Burn Tensor

Burn Tensor Library

This library provides the core abstractions required to run tensor operations with Burn.

Tensor<D, K> is generic over rank and tensor kind. Its Device selects the backend at runtime; operations pass through an opaque bridge and dispatch to backend primitives. With the autodiff feature, create tensors on device.autodiff() and call require_grad() on source leaves whose gradients you need. Models and tensor functions do not need a backend type parameter.