Skip to main content

quat_scalar_affine

burn_mamba::mamba3::step_constant

Function quat_scalar_affine 

Source 
fn quat_scalar_affine(
    a_bh11: Tensor<4>,
    b_bh11: Tensor<4>,
    q_bhj4: Tensor<4>,
) -> Tensor<4>
Expand description

a + b·q for per-head scalars a, b ([batch, nheads, 1, 1]) and a quaternion tensor q.