core.services.utils.local_context

local_context

Functions:

Name	Description
create_local_processing_context	Create a local processing context for running a processing pipeline outside of Picsellia.
create_local_training_context	Create a local training context for testing model training logic locally.

create_local_processing_context(processing_parameters_cls, api_token, organization_name, job_type, input_dataset_version_id, output_dataset_version_name=None, model_version_id=None, processing_parameters=None, working_dir=None, host=None)

Create a local processing context for running a processing pipeline outside of Picsellia.

This is typically used for development and testing, with full local control over input/output paths and parameter overrides.

Parameters:

Name	Type	Description	Default
processing_parameters_cls	type[TParameters]	A subclass of Parameters used to define typed inputs.	required
api_token	str	API token for authentication with Picsellia.	required
organization_name	str	Name of the Picsellia organization.	required
job_type	ProcessingType	Type of processing job (e.g., PRE_ANNOTATION, DATASET_VERSION_CREATION).	required
input_dataset_version_id	str	ID of the dataset version used as input.	required
output_dataset_version_name	str | None	Optional name for the output dataset version.	None
model_version_id	str | None	Optional ID of a model version to include in the context.	None
processing_parameters	dict[str, Any] | None	Raw values to override defaults in the processing parameters.	None
working_dir	str | None	Optional working directory for local file operations.	None
host	str | None	Optional Picsellia API host override.	None

Returns:

Type	Description
LocalProcessingContext	LocalProcessingContext[TParameters]: A fully initialized local processing context.

create_local_training_context(hyperparameters_cls, augmentation_parameters_cls, export_parameters_cls, api_token, organization_name, experiment_id, working_dir=None, host=None)

Create a local training context for testing model training logic locally.

This context allows for local execution of training steps with parameters pulled from experiment logs.

Parameters:

Name	Type	Description	Default
hyperparameters_cls	type	Class defining training hyperparameters.	required
augmentation_parameters_cls	type	Class defining augmentation strategy parameters.	required
export_parameters_cls	type	Class defining model export configuration.	required
api_token	str	API token to authenticate with Picsellia.	required
organization_name	str	Name of the organization linked to the experiment.	required
experiment_id	str	Experiment ID from which parameter logs are retrieved.	required
working_dir	str | None	Optional local working directory.	None
host	str | None	Optional Picsellia host override.	None

Returns:

Name	Type	Description
LocalTrainingContext	LocalTrainingContext	Fully initialized context for local model training.