def b1_p
import bendtorch-gpu@1.3.0.0/main.bend as BendtorchGpuimport Base
import bendtorch@1.3.0.0/main.bend as T
815def b1_p(+e: U32, +sc: F32, +qa: U32, +ka: U32, +hd: Nat, +t32: U32, v: Array<F32>, d: Array<F32>, p: Array<F32>, g: Array<F32>, r: T.D2) -> S6:
816 match r:
817 case T.D2{q, k, +s}: b1_g(e, q, k, Array.set(F32, p, e, (s * sc : F32)), g, T.dot0(hd, qa, ka, 1, t32, d, v))