Engine

Engine

class olive.engine.Engine(config: Dict[str, Any] | EngineConfig | None = None, search_strategy: SearchStrategy | None = None, host: OliveSystem | None = None, target: OliveSystem | None = None, evaluator_config: OliveEvaluatorConfig | None = None, execution_providers: List[str] | None = None)

The engine executes the registered Olive Steps, facilitate evaluation of the output models using provided evaluation criteria and produces output model(s).

register(pass_type: Type[Pass], config: Dict[str, Any] | None = None, disable_search=False, name: str | None = None, host: OliveSystem | None = None, evaluator_config: OliveEvaluatorConfig | None = None, clean_run_cache: bool = False)

Register a pass configuration so that it could be instantiated and executed later.

run(input_model: OliveModel, packaging_config: PackagingConfig | None = None, output_dir: str | None = None, output_name: str | None = None, evaluation_only: bool = False)

Run all the registered Olive passes on the input model and produce one or more candidate models.

if search strategy is None, all passes are run in the order they were registered. Save the final model to {output_dir}/{output_name}_model and json file to {output_dir}/{output_name}_model.json Save evaluation results of the final model, if any, to {output_dir}/{output_name}_metrics.json Return {“model”: final_model_json, “metrics”: evaluation_results}

if search strategy is not None, run the search strategy to find candidate models. TODO: save the results using updated RunResult

if evaluation_only is True, run the evaluation on the input model and return the results.

Note

All parameters that of type ...Config or ConfigBase class can be assigned dictionaries with keys corresponding to the fields of the class.

EngineConfig

pydantic settings olive.engine.EngineConfig

field azureml_client_config: AzureMLClientConfig | None = None

field cache_dir: Path | str = '.olive-cache'

field clean_cache: bool = False

field clean_evaluation_cache: bool = False

field evaluator: OliveEvaluatorConfig = None

field execution_providers: List[str] = None

field host: SystemConfig = None

field packaging_config: PackagingConfig = None

field plot_pareto_frontier: bool = False

field search_strategy: SearchStrategyConfig | bool = None

field target: SystemConfig = None

SearchStrategyConfig

pydantic settings olive.strategy.search_strategy.SearchStrategyConfig

field execution_order: str [Required]

field max_iter: int = None

field max_time: int = None

field output_model_num: int = None

field search_algorithm: str [Required]

field search_algorithm_config: ConfigBase = None

field stop_when_goals_met: bool = False

SystemConfig

pydantic settings olive.systems.system_config.SystemConfig

field config: TargetUserConfig = None

field type: SystemType [Required]

create_system()

OliveEvaluatorConfig

pydantic settings olive.evaluator.olive_evaluator.OliveEvaluatorConfig

field metrics: List[Metric] = []

SearchStrategy

class olive.strategy.search_strategy.SearchStrategy(config: Dict[str, Any] | SearchStrategyConfig)

Search strategy