Struct EfQuantizationInfo 

#[repr(C)]
pub struct EfQuantizationInfo {
    pub axis: i32,
    pub count: u32,
    pub has_quantization: u32,
}

Presence/shape summary of a tensor’s quantization metadata.

The first half of the two-call idiom ef_tensor_quantization_info / ef_tensor_quantization_get use: this scalar block tells the caller whether quantization is attached and how big a buffer the second call needs, without allocating on either side of the boundary – Quantization itself is variable-length (an axis plus per-axis scales and zero_points), so unlike ef_tensor_plane/ef_tensor_view_origin this view cannot carry the payload itself.

has_quantization is a presence flag, not a sentinel, for the same reason EfViewOrigin::has_origin is: axis 0 and a scale of 0.0 are both legitimate values, so there is no unused bit pattern to repurpose as “absent” without colliding with a real one.

Fields

axis: i32

Channel axis for per-channel quantization; -1 for per-tensor (no axis). Meaningful only when has_quantization != 0.

count: u32

Number of entries in the scale/zero_point arrays (1 for per-tensor). Meaningful only when has_quantization != 0.

has_quantization: u32

Non-zero when this tensor carries quantization metadata; 0 when it does not, in which case axis and count are both zeroed and unused.

Trait Implementations

impl Clone for EfQuantizationInfo

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for EfQuantizationInfo

impl Debug for EfQuantizationInfo

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

Formats the value using the given formatter.

impl Default for EfQuantizationInfo

fn default() -> Self

Returns the “default value” for a type.