Crate ghostflow_autograd 

GhostFlow Autograd - Automatic Differentiation

Tape-based reverse-mode automatic differentiation.

Re-exports

pub use tape::GradTape;

pub use backward::backward;

pub use dynamic_graph::DynamicGraph;

pub use dynamic_graph::DynamicContext;

pub use dynamic_graph::DynamicTensor;

pub use dynamic_graph::GraphNode;

Modules

backward

Backward pass implementation

dynamic_graph

Dynamic Computation Graph

tape

Gradient tape for recording operations

Functions

enable_grad

Enable gradient computation for a tensor

no_grad

Disable gradient computation (inference mode)