pub struct MoeGateParams {
pub hidden_dim: usize,
pub n_experts: usize,
pub top_k: usize,
pub seq_len: usize,
pub rms_eps: f32,
}Expand description
Parameters for MoE gate routing.
Fields§
Hidden state dimension (e.g. 2816 for Gemma 4).
n_experts: usizeTotal number of experts (e.g. 128 for Gemma 4).
top_k: usizeNumber of experts to select (e.g. 8 for Gemma 4).
seq_len: usizeNumber of tokens in the sequence (seq_len >= 1).
rms_eps: f32RMS norm epsilon (e.g. 1e-6).
Auto Trait Implementations§
impl Freeze for MoeGateParams
impl RefUnwindSafe for MoeGateParams
impl Send for MoeGateParams
impl Sync for MoeGateParams
impl Unpin for MoeGateParams
impl UnsafeUnpin for MoeGateParams
impl UnwindSafe for MoeGateParams
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