def lf_tk
import bendtorch-gpu@1.3.0.0/main.bend as BendtorchGpuimport Base
import bend-ml-tensor-gpu@0.1.5.0/main.bend as TG
708# x1 = k; kᵀ written (made here)
709def lf_tk(+t: U32, +l: U32, +c: EAC, x1: Array<F32>, x2: Array<F32>, x3: Array<F32>, x4: Array<F32>, o: Array<F32>) -> Array<F32>:
710 match c:
711 case EAC{+n, +tt, +t32, +h, +g, +gn, +hd, +hd32, +w, +kvw, +sc, +nw, +nk, +tr}: o2_b(tk_loop(TG.share(t, n, l), t, l, t32, U32.div(h, g), hd32, kvw, TG.O2{x1, o}))