Enum LmHeadChoice 

pub enum LmHeadChoice {
    EngineDefault,
    Declared(LmHeadFormat),
    PendingFp8Kernel,
}

2026-09-28: What --lm-head-dtype default resolves to (WeightQuantPolicy::lm_head).

Variants

EngineDefault

2026-09-28: The engine’s per-model default head: under nvfp4, or with nothing declared.

Declared(LmHeadFormat)

2026-09-28: The head format the checkpoint declares.

PendingFp8Kernel

2026-09-28: The checkpoint declares an FP8 head but the batched FP8 head kernel (KernelCaps::fp8_lm_head_batched) is absent: the engine’s default head, which is above or below the declared precision depending on the model, until it lands.

Trait Implementations

impl Clone for LmHeadChoice

fn clone(&self) -> LmHeadChoice

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LmHeadChoice

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

Formats the value using the given formatter.

impl PartialEq for LmHeadChoice

fn eq(&self, other: &LmHeadChoice) -> 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 Copy for LmHeadChoice

impl Eq for LmHeadChoice

impl StructuralPartialEq for LmHeadChoice