Struct QuantizationConfig 

pub struct QuantizationConfig {
    pub qtype: QuantizationType,
    pub granularity: QuantizationGranularity,
    pub scheme: QuantizationScheme,
    pub calibration: CalibrationMethod,
    pub quantize_weights: bool,
    pub quantize_activations: bool,
    pub quantize_biases: bool,
    pub skip_layers: Vec<String>,
    pub force_fp32_layers: Vec<String>,
}

Quantization configuration

Fields

qtype: QuantizationType

Target quantization type

granularity: QuantizationGranularity

Quantization granularity

scheme: QuantizationScheme

Quantization scheme

calibration: CalibrationMethod

Calibration method

quantize_weights: bool

Whether to quantize weights

quantize_activations: bool

Whether to quantize activations

quantize_biases: bool

Whether to quantize biases (usually kept in higher precision)

skip_layers: Vec<String>

Layers to skip quantization (by name)

force_fp32_layers: Vec<String>

Force specific layers to use FP32 (e.g., output layer)

Implementations

impl QuantizationConfig

pub fn new(qtype: QuantizationType) -> Self

Create a new quantization configuration

pub fn with_granularity(self, granularity: QuantizationGranularity) -> Self

Set quantization granularity

pub fn with_scheme(self, scheme: QuantizationScheme) -> Self

Set quantization scheme

pub fn with_calibration(self, calibration: CalibrationMethod) -> Self

Set calibration method

pub fn skip_layer(self, layer_name: String) -> Self

Add layer to skip

pub fn force_fp32_layer(self, layer_name: String) -> Self

Add layer to force FP32

Trait Implementations

impl Clone for QuantizationConfig

fn clone(&self) -> QuantizationConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for QuantizationConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for QuantizationConfig

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for QuantizationConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for QuantizationConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.