Enum HlaMode 

#[repr(u8)]
pub enum HlaMode {
    Standard = 0,
    Hla = 1,
    Ahla = 2,
}

Attention mode for HLA (Higher-order Linear Attention).

• Standard: SDPA with KV cache (default, backward-compatible).
• Hla: Symmetric second-order linear attention — O(1) per-token memory.
• Ahla: Asymmetric second-order linear attention — lower state cost than symmetric.

Variants

Standard = 0

Hla = 1

Symmetric second-order: SK, CQV, mQ accumulators.

Ahla = 2

Asymmetric second-order: PKV, mK accumulators.

Trait Implementations

impl Clone for HlaMode

fn clone(&self) -> HlaMode

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for HlaMode

impl Debug for HlaMode

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

Formats the value using the given formatter.

impl Default for HlaMode

fn default() -> HlaMode

Returns the “default value” for a type.

impl Eq for HlaMode

impl PartialEq for HlaMode

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

Equality operator ==.

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

Inequality operator !=.

impl StructuralPartialEq for HlaMode