/external/tensorflow/tensorflow/contrib/keras/api/keras/backend/ |
H A D | __init__.py | 53 from tensorflow.python.keras._impl.keras.backend import dropout namespace
|
/external/tensorflow/tensorflow/contrib/learn/python/learn/estimators/ |
H A D | dnn_linear_combined.py | 260 net = layers.dropout( 263 # TODO(b/31209633): Consider adding summary before dropout.
|
/external/tensorflow/tensorflow/python/keras/backend/ |
H A D | __init__.py | 53 from tensorflow.python.keras._impl.keras.backend import dropout namespace
|
/external/tensorflow/tensorflow/stream_executor/ |
H A D | stream_executor_pimpl.h | 375 dnn::RnnMode rnn_mode, dnn::DataType data_type, float dropout,
|
H A D | dnn.h | 1967 // dropout: the dropout threshold between layers. When it is 0., no dropout 1969 // seed: a seed for initializing the dropout layers. 1971 // for dropout layer. The user has to maintain the memory until the model 1978 float dropout, uint64 seed, 1974 createRnnDescriptor(int num_layers, int hidden_size, int input_size, dnn::RnnInputMode input_mode, dnn::RnnDirectionMode direction_mode, dnn::RnnMode rnn_mode, dnn::DataType data_type, float dropout, uint64 seed, ScratchAllocator* state_allocator) argument
|
/external/tensorflow/tensorflow/contrib/layers/python/layers/ |
H A D | layers_test.py | 1324 output = _layers.dropout(images) 1325 self.assertEqual(output.op.name, 'Dropout/dropout/mul') 1334 output = _layers.dropout(images, is_training=is_training) 1342 output = _layers.dropout(images, is_training=is_training) 1350 output = _layers.dropout(images, is_training=is_training) 1358 output = _layers.dropout(images, outputs_collections='outputs') 1369 output = _layers.dropout(images) 1381 output1 = _layers.dropout(images, seed=1) 1382 output2 = _layers.dropout(images, seed=1) 1391 output = _layers.dropout(image [all...] |
H A D | layers.py | 63 'dropout', 'elu', 'flatten', 'fully_connected', 'GDN', 'gdn', 1430 def dropout(inputs, function 1437 """Returns a dropout op applied to the input. 1444 inputs: The tensor to pass to the nn.dropout op. 1450 is in training mode. If so, dropout is applied and values scaled.
|
/external/tensorflow/tensorflow/stream_executor/cuda/ |
H A D | cuda_dnn.cc | 979 float dropout, uint64 seed, 984 CUDNN_RETURN_IF_FAIL(status, "Failed to create dropout descriptor"); 986 if (dropout == 0.f) { 995 CUDNN_RETURN_IF_FAIL(status, "Failed to query dropout state sizes"); 1002 port::StrCat("Failed to allocate Cudnn dropout state memory of ", 1010 dropout, state_memory.opaque(), 1014 "Failed to set dropout descriptor with state memory size: ", 1022 CUDNN_RETURN_IF_FAIL(status, "Failed to destroy Cudnn dropout handle: "); 1081 float dropout, uint64 seed, 1092 // Create the dropout handl 978 CudnnDropoutDescriptor(CUDAExecutor* parent, cudnnHandle_t cudnn_handle, float dropout, uint64 seed, ScratchAllocator* state_allocator) argument 1076 CudnnRnnDescriptor(CUDAExecutor* parent, cudnnHandle_t cudnn_handle, int num_layers, int hidden_size, int input_size, cudnnRNNInputMode_t input_mode, cudnnDirectionMode_t direction_mode, cudnnRNNMode_t rnn_mode, cudnnDataType_t data_type, float dropout, uint64 seed, ScratchAllocator* state_allocator) argument 1755 createRnnDescriptor(int num_layers, int hidden_size, int input_size, dnn::RnnInputMode input_mode, dnn::RnnDirectionMode direction_mode, dnn::RnnMode rnn_mode, dnn::DataType data_type, float dropout, uint64 seed, ScratchAllocator* state_allocator) argument [all...] |
H A D | cuda_dnn.h | 53 dnn::RnnMode rnn_mode, dnn::DataType data_type, float dropout,
|
/external/tensorflow/tensorflow/python/estimator/canned/ |
H A D | dnn_testing_utils.py | 512 dropout=None, 693 dropout=None,
|
/external/tensorflow/tensorflow/contrib/boosted_trees/kernels/ |
H A D | training_ops.cc | 302 // Determine whether dropout was used when building this tree. 306 dropout_config_ = learner_config_.learning_rate_tuner().dropout(); 346 // Read seed that was used for dropout. 393 // If the tree is fully built and dropout was applied, it also adjusts the 471 // It is possible that the tree was built with dropout. If it is the case,
|
/external/tensorflow/tensorflow/python/keras/_impl/keras/ |
H A D | backend.py | 3446 @tf_export('keras.backend.dropout') 3447 def dropout(x, level, noise_shape=None, seed=None): function 3466 return nn.dropout(x * 1., retain_prob, noise_shape, seed=seed)
|
/external/tensorflow/tensorflow/python/ops/ |
H A D | nn_ops.py | 2242 @tf_export("nn.dropout") 2243 def dropout(x, keep_prob, noise_shape=None, seed=None, name=None): # pylint: disable=invalid-name function 2244 """Computes dropout. 2276 with ops.name_scope(name, "dropout", [x]) as name:
|
/external/tensorflow/tensorflow/contrib/rnn/python/ops/ |
H A D | rnn_cell.py | 1335 """LSTM unit with layer normalization and recurrent dropout. 1337 This class adds layer normalization and recurrent dropout to a 1346 Recurrent dropout is base on: 1378 recurrent dropout probability value. If float and 1.0, no dropout will 1431 """LSTM cell with layer normalization and recurrent dropout.""" 1446 g = nn_ops.dropout(g, self._keep_prob, seed=self._seed)
|