Module tensor 

Tensor types and operations

This module provides the core Tensor type, which represents an n-dimensional array stored on a compute device (CPU, GPU, etc.).

Modules

id

Tensor ID generation for autograd graph tracking

Structs

Layout

Layout describes the memory layout of a tensor

Shape

Shape type: dimensions of a tensor

Storage

Storage for tensor data on a device

Strides

Strides type: element offsets between consecutive elements along each dimension Signed to support negative strides (e.g., for flip operations) NOTE: Strides are in ELEMENTS, not bytes

Tensor

N-dimensional array stored on a compute device

TensorId

Unique identifier for a tensor