Struct dfdx::nn::DropoutOneIn

pub struct DropoutOneIn<const N: usize> { /* private fields */ }

Does nothing as a Module, and calls dropout() as ModuleMut with probability 1.0 / N.

To prevent programmer error, Module and ModuleMut are only implemented for specific tapes:

1. Module requires that the input tensor has a NoneTape. i.e. that gradients are not being tracked.
2. ModuleMut requires that the tensor has a OwnedTape. i.e. that the gradients are being tracked

That means the following will fail to compile:

1. Using Module with OwnedTape fails to compile

let dropout: DropoutOneIn<2> = Default::default();
let x: Tensor1D<5, OwnedTape> = Tensor1D::zeros().trace();
dropout.forward(x);

2. Using ModuleMut with NoneTape fails to compile

let mut dropout: DropoutOneIn<2> = Default::default();
let x: Tensor1D<5, NoneTape> = TensorCreator::zeros();
dropout.forward_mut(x);

Generics:

• N: p is set as 1.0 / N

Examples:

let mut dropout: DropoutOneIn<2> = Default::default();
let t: Tensor2D<2, 5> = Tensor2D::ones();
let r = dropout.forward_mut(t.trace());
assert_eq!(r.data(), &[[2.0, 2.0, 2.0, 0.0, 0.0], [2.0, 2.0, 0.0, 0.0, 2.0]]);

Trait Implementations

impl<const N: usize> CanUpdateWithGradients for DropoutOneIn<N>

fn update<G: GradientProvider>(&mut self, _: &mut G, _: &mut UnusedTensors)

Does nothing.

impl<const N: usize> Clone for DropoutOneIn<N>

fn clone(&self) -> DropoutOneIn<N>

Returns a copy of the value.

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

Performs copy-assignment from source.

impl<const N: usize> Debug for DropoutOneIn<N>

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

Formats the value using the given formatter.

impl<const N: usize> Default for DropoutOneIn<N>

fn default() -> Self

Seeds StdRng with a new seed every time this is called. The seed is initialized with a deterministic value.

impl<const N: usize> LoadFromNpz for DropoutOneIn<N>

fn load<P: AsRef<Path>>(&mut self, path: P) -> Result<(), NpzError>

Loads data from a .npz zip archive at the specified path.

fn read<R>(
    &mut self,
    _filename_prefix: &str,
    _r: &mut ZipArchive<R>
) -> Result<(), NpzError>where
    R: Read + Seek,

Reads this object from a ZipArchive. r with a base filename of filename_prefix.

impl<const N: usize, T: Tensor<Dtype = f32, Tape = NoneTape>> Module<T> for DropoutOneIn<N>

fn forward(&self, input: T) -> Self::Output

Does nothing

type Output = T

The type that this unit produces given Input.

impl<const N: usize, T: Tensor<Dtype = f32, Tape = OwnedTape>> ModuleMut<T> for DropoutOneIn<N>

fn forward_mut(&mut self, input: T) -> Self::Output

Calls dropout() with p=1/N using self.rng.

type Output = T

The type that this unit produces given Input.

impl<const N: usize> ResetParams for DropoutOneIn<N>

fn reset_params<R: Rng>(&mut self, _: &mut R)

Does nothing.

impl<const N: usize> SaveToNpz for DropoutOneIn<N>

fn save<P: AsRef<Path>>(&self, path: P) -> ZipResult<()>

Save this object into the .npz file determined located at path.

fn write<W>(&self, _filename_prefix: &str, _w: &mut ZipWriter<W>) -> ZipResult<()>where
    W: Write + Seek,

Write this object into ZipWriter w with a base filename of filename_prefix.