Enum TensorRole 

pub enum TensorRole {
    Embedding,
    LmHead,
    FinalNorm,
    KdaProjection,
    KdaConv,
    KdaDecay,
    KdaGate,
    KdaNorm,
    MlaProjection,
    MlaNorm,
    Indexer,
    IndexerNorm,
    MoeRouter,
    MoeExpert,
    MoeShared,
    DenseFfn,
    LayerNorm,
    HyperConnection,
    MtpProjection,
    MtpNorm,
    Vision,
}

2026-09-25: What a tensor is for.

Variants

Embedding

LmHead

FinalNorm

KdaProjection

2026-09-25: KDA linear-attention projections. The reference checkpoint has 34 KDA layers.

KdaConv

KdaDecay

KdaGate

KdaNorm

MlaProjection

2026-09-25: Sparse-MLA projections, plus every mixer’s o_proj. The reference checkpoint has 12 sparse-MLA layers.

MlaNorm

Indexer

2026-09-25: The DSA top-k indexer of a sparse-MLA layer (self_attn.indexer.*, its k_norm excepted).

IndexerNorm

MoeRouter

MoeExpert

MoeShared

DenseFfn

2026-09-25: Dense FFN: mlp.gate_proj/up_proj/down_proj with no expert index. The reference checkpoint has 3 dense-FFN layers.

LayerNorm

2026-09-25: input_layernorm and post_attention_layernorm.

HyperConnection

MtpProjection

2026-09-25: The MTP layer’s eh_proj. The MTP layer is a numbered layers.# layer; the reference checkpoint has no mtp.0.* tensor.

MtpNorm

Vision

2026-09-25: The vision tower (model.visual.*, model.vision*), not part of the text model.

Implementations

impl TensorRole

pub fn is_norm(self) -> bool

2026-09-25: Norm-family roles. norm_sanity_pass in glm5next_tensor_accounting.rs requires every text tensor whose name contains norm to land in one of them.

pub fn is_text_model(self) -> bool

2026-09-25: Every role except Vision.

Trait Implementations

impl Clone for TensorRole

fn clone(&self) -> TensorRole

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TensorRole

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

Formats the value using the given formatter.

impl Ord for TensorRole

fn cmp(&self, other: &TensorRole) -> Ordering

This method returns an Ordering between self and other.

fn max(self, other: Self) -> Self

where Self: Sized,

Compares and returns the maximum of two values.

fn min(self, other: Self) -> Self

where Self: Sized,

Compares and returns the minimum of two values.

fn clamp(self, min: Self, max: Self) -> Self

where Self: Sized,

Restrict a value to a certain interval.

impl PartialEq for TensorRole

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl PartialOrd for TensorRole

fn partial_cmp(&self, other: &TensorRole) -> Option<Ordering>

This method returns an ordering between self and other values if one exists.

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

Tests less than (for self and other) and is used by the < operator.

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

Tests less than or equal to (for self and other) and is used by the <= operator.

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

Tests greater than (for self and other) and is used by the > operator.

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

Tests greater than or equal to (for self and other) and is used by the >= operator.

impl Copy for TensorRole

impl Eq for TensorRole

impl StructuralPartialEq for TensorRole