pub struct QuantWeights<'a, B: Backend> {
pub qweight: &'a B::Buffer,
pub scales: Option<&'a B::Buffer>,
pub zeros: Option<&'a B::Buffer>,
pub g_idx: Option<&'a B::Buffer>,
}Expand description
Packed quantized weight buffers passed to Backend::gemm_quant.
Not every field is used by every QuantKind — e.g. GGUF packs scales
inside qweight, so scales / zeros may be dummies. The Backend
implementation is expected to validate the shape for the kind it handles.
Fields§
§qweight: &'a B::Buffer§scales: Option<&'a B::Buffer>§zeros: Option<&'a B::Buffer>§g_idx: Option<&'a B::Buffer>Auto Trait Implementations§
impl<'a, B> Freeze for QuantWeights<'a, B>
impl<'a, B> RefUnwindSafe for QuantWeights<'a, B>
impl<'a, B> Send for QuantWeights<'a, B>
impl<'a, B> Sync for QuantWeights<'a, B>
impl<'a, B> Unpin for QuantWeights<'a, B>
impl<'a, B> UnsafeUnpin for QuantWeights<'a, B>
impl<'a, B> UnwindSafe for QuantWeights<'a, B>
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