Package io.github.kirstenali.deepj.autograd

package io.github.kirstenali.deepj.autograd

Class	Description
ActivationOps
AttentionOps
Autograd
AxisReductionOps
BackwardRule	A first-order vector-Jacobian product, in the same order as the operation's inputs.
BinaryOps
BroadcastOps
GradientState	Runtime-only state; never serialized with a tensor or checkpoint.
GradMode	Thread-local, nestable scope for computations that must not build a backward graph.
LayoutOps
NormalizationOps
ReductionOps
SelectionOps
UnaryOps
ViewOps