From f978fbce4eb6fa8583b92e86975fa095299a6ef2 Mon Sep 17 00:00:00 2001 From: hong19860320 <9973393+hong19860320@users.noreply.github.com> Date: Sat, 12 Oct 2019 14:09:00 +0800 Subject: [PATCH] Refine the en api doc of ones, zeros, reverse, assign, increment, hsigmoid and create_py_reader_by_data ops (#20343) (#20563) test=develop, test=release/1.6, test=document_fix * refine the en api doc of ones, zeros, reverse, increment, hsigmoid and create_py_reader_by_data ops test=develop, test=document_preview, test=document_fix * refine eng doc for hsigmoid and create_py_reader_by_data ops test=develop, test=document_preview, test=document_fix * update API.spec test=document_fix * Fix the parameter name axis of reverse op in eng doc test=develop, test=document_fix * Update API.spec test=develop, test=document_fix * Refine eng doc of zeros, ones, reverse and assign op test=develop, test=document_fix * Update API.spec for assign, ones, zeros and reverse test=develop, test=document_fix * Fix data type of reverse op in eng doc test=develop, test=document_fix * Update API.spec for reverse op test=develop, test=document_fix --- paddle/fluid/API.spec | 14 +-- python/paddle/fluid/layers/control_flow.py | 22 ++-- python/paddle/fluid/layers/io.py | 116 ++++++++++----------- python/paddle/fluid/layers/nn.py | 114 ++++++++++---------- python/paddle/fluid/layers/tensor.py | 93 ++++++++--------- 5 files changed, 179 insertions(+), 180 deletions(-) diff --git a/paddle/fluid/API.spec b/paddle/fluid/API.spec index 298f399a267..08f9a965d5f 100644 --- a/paddle/fluid/API.spec +++ b/paddle/fluid/API.spec @@ -180,7 +180,7 @@ paddle.fluid.layers.transpose (ArgSpec(args=['x', 'perm', 'name'], varargs=None, paddle.fluid.layers.im2sequence (ArgSpec(args=['input', 'filter_size', 'stride', 'padding', 'input_image_size', 'out_stride', 'name'], varargs=None, keywords=None, defaults=(1, 1, 0, None, 1, None)), ('document', 'fe352915a543cec434f74e9b32ac49da')) paddle.fluid.layers.nce (ArgSpec(args=['input', 'label', 'num_total_classes', 'sample_weight', 'param_attr', 'bias_attr', 'num_neg_samples', 'name', 'sampler', 'custom_dist', 'seed', 'is_sparse'], varargs=None, keywords=None, defaults=(None, None, None, None, None, 'uniform', None, 0, False)), ('document', '38297567127888e01542857839058d52')) paddle.fluid.layers.sampled_softmax_with_cross_entropy (ArgSpec(args=['logits', 'label', 'num_samples', 'num_true', 'remove_accidental_hits', 'use_customized_samples', 'customized_samples', 'customized_probabilities', 'seed'], varargs=None, keywords=None, defaults=(1, True, False, None, None, 0)), ('document', 'd4435a63d34203339831ee6a86ef9242')) -paddle.fluid.layers.hsigmoid (ArgSpec(args=['input', 'label', 'num_classes', 'param_attr', 'bias_attr', 'name', 'path_table', 'path_code', 'is_custom', 'is_sparse'], varargs=None, keywords=None, defaults=(None, None, None, None, None, False, False)), ('document', 'b83e7dfa81059b39bb137922dc914f50')) +paddle.fluid.layers.hsigmoid (ArgSpec(args=['input', 'label', 'num_classes', 'param_attr', 'bias_attr', 'name', 'path_table', 'path_code', 'is_custom', 'is_sparse'], varargs=None, keywords=None, defaults=(None, None, None, None, None, False, False)), ('document', '247de339879885526e7f4d271967088f')) paddle.fluid.layers.beam_search (ArgSpec(args=['pre_ids', 'pre_scores', 'ids', 'scores', 'beam_size', 'end_id', 'level', 'is_accumulated', 'name', 'return_parent_idx'], varargs=None, keywords=None, defaults=(0, True, None, False)), ('document', '2b505ddaa309fd7b9be5445e41ca76d5')) paddle.fluid.layers.row_conv (ArgSpec(args=['input', 'future_context_size', 'param_attr', 'act'], varargs=None, keywords=None, defaults=(None, None)), ('document', 'a6477957b44907787b3c74157400b80c')) paddle.fluid.layers.multiplex (ArgSpec(args=['inputs', 'index'], varargs=None, keywords=None, defaults=None), ('document', '2c4d1ae83da6ed35e3b36ba1b3b51d23')) @@ -312,7 +312,7 @@ paddle.fluid.layers.data (ArgSpec(args=['name', 'shape', 'append_batch_size', 'd paddle.fluid.layers.read_file (ArgSpec(args=['reader'], varargs=None, keywords=None, defaults=None), ('document', 'd5b41c7b2df1b064fbd42dcf435268cd')) paddle.fluid.layers.double_buffer (ArgSpec(args=['reader', 'place', 'name'], varargs=None, keywords=None, defaults=(None, None)), ('document', '556fa82daf62cbb0fb393f4125daba77')) paddle.fluid.layers.py_reader (ArgSpec(args=['capacity', 'shapes', 'dtypes', 'lod_levels', 'name', 'use_double_buffer'], varargs=None, keywords=None, defaults=(None, None, True)), ('document', 'd78a1c7344955c5caed8dc13adb7beb6')) -paddle.fluid.layers.create_py_reader_by_data (ArgSpec(args=['capacity', 'feed_list', 'name', 'use_double_buffer'], varargs=None, keywords=None, defaults=(None, True)), ('document', '2edf37d57862b24a7a26aa19a3573f73')) +paddle.fluid.layers.create_py_reader_by_data (ArgSpec(args=['capacity', 'feed_list', 'name', 'use_double_buffer'], varargs=None, keywords=None, defaults=(None, True)), ('document', '1321d4ce89d82f96fcfd5601f816b0f3')) paddle.fluid.layers.load (ArgSpec(args=['out', 'file_path', 'load_as_fp16'], varargs=None, keywords=None, defaults=(None,)), ('document', '309f9e5249463e1b207a7347b2a91134')) paddle.fluid.layers.create_tensor (ArgSpec(args=['dtype', 'name', 'persistable'], varargs=None, keywords=None, defaults=(None, False)), ('document', 'fdc2d964488e99fb0743887454c34e36')) paddle.fluid.layers.create_parameter (ArgSpec(args=['shape', 'dtype', 'name', 'attr', 'is_bias', 'default_initializer'], varargs=None, keywords=None, defaults=(None, None, False, None)), ('document', '021272f30e0cdf7503586815378abfb8')) @@ -321,15 +321,15 @@ paddle.fluid.layers.cast (ArgSpec(args=['x', 'dtype'], varargs=None, keywords=No paddle.fluid.layers.tensor_array_to_tensor (ArgSpec(args=['input', 'axis', 'name'], varargs=None, keywords=None, defaults=(1, None)), ('document', 'dd7d2f1e12a8a4225d017209866e5621')) paddle.fluid.layers.concat (ArgSpec(args=['input', 'axis', 'name'], varargs=None, keywords=None, defaults=(0, None)), ('document', 'ec7d6e716fb29ef1e73e1e3efa5ca46b')) paddle.fluid.layers.sums (ArgSpec(args=['input', 'out'], varargs=None, keywords=None, defaults=(None,)), ('document', '5df743d578638cd2bbb9369499b44af4')) -paddle.fluid.layers.assign (ArgSpec(args=['input', 'output'], varargs=None, keywords=None, defaults=(None,)), ('document', '8bd94aef4e123986d9a8c29f67b5532b')) +paddle.fluid.layers.assign (ArgSpec(args=['input', 'output'], varargs=None, keywords=None, defaults=(None,)), ('document', '98ce6e7c3659b8377c04cecfc72c2000')) paddle.fluid.layers.fill_constant_batch_size_like (ArgSpec(args=['input', 'shape', 'dtype', 'value', 'input_dim_idx', 'output_dim_idx'], varargs=None, keywords=None, defaults=(0, 0)), ('document', '37a288e4400f6d5510e982827461c11b')) paddle.fluid.layers.fill_constant (ArgSpec(args=['shape', 'dtype', 'value', 'force_cpu', 'out'], varargs=None, keywords=None, defaults=(False, None)), ('document', '66e1e468666dd47e5b2715226cebeac0')) paddle.fluid.layers.argmin (ArgSpec(args=['x', 'axis'], varargs=None, keywords=None, defaults=(0,)), ('document', '53629e27597e5dfb7020aac5bc639ebb')) paddle.fluid.layers.argmax (ArgSpec(args=['x', 'axis'], varargs=None, keywords=None, defaults=(0,)), ('document', 'd9a89fbedbaebd5f65897ac75ee636f3')) paddle.fluid.layers.argsort (ArgSpec(args=['input', 'axis', 'name'], varargs=None, keywords=None, defaults=(-1, None)), ('document', '8c7966eb4b37b2272a16717cac3a876c')) -paddle.fluid.layers.ones (ArgSpec(args=['shape', 'dtype', 'force_cpu'], varargs=None, keywords=None, defaults=(False,)), ('document', '812c623ed52610b9773f9fc05413bc34')) -paddle.fluid.layers.zeros (ArgSpec(args=['shape', 'dtype', 'force_cpu'], varargs=None, keywords=None, defaults=(False,)), ('document', '95379f9288c2d05356ec0e2375c6bc57')) -paddle.fluid.layers.reverse (ArgSpec(args=['x', 'axis'], varargs=None, keywords=None, defaults=None), ('document', '628135603692137d52bcf5a8d8d6816d')) +paddle.fluid.layers.ones (ArgSpec(args=['shape', 'dtype', 'force_cpu'], varargs=None, keywords=None, defaults=(False,)), ('document', '4c5de61e6920f59992d7d15452cf4248')) +paddle.fluid.layers.zeros (ArgSpec(args=['shape', 'dtype', 'force_cpu'], varargs=None, keywords=None, defaults=(False,)), ('document', '6fe7fc0956ffbfb00c05fb5ea5b575ea')) +paddle.fluid.layers.reverse (ArgSpec(args=['x', 'axis'], varargs=None, keywords=None, defaults=None), ('document', 'e735e3fe926f5cc9f68a2336fe229681')) paddle.fluid.layers.has_inf (ArgSpec(args=['x'], varargs=None, keywords=None, defaults=None), ('document', 'aca8a35516cef98af836fb6a64ac8acb')) paddle.fluid.layers.has_nan (ArgSpec(args=['x'], varargs=None, keywords=None, defaults=None), ('document', '99f4cf36db08a4e23c8c3857e2af1316')) paddle.fluid.layers.isfinite (ArgSpec(args=['x'], varargs=None, keywords=None, defaults=None), ('document', '9e40eab383fbe2d76e065345cb27f140')) @@ -346,7 +346,7 @@ paddle.fluid.layers.Switch ('paddle.fluid.layers.control_flow.Switch', ('documen paddle.fluid.layers.Switch.__init__ (ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=(None,)), ('document', '6adf97f83acf6453d4a6a4b1070f3754')) paddle.fluid.layers.Switch.case (ArgSpec(args=['self', 'condition'], varargs=None, keywords=None, defaults=None), ('document', '6adf97f83acf6453d4a6a4b1070f3754')) paddle.fluid.layers.Switch.default (ArgSpec(args=['self'], varargs=None, keywords=None, defaults=None), ('document', '6adf97f83acf6453d4a6a4b1070f3754')) -paddle.fluid.layers.increment (ArgSpec(args=['x', 'value', 'in_place'], varargs=None, keywords=None, defaults=(1.0, True)), ('document', 'f88b5787bb80ae6b8bf513a70dabbdc1')) +paddle.fluid.layers.increment (ArgSpec(args=['x', 'value', 'in_place'], varargs=None, keywords=None, defaults=(1.0, True)), ('document', 'c5fe17688e64c2d7e57312128d61da98')) paddle.fluid.layers.array_write (ArgSpec(args=['x', 'i', 'array'], varargs=None, keywords=None, defaults=(None,)), ('document', 'd357f71a280bf06aab4c79de9bd4facf')) paddle.fluid.layers.create_array (ArgSpec(args=['dtype'], varargs=None, keywords=None, defaults=None), ('document', '556de793fdf24d515f3fc91260e2c048')) paddle.fluid.layers.less_than (ArgSpec(args=['x', 'y', 'force_cpu', 'cond'], varargs=None, keywords=None, defaults=(None, None)), ('document', '329bdde01cba69463b08b8c13015560a')) diff --git a/python/paddle/fluid/layers/control_flow.py b/python/paddle/fluid/layers/control_flow.py index fb38b2286d2..591dbaabf41 100755 --- a/python/paddle/fluid/layers/control_flow.py +++ b/python/paddle/fluid/layers/control_flow.py @@ -896,26 +896,24 @@ def array_to_lod_tensor(x, table): def increment(x, value=1.0, in_place=True): """ - This function performs an operation that increments the value in the - input :math:`x` by an amount: :math:`value` as mentioned in the input - parameter. This operation is performed in-place by default. Notice that - the number of elements in :math:`x` must be equal to 1. + The OP is usually used for control flow to increment the data of :attr:`x` by an amount :attr:`value`. + Notice that the number of elements in :attr:`x` must be equal to 1. - Args: - x (Variable|list): The tensor that has the input values. - value (float): The amount by which the values should be incremented. - in_place (bool): If the increment should be performed in-place. + Parameters: + x (Variable): A tensor that must alway contain only one element, its data type supports + float32, float64, int32 and int64. + value (float, optional): The amount to increment the data of :attr:`x`. Default: 1.0. + in_place (bool, optional): Whether the OP should be performed in-place. Default: True. Returns: - Variable: The elementwise-incremented object. + Variable: The elementwise-incremented tensor with the same shape and data type as :attr:`x`. Examples: .. code-block:: python import paddle.fluid as fluid - data = fluid.layers.data(name='data', shape=[1], dtype='float32', - append_batch_size=False) - data = fluid.layers.increment(x=data, value=3.0, in_place=True) + counter = fluid.layers.zeros(shape=[1], dtype='float32') # [0.] + fluid.layers.increment(counter) # [1.] """ helper = LayerHelper("increment", **locals()) if not in_place: diff --git a/python/paddle/fluid/layers/io.py b/python/paddle/fluid/layers/io.py index 181e8792198..bf94555846c 100644 --- a/python/paddle/fluid/layers/io.py +++ b/python/paddle/fluid/layers/io.py @@ -682,70 +682,68 @@ def create_py_reader_by_data(capacity, name=None, use_double_buffer=True): """ - Create a Python reader for data feeding in Python - - This layer returns a Reader Variable. - - Works much like py_reader except that it's input is feed_list - instead of shapes, dtypes and lod_levels - - Args: - capacity(int): The buffer capacity maintained by :code:`py_reader`. - feed_list(list(Variable)): The data feed list. - name(basestring): The prefix Python queue name and Reader name. None will - be generated automatically. - use_double_buffer(bool): Whether use double buffer or not. + The OP creates a Python reader for data feeding in Python, it is similar + to :ref:`api_fluid_layers_py_reader` except that it can read data from + the list of feed variables. + + Parameters: + capacity (int): The buffer capacity maintained by :code:`py_reader`. Its unit + is batch number. Set larger :attr:`capacity` if the reader is fast. + feed_list (list(Variable)): The feed variables, are usually created by + :code:`fluid.data()`. + name (str, optional): Normally there is no need for user to set this property. + For more information, please refer to :ref:`api_guide_Name`. Default: None. + use_double_buffer (bool, optional): Whether use double buffer. If it's True, + the OP would prefetch next batch data asynchronously. Default: True. Returns: - Variable: A Reader from which we can get feeding data. + Reader: A Reader for data feeding. The data types of read data are the same as the data types of variables of :attr:`feed_list`. Examples: - .. code-block:: python - - import paddle - import paddle.fluid as fluid - import paddle.dataset.mnist as mnist - import paddle.fluid.compiler as compiler - - def network(img, label): - # User defined network. Here a simple regression as example - predict = fluid.layers.fc(input=img, size=10, act='softmax') - loss = fluid.layers.cross_entropy(input=predict, label=label) - return fluid.layers.mean(loss) - - MEMORY_OPT = False - USE_CUDA = False - - image = fluid.layers.data(name='image', shape=[1, 28, 28], dtype='float32') - label = fluid.layers.data(name='label', shape=[1], dtype='int64') - reader = fluid.layers.create_py_reader_by_data(capacity=64, - feed_list=[image, label]) - reader.decorate_paddle_reader( - paddle.reader.shuffle(paddle.batch(mnist.train(), batch_size=5), - buf_size=500)) - - img, label = fluid.layers.read_file(reader) - loss = network(img, label) # some network definition - - place = fluid.CUDAPlace(0) if USE_CUDA else fluid.CPUPlace() - exe = fluid.Executor(place) - exe.run(fluid.default_startup_program()) - - build_strategy = fluid.BuildStrategy() - build_strategy.memory_optimize = True if MEMORY_OPT else False - compiled_prog = compiler.CompiledProgram( - fluid.default_main_program()).with_data_parallel( - loss_name=loss.name, - build_strategy=build_strategy, - exec_strategy=exec_strategy) + .. code-block:: python - for epoch_id in range(2): - reader.start() - try: - while True: - exe.run(compiled_prog, fetch_list=[loss.name]) - except fluid.core.EOFException: - reader.reset() + import paddle + import paddle.fluid as fluid + import paddle.dataset.mnist as mnist + + def network(img, label): + # User defined network. Here a simple regression as example + predict = fluid.layers.fc(input=img, size=10, act='softmax') + loss = fluid.layers.cross_entropy(input=predict, label=label) + return fluid.layers.mean(loss) + + MEMORY_OPT = False + USE_CUDA = False + + image = fluid.data(name='image', shape=[None, 1, 28, 28], dtype='float32') + label = fluid.data(name='label', shape=[None, 1], dtype='int64') + reader = fluid.layers.create_py_reader_by_data(capacity=64, + feed_list=[image, label]) + reader.decorate_paddle_reader( + paddle.reader.shuffle(paddle.batch(mnist.train(), batch_size=5), buf_size=500)) + img, label = fluid.layers.read_file(reader) + loss = network(img, label) # The definition of custom network and the loss funtion + + place = fluid.CUDAPlace(0) if USE_CUDA else fluid.CPUPlace() + exe = fluid.Executor(place) + exe.run(fluid.default_startup_program()) + + build_strategy = fluid.BuildStrategy() + build_strategy.memory_optimize = True if MEMORY_OPT else False + exec_strategy = fluid.ExecutionStrategy() + compiled_prog = fluid.compiler.CompiledProgram( + fluid.default_main_program()).with_data_parallel( + loss_name=loss.name, + build_strategy=build_strategy, + exec_strategy=exec_strategy) + + for epoch_id in range(2): + reader.start() + try: + while True: + exe.run(compiled_prog, fetch_list=[loss.name]) + except fluid.core.EOFException: + reader.reset() """ logging.warn( 'paddle.fluid.layers.create_py_reader_by_data() may be deprecated in the near future. ' diff --git a/python/paddle/fluid/layers/nn.py b/python/paddle/fluid/layers/nn.py index 29f2d2057b2..a4d4a780d83 100755 --- a/python/paddle/fluid/layers/nn.py +++ b/python/paddle/fluid/layers/nn.py @@ -7528,70 +7528,74 @@ def hsigmoid(input, is_custom=False, is_sparse=False): """ - The hierarchical sigmoid operator is used to accelerate the training - process of language model. This operator organizes the classes into a - complete binary tree, or you can use is_custom to pass your own tree to - implement hierarchical. Each leaf node represents a class(a word) and each - internal node acts as a binary classifier. For each word there's a unique - path from root to it's leaf node, hsigmoid calculate the cost for each - internal node on the path, and sum them to get a total cost. hsigmoid can - achive a acceleration from :math:`O(N)` to :math:`O(logN)`, where :math:`N` - represents the size of word dict. - - Using default tree you can Refer to `Hierarchical Probabilistic Neural Network Language Model - `_ - - And if you want to use the costumed tree by set 'is_custom' as true you may need to do following things first: - - 1. using your word dict to build a binary tree, each leaf node should be an word of your word dict - 2. build a dict to store word_id -> word's leaf to root path, we call it path_table. - 3. build a dict to store word_id -> code of word's leaf to root path, we call it path_code. Code - means label of each binary classification, using 1 indicate true, 0 indicate false. - 4. now, each word should has its path and code along the path, you can pass a batch of path and code - related to the same batch of inputs. + The hierarchical sigmoid organizes the classes into a complete binary tree to reduce the computational complexity + and speed up the model training, especially the training of language model. + Each leaf node of the complete binary tree represents a class(word) and each non-leaf node acts as a binary classifier. + For each class(word), there's a unique path from root to itself, hsigmoid calculate the cost for each non-leaf node on + the path, and sum them to get a total cost. + Comparing to softmax, the OP can reduce the computational complexity from :math:`O(N)` to :math:`O(logN)`, where :math:`N` + represents the number of classes or the size of word dict. + + The OP supports default tree and custom tree. For the default tree, you can refer to `Hierarchical Probabilistic Neural + Network Language Model `. For the custom + tree, you need to set :attr:`is_custom` to True, and do the following steps (take the language model as an example): + + 1. Using a custom word dict to build a binary tree, each leaf node should be an word in the word dict. + 2. Creating a dict map word_id -> path that from the word to the root node, we call it path_table. + 3. Creating a dict map word_id -> code of path that from the word to the root node, we call it path_code. + Code means the label of each binary classifier, 1 indicate true, 0 indicate false. + 4. Now, each word should has its path and code along the path, you can pass a batch of path and code related + to the same batch of inputs. - Args: - input (Variable): The input tensor variable with shape - :math:`[N \\times D]`, where :math:`N` is the size of mini-batch, - and :math:`D` is the feature size. - label (Variable): The tensor variable contains labels of training data. - It's a tensor with shape is :math:`[N \\times 1]`. - num_classes: (int), The number of classes, must not be less than 2. with default tree this has to be set, - it should never be None under is_custom=False, but while is_custom is true, it should be non leaf num - which indicates the num of classes using by binary classify. - param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights - of hsigmoid. If it is set to None or one attribute of ParamAttr, hsigmoid - will create ParamAttr as param_attr. If the Initializer of the param_attr - is not set, the parameter is initialized with Xavier. Default: None. - bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of hsigmoid. - If it is set to False, no bias will be added to the output units. - If it is set to None or one attribute of ParamAttr, hsigmoid - will create ParamAttr as bias_attr. If the Initializer of the bias_attr - is not set, the bias is initialized zero. Default: None. - name (str|None): A name for this layer(optional). If set None, the layer - will be named automatically. Default: None. - path_table: (Variable|None) this variable can store each batch of samples' path to root, - it should be in leaf -> root order - path_table should have the same shape with path_code, and for each sample i path_table[i] indicates a np.array like - structure and each element in this array is indexes in parent nodes' Weight Matrix. - path_code: (Variable|None) this variable can store each batch of samples' code, - each code consist with every code of parent nodes. it should be in leaf -> root order - is_custom: (bool|False)using user defined binary tree instead of default complete binary tree, if costum is - set you need to set path_table/path_code/num_classes, otherwise num_classes should be set - is_sparse: (bool|False)using sparse update instead of dense update, if set, the gradient - of W and input will be sparse. + Parameters: + input (Variable): A tensor with the shape [N, D], where N is the size of mini-batch, + and D is the feature size. Its data type supports float32 and float64. + label (Variable): A tensor contains the labels of training data. Its shape is [N, 1] + and data type is int64. + num_classes (int): The number of classes or the size of word dict, must be greater than 2. + If the default tree is used (:attr:`is_custom` is set to False), :attr:`num_classes` + should not be None. If the custom tree is used (:attr:`is_custom` is set to True), + :attr:`num_classes` should be the number of non-leaf nodes, which indicates the num of + classes using by the binary classifier. + param_attr (ParamAttr, optional): The parameter attribute for the learnable parameters/weights + of hsigmoid. If it is set to None or one attribute of ParamAttr, hsigmoid will create a + ParamAttr as param_attr. If the Initializer of the param_attr is not set, the parameter is + initialized with Xavier. Default: None. + bias_attr (ParamAttr|bool, optional): The parameter attribute for the bias of hsigmoid. If it + is set to False, no bias will be added. If it is set to None or one attribute of ParamAttr, + hsigmoid will create a ParamAttr as bias_attr. If the Initializer of the bias_attr is not + set, the bias is initialized zero. Default: None. + name (str, optional): Normally there is no need for user to set this property. For more information, + please refer to :ref:`api_guide_Name`. Default: None. + path_table (Variable, optional): A tensor that stores each batch of samples' path from leaf to root + node, its shape is [N, L] and data type is int64, where L is the length of path. For each sample i, + path_table[i] is a np.array like structure and each element in this array is the indexes in parent + nodes' weight matrix. Default: None. + path_code (Variable, optional): A tensor that stores each batch of samples' code of path from leaf + to root node, its shape is [N, L] and data type is int64, which is the same as :attr:`path_table`. + Each code of path is consisted with the code of nodes from leaf to root node. Default: None. + is_custom (bool, optional): Whether use custom binary tree. If it's True, :attr:`path_table`, + :attr:`path_code` and :attr:`num_classes` should be set, otherwise :attr:`num_classes` should + be set. Default: False. + is_sparse (bool, optional): Whether use sparse updating instead of dense updating, if it's True, the + gradient of W and input will be sparse. Default: False. Returns: - Out: (LodTensor) The cost of hierarchical sigmoid operator. the shape is [N, 1] + Variable: A tensor with the cost of hierarchical sigmoid, its shape is [N, 1] and data type is the same as :attr:`input`. Examples: .. code-block:: python import paddle.fluid as fluid - x = fluid.layers.data(name='x', shape=[2], dtype='float32') - y = fluid.layers.data(name='y', shape=[1], dtype='int64') - out = fluid.layers.hsigmoid(input=x, label=y, num_classes=6) + x = fluid.layers.fill_constant(shape=[4, 3], value=0.9, dtype='float32') + # x = [[0.9, 0.9, 0.9], [0.9, 0.9, 0.9], [0.9, 0.9, 0.9], [0.9, 0.9, 0.9]] + y = fluid.layers.fill_constant( + shape=[4, 1], value=1, dtype='int64') + # y = [[1], [1], [1], [1]] + out = fluid.layers.hsigmoid(input=x, label=y, num_classes=2, param_attr=fluid.initializer.Constant( + value=0.05), bias_attr=fluid.initializer.Constant(value=.0)) + # out = [[0.62792355], [0.62792355], [0.62792355], [0.62792355]] """ helper = LayerHelper('hierarchical_sigmoid', **locals()) diff --git a/python/paddle/fluid/layers/tensor.py b/python/paddle/fluid/layers/tensor.py index 8b402ef39d8..efd4068515b 100644 --- a/python/paddle/fluid/layers/tensor.py +++ b/python/paddle/fluid/layers/tensor.py @@ -362,25 +362,27 @@ def sums(input, out=None): def assign(input, output=None): """ - **Assign** + The OP copies the :attr:`input` to the :attr:`output`. - This function copies the *input* Variable to the *output* Variable. - - Args: - input(Variable|numpy.ndarray): The source variable - output(Variable|None): The destination variable + Parameters: + input (Variable|numpy.ndarray): A tensor or numpy ndarray, its data type supports + float32, float64, int32 and int64. + output (Variable, optional): A tensor. If :attr:`output` is None, a new tensor will + be created as :attr:`output`. Default: None. Returns: - Variable: The destination variable that was supplied as the *output*. + Variable: A tensor with the same shape, data type and value as :attr:`input`. Examples: .. code-block:: python import paddle.fluid as fluid - data = fluid.layers.data(name="data", shape=[3, 32, 32], dtype="float32") - out = fluid.layers.create_tensor(dtype='float32') - hidden = fluid.layers.fc(input=data, size=10) - fluid.layers.assign(hidden, out) + import numpy as np + data = fluid.layers.fill_constant(shape=[3, 2], value=2.5, dtype='float64') # [[2.5, 2.5], [2.5, 2.5], [2.5, 2.5]] + result1 = fluid.layers.create_tensor(dtype='float64') + fluid.layers.assign(data, result1) # result1 = [[2.5, 2.5], [2.5, 2.5], [2.5, 2.5]] + result2 = fluid.layers.assign(data) # result2 = [[2.5, 2.5], [2.5, 2.5], [2.5, 2.5]] + result3 = fluid.layers.assign(np.array([[2.5, 2.5], [2.5, 2.5], [2.5, 2.5]], dtype='float32')) # result3 = [[2.5, 2.5], [2.5, 2.5], [2.5, 2.5]] """ helper = LayerHelper('assign', **locals()) if output is None: @@ -738,25 +740,25 @@ def argsort(input, axis=-1, name=None): def ones(shape, dtype, force_cpu=False): """ - **ones** + The OP creates a tensor of specified :attr:`shape` and :attr:`dtype`, and fills it with 1. + Its :attr:`stop_gradient` will be set to True to stop gradient computation. - This function creates a tensor of specified *shape* and - *dtype*, and initializes this with 1. - - It also sets *stop_gradient* to True. - - Args: - shape(tuple|list): Shape of output tensor - dtype(np.dtype|core.VarDesc.VarType|str): Data type of output tensor + Parameters: + shape (tuple|list): Shape of output tensor. + dtype (np.dtype|core.VarDesc.VarType|str): Data type of output tensor, it supports + bool, float16, float32, float64, int32 and int64. + force_cpu (bool, optional): Whether force to store the output tensor in CPU memory. + If :attr:`force_cpu` is False, the output tensor will be stored in running device memory. + Default: False. Returns: - Variable: The tensor variable storing the output + Variable: A tensor of data type :attr:`dtype` with shape :attr:`shape` and all elements set to 1. Examples: .. code-block:: python import paddle.fluid as fluid - data = fluid.layers.ones(shape=[1], dtype='int64') + data = fluid.layers.ones(shape=[2, 4], dtype='float32') # [[1., 1., 1., 1.], [1., 1., 1., 1.]] """ assert isinstance(shape, list) or isinstance( shape, tuple), "The shape's type should be list or tuple." @@ -767,53 +769,50 @@ def ones(shape, dtype, force_cpu=False): def zeros(shape, dtype, force_cpu=False): """ - **zeros** + The OP creates a tensor of specified :attr:`shape` and :attr:`dtype`, and fills it with 0. + Its :attr:`stop_gradient` will be set to True to stop gradient computation. - This function creates a tensor of specified *shape* and - *dtype*, and initializes this with 0. - - It also sets *stop_gradient* to True. - - Args: - shape(tuple|list|None): Shape of output tensor. - dtype(np.dtype|core.VarDesc.VarType|str): Data type of output tensor. - force_cpu(bool, default False): Whether to make output stay on CPU. + Parameters: + shape (tuple|list): Shape of output tensor. + dtype (np.dtype|core.VarDesc.VarType|str): Data type of output tensor, it supports + bool, float16, float32, float64, int32 and int64. + force_cpu (bool, optional): Whether force to store the output tensor in CPU memory. + If :attr:`force_cpu` is False, the output tensor will be stored in running device memory. + Default: False. Returns: - Variable: The tensor variable storing the output. + Variable: A tensor of data type :attr:`dtype` with shape :attr:`shape` and all elements set to 0. Examples: .. code-block:: python import paddle.fluid as fluid - data = fluid.layers.zeros(shape=[1], dtype='int64') + data = fluid.layers.zeros(shape=[3, 2], dtype='float32') # [[0., 0.], [0., 0.], [0., 0.]] """ return fill_constant(value=0.0, **locals()) def reverse(x, axis): """ - **reverse** + The OP reverses the tensor :attr:`x` along the given :attr:`axis`. - This function reverse the input 'x' along given axises. - - Args: - x(Vairbale): the input to be reversed. - axis(int|tuple|list): Axis that along which order of elements - is reversed. If it is a tuple or a list, reversing - will be apply on each axis in the tuple or list. + Parameters: + x (Variable): A tensor to be reversed, its data type supports bool, float32, float64, int32, int64 and uint8. + axis (int|tuple|list): A dimension or a set of dimensions of :attr:`x` to reverse. Must be + in the range [-rank( :attr:`x` ), rank( :attr:`x` )). If it is a tuple or a list, reversing + will be apply on each axis in the tuple or list. Returns: - Variable: The reversed tensor. + Variable: The reversed tensor with the same shape and data type as :attr:`x`. Examples: .. code-block:: python import paddle.fluid as fluid - data = fluid.layers.data(name="data", shape=[4, 8], dtype="float32") - out = fluid.layers.reverse(x=data, axis=0) - # or: - out = fluid.layers.reverse(x=data, axis=[0,1]) + import numpy as np + data = fluid.layers.assign(np.array([[0, 1, 2], [3, 4, 5], [6, 7, 8]], dtype='float32')) # [[0., 1., 2.], [3., 4., 5.], [6., 7., 8.]] + result1 = fluid.layers.reverse(data, 0) # [[6., 7., 8.], [3., 4., 5.], [0., 1., 2.]] + result2 = fluid.layers.reverse(data, [0, 1]) # [[8., 7., 6.], [5., 4., 3.], [2., 1., 0.]] """ if isinstance(axis, int): axis = [axis] -- GitLab