Enum GgmlType 

#[non_exhaustive]
pub enum GgmlType {
    F32,
    F16,
    Q4_0,
    Q4_1,
    Q5_0,
    Q5_1,
    Q8_0,
    Q8_1,
    Q2_K,
    Q3_K,
    Q4_K,
    Q5_K,
    Q6_K,
    Q8_K,
    IQ2_XXS,
    IQ2_XS,
    IQ3_XXS,
    IQ1_S,
    IQ4_NL,
    IQ3_S,
    IQ2_S,
    IQ4_XS,
    I8,
    I16,
    I32,
    I64,
    F64,
    IQ1_M,
    BF16,
}

A GGUF tensor element type, identified by its ggml_type integer ID. Each variant maps 1:1 to a [prism::tensor::dtype] shape.

Variant spellings deliberately mirror the GGML ggml_type enum and the prism::tensor::dtype type names (Q4_0, IQ4_NL, …) rather than Rust camel case, so the mapping reads 1:1 against the authoritative source.

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.

F32

GGML_TYPE_F32 (ID 0) → [F32].

F16

GGML_TYPE_F16 (ID 1) → [F16].

Q4_0

GGML_TYPE_Q4_0 (ID 2) → [Q4_0].

Q4_1

GGML_TYPE_Q4_1 (ID 3) → [Q4_1].

Q5_0

GGML_TYPE_Q5_0 (ID 6) → [Q5_0].

Q5_1

GGML_TYPE_Q5_1 (ID 7) → [Q5_1].

Q8_0

GGML_TYPE_Q8_0 (ID 8) → [Q8_0].

Q8_1

GGML_TYPE_Q8_1 (ID 9) → [Q8_1].

Q2_K

GGML_TYPE_Q2_K (ID 10) → [Q2_K].

Q3_K

GGML_TYPE_Q3_K (ID 11) → [Q3_K].

Q4_K

GGML_TYPE_Q4_K (ID 12) → [Q4_K].

Q5_K

GGML_TYPE_Q5_K (ID 13) → [Q5_K].

Q6_K

GGML_TYPE_Q6_K (ID 14) → [Q6_K].

Q8_K

GGML_TYPE_Q8_K (ID 15) → [Q8_K].

IQ2_XXS

GGML_TYPE_IQ2_XXS (ID 16) → [IQ2_XXS].

IQ2_XS

GGML_TYPE_IQ2_XS (ID 17) → [IQ2_XS].

IQ3_XXS

GGML_TYPE_IQ3_XXS (ID 18) → [IQ3_XXS].

IQ1_S

GGML_TYPE_IQ1_S (ID 19) → [IQ1_S].

IQ4_NL

GGML_TYPE_IQ4_NL (ID 20) → [IQ4_NL].

IQ3_S

GGML_TYPE_IQ3_S (ID 21) → [IQ3_S].

IQ2_S

GGML_TYPE_IQ2_S (ID 22) → [IQ2_S].

IQ4_XS

GGML_TYPE_IQ4_XS (ID 23) → [IQ4_XS].

I8

GGML_TYPE_I8 (ID 24) → [I8].

I16

GGML_TYPE_I16 (ID 25) → [I16].

I32

GGML_TYPE_I32 (ID 26) → [I32].

I64

GGML_TYPE_I64 (ID 27) → [I64].

F64

GGML_TYPE_F64 (ID 28) → [F64].

IQ1_M

GGML_TYPE_IQ1_M (ID 29) → [IQ1_M].

BF16

GGML_TYPE_BF16 (ID 30) → [BF16].

Implementations

impl GgmlType

pub const fn from_u32(id: u32) -> Option<Self>

Map a raw ggml_type integer ID to a GgmlType.

Returns None for IDs outside the GGUF v3 tensor-type set — including the deprecated IDs 4 and 5.

pub const fn id(self) -> u32

The canonical ggml_type integer ID for this dtype.

pub const fn block_bytes(self) -> usize

Bytes per block, sourced from the [prism::tensor::dtype] shape.

pub const fn block_elems(self) -> usize

Elements per block, sourced from the [prism::tensor::dtype] shape.

pub const fn tensor_data_bytes(self, num_elements: u64) -> Option<u64>

The total byte count for a tensor of num_elements of this dtype: (num_elements / BLOCK_ELEMS) * BLOCK_BYTES. Returns None if num_elements is not a whole multiple of BLOCK_ELEMS (a malformed quantized tensor) or on overflow.

Trait Implementations

impl Clone for GgmlType

fn clone(&self) -> GgmlType

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for GgmlType

impl Debug for GgmlType

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

Formats the value using the given formatter.

impl Eq for GgmlType

impl PartialEq for GgmlType

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

Equality operator ==.

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

Inequality operator !=.

impl StructuralPartialEq for GgmlType