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

Automated rollback of commit 23e33f871b2bf2879b40ebf3b883e104f30f389b. Revert #31450.

Browse Source 
PiperOrigin-RevId: 262675086
This commit is contained in:
Anna R 2019-08-09 19:01:23 -07:00 committed by TensorFlower Gardener
parent 019da8c2af
commit bf62fcec00
2 changed files with 3 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
tensorflow/python/keras/layers/core.py
View File

@ -26,7 +26,6 @@ import warnings
import numpy as np import numpy as np
from tensorflow.python.eager import context from tensorflow.python.eager import context
from tensorflow.python.framework import constant_op
from tensorflow.python.framework import dtypes from tensorflow.python.framework import dtypes
from tensorflow.python.framework import ops from tensorflow.python.framework import ops
from tensorflow.python.framework import tensor_shape from tensorflow.python.framework import tensor_shape
@ -581,29 +580,9 @@ class Flatten(Layer):
permutation.append(1) permutation.append(1)
inputs = array_ops.transpose(inputs, perm=permutation) inputs = array_ops.transpose(inputs, perm=permutation)
input_shape = inputs.shape outputs = array_ops.reshape(
if input_shape[1:].is_fully_defined(): inputs, (tensor_shape.dimension_value(inputs.shape[0]) or
flattened_dim = tensor_shape.dimension_value( array_ops.shape(inputs)[0], -1))
np.prod(input_shape[1:], dtype=int))
# Temporary fix for integer overflow issue.
if flattened_dim > np.iinfo(np.int32).max:
shape_dtype = dtypes.int64
else:
shape_dtype = dtypes.int32
outputs = array_ops.reshape(
inputs, constant_op.constant((-1, flattened_dim), dtype=shape_dtype))
else:
batch_size = tensor_shape.dimension_value(inputs.shape[0])
if batch_size:
# Temporary fix for integer overflow issue.
if batch_size > np.iinfo(np.int32).max:
shape_dtype = dtypes.int64
else:
shape_dtype = dtypes.int32
outputs = array_ops.reshape(
inputs, constant_op.constant((batch_size, -1), dtype=shape_dtype))
else:
outputs = array_ops.reshape(inputs, (array_ops.shape(inputs)[0], -1))
if not context.executing_eagerly(): if not context.executing_eagerly():
outputs.set_shape(self.compute_output_shape(inputs.shape)) outputs.set_shape(self.compute_output_shape(inputs.shape))
return outputs return outputs

14
tensorflow/python/layers/core_test.py
View File

@ -556,20 +556,6 @@ class FlattenTest(test.TestCase):
self.assertEqual(list(np_output.shape), [5, 6]) self.assertEqual(list(np_output.shape), [5, 6])
self.assertEqual(y.get_shape().as_list(), [5, None]) self.assertEqual(y.get_shape().as_list(), [5, None])
@test_util.run_deprecated_v1
def testFlattenLargeDim(self):
x = array_ops.placeholder(shape=(None, 21316, 21316, 80), dtype='float32')
y = core_layers.Flatten()(x)
self.assertEqual(y.shape.as_list(), [None, 21316 * 21316 * 80])
@test_util.run_deprecated_v1
def testFlattenLargeBatchDim(self):
batch_size = np.iinfo(np.int32).max + 10
x = array_ops.placeholder(
shape=(batch_size, None, None, 1), dtype='float32')
y = core_layers.Flatten()(x)
self.assertEqual(y.shape.as_list(), [batch_size, None])
if __name__ == '__main__': if __name__ == '__main__':
test.main() test.main()