TFLiteConfiguration

Options & Configuration · AiDotNet.Deployment.Mobile.TensorFlowLite

Configuration for TensorFlow Lite model export.

For Beginners

TFLiteConfiguration provides AI safety functionality. Default values follow the original paper settings.

Constructors

Constructor Summary
TFLiteConfiguration()

Properties

Property Summary
QuantizationMode QuantizationMode { get; set; } Gets or sets the quantization mode.
TFLiteTargetSpec TargetSpec { get; set; } Gets or sets the target specification for compatibility.
bool EnableConstantFolding { get; set; } Gets or sets whether to enable constant folding (default: true).
bool EnableGpuDelegate { get; set; } Gets or sets whether to enable GPU delegate (default: false).
bool EnableOperatorFusion { get; set; } Gets or sets whether to enable operator fusion (default: true).
bool UseDynamicRangeQuantization { get; set; } Gets or sets whether to use dynamic range quantization (default: false).
bool UseIntegerOnlyQuantization { get; set; } Gets or sets whether to use integer-only quantization (default: false).
bool UseNnapiDelegate { get; set; } Gets or sets whether to use NNAPI delegate for Android (default: false).
bool UseQuantization { get; set; } Gets or sets whether to use post-training quantization (default: true).
bool UseXnnpackDelegate { get; set; } Gets or sets whether to use XNNPACK delegate (default: true).
int NumThreads { get; set; } Gets or sets the number of threads for CPU inference (default: 4).
string ModelDescription { get; set; } Gets or sets the model description.
string ModelName { get; set; } Gets or sets the model name.

Methods

Method Summary
ExportConfiguration ToExportConfiguration() Converts to ExportConfiguration.
TFLiteConfiguration ForAndroid() Creates a configuration for Android deployment.
TFLiteConfiguration ForCPU(int numThreads) Creates a configuration optimized for CPU inference.
TFLiteConfiguration ForIOS() Creates a configuration for iOS deployment.
TFLiteConfiguration ForIntegerOnly() Creates a configuration with full integer quantization.