271f6bb49d
1. Clean up old code related to legacy scalars and vectors in TensorFlow. This CL does not bump the GrapDef version.
Legacy behavior is allowed for a small number of frequently used kernels and a TODO to fix this in a future CL added.
2. Move OpKernel::MakeShape into tensor_util.{h,cc}. Ideally it should go in TensorShapeUtils, but that would create a cyclical dependence.
The oldest branch of TensorFlow on github is r0.7, which is at GraphDef version 8 already, so this should have no impact on GraphDefs in the wild.
We also changed the scalar test to be strict in open source on 2017-04-03.
PiperOrigin-RevId: 292234389
Change-Id: Iee372607d9b9139d33ba7be5be8b792d9471e0f6
147 lines
4.9 KiB
Python
147 lines
4.9 KiB
Python
# Copyright 2015 The TensorFlow Authors. All Rights Reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
# ==============================================================================
|
|
"""Tests for scalar strictness and scalar leniency."""
|
|
|
|
from __future__ import absolute_import
|
|
from __future__ import division
|
|
from __future__ import print_function
|
|
|
|
import numpy as np
|
|
|
|
from tensorflow.python.framework import ops
|
|
from tensorflow.python.framework import test_util
|
|
from tensorflow.python.ops import array_ops
|
|
from tensorflow.python.ops import gen_io_ops
|
|
from tensorflow.python.ops import math_ops
|
|
from tensorflow.python.ops import random_ops
|
|
from tensorflow.python.ops import sparse_ops
|
|
import tensorflow.python.ops.nn_grad # pylint: disable=unused-import
|
|
from tensorflow.python.platform import test
|
|
|
|
|
|
# TODO(rmlarsen) : Remove this test completely after we stop supporting GraphDef
|
|
# version 5 and remove support of legacy scalars from Concat, Fill, Range,
|
|
# and Reshape.
|
|
class ScalarTest(test.TestCase):
|
|
|
|
def check(self, op, args, error, correct=None, lenient=None, strict=[5, 6]):
|
|
if lenient is None:
|
|
lenient = []
|
|
# Use placeholders to bypass shape inference, since only the C++
|
|
# G raphDef level is ever scalar lenient.
|
|
def placeholders(args, feed):
|
|
if isinstance(args, tuple):
|
|
return [placeholders(x, feed) for x in args]
|
|
else:
|
|
x = ops.convert_to_tensor(args).eval()
|
|
fake = array_ops.placeholder(np.asarray(x).dtype)
|
|
feed[fake] = x
|
|
return fake
|
|
|
|
# Test various GraphDef versions
|
|
for version in strict + lenient:
|
|
with ops.Graph().as_default() as g:
|
|
test_util.set_producer_version(g, version)
|
|
with self.session(graph=g) as sess:
|
|
feed = {}
|
|
xs = placeholders(args, feed)
|
|
x = op(*xs)
|
|
if version in strict:
|
|
with self.assertRaisesOpError(error):
|
|
sess.run(x, feed_dict=feed)
|
|
else:
|
|
r = sess.run(x, feed_dict=feed)
|
|
if correct is not None:
|
|
self.assertAllEqual(r, correct)
|
|
|
|
def testConcat(self):
|
|
for data in (2, [3], 7), ([2], 3, 7), ([2], [3], 7):
|
|
self.check(array_ops.concat, (data, 0),
|
|
r'Ranks of all input tensors should match', [2, 3, 7])
|
|
|
|
def testFill(self):
|
|
self.check(
|
|
array_ops.fill, (2, 3),
|
|
'dims must be a vector', [3, 3],
|
|
lenient=[5, 6],
|
|
strict=[])
|
|
self.check(
|
|
array_ops.fill, ([2], [3]),
|
|
'value must be a scalar', [3, 3],
|
|
lenient=[5, 6],
|
|
strict=[])
|
|
|
|
def testPad(self):
|
|
self.check(array_ops.pad, (7, [[1, 2]]),
|
|
'The first dimension of paddings must be the rank of inputs',
|
|
[0, 7, 0, 0])
|
|
|
|
def testRandom(self):
|
|
self.check(random_ops.random_uniform, (3,), 'shape must be a vector')
|
|
|
|
def testReshape(self):
|
|
self.check(
|
|
array_ops.reshape, (7, 1),
|
|
'sizes input must be 1-D', [7],
|
|
lenient=[5, 6],
|
|
strict=[])
|
|
|
|
def testShardedFilename(self):
|
|
self.check(gen_io_ops.sharded_filename, ('foo', 4, [100]),
|
|
'must be a scalar', b'foo-00004-of-00100')
|
|
|
|
def testShardedFilespec(self):
|
|
self.check(gen_io_ops.sharded_filespec, ('foo', [100]), 'must be a scalar',
|
|
b'foo-?????-of-00100')
|
|
|
|
def testUnsortedSegmentSum(self):
|
|
self.check(math_ops.unsorted_segment_sum, (7, 1, [4]),
|
|
'num_segments should be a scalar', [0, 7, 0, 0])
|
|
|
|
def testRange(self):
|
|
self.check(
|
|
math_ops.range, ([0], 3, 2),
|
|
'start must be a scalar', [0, 2],
|
|
lenient=[5, 6],
|
|
strict=[])
|
|
self.check(
|
|
math_ops.range, (0, [3], 2),
|
|
'limit must be a scalar', [0, 2],
|
|
lenient=[5, 6],
|
|
strict=[])
|
|
self.check(
|
|
math_ops.range, (0, 3, [2]),
|
|
'delta must be a scalar', [0, 2],
|
|
lenient=[5, 6],
|
|
strict=[])
|
|
|
|
def testSlice(self):
|
|
data = np.arange(10)
|
|
error = 'Expected begin and size arguments to be 1-D tensors'
|
|
self.check(array_ops.slice, (data, 2, 3), error, [2, 3, 4])
|
|
self.check(array_ops.slice, (data, [2], 3), error, [2, 3, 4])
|
|
self.check(array_ops.slice, (data, 2, [3]), error, [2, 3, 4])
|
|
|
|
def testSparseToDense(self):
|
|
self.check(sparse_ops.sparse_to_dense, (1, 4, 7),
|
|
'output_shape should be a vector', [0, 7, 0, 0])
|
|
|
|
def testTile(self):
|
|
self.check(array_ops.tile, ([7], 2), 'Expected multiples to be 1-D', [7, 7])
|
|
|
|
|
|
if __name__ == '__main__':
|
|
test.main()
|