Enum QuantizationPrecision 

pub enum QuantizationPrecision {
    FP16,
    FP8,
    INT8,
    INT4,
    INT2,
    INT1,
    Mixed,
    Adaptive,
}

Quantization precision levels

Variants

FP16

16-bit floating point

FP8

8-bit floating point (E4M3 or E5M2 format)

INT8

8-bit integer

INT4

4-bit integer

INT2

2-bit integer (experimental)

INT1

1-bit binary quantization

Mixed

Mixed precision (FP16 for outliers, INT8 for normal weights)

Adaptive

Adaptive precision based on layer importance

Implementations

impl QuantizationPrecision

pub fn bits(&self) -> u32

Number of bits used to represent one quantized element for this precision.

Mixed (FP16 for a minority of outlier weights, INT8 for the rest) and Adaptive (a runtime-selected, per-layer bit-width) do not have one single fixed bit-width by design. Both nominally report 8 bits here as a documented approximation used for size/statistics purposes: for Mixed, INT8 is the majority-path width (outliers are expected to be a small minority of elements); for Adaptive, INT8 is the default/starting width used before any runtime adaptation narrows individual layers further. This is intentionally explicit (not a silently-defaulted wildcard arm).

pub fn bytes_per_element(&self) -> f32

Storage size in bytes per element (fractional for sub-byte widths like INT4/INT2/INT1).

Trait Implementations

impl Clone for QuantizationPrecision

fn clone(&self) -> QuantizationPrecision

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for QuantizationPrecision

impl Debug for QuantizationPrecision

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for QuantizationPrecision

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Eq for QuantizationPrecision

impl From<QuantizationPrecision> for JsValue

fn from(value: QuantizationPrecision) -> Self

Converts to this type from the input type.

impl FromWasmAbi for QuantizationPrecision

type Abi = u32

The Wasm ABI type that this converts from when coming back out from the ABI boundary.

unsafe fn from_abi(js: u32) -> Self

Recover a Self from Self::Abi.

impl IntoWasmAbi for QuantizationPrecision

type Abi = u32

The Wasm ABI type that this converts into when crossing the ABI boundary.

fn into_abi(self) -> u32

Convert self into Self::Abi so that it can be sent across the wasm ABI boundary.

impl OptionFromWasmAbi for QuantizationPrecision

fn is_none(val: &Self::Abi) -> bool

Tests whether the argument is a “none” instance. If so it will be deserialized as None, and otherwise it will be passed to FromWasmAbi.

impl OptionIntoWasmAbi for QuantizationPrecision

fn none() -> Self::Abi

Returns an ABI instance indicating “none”, which JS will interpret as the None branch of this option.

impl PartialEq for QuantizationPrecision

fn eq(&self, other: &QuantizationPrecision) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for QuantizationPrecision

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for QuantizationPrecision

impl TryFromJsValue for QuantizationPrecision

fn try_from_js_value_ref(value: &JsValue) -> Option<Self>

Performs the conversion.

fn try_from_js_value(value: JsValue) -> Result<Self, JsValue>

Performs the conversion.

impl VectorFromWasmAbi for QuantizationPrecision

type Abi = <Box<[JsValue]> as FromWasmAbi>::Abi

unsafe fn vector_from_abi(js: Self::Abi) -> Box<[QuantizationPrecision]>

impl VectorIntoWasmAbi for QuantizationPrecision

type Abi = <Box<[JsValue]> as IntoWasmAbi>::Abi

fn vector_into_abi(vector: Box<[QuantizationPrecision]>) -> Self::Abi

impl WasmDescribe for QuantizationPrecision

fn describe()

impl WasmDescribeVector for QuantizationPrecision

fn describe_vector()