BendMlAutograd.ce_grad: def of bend-ml-autograd@0.1.1.0
Import: import bend-ml-autograd@0.1.1.0/main.bend as BendMlAutograd
Lines 291-293 of bend-ml-autograd@0.1.1.0/main.bend: https://jend.leeeet.dev/read.txt?path=bend-ml-autograd@0.1.1.0/main.bend&start_line=291&end_line=293

import Base
import bend-ml-tensor@0.1.0.0/main.bend as T

# gradient of the loss with respect to the logits: (softmax - one-hot) / n
def ce_grad(+n: Nat, +c: Nat, logits: T.Mat<n, c>, labels: List<&2, Nat>) -> T.Mat<n, c>:
  T.Mat.scale(n, c, (1.0 / F32.from_nat(n) : F32), T.Mat.sub(n, c, T.Mat.softmax(n, c, logits), T.Mat{onehot.rows(labels, c)}))

Uses (10):
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=F32
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=F32.div
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=F32.from_nat
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=List
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=Nat
  https://jend.leeeet.dev/definition.txt?path=bend-ml-autograd@0.1.1.0/main.bend&name=onehot.rows
  https://jend.leeeet.dev/definition.txt?path=bend-ml-tensor@0.1.2.0/main.bend&name=Mat
  https://jend.leeeet.dev/definition.txt?path=bend-ml-tensor@0.1.2.0/main.bend&name=Mat.scale
  https://jend.leeeet.dev/definition.txt?path=bend-ml-tensor@0.1.2.0/main.bend&name=Mat.softmax
  https://jend.leeeet.dev/definition.txt?path=bend-ml-tensor@0.1.2.0/main.bend&name=Mat.sub