Module grad 

Gradient computation support for the array protocol.

This module provides automatic differentiation capabilities for arrays using the array protocol. It enables gradient computation for any array type that implements the ArrayProtocol trait.

Structs

Adam

Adam optimizer.

GradientDict

Dictionary for storing parameter gradients

GradientTensor

Tensor with gradient tracking capabilities.

SGD

Stochastic Gradient Descent optimizer.

Variable

Gradient-aware variable that can be optimized.

Traits

Optimizer

Trait for optimizers that update variables.

Functions

grad_add

Implementations of gradient-aware operations Addition operation with gradient tracking.

grad_divide

Division with gradient tracking.

grad_matmul

Matrix multiplication with gradient tracking.

grad_mean

Mean reduction with gradient tracking.

grad_multiply

Element-wise multiplication with gradient tracking.

grad_sigmoid

Sigmoid activation with gradient tracking.

grad_subtract

Subtraction with gradient tracking.