Trait FloatProtoElem 

pub trait FloatProtoElem: Copy + 'static {
    // Required methods
    fn slice_into_tensor_dyn(
        values: &[Self],
        shape: &[usize],
    ) -> Result<TensorDyn>;
    fn arrayview3_into_tensor_dyn(
        view: ArrayView3<'_, Self>,
    ) -> Result<TensorDyn>;
}

Per-float-dtype construction of an edgefirst_tensor::TensorDyn from a flat slice or a 3-D ArrayView. Each implementor either passes its element type straight to Tensor::from_slice / Tensor::from_arrayview3, or narrows f64 to f32 — there is no native f64 kernel path, so f64 protos lose precision here rather than at some later, less obvious point.

Implemented for f32, f64, and half::f16. It is public because it appears in the bounds of the proto-extraction paths; there is no reason to implement it outside this crate.

Required Methods

fn slice_into_tensor_dyn(values: &[Self], shape: &[usize]) -> Result<TensorDyn>

Wrap a flat slice with the given shape as a tensor of this element type.

fn arrayview3_into_tensor_dyn(view: ArrayView3<'_, Self>) -> Result<TensorDyn>

Wrap a 3-D ArrayView (the proto-mask layout) as a tensor of this element type.

Dyn Compatibility

This trait is not dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementations on Foreign Types

impl FloatProtoElem for f16

fn slice_into_tensor_dyn(values: &[f16], shape: &[usize]) -> Result<TensorDyn>

fn arrayview3_into_tensor_dyn(view: ArrayView3<'_, f16>) -> Result<TensorDyn>

impl FloatProtoElem for f32

fn slice_into_tensor_dyn(values: &[f32], shape: &[usize]) -> Result<TensorDyn>

fn arrayview3_into_tensor_dyn(view: ArrayView3<'_, f32>) -> Result<TensorDyn>

impl FloatProtoElem for f64

fn slice_into_tensor_dyn(values: &[f64], shape: &[usize]) -> Result<TensorDyn>

fn arrayview3_into_tensor_dyn(view: ArrayView3<'_, f64>) -> Result<TensorDyn>

Implementors