Enum SourceTensorEncoding 

#[non_exhaustive]
pub enum SourceTensorEncoding {
    Safetensors(StoredDtype),
    Gguf {
        ggml_type: GgmlType,
        endian: Endian,
    },
    RecipeOutput(StoredDtype),
}

Physical tensor encoding recorded by an admitted artifact catalog.

This remains distinct from LinearFormat, which describes the format selected for an executable neural operator.

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

Safetensors(StoredDtype)

Scalar storage in a SafeTensors payload.

Gguf

One physical GGML block encoding in a GGUF shard.

Fields

ggml_type: GgmlType

Exact GGML tensor encoding.

endian: Endian

Byte order declared by the containing shard.

RecipeOutput(StoredDtype)

Scalar tensor produced by an admitted architecture recipe before an optional executable-format lowering.

Implementations

impl SourceTensorEncoding

pub fn scalar_dtype(&self) -> Option<StoredDtype>

Returns the scalar dtype when this encoding is an unpacked tensor.

Trait Implementations

impl Clone for SourceTensorEncoding

fn clone(&self) -> SourceTensorEncoding

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SourceTensorEncoding

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

Formats the value using the given formatter.

impl Eq for SourceTensorEncoding

impl PartialEq for SourceTensorEncoding

fn eq(&self, other: &SourceTensorEncoding) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for SourceTensorEncoding