| /external/tensorflow/tensorflow/contrib/kernel_methods/python/ |
| D | losses_test.py | 36 logits = constant_op.constant([-1.0, 2.1], shape=(2,))
39 _ = losses.sparse_multiclass_hinge_loss(labels, logits)
44 logits = constant_op.constant([-1.0, 2.1], shape=(2, 1))
47 _ = losses.sparse_multiclass_hinge_loss(labels, logits)
52 logits = constant_op.constant([-1.0, 2.1], shape=(2, 1))
56 _ = losses.sparse_multiclass_hinge_loss(labels, logits, weights)
61 logits = constant_op.constant([-1.0, 2.1], shape=(2, 1))
64 _ = losses.sparse_multiclass_hinge_loss(labels, logits)
69 logits = constant_op.constant([-1.0, 2.1], shape=(2, 1))
72 _ = losses.sparse_multiclass_hinge_loss(labels, logits, weights=None)
[all …]
|
| /external/tensorflow/tensorflow/python/ops/ |
| D | nn_xent_test.py | 39 def _SigmoidCrossEntropyWithLogits(self, logits, targets): argument
40 assert len(logits) == len(targets)
41 pred = [1 / (1 + exp(-x)) for x in logits]
51 logits = constant_op.constant(x, shape=sizes, dtype=dtype, name="logits")
54 return logits, targets, losses
58 logits, targets, _ = self._Inputs()
60 labels=targets, logits=logits, name="mylogistic")
67 logits, targets, losses = self._Inputs(dtype=dtype)
69 labels=targets, logits=logits)
78 logits, targets, losses = self._Inputs(dtype=dtype, sizes=[2, 2, 2])
[all …]
|
| /external/tensorflow/tensorflow/contrib/distributions/python/ops/ |
| D | onehot_categorical.py | 88 logits=None, argument
119 with ops.name_scope(name, values=[logits, probs]):
121 name=name, logits=logits, probs=probs, validate_args=validate_args,
153 def logits(self): member in OneHotCategorical
163 return array_ops.shape(self.logits)[:-1]
166 return self.logits.get_shape()[:-1]
169 return array_ops.shape(self.logits)[-1:]
172 return self.logits.get_shape().with_rank_at_least(1)[-1:]
175 sample_shape = array_ops.concat([[n], array_ops.shape(self.logits)], 0)
176 logits = self.logits
[all …]
|
| D | relaxed_onehot_categorical.py | 131 logits=None, argument
166 with ops.name_scope(name, values=[logits, probs, temperature]):
169 name=name, logits=logits, probs=probs, validate_args=validate_args,
218 def logits(self): member in ExpRelaxedOneHotCategorical
231 return self.logits.get_shape()[:-1]
234 return array_ops.shape(self.logits)[-1:]
237 return self.logits.get_shape().with_rank_at_least(1)[-1:]
240 sample_shape = array_ops.concat([[n], array_ops.shape(self.logits)], 0)
241 logits = self.logits * array_ops.ones(sample_shape, dtype=self.dtype)
242 logits_2d = array_ops.reshape(logits, [-1, self.event_size])
[all …]
|
| /external/tensorflow/tensorflow/contrib/slim/python/slim/nets/ |
| D | vgg_test.py | 39 logits, _ = vgg.vgg_a(inputs, num_classes)
40 self.assertEquals(logits.op.name, 'vgg_a/fc8/squeezed')
41 self.assertListEqual(logits.get_shape().as_list(),
50 logits, _ = vgg.vgg_a(inputs, num_classes, spatial_squeeze=False)
51 self.assertEquals(logits.op.name, 'vgg_a/fc8/BiasAdd')
52 self.assertListEqual(logits.get_shape().as_list(),
112 logits, _ = vgg.vgg_a(eval_inputs, is_training=False)
113 self.assertListEqual(logits.get_shape().as_list(),
115 predictions = math_ops.argmax(logits, 1)
127 logits, _ = vgg.vgg_a(train_inputs)
[all …]
|
| D | overfeat_test.py | 38 logits, _ = overfeat.overfeat(inputs, num_classes)
39 self.assertEquals(logits.op.name, 'overfeat/fc8/squeezed')
40 self.assertListEqual(logits.get_shape().as_list(),
49 logits, _ = overfeat.overfeat(inputs, num_classes, spatial_squeeze=False)
50 self.assertEquals(logits.op.name, 'overfeat/fc8/BiasAdd')
51 self.assertListEqual(logits.get_shape().as_list(),
103 logits, _ = overfeat.overfeat(eval_inputs, is_training=False)
104 self.assertListEqual(logits.get_shape().as_list(),
106 predictions = math_ops.argmax(logits, 1)
118 logits, _ = overfeat.overfeat(train_inputs)
[all …]
|
| D | alexnet_test.py | 38 logits, _ = alexnet.alexnet_v2(inputs, num_classes)
39 self.assertEquals(logits.op.name, 'alexnet_v2/fc8/squeezed')
40 self.assertListEqual(logits.get_shape().as_list(),
49 logits, _ = alexnet.alexnet_v2(inputs, num_classes, spatial_squeeze=False)
50 self.assertEquals(logits.op.name, 'alexnet_v2/fc8/BiasAdd')
51 self.assertListEqual(logits.get_shape().as_list(),
103 logits, _ = alexnet.alexnet_v2(eval_inputs, is_training=False)
104 self.assertListEqual(logits.get_shape().as_list(),
106 predictions = math_ops.argmax(logits, 1)
118 logits, _ = alexnet.alexnet_v2(train_inputs)
[all …]
|
| /external/tensorflow/tensorflow/contrib/estimator/python/estimator/ |
| D | head_test.py | 80 def _sigmoid(logits): argument
81 return 1 / (1 + np.exp(-logits))
84 def _sigmoid_cross_entropy(labels, logits): argument
86 sigmoid_logits = _sigmoid(logits)
147 def _loss_fn(logits): argument
148 del logits # Unused
165 def _loss_fn(labels, logits, features): argument
166 del labels, logits, features # Unused
170 def _loss_fn(labels, logits, name=None): argument
171 del labels, logits, name # Unused
[all …]
|
| /external/tensorflow/tensorflow/python/ops/distributions/ |
| D | bernoulli.py | 43 logits=None, argument
77 logits=logits,
95 def logits(self): member in Bernoulli
130 event = math_ops.cast(event, self.logits.dtype)
131 logits = self.logits
135 def _broadcast(logits, event): argument
136 return (array_ops.ones_like(event) * logits,
137 array_ops.ones_like(logits) * event)
140 logits.get_shape().is_fully_defined() and
141 event.get_shape() == logits.get_shape()):
[all …]
|
| /external/tensorflow/tensorflow/contrib/distributions/python/kernel_tests/ |
| D | relaxed_onehot_categorical_test.py | 33 logits = random_ops.random_uniform(
38 temperatures, logits, dtype=dtype)
45 logits = [2.0, 3.0, -4.0]
47 logits)
48 expected_p = np.exp(logits)/np.sum(np.exp(logits))
55 logits = [.3, .1, .4]
56 k = len(logits)
57 p = np.exp(logits)/np.sum(np.exp(logits))
59 logits)
74 logits = [2.0, 3.0, -4.0]
[all …]
|
| D | onehot_categorical_test.py | 34 logits = random_ops.random_uniform(
36 return onehot_categorical.OneHotCategorical(logits, dtype=dtype)
49 self.assertAllEqual([2], dist.logits.get_shape())
53 logits = np.log(p) - 50.
54 dist = onehot_categorical.OneHotCategorical(logits=logits)
57 self.assertAllEqual([2], dist.logits.get_shape())
59 self.assertAllClose(dist.logits.eval(), logits)
92 self.assertEqual(dist.logits.dtype, dtypes.float32)
93 self.assertEqual(dist.logits.dtype, dist.entropy().dtype)
94 self.assertEqual(dist.logits.dtype, dist.prob(
[all …]
|
| D | estimator_test.py | 54 def actual_loss(logits, labels): argument
55 mu = actual_mean(logits)
56 sigma = actual_stddev(logits)
62 def actual_mean(logits): argument
63 return logits[..., 0]
65 def actual_stddev(logits): argument
66 return softplus(logits[..., 1] + scale_bias)
68 def make_distribution_fn(logits): argument
70 loc=logits[..., 0],
71 scale=nn_ops.softplus(logits[..., 1] + scale_bias))
[all …]
|
| /external/tensorflow/tensorflow/contrib/kfac/python/kernel_tests/ |
| D | loss_functions_test.py | 52 logits = np.asarray([
57 array_ops.constant(logits))
65 logits = np.asarray([
71 array_ops.constant(logits), targets=array_ops.constant(targets))
76 probs = np.exp(logits) / np.sum(np.exp(logits), axis=1, keepdims=True)
88 logits = np.asarray([
93 array_ops.constant(logits))
107 logits = random_ops.random_uniform(shape=[2, 3])
108 tower_logits.append(logits)
110 loss = loss_functions.CategoricalLogitsNegativeLogProbLoss(logits)
[all …]
|
| /external/tensorflow/tensorflow/python/estimator/canned/ |
| D | head_test.py | 85 def _sigmoid(logits): argument
86 return 1 / (1 + np.exp(-logits))
115 def _loss_fn(logits): argument
116 del logits # Unused
135 def _loss_fn(labels, logits, features): argument
136 del labels, logits, features # Unused
141 def _loss_fn(labels, logits, name=None): argument
142 del labels, logits, name # Unused
162 logits=logits_2x2)
169 logits=logits_placeholder)
[all …]
|
| D | head.py | 149 def create_loss(self, features, mode, logits, labels): argument
177 self, features, mode, logits, labels=None, train_op_fn=None, argument
204 labels, logits, expected_labels_dimension):
229 with ops.name_scope(None, 'labels', (labels, logits)) as scope:
243 if (labels.shape.ndims is not None and logits.shape.ndims is not None and
244 labels.shape.ndims == logits.shape.ndims - 1):
247 logits_shape = array_ops.shape(logits)
275 features, weight_column, logits, allow_per_logit_weights=False):
305 values=tuple(six.itervalues(features)) + (logits,)) as scope:
324 logits_shape = array_ops.shape(logits, name='logits_shape')
[all …]
|
| /external/tensorflow/tensorflow/contrib/learn/python/learn/estimators/ |
| D | head.py | 144 logits=None, argument
542 def _mean_squared_loss(labels, logits, weights=None): argument
543 with ops.name_scope(None, "mean_squared_loss", (logits, labels)) as name:
544 logits = ops.convert_to_tensor(logits)
550 if len(logits.get_shape()) == 1:
551 logits = array_ops.expand_dims(logits, dim=(1,))
552 logits.get_shape().assert_is_compatible_with(labels.get_shape())
553 loss = math_ops.square(logits - math_ops.to_float(labels), name=name)
557 def _poisson_loss(labels, logits, weights=None): argument
559 with ops.name_scope(None, "_poisson_loss", (logits, labels)) as name:
[all …]
|
| /external/tensorflow/tensorflow/contrib/layers/python/layers/ |
| D | target_column.py | 163 def logits_to_predictions(self, logits, proba=False): argument
167 def get_eval_ops(self, features, logits, labels, metrics=None): argument
202 def training_loss(self, logits, target, features, name="training_loss"): argument
224 loss_unweighted = self._loss_fn(logits, target)
232 def loss(self, logits, target, features): argument
249 loss_unweighted = self._loss_fn(logits, target)
271 def logits_to_predictions(self, logits, proba=False): argument
273 return array_ops.squeeze(logits, squeeze_dims=[1])
274 return logits
276 def get_eval_ops(self, features, logits, labels, metrics=None): argument
[all …]
|
| /external/tensorflow/tensorflow/contrib/boosted_trees/python/utils/ |
| D | losses.py | 42 labels=labels, logits=predictions)
49 def per_example_maxent_loss(labels, weights, logits, num_classes, eps=1e-15): argument
80 unnormalized_probs = math_ops.exp(logits)
89 zeros = array_ops.zeros_like(probs_for_real_class, dtype=logits.dtype) + eps
91 probs_for_real_class, dtype=logits.dtype) - eps
154 def exp_with_logits(name, eps, labels=None, logits=None): argument
175 with ops.name_scope(name, "exp_loss", [logits, labels]) as name:
176 logits = ops.convert_to_tensor(logits, name="logits")
179 labels.get_shape().merge_with(logits.get_shape())
182 % (logits.get_shape(), labels.get_shape()))
[all …]
|
| /external/tensorflow/tensorflow/contrib/losses/python/losses/ |
| D | loss_ops_test.py | 116 logits = constant_op.constant([[10.0, 0.0, 0.0],
124 loss_ops.softmax_cross_entropy(logits, labels, weights=None)
128 logits = constant_op.constant([[10.0, 0.0, 0.0],
134 loss = loss_ops.softmax_cross_entropy(logits, labels)
139 logits = constant_op.constant([[10.0, 0.0, 0.0],
147 loss = loss_ops.softmax_cross_entropy(logits, labels)
152 logits = constant_op.constant([[10.0, 0.0, 0.0],
160 loss = loss_ops.softmax_cross_entropy(logits, labels, weights)
164 logits = constant_op.constant([[10.0, 0.0, 0.0],
172 loss = loss_ops.softmax_cross_entropy(logits, labels,
[all …]
|
| /external/tensorflow/tensorflow/contrib/sparsemax/python/ops/ |
| D | sparsemax.py | 30 def sparsemax(logits, name=None): argument
47 with ops.name_scope(name, "sparsemax", [logits]) as name:
48 logits = ops.convert_to_tensor(logits, name="logits")
49 obs = array_ops.shape(logits)[0]
50 dims = array_ops.shape(logits)[1]
52 z = logits - math_ops.reduce_mean(logits, axis=1)[:, array_ops.newaxis]
60 1, math_ops.cast(dims, logits.dtype) + 1, dtype=logits.dtype)
69 tau_z = (tau_sum - 1) / math_ops.cast(k_z, logits.dtype)
73 math_ops.cast(0, logits.dtype), z - tau_z[:, array_ops.newaxis])
|
| /external/tensorflow/tensorflow/python/kernel_tests/ |
| D | losses_test.py | 110 logits = constant_op.constant([[10.0, 0.0, 0.0], [0.0, 10.0, 0.0],
115 losses.softmax_cross_entropy(labels, logits, weights=None)
119 logits = constant_op.constant([[10.0, 0.0, 0.0], [0.0, 10.0, 0.0],
122 loss = losses.softmax_cross_entropy(labels, logits)
127 logits = constant_op.constant([[10.0, 0.0, 0.0], [0.0, 10.0, 0.0],
132 loss = losses.softmax_cross_entropy(labels, logits)
137 logits = constant_op.constant([[10.0, 0.0, 0.0], [0.0, 10.0, 0.0],
142 loss = losses.softmax_cross_entropy(labels, logits, weights)
146 logits = constant_op.constant([[10.0, 0.0, 0.0], [0.0, 10.0, 0.0],
151 loss = losses.softmax_cross_entropy(labels, logits,
[all …]
|
| D | sparse_xent_op_test.py | 146 labels=[[0, 2]], logits=[[0., 1.], [2., 3.], [2., 3.]])
152 labels=constant_op.constant(0), logits=constant_op.constant(1.0))
158 labels=labels, logits=[[7.]])
165 labels=constant_op.constant(0), logits=constant_op.constant([1.0]))
198 labels=l, logits=f, name="xent")
209 logits = math_ops.matmul(images_placeholder, weights_with_zeros)
211 labels=labels_placeholder, logits=logits)
226 labels=labels, logits=features)
248 logits = array_ops.placeholder(dtypes.float32, shape=[None, 3])
250 labels=array_ops.squeeze(labels), logits=logits)
[all …]
|
| /external/tensorflow/tensorflow/python/kernel_tests/random/ |
| D | multinomial_op_test.py | 40 def composed_sampler(logits, num_samples): argument
42 unif = random_ops.random_uniform(logits.get_shape().concatenate(
46 logits = array_ops.expand_dims(logits, -1)
49 return math_ops.argmax(logits + noise, dimension=1)
63 logits = constant_op.constant([[-10., 10., -10.], [-10., -10., 10.]])
66 logits, num_samples, output_dtype=output_dtype))
100 logits = np.array([[1000.] * 5])
102 logits *= -1
103 samples = random_ops.multinomial(logits, 10).eval()
119 logits = np.log(probs).astype(np.float32)
[all …]
|
| /external/tensorflow/tensorflow/contrib/nn/python/ops/ |
| D | cross_entropy.py | 28 def deprecated_flipped_softmax_cross_entropy_with_logits(logits, argument
68 labels=labels, logits=logits, dim=dim, name=name)
75 def deprecated_flipped_sparse_softmax_cross_entropy_with_logits(logits, argument
122 labels=labels, logits=logits, name=name)
129 def deprecated_flipped_sigmoid_cross_entropy_with_logits(logits, argument
177 labels=targets, logits=logits, name=name)
|
| /external/tensorflow/tensorflow/python/ops/losses/ |
| D | losses_impl.py | 320 def hinge_loss(labels, logits, weights=1.0, scope=None, argument
346 if logits is None:
348 with ops.name_scope(scope, "hinge_loss", (logits, labels, weights)) as scope:
349 logits = math_ops.to_float(logits)
351 logits.get_shape().assert_is_compatible_with(labels.get_shape())
356 math_ops.subtract(all_ones, math_ops.multiply(labels, logits)))
620 multi_class_labels, logits, weights=1.0, label_smoothing=0, scope=None, argument
658 if logits is None:
661 (logits, multi_class_labels, weights)) as scope:
662 logits = ops.convert_to_tensor(logits)
[all …]
|