Enum Gating 

pub enum Gating {
    Softmax,
    SoftmaxOfTopK,
    Sigmoid {
        bias: bool,
    },
    SigmoidScaledInput,
}

How a router turns logits into expert weights, and where a weight applies: to its expert’s output, unless the gating says otherwise.

Variants

Softmax

Softmax over all experts, then the top-k.

SoftmaxOfTopK

The top-k logits, then a softmax over them (gpt-oss).

Sigmoid

Sigmoid scores, optionally with a learned selection bias (DeepSeek V3).

Fields

bias: bool

SigmoidScaledInput

The top-k logits, each chosen expert’s INPUT scaled by the sigmoid of its logit and its output summed as it stands (Llama 4; llama.cpp’s weight_before_ffn).

Trait Implementations

impl Clone for Gating

fn clone(&self) -> Gating

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Gating

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

Formats the value using the given formatter.

impl PartialEq for Gating

fn eq(&self, other: &Gating) -> 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 StructuralPartialEq for Gating