example, suppose y = f(x) and we wish to apply a custom function g for backprop
such that dx = g(dy). In Python,

```python
with tf.get_default_graph().gradient_override_map(
    {'IdentityN': 'OverrideGradientWithG'}):
  y, _ = identity_n([f(x), x])

@tf.RegisterGradient('OverrideGradientWithG')
def ApplyG(op, dy, _):
  return [None, g(dy)]  # Do not backprop to f(x).
```
  }];

  let arguments = (ins

257p\022Y\334?l \023\216\340\373\336?\340\025\305\265\345y\356?\034#\252\324\017\247\352?\264\323\262{G\215\307?pv\0359\001\315\306?\225\3428z\204c\355?^\341\343\017S(\344?\261\343\027\262\327\236\342?3\t\322%:+\344?d\035\232\232rD\347?\034\366\021,dY\315?%\356<L\273n\346?&H\230\262\254\000\324?\300\225\266\263\036\313\206?\210\347;\005\270n\337?\266\377\032\315`G\325?\260\316%\000\021\r\301?-\367\353\375\325n\340?\016\362>\244\340\013\340?8v\000JQ\231\305?66\326:\035\276\353?_\246\207\253\377\3...

x6twUmXV1yKjjtGrO+9fHvTOWBfSSP+fP9KTaTQYSasoJq2Mw4cQDy1i0zrxNZUw
N4/BiyjQA25sdfaOolhO0sFlZuTZpYy5wG72KkA1ygNq0L+8aBKhEF6zDU61YzCC
AxjcgTftgTeeoqkJtYa06lNz3jmJDN+zUQignfRa3ymoGtFHTzoXR9maE8RWDty4
y+DY+8ibdGgSgKPZ0byTCDyNojgU1YTlADa/1/NY1ShYg617O1xicLNo0JEJlf2U
Tu4Ymql36+xSkYSISU97Q6Utgq27XMuZvDUDABEBAAG0IkNvbGluIERlY2tlciA8
Y2dkZWNrZXJAZ29vZ2xlLmNvbT65Ag0EWSB9/gEQAMd2uqUse7rkjD/IAHA5yrxW

		{"(Rectangle).At", Method, 5},
		{"(Rectangle).Bounds", Method, 5},
		{"(Rectangle).Canon", Method, 0},
		{"(Rectangle).ColorModel", Method, 5},
		{"(Rectangle).Dx", Method, 0},
		{"(Rectangle).Dy", Method, 0},
		{"(Rectangle).Empty", Method, 0},
		{"(Rectangle).Eq", Method, 0},
		{"(Rectangle).In", Method, 0},
		{"(Rectangle).Inset", Method, 0},
		{"(Rectangle).Intersect", Method, 0},