pub struct SharedRuntime {
pub compress_kv: Vec<f32>,
pub index_k: Vec<f32>,
pub topk_idxs: Vec<i32>,
pub topk: usize,
pub candidates: Vec<bool>,
pub cand_width: usize,
}Expand description
SharedAttentionRuntime: what source layers publish and later layers read.
Fields§
§compress_kv: Vec<f32>the source’s full cache [max_groups][hd]
index_k: Vec<f32>the source’s full index-key cache [max_groups][index_hd]
topk_idxs: Vec<i32>[queries][topk], already offset for the window rows
topk: usize§candidates: Vec<bool>[queries][width]
cand_width: usizeTrait Implementations§
Source§fn default() -> SharedRuntime
fn default() -> SharedRuntime
Returns the “default value” for a type. Read more
Auto Trait Implementations§
Blanket Implementations§
Source§impl<T> BorrowMut<T> for Twhere
T: ?Sized,
impl<T> BorrowMut<T> for Twhere
T: ?Sized,
Source§fn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Mutably borrows from an owned value. Read more