Struct WeightTensor 

pub struct WeightTensor {
    pub data: Vec<f32>,
    pub shape: Vec<usize>,
    pub name: String,
}

Weight tensor for pruning operations

Represents a named tensor with shape information, suitable for layer-wise or global pruning operations.

Fields

data: Vec<f32>

Weight data in row-major order

shape: Vec<usize>

Shape of the tensor (e.g., [out_channels, in_channels, kernel_h, kernel_w])

name: String

Layer name for identification

Implementations

impl WeightTensor

pub fn new(data: Vec<f32>, shape: Vec<usize>, name: String) -> Self

Creates a new weight tensor

pub fn numel(&self) -> usize

Returns the total number of elements

pub fn is_empty(&self) -> bool

Returns true if the tensor is empty

pub fn validate(&self) -> Result<()>

Validates that shape matches data length

Errors

Returns an error if shape product does not match data length

pub fn sparsity(&self) -> f32

Computes sparsity (fraction of zero weights)

pub fn l1_norm(&self) -> f32

Returns the L1 norm (sum of absolute values)

pub fn l2_norm(&self) -> f32

Returns the L2 norm (Euclidean norm)

pub fn statistics(&self) -> WeightStatistics

Returns statistics about the weight distribution

Trait Implementations

impl Clone for WeightTensor

fn clone(&self) -> WeightTensor

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for WeightTensor

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

Formats the value using the given formatter.