def test_sparse_cross_dense(self):
"""Tests sparse and dense inputs."""
sp_inp = self._sparse_tensor([['batch1-FC1-F1'],
['batch2-FC1-F1', 'batch2-FC1-F2']])
dense_inp = constant_op.constant([['batch1-FC2-F1', 'batch1-FC2-F2'],
['batch2-FC2-F1', 'batch2-FC2-F2']],
dtypes.string)
inds, vals, shapes = gen_sparse_ops.sparse_cross_v2(
indices=[sp_inp.indices],
values=[sp_inp.values],
shapes=[sp_inp.dense_shape],
dense_inputs=[dense_inp],
sep='_X_')
expected_out = self._sparse_tensor([
['batch1-FC1-F1_X_batch1-FC2-F1', 'batch1-FC1-F1_X_batch1-FC2-F2'],
[
'batch2-FC1-F1_X_batch2-FC2-F1',
'batch2-FC1-F1_X_batch2-FC2-F2',
'batch2-FC1-F2_X_batch2-FC2-F1',
'batch2-FC1-F2_X_batch2-FC2-F2'
]
])
out = sparse_tensor.SparseTensor(inds, vals, shapes)
with self.cached_session():
self._assert_sparse_tensor_equals(expected_out, self.evaluate(out))
def test_some_columns_empty(self):
"""Tests when more than one columns are empty.
Cross for the corresponding batch should be empty.
"""
sp_inp_1 = self._sparse_tensor([['batch1-FC1-F1', 'batch1-FC1-F2']], 2)
sp_inp_2 = self._sparse_tensor([['batch1-FC2-F1'], ['batch2-FC2-F1']],
2)
sp_inp_3 = self._sparse_tensor([['batch1-FC3-F1', 'batch1-FC3-F2']], 2)
inds, vals, shapes = gen_sparse_ops.sparse_cross_v2(
indices=[sp_inp_1.indices, sp_inp_2.indices, sp_inp_3.indices],
values=[sp_inp_1.values, sp_inp_2.values, sp_inp_3.values],
shapes=[
sp_inp_1.dense_shape, sp_inp_2.dense_shape,
sp_inp_3.dense_shape
],
dense_inputs=[],
sep='_X_')
expected_out = self._sparse_tensor([[
'batch1-FC1-F1_X_batch1-FC2-F1_X_batch1-FC3-F1',
'batch1-FC1-F1_X_batch1-FC2-F1_X_batch1-FC3-F2',
'batch1-FC1-F2_X_batch1-FC2-F1_X_batch1-FC3-F1',
'batch1-FC1-F2_X_batch1-FC2-F1_X_batch1-FC3-F2'
]], 2)
out = sparse_tensor.SparseTensor(inds, vals, shapes)
with self.cached_session():
self._assert_sparse_tensor_equals(expected_out, self.evaluate(out))
def test_permutation_3x1x2(self):
"""Tests 3x1x2 permutation."""
sp_inp_1 = self._sparse_tensor(
[['batch1-FC1-F1', 'batch1-FC1-F2', 'batch1-FC1-F3']])
sp_inp_2 = self._sparse_tensor([['batch1-FC2-F1']])
sp_inp_3 = self._sparse_tensor([['batch1-FC3-F1', 'batch1-FC3-F2']])
inds, vals, shapes = gen_sparse_ops.sparse_cross_v2(
indices=[sp_inp_1.indices, sp_inp_2.indices, sp_inp_3.indices],
values=[sp_inp_1.values, sp_inp_2.values, sp_inp_3.values],
shapes=[
sp_inp_1.dense_shape, sp_inp_2.dense_shape,
sp_inp_3.dense_shape
],
dense_inputs=[],
sep='_X_')
expected_out = self._sparse_tensor([[
'batch1-FC1-F1_X_batch1-FC2-F1_X_batch1-FC3-F1',
'batch1-FC1-F1_X_batch1-FC2-F1_X_batch1-FC3-F2',
'batch1-FC1-F2_X_batch1-FC2-F1_X_batch1-FC3-F1',
'batch1-FC1-F2_X_batch1-FC2-F1_X_batch1-FC3-F2',
'batch1-FC1-F3_X_batch1-FC2-F1_X_batch1-FC3-F1',
'batch1-FC1-F3_X_batch1-FC2-F1_X_batch1-FC3-F2'
]])
out = sparse_tensor.SparseTensor(inds, vals, shapes)
with self.cached_session():
self._assert_sparse_tensor_equals(expected_out, self.evaluate(out))
def test_integer_mixed_string_dense(self):
"""Tests mixed dense inputs."""
dense_inp_1 = constant_op.constant([[11, 333], [55555, 999999]],
dtypes.int64)
dense_inp_2 = constant_op.constant(
[['batch1-FC2-F1', 'batch1-FC2-F2'],
['batch2-FC2-F1', 'batch2-FC2-F2']], dtypes.string)
inds, vals, shapes = gen_sparse_ops.sparse_cross_v2(
indices=[],
values=[],
shapes=[],
dense_inputs=[dense_inp_1, dense_inp_2],
sep='_X_')
out = sparse_tensor.SparseTensor(inds, vals, shapes)
# pyformat: disable
expected_out = self._sparse_tensor([[
'11_X_batch1-FC2-F1', '11_X_batch1-FC2-F2', '333_X_batch1-FC2-F1',
'333_X_batch1-FC2-F2'
],
[
'55555_X_batch2-FC2-F1',
'55555_X_batch2-FC2-F2',
'999999_X_batch2-FC2-F1',
'999999_X_batch2-FC2-F2'
]])
# pyformat: enable
with self.cached_session():
self._assert_sparse_tensor_equals(expected_out, self.evaluate(out))
def test_dense_sep(self):
"""Tests only dense inputs."""
dense_inp_1 = constant_op.constant(
[['batch1-FC1-F1', 'batch1-FC1-F2'],
['batch2-FC1-F1', 'batch2-FC1-F2']], dtypes.string)
dense_inp_2 = constant_op.constant(
[['batch1-FC2-F1', 'batch1-FC2-F2'],
['batch2-FC2-F1', 'batch2-FC2-F2']], dtypes.string)
inds, vals, shapes = gen_sparse_ops.sparse_cross_v2(
indices=[],
values=[],
shapes=[],
dense_inputs=[dense_inp_1, dense_inp_2],
sep='_')
out = sparse_tensor.SparseTensor(inds, vals, shapes)
# pyformat: disable
expected_out = self._sparse_tensor([[
'batch1-FC1-F1_batch1-FC2-F1', 'batch1-FC1-F1_batch1-FC2-F2',
'batch1-FC1-F2_batch1-FC2-F1', 'batch1-FC1-F2_batch1-FC2-F2'
],
[
'batch2-FC1-F1_batch2-FC2-F1',
'batch2-FC1-F1_batch2-FC2-F2',
'batch2-FC1-F2_batch2-FC2-F1',
'batch2-FC1-F2_batch2-FC2-F2'
]])
# pyformat: enable
with self.cached_session():
self._assert_sparse_tensor_equals(expected_out, self.evaluate(out))
def test_sparse_sep(self):
"""Tests a simple scenario."""
sp_inp_1 = self._sparse_tensor([['batch1-FC1-F1'],
['batch2-FC1-F1', 'batch2-FC1-F2']])
sp_inp_2 = self._sparse_tensor([['batch1-FC2-F1'],
['batch2-FC2-F1', 'batch2-FC2-F2']])
inds, vals, shapes = gen_sparse_ops.sparse_cross_v2(
indices=[sp_inp_1.indices, sp_inp_2.indices],
values=[sp_inp_1.values, sp_inp_2.values],
shapes=[sp_inp_1.dense_shape, sp_inp_2.dense_shape],
dense_inputs=[],
sep='_Y_')
out = sparse_tensor.SparseTensor(inds, vals, shapes)
# pyformat: disable
expected_out = self._sparse_tensor(
[['batch1-FC1-F1_Y_batch1-FC2-F1'],
[
'batch2-FC1-F1_Y_batch2-FC2-F1',
'batch2-FC1-F1_Y_batch2-FC2-F2',
'batch2-FC1-F2_Y_batch2-FC2-F1',
'batch2-FC1-F2_Y_batch2-FC2-F2'
]])
# pyformat: enable
with self.cached_session():
self._assert_sparse_tensor_equals(expected_out, self.evaluate(out))
def test_large_batch(self):
"""Tests with large batch size to force multithreading."""
batch_size = 5000
col1 = []
col2 = []
col3 = []
for b in range(batch_size):
col1.append([
'batch%d-FC1-F1' % b,
'batch%d-FC1-F2' % b,
'batch%d-FC1-F3' % b
])
col2.append(['batch%d-FC2-F1' % b])
col3.append(['batch%d-FC3-F1' % b, 'batch%d-FC3-F2' % b])
sp_inp_1 = self._sparse_tensor(col1)
sp_inp_2 = self._sparse_tensor(col2)
sp_inp_3 = self._sparse_tensor(col3)
inds, vals, shapes = gen_sparse_ops.sparse_cross_v2(
indices=[sp_inp_1.indices, sp_inp_2.indices, sp_inp_3.indices],
values=[sp_inp_1.values, sp_inp_2.values, sp_inp_3.values],
shapes=[
sp_inp_1.dense_shape, sp_inp_2.dense_shape,
sp_inp_3.dense_shape
],
dense_inputs=[],
sep='_X_')
col_out = []
for b in range(batch_size):
col_out.append([
'batch%d-FC1-F1_X_batch%d-FC2-F1_X_batch%d-FC3-F1' % (b, b, b),
'batch%d-FC1-F1_X_batch%d-FC2-F1_X_batch%d-FC3-F2' % (b, b, b),
'batch%d-FC1-F2_X_batch%d-FC2-F1_X_batch%d-FC3-F1' % (b, b, b),
'batch%d-FC1-F2_X_batch%d-FC2-F1_X_batch%d-FC3-F2' % (b, b, b),
'batch%d-FC1-F3_X_batch%d-FC2-F1_X_batch%d-FC3-F1' % (b, b, b),
'batch%d-FC1-F3_X_batch%d-FC2-F1_X_batch%d-FC3-F2' % (b, b, b)
])
expected_out = self._sparse_tensor(col_out)
out = sparse_tensor.SparseTensor(inds, vals, shapes)
with self.cached_session():
self._assert_sparse_tensor_equals(expected_out, self.evaluate(out))
def test_all_columns_empty(self):
"""Tests when all columns are empty.
The crossed tensor should be empty.
"""
sp_inp_1 = self._sparse_tensor([])
sp_inp_2 = self._sparse_tensor([])
sp_inp_3 = self._sparse_tensor([])
inds, vals, shapes = gen_sparse_ops.sparse_cross_v2(
indices=[sp_inp_1.indices, sp_inp_2.indices, sp_inp_3.indices],
values=[sp_inp_1.values, sp_inp_2.values, sp_inp_3.values],
shapes=[
sp_inp_1.dense_shape, sp_inp_2.dense_shape, sp_inp_3.dense_shape
],
dense_inputs=[],
sep='_X_')
out = sparse_tensor.SparseTensor(inds, vals, shapes)
with self.cached_session():
self._assert_sparse_tensor_empty(self.evaluate(out))
def test_integer_mixed_string_sparse(self):
"""Tests mixed type."""
sp_inp_1 = self._sparse_tensor([[11], [333, 55555]])
sp_inp_2 = self._sparse_tensor([['batch1-FC2-F1'],
['batch2-FC2-F1', 'batch2-FC2-F2']])
inds, vals, shapes = gen_sparse_ops.sparse_cross_v2(
indices=[sp_inp_1.indices, sp_inp_2.indices],
values=[sp_inp_1.values, sp_inp_2.values],
shapes=[sp_inp_1.dense_shape, sp_inp_2.dense_shape],
dense_inputs=[],
sep='_X_')
out = sparse_tensor.SparseTensor(inds, vals, shapes)
# pyformat: disable
expected_out = self._sparse_tensor([
['11_X_batch1-FC2-F1'],
['333_X_batch2-FC2-F1', '333_X_batch2-FC2-F2',
'55555_X_batch2-FC2-F1', '55555_X_batch2-FC2-F2'
]])
# pyformat: enable
with self.cached_session():
self._assert_sparse_tensor_equals(expected_out, self.evaluate(out))
