pub struct Q8KActsX4 {
pub na: usize,
pub n_blocks: usize,
pub qs: Vec<i8>,
pub bsums: Vec<i16>,
pub d: Vec<f32>,
}Fields§
§na: usizeReal activations in the quad (≤ 4); rows na..4 are zero padding.
n_blocks: usizeQ8_K super-blocks per activation (n_cols / 256).
qs: Vec<i8>Interleaved quants, n_blocks * 1024 long. Block b, 8-element run
c, quad row a, lane k ↦
qs[b*1024 + c*32 + a*8 + k] = acts[a].q[b*256 + c*8 + k].
bsums: Vec<i16>Folded bsums pairs, n_blocks * 4 * 8 long:
bsums[(b*4 + a)*8 + i] = acts[a].bsums[b*16 + 2i] + acts[a].bsums[b*16 + 2i + 1].
d: Vec<f32>Activation scales, n_blocks * 4 long: d[b*4 + a] = acts[a].d[b].
Auto Trait Implementations§
impl Freeze for Q8KActsX4
impl RefUnwindSafe for Q8KActsX4
impl Send for Q8KActsX4
impl Sync for Q8KActsX4
impl Unpin for Q8KActsX4
impl UnsafeUnpin for Q8KActsX4
impl UnwindSafe for Q8KActsX4
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