Struct GradientCheckConfig 

pub struct GradientCheckConfig<T: Float> {
    pub epsilon: T,
    pub relative_tolerance: T,
    pub absolute_tolerance: T,
    pub use_centered_differences: bool,
    pub max_params_to_check: Option<usize>,
    pub random_seed: Option<u64>,
}

Numerical gradient checking utilities for neural networks

This module provides tools for validating analytical gradients by comparing them with numerically computed gradients using finite differences. Gradient checking configuration

Fields

epsilon: T

Finite difference step size (epsilon)

relative_tolerance: T

Relative tolerance for gradient comparison

absolute_tolerance: T

Absolute tolerance for gradient comparison

use_centered_differences: bool

Whether to use centered differences (more accurate but 2x slower)

max_params_to_check: Option<usize>

Maximum number of parameters to check (for efficiency)

random_seed: Option<u64>

Random seed for parameter sampling

Trait Implementations

impl<T: Clone + Float> Clone for GradientCheckConfig<T>

fn clone(&self) -> GradientCheckConfig<T>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<T: Debug + Float> Debug for GradientCheckConfig<T>

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

Formats the value using the given formatter.

impl<T: Float> Default for GradientCheckConfig<T>

fn default() -> Self

Returns the “default value” for a type.