Struct Value 

pub struct Value<V>
where
    V: Differentiable,
{ /* private fields */ }

Public value handle for reverse-mode AD.

Value exposes a torch-like surface while keeping graph ownership hidden. Internally it carries either a leaf gradient sink or an edge into a reverse graph.

Implementations

impl<V> Value<V>

where V: Differentiable + 'static,

pub fn new(primal: V) -> Value<V>

Create a detached value.

pub fn primal(&self) -> &V

Borrow the primal value.

pub fn requires_grad(&self) -> bool

Return whether this value participates in reverse-mode AD.

pub fn with_requires_grad(self, enabled: bool) -> Value<V>

Return a new value handle with gradient tracking enabled or disabled.

pub fn grad( &self, ) -> Result<Option<<V as Differentiable>::Tangent>, AutodiffError>

where <V as Differentiable>::Tangent: Clone,

Read the accumulated leaf gradient, if available.

pub fn zero_grad(&self) -> Result<(), AutodiffError>

Clear the accumulated leaf gradient.

pub fn backward(&self) -> Result<(), AutodiffError>

where <V as Differentiable>::Tangent: Clone,

Run reverse-mode backward with the default scalar cotangent seed.

pub fn backward_with_seed( &self, seed: <V as Differentiable>::Tangent, ) -> Result<(), AutodiffError>

where <V as Differentiable>::Tangent: Clone,

Run reverse-mode backward with an explicit cotangent seed.

pub fn grad_wrt_with_seed( &self, seed: <V as Differentiable>::Tangent, wrt: &[&Value<V>], ) -> Result<Vec<Option<<V as Differentiable>::Tangent>>, AutodiffError>

where <V as Differentiable>::Tangent: Clone,

Compute functional gradients with respect to the requested inputs.

pub fn shares_reverse_graph(&self, other: &Value<V>) -> bool

Return whether self and other share any reachable reverse graph.