AdvancedQuantization Configuration

AdvancedQuantization Configuration#

Mobilint SDK qb Compiler: AdvancedQuantization Configuration

Mobilint SDK qb Compiler v0.12.0.0

MCS002-KR

AdvancedQuantization configuration definitions used by Mobilint compiler for model quantization. More...

Classes
class	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig
	Configuration for advanced quantization options used during quantization. More...

Functions
	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.__init__ (self, Optional[OptqConfig] optq=None, Optional[ModConfig] mod=None, Optional[EquivalentTransformationConfig] equivalent_transformation=None, Optional[SearchWeightScaleConfig] search_weight_scale=None, bool optional=optional)
	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.default_config (cls)
	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.from_dict (cls, dict config_dict)
	Create AdvancedQuantizationConfig from dictionary (JSON structure)
	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.to_dict (self)
AdvancedQuantizationConfig	qubee.configs.advanced_quantization_config.get_advanced_quantization_config (Optional[OptqConfig] optq=None, Optional[ModConfig] mod=None, Optional[EquivalentTransformationConfig] equivalent_transformation=None, Optional[SearchWeightScaleConfig] search_weight_scale=None)
	Create AdvancedQuantizationConfig with EquivalentTransformationConfig and SearchWeightScaleConfig.

Variables
bool	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.optional = False
	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.optq = optq or OptqConfig.default_config()
	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.mod = mod or ModConfig.default_config()
tuple	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.equivalent_transformation
tuple	qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.search_weight_scale

Detailed Description

AdvancedQuantization configuration definitions used by Mobilint compiler for model quantization.

Function Documentation

◆ init()

qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.__init__	(		self,
		Optional[OptqConfig]	optq = None,
		Optional[ModConfig]	mod = None,
		Optional[EquivalentTransformationConfig]	equivalent_transformation = None,
		Optional[SearchWeightScaleConfig]	search_weight_scale = None,
		bool	optional = optional )

Parameters

optq	OptqConfig. OPTQ configuration.
mod	ModConfig. MOD configuration.
equivalent_transformation	EquivalentTransformationConfig. Equivalent transformation configuration.
search_weight_scale	SearchWeightScaleConfig. Weight scale search configuration.
optional	bool. Indicates whether this config is optional. Use the default value defined as a class variable.

Definition at line 27 of file advanced_quantization_config.py.

◆ default_config()

qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.default_config ( cls )

Definition at line 53 of file advanced_quantization_config.py.

◆ from_dict()

qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.from_dict	(		cls,
		dict	config_dict )

Create AdvancedQuantizationConfig from dictionary (JSON structure)

Definition at line 63 of file advanced_quantization_config.py.

◆ to_dict()

qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.to_dict ( self )

Definition at line 77 of file advanced_quantization_config.py.

◆ get_advanced_quantization_config()

AdvancedQuantizationConfig qubee.configs.advanced_quantization_config.get_advanced_quantization_config	(	Optional[OptqConfig]	optq = None,
		Optional[ModConfig]	mod = None,
		Optional[EquivalentTransformationConfig]	equivalent_transformation = None,
		Optional[SearchWeightScaleConfig]	search_weight_scale = None )

Create AdvancedQuantizationConfig with EquivalentTransformationConfig and SearchWeightScaleConfig.

Definition at line 88 of file advanced_quantization_config.py.

Variable Documentation

◆ optional

bool qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.optional = False

static

Definition at line 24 of file advanced_quantization_config.py.

◆ optq

qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.optq = optq or OptqConfig.default_config()

Definition at line 43 of file advanced_quantization_config.py.

◆ mod

qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.mod = mod or ModConfig.default_config()

Definition at line 44 of file advanced_quantization_config.py.

◆ equivalent_transformation

tuple qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.equivalent_transformation

Initial value:

=  (
            equivalent_transformation or EquivalentTransformationConfig.default_config()
        )

Definition at line 45 of file advanced_quantization_config.py.

◆ search_weight_scale

tuple qubee.configs.advanced_quantization_config.AdvancedQuantizationConfig.search_weight_scale