Enum PruningStrategy 

pub enum PruningStrategy {
    Magnitude(f64),
    PercentileMagnitude(f64),
    StructuredL1(f64),
    RandomPruning(f64),
    GradualPruning {
        initial_sparsity: f64,
        final_sparsity: f64,
        begin_step: usize,
        end_step: usize,
    },
}

Selects the algorithm used to decide which weights to prune.

Variants

Magnitude(f64)

Zero out every weight whose absolute value is strictly below threshold.

PercentileMagnitude(f64)

Zero out the bottom percentile% of weights ranked by absolute magnitude. percentile must be in [0, 100].

StructuredL1(f64)

Prune entire neurons (rows) whose mean L1 norm is below threshold.

RandomPruning(f64)

Randomly zero out fraction% of weights using the pruner’s seeded PRNG. fraction must be in [0, 1].

GradualPruning

Linearly increase sparsity from initial_sparsity to final_sparsity between begin_step and end_step.

Fields

initial_sparsity: f64

Starting sparsity (fraction in [0, 1]).

final_sparsity: f64

Target sparsity (fraction in [0, 1]).

begin_step: usize

Step at which ramping begins.

end_step: usize

Step at which ramping ends (and final_sparsity is held).

Trait Implementations

impl Clone for PruningStrategy

fn clone(&self) -> PruningStrategy

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for PruningStrategy

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

Formats the value using the given formatter.

impl PartialEq for PruningStrategy

fn eq(&self, other: &PruningStrategy) -> 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 StructuralPartialEq for PruningStrategy