Struct CrossEntropyLoss

pub struct CrossEntropyLoss;

Cross-Entropy Loss with softmax

Trait Implementations

impl LossFunction for CrossEntropyLoss

fn compute_loss(&self, predictions: &Array2<f64>, targets: &Array2<f64>) -> f64

Compute the loss between predictions and targets

fn compute_gradient( &self, predictions: &Array2<f64>, targets: &Array2<f64>, ) -> Array2<f64>

Compute the gradient of the loss with respect to predictions

fn compute_batch_loss( &self, predictions: &Array2<f64>, targets: &Array2<f64>, ) -> f64

Compute batch loss for multiple predictions and targets Default implementation computes average loss across batch

fn compute_batch_gradient( &self, predictions: &Array2<f64>, targets: &Array2<f64>, ) -> Array2<f64>

Compute batch gradients for multiple predictions and targets Default implementation computes gradients for each sample and concatenates