id, permissions)
>>> if err:
...     print(f"Error: {err}")
... else:
...     print("Permissions updated successfully")
On this page
Show Source