Trace one MLP backward pass: the loss gradient at the output is 1, and the output is 1·h. What is the gradient at h?

prerequisite: NN-003

Context fixture

{
  "dLdy": 1,
  "y": "W2·h",
  "w2": [
    1,
    1
  ]
}

Hints · 0/3

Your answer

Opening…