experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Automated rollback of commit a7e7582e9a2b698054bf93aa27e53ebbc081d1a6. Revert #31106.

Browse Source 
PiperOrigin-RevId: 263621326
This commit is contained in:
Emily Glanz 2019-08-15 12:46:03 -07:00 committed by TensorFlower Gardener
parent e3c100fdce
commit 9cfcec4123
1 changed files with 14 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
tensorflow/python/ops/math_grad.py
View File

@ -192,26 +192,22 @@ def _SumGrad(op, grad):
return [array_ops.tile(grad, tile_scaling), None]
input_shape = array_ops.shape(op.inputs[0])
if not op.get_attr("keep_dims"):
# TODO(apassos) remove this once device placement for eager ops makes more
# sense.
with ops.colocate_with(input_shape):
output_shape_kept_dims = math_ops.reduced_shape(input_shape, op.inputs[1])
grad = array_ops.reshape(grad, output_shape_kept_dims)
return [array_ops.broadcast_to(grad, input_shape), None]
# TODO(apassos) remove this once device placement for eager ops makes more
# sense.
with ops.colocate_with(input_shape):
output_shape_kept_dims = math_ops.reduced_shape(input_shape, op.inputs[1])
tile_scaling = _safe_shape_div(input_shape, output_shape_kept_dims)
grad = array_ops.reshape(grad, output_shape_kept_dims)
return [array_ops.tile(grad, tile_scaling), None]
def _MinOrMaxGrad(op, grad):
"""Gradient for Min or Max. Amazingly it's precisely the same code."""
input_shape = array_ops.shape(op.inputs[0])
output_shape_kept_dims = math_ops.reduced_shape(input_shape, op.inputs[1])
y = op.outputs[0]
if not op.get_attr("keep_dims"):
output_shape_kept_dims = math_ops.reduced_shape(input_shape, op.inputs[1])
y = array_ops.reshape(y, output_shape_kept_dims)
grad = array_ops.reshape(grad, output_shape_kept_dims)
else:
output_shape_kept_dims = array_ops.shape(y)
y = array_ops.reshape(y, output_shape_kept_dims)
grad = array_ops.reshape(grad, output_shape_kept_dims)
# Compute the number of selected (maximum or minimum) elements in each
# reduction dimension. If there are multiple minimum or maximum elements
@ -267,11 +263,10 @@ def _ProdGrad(op, grad):
reduction_indices = array_ops.reshape(op.inputs[1], [-1])
# Expand grad to full input shape
if not op.get_attr("keep_dims"):
output_shape_kept_dims = math_ops.reduced_shape(input_shape, op.inputs[1])
grad = array_ops.reshape(grad, output_shape_kept_dims)
grad = array_ops.broadcast_to(grad, input_shape)
output_shape_kept_dims = math_ops.reduced_shape(input_shape, op.inputs[1])
tile_scaling = _safe_shape_div(input_shape, output_shape_kept_dims)
grad = array_ops.reshape(grad, output_shape_kept_dims)
grad = array_ops.tile(grad, tile_scaling)
# Pack all reduced dimensions into a single one, so we can perform the
# cumprod ops. If the reduction dims list is empty, it defaults to float32,