Enum TensorFiniteValidation 

pub enum TensorFiniteValidation {
    Strict,
    OutputOnly,
    Trusted,
}

How aggressively a selector validates that f32 values are finite.

Finiteness validation is a full pass over the tensor copy. For a ReadWriteF32 selector the strict policy scans the native values and the committed output, so a large mutable capture pays for two passes per callback. Relax the policy when the caller already trusts the data.

Variants

Strict

Reject a non-finite native value before the handler runs and a non-finite handler output before commit. This is the default.

OutputOnly

Skip the pre-handler scan; still reject a non-finite committed output.

Trusted

Skip both scans. The caller guarantees finite values.

Trait Implementations

impl Clone for TensorFiniteValidation

fn clone(&self) -> TensorFiniteValidation

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for TensorFiniteValidation

impl Debug for TensorFiniteValidation

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for TensorFiniteValidation

fn default() -> TensorFiniteValidation

Returns the “default value” for a type.

impl Eq for TensorFiniteValidation

impl PartialEq for TensorFiniteValidation

fn eq(&self, other: &TensorFiniteValidation) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for TensorFiniteValidation