Struct DispatchTensor 

pub struct DispatchTensor {
    pub kind: DispatchTensorKind,
    pub checkpointing: Option<CheckpointingStrategy>,
}

A tensor that can dispatch operations to any enabled backend at runtime.

When the autodiff feature is enabled, tensors may carry a checkpointing strategy used to control gradient computation. This is derived from the device used to create the tensor.

Fields

kind: DispatchTensorKind

Tensor kind primitive.

checkpointing: Option<CheckpointingStrategy>

Holds the autodiff checkpointing strategy.

• None: tensor is not tracked by autodiff
• Some(strategy): tensor is tracked by autodiff, and uses the checkpointing strategy

Trait Implementations

impl Clone for DispatchTensor

fn clone(&self) -> DispatchTensor

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for DispatchTensor

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

Formats the value using the given formatter.

impl DispatchKindConversion<Flex> for DispatchTensor

Available on default_backend or crate feature flex only.

fn try_into_backend( tensor: DispatchTensor, ) -> Result<BackendTensor<Flex>, String>

Attempts to extract a backend-specific BackendTensor wrapper from a generic, dynamically-routed DispatchTensor.

fn from_backend(tensor: BackendTensor<Flex>) -> DispatchTensor

Encapsulates a backend-specific tensor variant back into a globally routing DispatchTensor.

impl TensorMetadata for DispatchTensor

type Device = DispatchDevice

The device type associated with the tensor.

fn dtype(&self) -> DType

Get the dtype of the tensor.

fn shape(&self) -> Shape

Get the shape of the tensor.

fn can_mut(&self) -> bool

Whether the tensor’s buffer can be mutated in place — i.e. this handle uniquely owns it, so an in-place op (slice_assign, an inplace kernel) writes the existing allocation instead of copying it first.

fn device(&self) -> Self::Device

Get the device associated with the tensor.

fn rank(&self) -> usize

Get the number of dimensions of the tensor.

fn scheme(&self) -> QuantScheme

Get the quantization scheme for a quantized float tensor.