Struct TensorMetadata 

pub struct TensorMetadata {
    pub size: usize,
    pub shape: Vec<usize>,
    pub dtype: String,
}

Metadata about a tensor source (without the actual tensor data).

Used for validation and introspection without requiring Python GIL.

Fields

size: usize

Number of samples in the tensor (first dimension)

shape: Vec<usize>

Shape of each sample (excluding batch dimension)

dtype: String

Data type as string (e.g., “float32”, “float64”)

Implementations

impl TensorMetadata

pub fn new(size: usize, shape: Vec<usize>) -> Self

Create metadata with default dtype (float32).

pub fn with_dtype(size: usize, shape: Vec<usize>, dtype: &str) -> Self

Create metadata with explicit dtype.

pub fn sample_numel(&self) -> usize

Total number of elements per sample.

pub fn full_shape(&self) -> Vec<usize>

Full shape including batch dimension.

Trait Implementations

impl Clone for TensorMetadata

fn clone(&self) -> TensorMetadata

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TensorMetadata

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

Formats the value using the given formatter.