Struct Latent 

pub struct Latent {

    pub qr: String,
    pub qpp: String,
    pub kc: String,
    pub pos: Option<String>,
    pub heads: usize,
    pub kw: usize,
    pub lora: usize,
    pub nope: usize,
    pub len: usize,
    pub rows: usize,
    pub scale: f64,
    pub wkb: String,
    pub fk: GgmlType,
    pub kro: bool,
    pub wvb: String,
    pub fv: GgmlType,
    pub vd: usize,
    pub ct: String,
}

A latent attention’s shape and names, as its expanded launches read them: the query projection qr (each head’s unrotated part its first nope values) and the rotated part qpp; the cache kc, kw wide a row — the latent lora values then the shared rotated key — len rows; the key and value halves of the latent’s expansion and their formats (kro: the key half’s checkpoint rows are its outputs); the result ct.

Fields

qr: String

qpp: String

kc: String

pos: Option<String>

heads: usize

kw: usize

lora: usize

nope: usize

len: usize

rows: usize

scale: f64

wkb: String

fk: GgmlType

kro: bool

wvb: String

fv: GgmlType

vd: usize

ct: String