afe.apis.loaded_net

Source: afe/apis/loaded_net.py

Back to index

Imports

• afe._tvm._tvm_graph_partition.CompileMode
• afe._tvm._utils.create_ir_evaluator
• afe._tvm._utils.run_ir_evaluator
• afe.apis.defines
• afe.apis.defines.ExceptionFuncType
• afe.apis.defines.HistogramEntropyMethod
• afe.apis.defines.HistogramMSEMethod
• afe.apis.defines.HistogramPercentileMethod
• afe.apis.defines.InputValues
• afe.apis.defines.MinMaxMethod
• afe.apis.defines.MovingAverageMinMaxMethod
• afe.apis.defines.QuantizationParams
• afe.apis.defines.SkipCalibration
• afe.apis.defines.gen1_target
• afe.apis.defines.gen2_target
• afe.apis.defines.gen_custom_target
• afe.apis.defines.quantization_scheme
• afe.apis.model.Model
• afe.core.configs.AfeProcessingConfigs
• afe.core.configs.ModelConfigs
• afe.core.configs.OptimizationConfigs
• afe.core.configs.QuantizationPrecision
• afe.core.configs.TransformerConfigs
• afe.core.configs.api_calibration_configs
• afe.core.configs.create_quantization_configs
• afe.driver.statistic.Statistic
• afe.ir.defines.RequantizationMode
• afe.ir.defines.Status
• afe.ir.tensor_type.ScalarType
• afe.ir.tensor_type.scalar_type_from_dtype
• afe.ir.utils.transpose_tensor_according_to_layout_strings
• afe.load.importers.general_importer.ImporterParams
• afe.load.importers.general_importer.ModelFormat
• afe.load.importers.general_importer.default_layout
• afe.load.importers.general_importer.detect_format
• afe.load.importers.general_importer.keras_source
• afe.load.importers.general_importer.onnx_source
• afe.load.importers.general_importer.pytorch_source
• afe.load.importers.general_importer.tensorflow2_source
• afe.load.importers.general_importer.tensorflow_source
• afe.load.importers.general_importer.tflite_source
• copy
• dataclasses
• logging
• numpy as np
• os.path
• sima_utils.common.CustomPlatformParams
• sima_utils.common.Platform
• sima_utils.logging.sima_logger
• tempfile
• typing.Callable
• typing.Iterable
• typing.TypeVar

Constants

• GroundTruth (line 54) [default/value TypeVar('GroundTruth')]

Classes

• LoadedNet (line 252)

  • execute(inputs: InputValues, *, log_level: int = logging.NOTSET) -> list[np.ndarray] (line 287) Decorators: _sanitize_exceptions(ExceptionFuncType.LOADED_NET_EXECUTE).

    Parameters:

    • inputs: type InputValues
    • log_level: type int, default logging.NOTSET

    Returns: list[np.ndarray]

  • quantize(calibration_data: Iterable[InputValues] | None, quantization_config: QuantizationParams, *, automatic_layout_conversion: bool = False, arm_only: bool = False, simulated_arm: bool = False, model_name: str | None = None, any_shape_on_mla: bool = False, log_level: int = logging.NOTSET) -> Model (line 332) Decorators: _sanitize_exceptions(ExceptionFuncType.LOADED_NET_QUANTIZE).

    Parameters:

    • calibration_data: type Iterable[InputValues] | None
    • quantization_config: type QuantizationParams
    • automatic_layout_conversion: type bool, default False
    • arm_only: type bool, default False
    • simulated_arm: type bool, default False
    • model_name: type str | None, default None
    • any_shape_on_mla: type bool, default False
    • log_level: type int, default logging.NOTSET

    Returns: Model

  • quantize_with_accuracy_feedback(calibration_data: Iterable[InputValues], evaluation_data: Iterable[tuple[InputValues, GroundTruth]], quantization_config: QuantizationParams, *, accuracy_score: Statistic[tuple[list[np.ndarray], GroundTruth], float], target_accuracy: float, automatic_layout_conversion: bool = False, max_optimization_steps: int | None = None, model_name: str | None = None, any_shape_on_mla: bool = False, log_level: int = logging.NOTSET) -> Model (line 390) Decorators: _sanitize_exceptions(ExceptionFuncType.LOADED_NET_QUANTIZE).

    Parameters:

    • calibration_data: type Iterable[InputValues]
    • evaluation_data: type Iterable[tuple[InputValues, GroundTruth]]
    • quantization_config: type QuantizationParams
    • accuracy_score: type Statistic[tuple[list[np.ndarray], GroundTruth], float]
    • target_accuracy: type float
    • automatic_layout_conversion: type bool, default False
    • max_optimization_steps: type int | None, default None
    • model_name: type str | None, default None
    • any_shape_on_mla: type bool, default False
    • log_level: type int, default logging.NOTSET

    Returns: Model

  • convert_to_sima_quantization(*, requantization_mode: RequantizationMode = RequantizationMode.sima, model_name: str | None = None, any_shape_on_mla: bool = False, log_level: int = logging.NOTSET) -> Model (line 448) Decorators: _sanitize_exceptions(ExceptionFuncType.LOADED_NET_CONVERT).

    Parameters:

    • requantization_mode: type RequantizationMode, default RequantizationMode.sima
    • model_name: type str | None, default None
    • any_shape_on_mla: type bool, default False
    • log_level: type int, default logging.NOTSET

    Returns: Model

Functions

• load_model(params: ImporterParams, *, target: Platform = gen1_target, custom_param_data: CustomPlatformParams | None = None, log_level: int = logging.NOTSET) -> LoadedNet (line 487) Decorators: _sanitize_exceptions(ExceptionFuncType.LOADED_NET_LOAD).

  Parameters:

  • params: type ImporterParams
  • target: type Platform, default gen1_target
  • custom_param_data: type CustomPlatformParams | None, default None
  • log_level: type int, default logging.NOTSET

  Returns: LoadedNet