BendtorchGpu.RMSNorm: def of bendtorch-gpu@1.3.0.0
Import: import bendtorch-gpu@1.3.0.0/main.bend as BendtorchGpu
Lines 409-411 of bendtorch-gpu@1.3.0.0/main.bend: https://jend.leeeet.dev/read.txt?path=bendtorch-gpu@1.3.0.0/main.bend&start_line=409&end_line=411

import Base
import bend-ml-tensor-array@0.1.18.0/main.bend as TA
import bendtorch@1.3.0.0/main.bend as T

# RMSNorm over the last dimension, its gain trained (main.bend's RMSNorm)
def RMSNorm(+n: Nat, +d: Nat, +eps: F32) -> T.Layer<TA.Mat<1n, d>, TA.Mat<n, d>, TA.Mat<n, d>, TA.Mat<n, d>>:
  T.Layer{p => x => rms_f(n, d, eps, p, x), p => s => dy => g => rms_b(n, d, eps, p, s, dy, g), op => p => g => st => mat_step(1n, d, op, p, g, st), u => TA.Mat.zeros(1n, d), p => T.Mat.dump(1n, d, p), p => xs => off => T.Mat.restore(1n, d, p, xs, off), p => xs => off => T.Mat.store(1n, d, p, xs, off), u => U32.from_nat(d)}

Uses (12):
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=F32
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=Nat
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=U32.from_nat
  https://jend.leeeet.dev/definition.txt?path=bend-ml-tensor-array@0.1.19.0/main.bend&name=Mat
  https://jend.leeeet.dev/definition.txt?path=bend-ml-tensor-array@0.1.19.0/main.bend&name=Mat.zeros
  https://jend.leeeet.dev/definition.txt?path=bendtorch-gpu@1.3.0.0/main.bend&name=mat_step
  https://jend.leeeet.dev/definition.txt?path=bendtorch-gpu@1.3.0.0/main.bend&name=rms_b
  https://jend.leeeet.dev/definition.txt?path=bendtorch-gpu@1.3.0.0/main.bend&name=rms_f
  https://jend.leeeet.dev/definition.txt?path=bendtorch@1.3.0.0/main.bend&name=Layer
  https://jend.leeeet.dev/definition.txt?path=bendtorch@1.3.0.0/main.bend&name=Mat.dump
  https://jend.leeeet.dev/definition.txt?path=bendtorch@1.3.0.0/main.bend&name=Mat.restore
  https://jend.leeeet.dev/definition.txt?path=bendtorch@1.3.0.0/main.bend&name=Mat.store