diff options
| -rw-r--r-- | tensorflow/contrib/grid_rnn/python/ops/grid_rnn_cell.py | 3 | ||||
| -rw-r--r-- | tensorflow/contrib/linear_optimizer/python/sdca_optimizer.py | 35 | ||||
| -rw-r--r-- | tensorflow/contrib/rnn/python/ops/rnn_cell.py | 23 | ||||
| -rw-r--r-- | tensorflow/contrib/tensor_forest/python/tensor_forest.py | 3 | ||||
| -rw-r--r-- | tensorflow/python/kernel_tests/split_op_test.py | 76 | ||||
| -rw-r--r-- | tensorflow/python/ops/array_ops.py | 61 | ||||
| -rw-r--r-- | tensorflow/python/ops/gradients_test.py | 2 | ||||
| -rw-r--r-- | tensorflow/python/ops/rnn_cell_impl.py | 13 | ||||
| -rw-r--r-- | tensorflow/python/ops/split_benchmark.py | 2 |
9 files changed, 73 insertions, 145 deletions
diff --git a/tensorflow/contrib/grid_rnn/python/ops/grid_rnn_cell.py b/tensorflow/contrib/grid_rnn/python/ops/grid_rnn_cell.py index 6247cad380..aef4ce3fdb 100644 --- a/tensorflow/contrib/grid_rnn/python/ops/grid_rnn_cell.py +++ b/tensorflow/contrib/grid_rnn/python/ops/grid_rnn_cell.py @@ -170,7 +170,8 @@ class GridRNNCell(rnn.RNNCell): # project input if inputs is not None and sum(inputs.get_shape().as_list()) > 0 and len( conf.inputs) > 0: - input_splits = array_ops.split(1, len(conf.inputs), inputs) + input_splits = array_ops.split( + value=inputs, num_or_size_splits=len(conf.inputs), axis=1) input_sz = input_splits[0].get_shape().as_list()[1] for i, j in enumerate(conf.inputs): diff --git a/tensorflow/contrib/linear_optimizer/python/sdca_optimizer.py b/tensorflow/contrib/linear_optimizer/python/sdca_optimizer.py index 644347f0b5..9edb00e7b0 100644 --- a/tensorflow/contrib/linear_optimizer/python/sdca_optimizer.py +++ b/tensorflow/contrib/linear_optimizer/python/sdca_optimizer.py @@ -89,11 +89,12 @@ class SDCAOptimizer(object): # very sparse features with weights and not weights. return SparseFeatureColumn( array_ops.reshape( - array_ops.split(1, 2, sparse_indices)[0], [-1]), + array_ops.split( + value=sparse_indices, num_or_size_splits=2, axis=1)[0], [-1]), array_ops.reshape( - array_ops.split(1, 2, sparse_indices)[1], [-1]), - array_ops.reshape( - math_ops.to_float(sparse_values), [-1])) + array_ops.split( + value=sparse_indices, num_or_size_splits=2, axis=1)[1], [-1]), + array_ops.reshape(math_ops.to_float(sparse_values), [-1])) def _training_examples_and_variables(): """Returns dictionaries for training examples and variables.""" @@ -135,19 +136,27 @@ class SDCAOptimizer(object): columns_to_variables[column][0]) elif isinstance(column, (layers.feature_column._CrossedColumn, layers.feature_column._SparseColumn)): - sparse_features.append(SparseFeatureColumn( - array_ops.reshape( - array_ops.split(1, 2, transformed_tensor.indices)[0], [-1]), - array_ops.reshape(transformed_tensor.values, [-1]), None)) + sparse_features.append( + SparseFeatureColumn( + array_ops.reshape( + array_ops.split( + value=transformed_tensor.indices, + num_or_size_splits=2, + axis=1)[0], [-1]), + array_ops.reshape(transformed_tensor.values, [-1]), + None)) sparse_feature_weights.append(columns_to_variables[column][0]) elif isinstance(column, layers.feature_column._WeightedSparseColumn): id_tensor = column.id_tensor(transformed_tensor) weight_tensor = column.weight_tensor(transformed_tensor) - sparse_feature_with_values.append(SparseFeatureColumn( - array_ops.reshape( - array_ops.split(1, 2, id_tensor.indices)[0], [-1]), - array_ops.reshape(id_tensor.values, [-1]), array_ops.reshape( - weight_tensor.values, [-1]))) + sparse_feature_with_values.append( + SparseFeatureColumn( + array_ops.reshape( + array_ops.split( + value=id_tensor.indices, num_or_size_splits=2, axis=1) + [0], [-1]), + array_ops.reshape(id_tensor.values, [-1]), + array_ops.reshape(weight_tensor.values, [-1]))) sparse_feature_with_values_weights.append( columns_to_variables[column][0]) else: diff --git a/tensorflow/contrib/rnn/python/ops/rnn_cell.py b/tensorflow/contrib/rnn/python/ops/rnn_cell.py index 6f6f19b41a..761fabba8d 100644 --- a/tensorflow/contrib/rnn/python/ops/rnn_cell.py +++ b/tensorflow/contrib/rnn/python/ops/rnn_cell.py @@ -216,7 +216,7 @@ class CoupledInputForgetGateLSTMCell(rnn_cell.RNNCell): # j = new_input, f = forget_gate, o = output_gate cell_inputs = array_ops.concat_v2([inputs, m_prev], 1) lstm_matrix = nn_ops.bias_add(math_ops.matmul(cell_inputs, concat_w), b) - j, f, o = array_ops.split(1, 3, lstm_matrix) + j, f, o = array_ops.split(value=lstm_matrix, num_or_size_splits=3, axis=1) # Diagonal connections if self._use_peepholes: @@ -363,7 +363,8 @@ class TimeFreqLSTMCell(rnn_cell.RNNCell): cell_inputs = array_ops.concat_v2( [freq_inputs[fq], m_prev, m_prev_freq], 1) lstm_matrix = nn_ops.bias_add(math_ops.matmul(cell_inputs, concat_w), b) - i, j, f, o = array_ops.split(1, 4, lstm_matrix) + i, j, f, o = array_ops.split( + value=lstm_matrix, num_or_size_splits=4, axis=1) if self._use_peepholes: c = (sigmoid(f + self._forget_bias + w_f_diag * c_prev) * c_prev + @@ -670,12 +671,12 @@ class GridLSTMCell(rnn_cell.RNNCell): lstm_matrix_freq = nn_ops.bias_add(math_ops.matmul(cell_inputs, concat_w_f), b_f) if self._couple_input_forget_gates: - i_freq, j_freq, o_freq = array_ops.split(1, num_gates, - lstm_matrix_freq) + i_freq, j_freq, o_freq = array_ops.split( + value=lstm_matrix_freq, num_or_size_splits=num_gates, axis=1) f_freq = None else: - i_freq, j_freq, f_freq, o_freq = array_ops.split(1, num_gates, - lstm_matrix_freq) + i_freq, j_freq, f_freq, o_freq = array_ops.split( + value=lstm_matrix_freq, num_or_size_splits=num_gates, axis=1) # T-LSTM if self._share_time_frequency_weights: i_time = i_freq @@ -686,12 +687,12 @@ class GridLSTMCell(rnn_cell.RNNCell): lstm_matrix_time = nn_ops.bias_add(math_ops.matmul(cell_inputs, concat_w_t), b_t) if self._couple_input_forget_gates: - i_time, j_time, o_time = array_ops.split(1, num_gates, - lstm_matrix_time) + i_time, j_time, o_time = array_ops.split( + value=lstm_matrix_time, num_or_size_splits=num_gates, axis=1) f_time = None else: - i_time, j_time, f_time, o_time = array_ops.split(1, 4, - lstm_matrix_time) + i_time, j_time, f_time, o_time = array_ops.split( + value=lstm_matrix_time, num_or_size_splits=num_gates, axis=1) # F-LSTM c_freq # input gate activations @@ -1229,7 +1230,7 @@ class LayerNormBasicLSTMCell(rnn_cell.RNNCell): args = array_ops.concat_v2([inputs, h], 1) concat = self._linear(args) - i, j, f, o = array_ops.split(1, 4, concat) + i, j, f, o = array_ops.split(value=concat, num_or_size_splits=4, axis=1) if self._layer_norm: i = self._norm(i, "input") j = self._norm(j, "transform") diff --git a/tensorflow/contrib/tensor_forest/python/tensor_forest.py b/tensorflow/contrib/tensor_forest/python/tensor_forest.py index 84fb8dab61..5dff5f5f31 100644 --- a/tensorflow/contrib/tensor_forest/python/tensor_forest.py +++ b/tensorflow/contrib/tensor_forest/python/tensor_forest.py @@ -334,7 +334,8 @@ class RandomForestGraphs(object): ] def _bag_features(self, tree_num, input_data): - split_data = array_ops.split(1, self.params.num_features, input_data) + split_data = array_ops.split( + value=input_data, num_or_size_splits=self.params.num_features, axis=1) return array_ops.concat_v2( [split_data[ind] for ind in self.params.bagged_features[tree_num]], 1) diff --git a/tensorflow/python/kernel_tests/split_op_test.py b/tensorflow/python/kernel_tests/split_op_test.py index 60a1f4bcf3..a76acea4cd 100644 --- a/tensorflow/python/kernel_tests/split_op_test.py +++ b/tensorflow/python/kernel_tests/split_op_test.py @@ -22,7 +22,7 @@ import numpy as np import tensorflow as tf -class SplitVOpTest(tf.test.TestCase): +class SplitOpTest(tf.test.TestCase): def testExplicitNum(self): size_splits = tf.placeholder(dtype=tf.int32, shape=[None]) @@ -31,11 +31,11 @@ class SplitVOpTest(tf.test.TestCase): with self.test_session(use_gpu=False) as sess: with self.assertRaises(ValueError) as context: - sess.run(tf.split_v(value, size_splits), {size_splits: [2, 2, 6]}) + sess.run(tf.split(value, size_splits), {size_splits: [2, 2, 6]}) self.assertTrue("Cannot infer num from shape" in str(context.exception)) - result = sess.run(tf.split_v(value, size_splits, num=3), + result = sess.run(tf.split(value, size_splits, num=3), {size_splits: [2, 2, 6]}) self.assertAllEqual(result[0], value[0:2]) @@ -49,12 +49,12 @@ class SplitVOpTest(tf.test.TestCase): value = np.random.rand(11, 11) with self.test_session(use_gpu=False) as sess: - result = sess.run(tf.split_v(value, [a, b])) + result = sess.run(tf.split(value, [a, b])) self.assertAllEqual(result[0], value[0:5, :]) self.assertAllEqual(result[1], value[5:, :]) - def _RunAndVerify(self, use_gpu, large_num_splits=False): + def _RunAndVerifyVariable(self, use_gpu, large_num_splits=False): # Random dims of rank 5 shape = np.random.randint(1, 5, size=5) split_dim = np.random.randint(0, 5) @@ -66,7 +66,7 @@ class SplitVOpTest(tf.test.TestCase): shape[split_dim] = np.sum(size_splits) inp = np.random.rand(*shape).astype("f") with self.test_session(use_gpu=use_gpu) as sess: - result = sess.run(tf.split_v(inp, size_splits, split_dim)) + result = sess.run(tf.split(inp, size_splits, split_dim)) slices = [slice(0, x) for x in shape] offset = 0 for i in range(num_split): @@ -74,53 +74,25 @@ class SplitVOpTest(tf.test.TestCase): offset += size_splits[i] self.assertAllEqual(result[i], inp[slices]) - def _RunAndVerifyScalar(self, use_gpu, large_num_splits=False): - shape = np.random.randint(0, 5, size=5) - split_dim = np.random.randint(0, 5) - if large_num_splits: - num_split = np.random.randint(16, 25) - else: - num_split = np.random.randint(2, 8) - shape[split_dim] = np.random.randint(2, 5) * num_split - inp = np.random.rand(*shape).astype("f") - with self.test_session(use_gpu=use_gpu) as sess: - result = sess.run(tf.split_v(inp, num_split, split_dim)) - slices = [slice(0, x) for x in shape] - offset = 0 - length = shape[split_dim] // num_split - for i in range(num_split): - slices[split_dim] = slice(offset, offset + length) - offset += length - self.assertAllEqual(result[i], inp[slices]) - - def testRandom(self): - for _ in range(5): - self._RunAndVerify(use_gpu=False) - self._RunAndVerify(use_gpu=True) - self._RunAndVerify(use_gpu=True, large_num_splits=True) - self._RunAndVerifyScalar(use_gpu=False) - self._RunAndVerifyScalar(use_gpu=True) - self._RunAndVerifyScalar(use_gpu=True, large_num_splits=True) - - def _testSpecialCases(self, use_gpu): + def _testSpecialCasesVariable(self, use_gpu): inp = np.random.rand(4, 4).astype("f") with self.test_session(use_gpu=use_gpu) as sess: - result = sess.run(tf.split_v(inp, [4], 0)) + result = sess.run(tf.split(inp, [4], 0)) self.assertAllEqual(result[0], inp) - result = sess.run(tf.split_v(inp, [-1, 3], 0)) + result = sess.run(tf.split(inp, [-1, 3], 0)) self.assertAllEqual(result[0], inp[0:1, :]) self.assertAllEqual(result[1], inp[1:4, :]) - def _testHugeNumberOfTensors(self, use_gpu): + def _testHugeNumberOfTensorsVariable(self, use_gpu): num_split = 10000 size_splits = np.random.randint(1, 3, num_split) shape = [3, np.sum(size_splits)] split_dim = 1 inp = np.random.rand(*shape).astype("f") with self.test_session(use_gpu=use_gpu) as sess: - result = sess.run(tf.split_v(inp, size_splits, split_dim)) + result = sess.run(tf.split(inp, size_splits, split_dim)) slices = [slice(0, x) for x in shape] offset = 0 for i in range(num_split): @@ -128,17 +100,17 @@ class SplitVOpTest(tf.test.TestCase): offset += size_splits[i] self.assertAllEqual(result[i], inp[slices]) - def testSpecialCases(self): - self._testSpecialCases(False) - self._testSpecialCases(True) - self._testHugeNumberOfTensors(False) - self._testHugeNumberOfTensors(True) + def testSpecialCasesVariable(self): + self._testSpecialCasesVariable(False) + self._testSpecialCasesVariable(True) + self._testHugeNumberOfTensorsVariable(False) + self._testHugeNumberOfTensorsVariable(True) - def _testGradientsSimple(self, use_gpu): + def _testGradientsSimpleVariable(self, use_gpu): inp = np.random.rand(4, 4).astype("f") with self.test_session(use_gpu=use_gpu): inp_tensor = tf.convert_to_tensor(inp) - s = tf.split_v(inp_tensor, [1, 4], 1) + s = tf.split(inp_tensor, [1, 4], 1) inp_grads = [ np.random.rand(4, 1).astype("f"), np.random.rand(4, 3).astype("f") ] @@ -149,13 +121,6 @@ class SplitVOpTest(tf.test.TestCase): self.assertAllEqual(result[:, 0:1], inp_grads[0]) self.assertAllEqual(result[:, 1:4], inp_grads[1]) - def testGradientsAll(self): - self._testGradientsSimple(use_gpu=False) - self._testGradientsSimple(use_gpu=True) - - -class SplitOpTest(tf.test.TestCase): - def _compare(self, x, dim, num, use_gpu): np_ans = np.split(x, num, dim) with self.test_session(use_gpu=use_gpu) as sess: @@ -244,6 +209,9 @@ class SplitOpTest(tf.test.TestCase): self._RunAndVerify(use_gpu=False) self._RunAndVerify(use_gpu=True) self._RunAndVerify(use_gpu=True, large_num_splits=True) + self._RunAndVerifyVariable(use_gpu=False) + self._RunAndVerifyVariable(use_gpu=True) + self._RunAndVerifyVariable(use_gpu=True, large_num_splits=True) def _testGradientsSimple(self, use_gpu): inp = np.random.rand(4, 4).astype("f") @@ -260,6 +228,8 @@ class SplitOpTest(tf.test.TestCase): def testGradientsAll(self): self._testGradientsSimple(use_gpu=False) self._testGradientsSimple(use_gpu=True) + self._testGradientsSimpleVariable(use_gpu=False) + self._testGradientsSimpleVariable(use_gpu=True) def testShapeFunctionEdgeCases(self): # split_dim greater than rank of input. diff --git a/tensorflow/python/ops/array_ops.py b/tensorflow/python/ops/array_ops.py index 02bea9aa57..d78d7817f1 100644 --- a/tensorflow/python/ops/array_ops.py +++ b/tensorflow/python/ops/array_ops.py @@ -50,7 +50,6 @@ or join multiple tensors together. @@slice @@strided_slice @@split -@@split_v @@tile @@pad @@concat @@ -1226,63 +1225,7 @@ def sparse_mask(a, mask_indices, name=None): return ops.IndexedSlices(out_values, out_indices, a.dense_shape) -def split(axis=None, - num_or_size_splits=None, - value=None, - name="split", - split_dim=None): - """DEPRECATED: use split_v; split_v rename to split happening soon. - - Splits `value` along dimension `axis` into `num_or_size_splits` smaller - tensors. Requires that `num_or_size_splits` evenly divide `value.shape[axis]`. - - For example: - - ```python - # 'value' is a tensor with shape [5, 30] - # Split 'value' into 3 tensors along dimension 1 - split0, split1, split2 = tf.split(value=value, num_or_size_splits=3, axis=1) - tf.shape(split0) ==> [5, 10] - ``` - - Note: If you are splitting along an axis by the length of that axis, consider - using unpack, e.g. - - ```python - num_items = t.get_shape()[axis].value - [tf.squeeze(s, [axis]) for s in - tf.split(value=t, num_or_size_splits=num_items, axis=axis)] - ``` - - can be rewritten as - - ```python - tf.unpack(t, axis=axis) - ``` - - Args: - axis: A 0-D `int32` `Tensor`. The dimension along which to split. - Must be in the range `[0, rank(value))`. - num_or_size_splits: A Python integer. The number of ways to split. Has a - different meaning in split_v (see docs). - value: The `Tensor` to split. - name: A name for the operation (optional). - split_dim: The old (deprecated) name for axis. - - Returns: - `num_or_size_splits` `Tensor` objects resulting from splitting `value`. - """ - axis = deprecation.deprecated_argument_lookup("axis", axis, "split_dim", - split_dim) - return gen_array_ops._split( - split_dim=axis, num_split=num_or_size_splits, value=value, name=name) - - -def split_v(value=None, - num_or_size_splits=None, - axis=0, - num=None, - name="split_v"): +def split(value, num_or_size_splits, axis=0, num=None, name="split"): """Splits a tensor into sub tensors. If `num_or_size_splits` is a scalar, `num_split`, then splits `value` along @@ -1298,7 +1241,7 @@ def split_v(value=None, ```python # 'value' is a tensor with shape [5, 30] # Split 'value' into 3 tensors with sizes [4, 15, 11] along dimension 1 - split0, split1, split2 = tf.split_v(value, [4, 15, 11], 1) + split0, split1, split2 = tf.split(value, [4, 15, 11], 1) tf.shape(split0) ==> [5, 4] tf.shape(split1) ==> [5, 15] tf.shape(split2) ==> [5, 11] diff --git a/tensorflow/python/ops/gradients_test.py b/tensorflow/python/ops/gradients_test.py index 7272b1729a..8b4fbeb7a9 100644 --- a/tensorflow/python/ops/gradients_test.py +++ b/tensorflow/python/ops/gradients_test.py @@ -151,7 +151,7 @@ class GradientsTest(test_util.TensorFlowTestCase): w = constant(1.0, shape=[2, 2]) x = constant(1.0, shape=[2, 2]) wx = math_ops.matmul(w, x) - split_wx = array_ops.split(0, 2, wx) + split_wx = array_ops.split(value=wx, num_or_size_splits=2, axis=0) c = math_ops.reduce_sum(split_wx[1]) gw = gradients.gradients(c, [w])[0] self.assertEquals("MatMul", gw.op.type) diff --git a/tensorflow/python/ops/rnn_cell_impl.py b/tensorflow/python/ops/rnn_cell_impl.py index 7e0abf854d..c448cbd315 100644 --- a/tensorflow/python/ops/rnn_cell_impl.py +++ b/tensorflow/python/ops/rnn_cell_impl.py @@ -203,8 +203,10 @@ class GRUCell(RNNCell): with vs.variable_scope("gates"): # Reset gate and update gate. # We start with bias of 1.0 to not reset and not update. r, u = array_ops.split( - 1, 2, _linear([inputs, state], 2 * self._num_units, True, 1.0, - scope=scope)) + value=_linear( + [inputs, state], 2 * self._num_units, True, 1.0, scope=scope), + num_or_size_splits=2, + axis=1) r, u = sigmoid(r), sigmoid(u) with vs.variable_scope("candidate"): c = self._activation(_linear([inputs, r * state], @@ -288,11 +290,11 @@ class BasicLSTMCell(RNNCell): if self._state_is_tuple: c, h = state else: - c, h = array_ops.split(1, 2, state) + c, h = array_ops.split(value=state, num_or_size_splits=2, axis=1) concat = _linear([inputs, h], 4 * self._num_units, True, scope=scope) # i = input_gate, j = new_input, f = forget_gate, o = output_gate - i, j, f, o = array_ops.split(1, 4, concat) + i, j, f, o = array_ops.split(value=concat, num_or_size_splits=4, axis=1) new_c = (c * sigmoid(f + self._forget_bias) + sigmoid(i) * self._activation(j)) @@ -449,7 +451,8 @@ class LSTMCell(RNNCell): # i = input_gate, j = new_input, f = forget_gate, o = output_gate lstm_matrix = _linear([inputs, m_prev], 4 * self._num_units, bias=True, scope=scope) - i, j, f, o = array_ops.split(1, 4, lstm_matrix) + i, j, f, o = array_ops.split( + value=lstm_matrix, num_or_size_splits=4, axis=1) # Diagonal connections if self._use_peepholes: diff --git a/tensorflow/python/ops/split_benchmark.py b/tensorflow/python/ops/split_benchmark.py index d0647ecc90..5da0eac4b7 100644 --- a/tensorflow/python/ops/split_benchmark.py +++ b/tensorflow/python/ops/split_benchmark.py @@ -41,7 +41,7 @@ def build_graph(device, input_shape, output_sizes, axis): outputs = [] for _ in range(100): - outputs.extend(tf.split_v(inp, output_sizes, axis)) + outputs.extend(tf.split(inp, output_sizes, axis)) return tf.group(*outputs) |