pub struct LocalResponseNorm {
pub size: usize,
pub alpha: f64,
pub beta: f64,
pub k: f64,
}Expand description
Local Response Normalization (LRN).
Normalises within a local neighbourhood across channels:
b[n,c,h,w] = a[n,c,h,w] / (k + alpha/size * sum(a[n,j,h,w]^2))^beta
where j runs over [max(0,c-size/2), min(C,c+size/2+1)).
Matches torch.nn.LocalResponseNorm(size, alpha, beta, k).
Fields§
§size: usizeNumber of adjacent channels to normalise across.
alpha: f64Scaling factor in the denominator.
beta: f64Exponent.
k: f64Additive constant for numerical stability.
Implementations§
Trait Implementations§
Source§impl Clone for LocalResponseNorm
impl Clone for LocalResponseNorm
Source§impl Debug for LocalResponseNorm
impl Debug for LocalResponseNorm
Source§impl<T: Float + FloatElement + Neg<Output = T> + RealScalar, B: BackendOps<T> + Default> Module<T, B> for LocalResponseNorm
impl<T: Float + FloatElement + Neg<Output = T> + RealScalar, B: BackendOps<T> + Default> Module<T, B> for LocalResponseNorm
Source§fn forward(&self, input: &Var<T, B>) -> Result<Var<T, B>, ModuleError<B::Error>>
fn forward(&self, input: &Var<T, B>) -> Result<Var<T, B>, ModuleError<B::Error>>
LRN forward pass — differentiable.
Supports 2D ([N, C]), 3D ([N, C, L]), and 4D ([N, C, H, W]) inputs.
Built entirely from autograd ops so gradients flow to the input: the
cross-channel windowed sum-of-squares is a constant band-matrix product
(differentiable through the squared activations), and the ^beta
denominator uses the differentiable pow (the base k + .. >= k > 0,
so the exp(beta*ln(.)) it computes is well defined).
Source§fn parameters(&self) -> Vec<Var<T, B>>
fn parameters(&self) -> Vec<Var<T, B>>
Source§fn named_parameters(&self) -> Vec<Parameter<T, B>>
fn named_parameters(&self) -> Vec<Parameter<T, B>>
Source§fn load_parameters(&mut self, _params: &[Var<T, B>])
fn load_parameters(&mut self, _params: &[Var<T, B>])
params in the same order parameters() enumerates
them (i.e. params.len() == parameters().len()). Read moreSource§fn load_named_parameters(
&mut self,
parameters: &[Parameter<T, B>],
) -> Result<(), ParameterLoadError>
fn load_named_parameters( &mut self, parameters: &[Parameter<T, B>], ) -> Result<(), ParameterLoadError>
Auto Trait Implementations§
impl Forget for LocalResponseNorm
impl Freeze for LocalResponseNorm
impl Move for LocalResponseNorm
impl RefUnwindSafe for LocalResponseNorm
impl Send for LocalResponseNorm
impl Sync for LocalResponseNorm
impl Unpin for LocalResponseNorm
impl UnsafeUnpin for LocalResponseNorm
impl UnwindSafe for LocalResponseNorm
Blanket Implementations§
Source§impl<T> ArchivePointee for T
impl<T> ArchivePointee for T
Source§type ArchivedMetadata = ()
type ArchivedMetadata = ()
Source§fn pointer_metadata(
_: &<T as ArchivePointee>::ArchivedMetadata,
) -> <T as Pointee>::Metadata
fn pointer_metadata( _: &<T as ArchivePointee>::ArchivedMetadata, ) -> <T as Pointee>::Metadata
Source§impl<T> BorrowMut<T> for Twhere
T: ?Sized,
impl<T> BorrowMut<T> for Twhere
T: ?Sized,
Source§fn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Source§impl<T> CloneToUninit for Twhere
T: Clone,
impl<T> CloneToUninit for Twhere
T: Clone,
Source§impl<T> Instrument for T
impl<T> Instrument for T
Source§fn instrument(self, span: Span) -> Instrumented<Self> ⓘ
fn instrument(self, span: Span) -> Instrumented<Self> ⓘ
Source§fn in_current_span(self) -> Instrumented<Self> ⓘ
fn in_current_span(self) -> Instrumented<Self> ⓘ
Source§impl<T> LayoutRaw for T
impl<T> LayoutRaw for T
Source§fn layout_raw(_: <T as Pointee>::Metadata) -> Result<Layout, LayoutError>
fn layout_raw(_: <T as Pointee>::Metadata) -> Result<Layout, LayoutError>
Source§impl<T, N1, N2> Niching<NichedOption<T, N1>> for N2
impl<T, N1, N2> Niching<NichedOption<T, N1>> for N2
Source§unsafe fn is_niched(niched: *const NichedOption<T, N1>) -> bool
unsafe fn is_niched(niched: *const NichedOption<T, N1>) -> bool
Source§fn resolve_niched(out: Place<NichedOption<T, N1>>)
fn resolve_niched(out: Place<NichedOption<T, N1>>)
out indicating that a T is niched.