Struct QuantizedAvgPool2d 

pub struct QuantizedAvgPool2d { /* private fields */ }

Quantized Average Pooling 2D

Uses i16 intermediate precision to accumulate sums before division. Output may have different scale than input due to averaging.

Implementations

impl QuantizedAvgPool2d

pub fn new(kernel_size: usize, stride: usize, padding: usize) -> Self

Create a new quantized average pooling layer

pub fn forward_int8( &self, input: &[u8], input_shape: &[usize], input_scale: f32, input_zero_point: u8, ) -> CnnResult<(Vec<u8>, Vec<usize>, f32, u8)>

Forward pass with INT8 input

Arguments

• input - Quantized u8 input tensor (NHWC layout)
• input_shape - Input shape [N, H, W, C]
• input_scale - Input scale
• input_zero_point - Input zero point

Returns

(output, output_shape, output_scale, output_zero_point)

Trait Implementations

impl Clone for QuantizedAvgPool2d

fn clone(&self) -> QuantizedAvgPool2d

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for QuantizedAvgPool2d

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

Formats the value using the given formatter.