Struct LotteryTicketPruner 

pub struct LotteryTicketPruner { /* private fields */ }

Lottery Ticket Hypothesis pruner.

Implements Iterative Magnitude Pruning (IMP) with weight rewinding to find sparse, trainable subnetworks.

Implementations

impl LotteryTicketPruner

pub fn new() -> Self

Create a new LTH pruner with default configuration. Default: 90% sparsity over 10 rounds with init rewinding.

pub fn with_config(config: LotteryTicketConfig) -> Self

Create a pruner with custom configuration.

pub fn builder() -> LotteryTicketPrunerBuilder

Get a builder for configuring the pruner.

pub fn config(&self) -> &LotteryTicketConfig

Get the configuration.

pub fn find_ticket( &self, module: &dyn Module, ) -> Result<WinningTicket, PruningError>

Find a winning ticket from the given module.

This performs iterative magnitude pruning:

1. Compute importance scores
2. Generate mask for current round’s pruning rate
3. Accumulate into overall mask
4. Record weight state for rewinding

Note: This is a simplified version that doesn’t require training. For full LTH, use find_ticket_with_training.

pub fn apply_ticket( &self, module: &mut dyn Module, ticket: &WinningTicket, ) -> Result<PruningResult, PruningError>

Apply a winning ticket mask to a module.

This zeros out the pruned weights according to the mask.

Trait Implementations

impl Clone for LotteryTicketPruner

fn clone(&self) -> LotteryTicketPruner

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LotteryTicketPruner

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

Formats the value using the given formatter.

impl Default for LotteryTicketPruner

fn default() -> Self

Returns the “default value” for a type.

impl Pruner for LotteryTicketPruner

fn generate_mask( &self, scores: &ImportanceScores, target_sparsity: f32, pattern: SparsityPattern, ) -> Result<SparsityMask, PruningError>

Generate a sparsity mask based on importance scores.

fn apply_mask( &self, module: &mut dyn Module, mask: &SparsityMask, ) -> Result<PruningResult, PruningError>

Apply a sparsity mask to a module, zeroing pruned weights.

fn importance(&self) -> &dyn Importance

Get the importance estimator used by this pruner.

fn name(&self) -> &'static str

Name of this pruner for logging.