Home
last modified time | relevance | path

Searched refs:activations (Results 1 – 25 of 96) sorted by relevance

1234

/external/tensorflow/tensorflow/contrib/keras/api/keras/activations/
D__init__.py22 from tensorflow.python.keras.activations import elu
23 from tensorflow.python.keras.activations import hard_sigmoid
24 from tensorflow.python.keras.activations import linear
25 from tensorflow.python.keras.activations import relu
26 from tensorflow.python.keras.activations import selu
27 from tensorflow.python.keras.activations import sigmoid
28 from tensorflow.python.keras.activations import softmax
29 from tensorflow.python.keras.activations import softplus
30 from tensorflow.python.keras.activations import softsign
31 from tensorflow.python.keras.activations import tanh
[all …]
/external/tensorflow/tensorflow/python/keras/
Dactivations_test.py43 fn = keras.activations.get(name)
44 ref_fn = getattr(keras.activations, name)
46 config = keras.activations.serialize(fn)
47 fn = keras.activations.deserialize(config)
53 fn_v2 = keras.activations.get(fn_v2_key)
54 config = keras.activations.serialize(fn_v2)
55 fn = keras.activations.deserialize(config)
60 f = keras.backend.function([x], [keras.activations.softmax(x)])
69 keras.activations.softmax(x)
73 f = keras.backend.function([x], [keras.activations.softmax(x)])
[all …]
/external/tensorflow/tensorflow/core/kernels/
Drelu_op_functor.h34 typename TTypes<T>::Tensor activations) { in operator()
35 activations.device(d) = features.cwiseMax(static_cast<T>(0)); in operator()
67 typename TTypes<T>::Tensor activations) { in operator()
68 activations.device(d) = in operator()
102 T alpha, typename TTypes<T>::Tensor activations) { in operator()
103 activations.device(d) = features.cwiseMax(features * alpha); in operator()
132 typename TTypes<T>::Tensor activations) { in operator()
134 activations.device(d) = in operator()
150 typename TTypes<T>::ConstTensor activations, in operator()
153 (activations < static_cast<T>(0)) in operator()
[all …]
Drelu_op.cc84 typename TTypes<T>::Tensor activations); \
97 typename TTypes<T>::Tensor activations); \
110 typename TTypes<T>::Tensor activations); \
123 typename TTypes<T>::Tensor activations); \
129 typename TTypes<T>::ConstTensor activations, \
136 typename TTypes<T>::Tensor activations); \
142 typename TTypes<T>::ConstTensor activations, \
149 typename TTypes<qint8>::Tensor activations);
Dsoftsign_op.h35 typename TTypes<T>::Tensor activations) { in operator()
36 activations.device(d) = in operator()
Dsoftplus_op.h35 typename TTypes<T>::Tensor activations) { in operator()
50 activations.device(d) = too_large.select( in operator()
/external/tensorflow/tensorflow/contrib/learn/python/learn/estimators/
Drnn_common.py182 def select_last_activations(activations, sequence_lengths): argument
196 'select_last_activations', values=[activations, sequence_lengths]):
197 activations_shape = array_ops.shape(activations)
203 reshaped_activations = array_ops.reshape(activations,
208 [activations.get_shape()[0], activations.get_shape()[2]])
212 def mask_activations_and_labels(activations, labels, sequence_lengths): argument
232 values=[activations, labels, sequence_lengths]):
238 activations_masked = array_ops.reshape(activations,
243 activations_masked = array_ops.boolean_mask(activations, mask)
248 def multi_value_predictions(activations, target_column, problem_type, argument
[all …]
Drnn_common_test.py39 activations = np.random.rand(batch_size, padded_length, num_classes)
43 constant_op.constant(activations, dtype=dtypes.float32),
65 expected_activations = activations[i, j, :]
90 activations = np.random.rand(batch_size, padded_length, num_classes)
92 constant_op.constant(activations, dtype=dtypes.float32),
106 expected_activations = activations[i, sequence_length[i] - 1, :]
Ddynamic_rnn_estimator.py252 activations = layers.fully_connected(
257 return activations, final_state
260 def _single_value_predictions(activations, argument
290 activations, sequence_length)
308 activations, labels, sequence_length, target_column, features): argument
326 activations, labels, sequence_length)
331 activations, labels, sequence_length, target_column, features): argument
350 activations, sequence_length)
Dstate_saving_rnn_estimator.py81 activations = layers.fully_connected(
89 return activations, final_state
93 activations, labels, sequence_length, target_column, features): argument
111 activations, labels, sequence_length)
/external/tensorflow/tensorflow/lite/kernels/
Dactivations.cc38 namespace activations { namespace
967 activations::GenericPrepare, in Register_ELU()
968 activations::EluEval}; in Register_ELU()
974 activations::GenericPrepare, in Register_RELU()
975 activations::ReluEval}; in Register_RELU()
981 activations::GenericPrepare, in Register_RELU_N1_TO_1()
982 activations::Relu1Eval}; in Register_RELU_N1_TO_1()
988 activations::GenericPrepare, in Register_RELU6()
989 activations::Relu6Eval}; in Register_RELU6()
995 activations::Init, activations::Free, activations::TanhPrepare, in Register_TANH_REF()
[all …]
/external/tensorflow/tensorflow/python/tpu/
Dtpu_embedding_gradient.py30 def get_gradients_through_compute_gradients(optimizer, loss, activations): argument
43 activation_list = activations.values()
47 zip(activations.keys(), grads))
101 def hook_dummy_table_variables_to_activations(tpu_embedding, activations, argument
117 for feature in activations:
121 activations[feature],
/external/tensorflow/tensorflow/compiler/xla/service/gpu/
Dcudnn_conv_rewriter_test.cc100 HloInstruction* activations = in TEST_F() local
111 activations->shape(), gradients->shape(), /*feature_group_count=*/1, in TEST_F()
115 activations, gradients, /*feature_group_count=*/1, in TEST_F()
141 HloInstruction* activations = in TEST_F() local
151 activations->shape(), gradients->shape(), /*feature_group_count=*/1, in TEST_F()
155 activations, gradients, /*feature_group_count=*/1, in TEST_F()
171 HloInstruction* activations = in TEST_F() local
185 ShapeUtil::MakeShape(F32, {32, 3, 3, 32}), activations, gradients, in TEST_F()
201 HloInstruction* activations = in TEST_F() local
215 ShapeUtil::MakeShape(F32, {320, 3, 3, 192}), activations, gradients, in TEST_F()
[all …]
/external/tensorflow/tensorflow/lite/experimental/micro/kernels/
Dsoftmax.cc28 namespace activations { namespace
205 static TfLiteRegistration r = {activations::Init, activations::Free, in Register_SOFTMAX()
206 activations::SoftmaxPrepare, in Register_SOFTMAX()
207 activations::SoftmaxEval}; in Register_SOFTMAX()
/external/tensorflow/tensorflow/core/api_def/base_api/
Dapi_def_TPUEmbeddingActivations.pbtxt13 The embedding activations Tensor to return.
20 these activations were computed.
27 activations.
Dapi_def_RecvTPUEmbeddingActivations.pbtxt7 A TensorList of embedding activations containing one Tensor per
24 summary: "An op that receives embedding activations on the TPU."
30 one Tensor of activations per table specified in the model. There can be at
/external/tensorflow/tensorflow/lite/tools/optimize/testdata/
DREADME.md11 All activations have min maxes and activations are in range [0,10].
16 as 127. The activations are all in range: [-128, 127].
/external/tensorflow/tensorflow/contrib/gan/python/eval/python/
Dclassifier_metrics_impl.py270 activations = run_image_classifier(images, graph_def, input_tensor,
272 if isinstance(activations, list):
273 for i, activation in enumerate(activations):
275 activations[i] = layers.flatten(activation)
277 if array_ops.rank(activations) != 2:
278 activations = layers.flatten(activations)
280 return activations
/external/tensorflow/tensorflow/lite/g3doc/performance/
Dmodel_optimization.md37 representations of weights and, optionally, activations for both storage and
41 * Quantization of activations reduces memory access costs for reading and storing intermediate acti…
46 * [Post-training quantization](post_training_quantization.md) quantizes weights and activations pos…
/external/tensorflow/tensorflow/compiler/tf2xla/kernels/
Dbatch_norm_op.cc135 auto activations = in Compile() local
166 xla::BatchNormGrad(activations, scale, mean, var, grad_backprop, in Compile()
198 xla::Mul(grad_backprop, xla::Sub(activations, mean, {feature_index})); in Compile()
Dconv_op_helpers.cc395 StringPiece type_string, xla::XlaOp activations, in MakeXlaBackpropFilterConvOp() argument
400 auto* builder = activations.builder(); in MakeXlaBackpropFilterConvOp()
402 builder->GetShape(activations)); in MakeXlaBackpropFilterConvOp()
533 activations, gradients, window_strides, padding, /*lhs_dilation=*/ones, in MakeXlaBackpropFilterConvOp()
540 filter_shape, filter_backprop, activations.builder()); in MakeXlaBackpropFilterConvOp()
/external/tensorflow/tensorflow/examples/tutorials/mnist/
Dmnist_with_summaries.py97 activations = act(preactivate, name='activation')
98 tf.summary.histogram('activations', activations)
99 return activations
/external/tensorflow/tensorflow/python/keras/layers/
Dlocal.py21 from tensorflow.python.keras import activations
146 self.activation = activations.get(activation)
272 activations.serialize(self.activation),
421 self.activation = activations.get(activation)
560 activations.serialize(self.activation),
/external/gemmlowp/doc/
Dpublic.md120 multiplication), while the RHS and result are neural network activations,
121 respectively the input and output activations of the layer.
123 Because the RHS and result are activations, we want them to share the same
124 storage order -- so that one layer's output activations can be readily used as
125 the next layer's input activations. Thus, we focus on `RhsOrder=ResultOrder`.
/external/tensorflow/tensorflow/core/protobuf/tpu/
Dtpu_embedding_configuration.proto35 // Number of samples in each batch of embedding layer activations sent to
73 // that the activations on every step observe the gradient updates from the
83 // is complete. The drawback is that embedding activations for step N+1 do not

1234