Enum DataType 

#[repr(u8)]
pub enum DataType {
    Undefined = 0,
    Float32 = 1,
    Uint8 = 2,
    Int8 = 3,
    Uint16 = 4,
    Int16 = 5,
    Int32 = 6,
    Int64 = 7,
    String = 8,
    Bool = 9,
    Float16 = 10,
    Float64 = 11,
    Uint32 = 12,
    Uint64 = 13,
    Complex64 = 14,
    Complex128 = 15,
    BFloat16 = 16,
    Float8E4M3FN = 17,
    Float8E4M3FNUZ = 18,
    Float8E5M2 = 19,
    Float8E5M2FNUZ = 20,
    Uint4 = 21,
    Int4 = 22,
    Float4E2M1 = 23,
    Float8E8M0 = 24,
    Uint2 = 25,
    Int2 = 26,
}

Supported tensor element types.

The discriminant values match ONNX TensorProto.DataType so that the loader can cast the protobuf integer directly (see DataType::from_onnx).

Variants

Undefined = 0

Float32 = 1

Uint8 = 2

Int8 = 3

Uint16 = 4

Int16 = 5

Int32 = 6

Int64 = 7

String = 8

Bool = 9

Float16 = 10

Float64 = 11

Uint32 = 12

Uint64 = 13

Complex64 = 14

Complex128 = 15

BFloat16 = 16

Float8E4M3FN = 17

Float8E4M3FNUZ = 18

Float8E5M2 = 19

Float8E5M2FNUZ = 20

Uint4 = 21

Int4 = 22

Float4E2M1 = 23

Float8E8M0 = 24

Uint2 = 25

Int2 = 26

Implementations

impl DataType

pub fn byte_size(self) -> usize

Byte size per element. Sub-byte (packed) and variable-width types return 0; use DataType::bit_size for those.

pub fn bit_size(self) -> usize

Bit size per element. Sub-byte types return their true width (e.g. 4).

pub fn is_float(self) -> bool

Whether this is a floating-point type (any precision).

pub fn is_int(self) -> bool

Whether this is a signed or unsigned integer type (excludes Bool).

pub fn is_complex(self) -> bool

Whether elements contain real and imaginary floating-point components.

pub fn is_sub_byte(self) -> bool

Whether elements are packed multiple-per-byte.

pub fn storage_bytes(self, count: usize) -> usize

Number of bytes needed to store count elements, accounting for sub-byte packing.

Panics on usize overflow of the count * byte_size product. Callers that size heap allocations from an untrusted shape MUST use DataType::checked_storage_bytes instead so a crafted static shape cannot wrap the multiply and under-allocate.

pub fn checked_storage_bytes(self, count: usize) -> Option<usize>

Number of bytes needed to store count elements, returning None when the count * byte_size product overflows usize.

This is the overflow-safe counterpart of DataType::storage_bytes: even though an element count may fit in usize, the element-count → bytes multiply can still wrap for a fixed-width dtype (e.g. 2^61 elements of an 8-byte type). Allocation sites route through this so a wrapped product becomes a clean error instead of a 1-byte allocation followed by an out-of-bounds access.

pub fn from_onnx(raw: i32) -> Option<Self>

Convert from the raw ONNX TensorProto.DataType integer.

Returns None for UNDEFINED (0) and any out-of-range or future value the runtime does not model. The discriminants below mirror the vendored onnx.proto3 TensorProto.DataType enum verbatim.

pub fn to_onnx(self) -> i32

The raw ONNX TensorProto.DataType integer for this type.

Trait Implementations

impl Clone for DataType

fn clone(&self) -> DataType

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for DataType

impl Debug for DataType

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

Formats the value using the given formatter.

impl Eq for DataType

impl Hash for DataType

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for DataType

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

Equality operator ==.

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

Inequality operator !=.

impl StructuralPartialEq for DataType