Struct DropBlock 

pub struct DropBlock { /* private fields */ }

DropBlock regularization (Ghiasi et al., 2018).

Drops contiguous regions (blocks) rather than individual elements. More effective for convolutional networks than standard dropout.

Reference

Ghiasi, G., et al. (2018). DropBlock: A regularization technique for CNNs.

Implementations

impl DropBlock

pub fn new(block_size: usize, p: f32) -> Self

Create DropBlock with given block size and drop probability.

pub fn with_seed(block_size: usize, p: f32, seed: u64) -> Self

Create with specific seed.

pub fn block_size(&self) -> usize

pub fn p(&self) -> f32

Trait Implementations

impl Debug for DropBlock

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

Formats the value using the given formatter.

impl Module for DropBlock

fn forward(&self, input: &Tensor) -> Tensor

Perform forward computation.

fn train(&mut self)

Set the module to training mode.

fn eval(&mut self)

Set the module to evaluation mode.

fn training(&self) -> bool

Check if the module is in training mode.

fn parameters(&self) -> Vec<&Tensor>

Get references to all learnable parameters.

fn parameters_mut(&mut self) -> Vec<&mut Tensor>

Get mutable references to all learnable parameters.

fn refresh_caches(&mut self)

Refresh any cached computations after parameters have been modified.

fn zero_grad(&mut self)

Zero out gradients for all parameters.

fn num_parameters(&self) -> usize

Get the number of learnable parameters.