-
DataFeeder¶
-
-Evaluator — PaddlePaddle documentation
+ evaluator — PaddlePaddle documentation
@@ -34,8 +34,8 @@
-
-
+
+
@@ -162,17 +162,17 @@
Training and Inference
Fluid
@@ -207,7 +207,7 @@
Fluid >
- Evaluator
+ evaluator
@@ -217,76 +217,24 @@
DataFeeder¶
+
+
@@ -230,10 +235,10 @@
diff --git a/develop/doc/api/v2/fluid/evaluator.html b/develop/doc/api/v2/fluid/evaluator.html
index a44ce3e959248a2e9571b73618657d77a065b379..a5230f3ccf5015fdb066fcac03fb23c8b2c7a816 100644
--- a/develop/doc/api/v2/fluid/evaluator.html
+++ b/develop/doc/api/v2/fluid/evaluator.html
@@ -8,7 +8,7 @@
- data_feeder¶
+
+
DataFeeder¶
+-
+
-
+class
paddle.v2.fluid.data_feeder.DataFeeder(feed_list, place, program=None)
+
-
Initializer — PaddlePaddle documentation
+ initializer — PaddlePaddle documentation
@@ -34,8 +34,8 @@
-
-
+
+
@@ -162,17 +162,17 @@
Training and Inference
Fluid
@@ -207,7 +207,7 @@
Fluid >
- Initializer
+ initializer
@@ -217,91 +217,40 @@
Evaluator¶
-
-Executor — PaddlePaddle documentation
+ executor — PaddlePaddle documentation
@@ -34,8 +34,8 @@
-
-
+
+
@@ -162,17 +162,17 @@
Training and Inference
Fluid
@@ -207,7 +207,7 @@
Fluid >
- Executor
+ executor
@@ -217,9 +217,38 @@
Evaluator¶
+evaluator¶
+
+
+Accuracy¶
-
-class
paddle.v2.fluid.evaluator.Evaluator(name, **kwargs)
- Base Class for all evaluators
--
-- - - Parameters: -
-
- name (str) – The name of evaluator. such as, “accuracy”. Used for generate -temporary variable name. -
- main_program (Program, optional) – The evaluator should be added to this -main_program. Default default_main_program() -
- startup_program (Program, optional) – The parameter should be added to this -startup_program. Default default_startup_program() -
-
-
-
-
states¶
- list – The list of state variables. states will be reset to zero -when reset is invoked.
-
-
-
-
-
metrics¶
- list – The list of metrics variables. They will be calculate -every mini-batch
-
-
-
-
-
reset(executor, reset_program=None)
- reset metric states at the begin of each pass/user specified batch
-
-
-
-
-
eval(executor, eval_program=None)
- Evaluate the statistics merged by multiple mini-batches.
+classpaddle.v2.fluid.evaluator.Accuracy(input, label, k=1, **kwargs) +Average Accuracy for multiple mini-batches.
-
+
+
@@ -299,10 +247,10 @@ every mini-batch
diff --git a/develop/doc/api/v2/fluid/executor.html b/develop/doc/api/v2/fluid/executor.html
index aa07c2f1e08b7fecdb24f542c4693adae3b1bf94..a81a5d9ab259312ca1049c01f47fd405c7febaa5 100644
--- a/develop/doc/api/v2/fluid/executor.html
+++ b/develop/doc/api/v2/fluid/executor.html
@@ -8,7 +8,7 @@
- ChunkEvaluator¶
+-
-
create_state(suffix, dtype, shape)
- Create state variable.
-NOTE: It is not a public API.
--
-- - - Parameters: -
-
- suffix (str) – the state suffix. -
- dtype (str|core.DataType) – the state data type -
- shape (tuple|list) – the shape of state -
Returns: State variable
-
paddle.v2.fluid.evaluator.ChunkEvaluator(input, label, chunk_scheme, num_chunk_types, excluded_chunk_types=None)
+Accumulate counter numbers output by chunk_eval from mini-batches and +compute the precision recall and F1-score using the accumulated counter +numbers.
-
+
@@ -230,10 +259,10 @@
diff --git a/develop/doc/api/v2/fluid/initializer.html b/develop/doc/api/v2/fluid/initializer.html
index d248f2294035d199826e93344386875d86fadd34..c8cf498be9c5564132e660d2b20658010393883a 100644
--- a/develop/doc/api/v2/fluid/initializer.html
+++ b/develop/doc/api/v2/fluid/initializer.html
@@ -8,7 +8,7 @@
- Executor¶
+executor¶
Executor¶
+-
+
-
+class
paddle.v2.fluid.executor.Executor(places)
+
+
+global_scope¶
+-
+
-
+
paddle.v2.fluid.executor.global_scope()
+
+
+scope_guard¶
+-
+
-
+
paddle.v2.fluid.executor.scope_guard(*args, **kwds)
+
+
switch_scope¶
+-
+
-
+
paddle.v2.fluid.executor.switch_scope(scope)
+
-
Initializer¶
-
-
Initializer¶
--
+
-
-class
paddle.v2.fluid.initializer.Initializer
- Base class for variable initializers
-Defines the common interface of variable initializers. -They add operations to the init program that are used -to initialize variables. Users should not use this class -directly, but need to use one of its implementations.
+paddle.v2.fluid.initializer.Constant+alias of
ConstantInitializer-
-class
paddle.v2.fluid.initializer.ConstantInitializer(value=0.0)
- Implements the constant initializer
+paddle.v2.fluid.initializer.Uniform+alias of
UniformInitializer-
-class
paddle.v2.fluid.initializer.UniformInitializer(low=-1.0, high=1.0, seed=0)
- Implements the random uniform distribution initializer
+paddle.v2.fluid.initializer.Normal+alias of
NormalInitializer-
-class
paddle.v2.fluid.initializer.NormalInitializer(loc=0.0, scale=1.0, seed=0)
- Implements the random Normal(Gaussian) distribution initializer
--
-class
paddle.v2.fluid.initializer.XavierInitializer(uniform=True, fan_in=None, fan_out=None, seed=0)
- Implements the Xavier initializer
-This class implements the Xavier weight initializer from the paper -Understanding the difficulty of training deep feedforward neural -networks[1] by Xavier Glorot and Yoshua Bengio.
-This initializer is designed to keep the scale of the gradients -approximately same in all the layers. In case of Uniform distribution, -the range is [-x, x], where x = sqrt(6 / (fan_in + fan_out)). -In case of Normal distribution, the mean is 0 and the standard deviation -is sqrt(2/ (fan_in + fan_out)).
-References
--
-
- [1] Understanding the difficulty of training deep feedforward neural -
- networks. International conference on artificial intelligence and -statistics. -(http://proceedings.mlr.press/v9/glorot10a.html) -
-
-class
paddle.v2.fluid.initializer.MSRAInitializer(uniform=True, fan_in=None, seed=0)
- Implements the MSRA initializer a.k.a. Kaiming Initializer
-This class implements the weight initialization from the paper -Delving Deep into Rectifiers: Surpassing Human-Level Performance on -ImageNet Classification[1] by Kaiming He, Xiangyu Zhang, Shaoqing Ren -and Jian Sun. This is a robust initialization method that particularly -considers the rectifier nonlinearities. In case of Uniform distribution, -the range is [-x, x], where x = sqrt(6 / fan_in). In case of Normal -distribution, the mean is 0 and the standard deviation -is sqrt(2/ fan_in).
-References
--
-
- [1] Delving Deep into Rectifiers: Surpassing Human-Level Performance -
- on ImageNet Classification -(https://arxiv.org/abs/1502.01852) -
paddle.v2.fluid.initializer.Xavier+alias of
XavierInitializer- Training and Inference
- Fluid @@ -207,7 +207,7 @@
- Fluid > -
- IO +
- io
-
-
paddle.v2.fluid.io.is_parameter(var)
- Check whether the variable is a Parameter.
-This function checks whether the input variable is a Parameter.
+paddle.v2.fluid.io.save_vars(executor, dirname, main_program=None, vars=None, predicate=None) +Save variables to directory by executor.
+- Parameters: var – The input variable. +Parameters: -
+
- executor – executor that save variable +
- dirname – directory path +
- main_program – program. If vars is None, then filter all variables in this +
Returns: boolean result whether the variable is a Parameter. +program which fit predicate. Default default_main_program. +:param predicate: The Predicate describes a callable that returns a variable +as a bool. If it returns true, the variables will be saved. +:param vars: variables need to be saved. If specify vars, program & predicate +will be ignored +:return: None
+-
+
paddle.v2.fluid.io.save_params(executor, dirname, main_program=None)
+ Save all parameters to directory with executor.
+-
+
paddle.v2.fluid.io.save_persistables(executor, dirname, main_program=None)
+ Save all persistables to directory with executor.
+-
+
paddle.v2.fluid.io.load_vars(executor, dirname, main_program=None, vars=None, predicate=None)
+ Load variables from directory by executor.
++
++ + + Parameters: -
+
- executor – executor that save variable +
- dirname – directory path +
- main_program – program. If vars is None, then filter all variables in this +
program which fit predicate. Default default_main_program(). +:param predicate: The Predicate describes a callable that returns a variable +as a bool. If it returns true, the variables will be loaded. +:param vars: variables need to be loaded. If specify vars, program & +predicate will be ignored +:return: None
+-
+
paddle.v2.fluid.io.load_params(executor, dirname, main_program=None)
+ load all parameters from directory by executor.
+-
+
paddle.v2.fluid.io.load_persistables(executor, dirname, main_program=None)
+ load all persistables from directory by executor.
-
+
paddle.v2.fluid.io.save_inference_model(dirname, feeded_var_names, target_vars, executor, main_program=None)
+ Build a model especially for inference, +and save it to directory by the executor.
++
++ + + Parameters: -
+
- dirname – directory path +
- feeded_var_names – Names of variables that need to be feeded data during inference +
- target_vars – Variables from which we can get inference results. +
- executor – executor that save inference model +
- main_program – original program, which will be pruned to build the inference model. +Default default_main_program(). +
Returns: None
+-
+
paddle.v2.fluid.io.load_inference_model(dirname, executor)
+ Load inference model from a directory
++
++ + + Parameters: -
+
- dirname – directory path +
- executor – executor that load inference model +
Returns: [program, feed_target_names, fetch_targets] +program: program especially for inference. +feed_target_names: Names of variables that need to feed data +fetch_targets: Variables from which we can get inference results.
+-
+
paddle.v2.fluid.io.get_inference_program(target_vars, main_program=None)
+ - Training and Inference
- Fluid @@ -207,7 +207,7 @@
- Fluid > -
- Layers +
- layers
-
-
paddle.v2.fluid.layers.fc(input, size, num_flatten_dims=1, param_attr=None, bias_attr=None, act=None, name=None)
- Fully Connected Layer
-The fully connected layer can take multiple tensors as its inputs. It -creates a variable (one for each input tensor) called weights for each -input tensor, which represents a fully connected weight matrix from -each input unit to each output unit. The fully connected layer -multiplies each input tensor with its coresponding weight to produce -an output Tensor. If multiple input tensors are given, the results of -multiple multiplications will be sumed up. If bias_attr is not None, -a biases variable will be created and added to the output. Finally, -if activation is not None, it will be applied to the output as well.
-This process can be formulated as follows:
--\[Out = Act({\sum_{i=0}^{N-1}W_iX_i + b})\]-In the above equation:
--
-
- \(N\): Number of the input. -
- \(X_i\): The input tensor. -
- \(W\): The weights created by this layer. -
- \(b\): The bias parameter created by this layer (if needed). -
- \(Act\): The activation funtion. -
- \(Out\): The output tensor. -
paddle.v2.fluid.layers.split_lod_tensor(input, mask, level=0) +split_lod_tensor
+This function takes in an input that contains the complete lod information, +and takes in a mask which is used to mask certain parts of the input. +The output is the true branch and the false branch with the mask applied to +the input at a certain level in the tensor.
Parameters: -
-
- input (Variable|list) – The input tensor(s) to the fully connected layer. -
- size (int) – The number of output units in the fully connected layer. -
- num_flatten_dims (int) – The fc layer can accept an input tensor with more -than two dimensions. If this happens, the -multidimensional tensor will first be flattened -into a 2-dimensional matrix. The parameter -num_flatten_dims determines how the input tensor -is flattened: the first num_flatten_dims -(inclusive, index starts from 1) dimensions will -be flatten to form the first dimension of the -final matrix (height of the matrix), and the rest -rank(X) - num_flatten_dims dimensions are -flattened to form the second dimension of the -final matrix (width of the matrix). For example, -suppose X is a 6-dimensional tensor with a shape -[2, 3, 4, 5, 6], and num_flatten_dims = 3. Then, -the flattened matrix will have a shape -[2 x 3 x 4, 5 x 6] = [24, 30]. By default, -num_flatten_dims is set to 1. -
- param_attr (ParamAttr|list) – The parameter attribute for learnable -parameters/weights of the fully connected -layer. -
- param_initializer (ParamAttr|list) – The initializer used for the -weight/parameter. If set None, -XavierInitializer() will be used. -
- bias_attr (ParamAttr|list) – The parameter attribute for the bias parameter -for this layer. If set None, no bias will be -added to the output units. -
- bias_initializer (ParamAttr|list) – The initializer used for the bias. -If set None, then ConstantInitializer() -will be used. -
- act (str) – Activation to be applied to the output of the fully connected -layer. -
- name (str) – Name/alias of the fully connected layer. +
- input (tuple|list|None) – The input tensor that contains complete +lod information needed to construct the output. +
- mask (list) – A bool column vector which masks the input. +
- level (int) – The specific lod level to rank.
Returns: The output tensor variable.
-Return type: Variable
+Returns: The true branch of tensor as per the mask applied to input. +Variable: The false branch of tensor as per the mask applied to input.
Raises:
+ValueError– If rank of the input tensor is less than 2.Return type: Variable
Examples
--data = fluid.layers.data(name="data", shape=[32, 32], dtype="float32") -fc = fluid.layers.fc(input=data, size=1000, act="tanh") +
x = layers.data(name='x', shape=[1]) +x.persistable = True + +y = layers.data(name='y', shape=[1]) +y.persistable = True + +out_true, out_false = layers.split_lod_tensor( + input=x, mask=y, level=level)
-embedding¶
++-merge_lod_tensor¶
-
-
paddle.v2.fluid.layers.embedding(input, size, is_sparse=False, padding_idx=None, param_attr=None, dtype='float32')
- Embedding Layer
-This layer is used to lookup embeddings of IDs, provided by
-input, in -a lookup table. The result of this lookup is the embedding of each ID in the -input.All the input variables are passed in as local variables to the LayerHelper -constructor.
+paddle.v2.fluid.layers.merge_lod_tensor(in_true, in_false, x, mask, level=0) +merge_lod_tensor
+This function takes in an input \(x\), the True branch, the False +branch and a binary \(mask\). Using this information, this function +merges the True and False branches of the tensor into a single Output +at a certain lod level indiacted by \(level\).
+Parameters: -
-
- input (Variable) – The tensor variable containing the IDs. -
- size (tuple|list) – The shape of the look up table parameter. It should -have two elements which indicate the size of the dictionary of -embeddings and the size of each embedding vector respectively. -
- is_sparse (bool) – The flag indicating whether to use sparse update. -
- padding_idx (int|long|None) – If
None, it makes no effect to lookup. -Otherwise the givenpadding_idxindicates padding the output -with zeros whenever lookup encounters it ininput. If -\(padding_idx < 0\), the padding_idx to use in lookup is -\(size[0] + dim\).
- - param_attr (ParamAttr) – Parameters for this layer -
- dtype (np.dtype|core.DataType|str) – The type of data : float32, float_16, int etc +
- in_true (tuple|list|None) – The True branch to be merged. +
- in_false (tuple|list|None) – The False branch to be merged. +
- x (tuple|list|None) – The input tensor that contains complete +lod information needed to construct the output. +
- mask (list) – A bool column vector which masks the input. +
- level (int) – The specific lod level to rank.
Returns: The tensor variable storing the embeddings of the supplied inputs.
+Returns: The merged output tensor.
Return type: Variable
@@ -346,344 +298,209 @@ with zeros whenever lookup encounters it inExamples --dict_size = len(dataset.ids) -data = fluid.layers.data(name='ids', shape=[32, 32], dtype='float32') -fc = fluid.layers.embedding(input=data, size=[dict_size, 16]) +
x = layers.data( + name='x', shape=[1], dtype='float32', stop_gradient=False) +y = layers.data( + name='y', shape=[1], dtype='bool', stop_gradient=False) + +level = 0 + +out_true, out_false = layers.split_lod_tensor( + input=x, mask=y, level=level) +out = layers.merge_lod_tensor( + in_true=out_true, in_false=out_false, mask=y, x=x, level=level)
-dynamic_lstm¶
--
+
-
-
paddle.v2.fluid.layers.dynamic_lstm(input, size, param_attr=None, bias_attr=None, use_peepholes=True, is_reverse=False, gate_activation='sigmoid', cell_activation='tanh', candidate_activation='tanh', dtype='float32', name=None)
- Dynamic LSTM Layer
-The defalut implementation is diagonal/peephole connection -(https://arxiv.org/pdf/1402.1128.pdf), the formula is as follows:
--\[ \begin{align}\begin{aligned}i_t & = \sigma(W_{ix}x_{t} + W_{ih}h_{t-1} + W_{ic}c_{t-1} + b_i)\\f_t & = \sigma(W_{fx}x_{t} + W_{fh}h_{t-1} + W_{fc}c_{t-1} + b_f)\\\tilde{c_t} & = act_g(W_{cx}x_t + W_{ch}h_{t-1} + b_c)\\o_t & = \sigma(W_{ox}x_{t} + W_{oh}h_{t-1} + W_{oc}c_t + b_o)\\c_t & = f_t \odot c_{t-1} + i_t \odot \tilde{c_t}\\h_t & = o_t \odot act_h(c_t)\end{aligned}\end{align} \]-where the \(W\) terms denote weight matrices (e.g. \(W_{xi}\) is -the matrix of weights from the input gate to the input), \(W_{ic}, W_{fc}, W_{oc}\) are diagonal weight matrices for peephole connections. In -our implementation, we use vectors to reprenset these diagonal weight -matrices. The \(b\) terms denote bias vectors (\(b_i\) is the input -gate bias vector), \(\sigma\) is the non-linear activations, such as -logistic sigmoid function, and \(i, f, o\) and \(c\) are the input -gate, forget gate, output gate, and cell activation vectors, respectively, -all of which have the same size as the cell output activation vector \(h\).
-The \(\odot\) is the element-wise product of the vectors. \(act_g\) -and \(act_h\) are the cell input and cell output activation functions -and tanh is usually used for them. \(\tilde{c_t}\) is also called -candidate hidden state, which is computed based on the current input and -the previous hidden state.
-Set use_peepholes to False to disable peephole connection. The formula -is omitted here, please refer to the paper -http://www.bioinf.jku.at/publications/older/2604.pdf for details.
-Note that these \(W_{xi}x_{t}, W_{xf}x_{t}, W_{xc}x_{t}, W_{xo}x_{t}\) -operations on the input \(x_{t}\) are NOT included in this operator. -Users can choose to use fully-connect layer before LSTM layer.
+classpaddle.v2.fluid.layers.BlockGuard(main_program) +BlockGuard class.
+BlockGuard class is used to create a sub-block in a program by +using the Python with keyword.
+-
+class
paddle.v2.fluid.layers.BlockGuardWithCompletion(rnn)
+ BlockGuardWithCompletion class.
+BlockGuardWithCompletion class is used to create an op with a block in a program.
+-
+class
paddle.v2.fluid.layers.StaticRNNMemoryLink(init, pre_mem, mem=None)
+ StaticRNNMemoryLink class.
-- Parameters: -
-
- input (Variable) – The input of dynamic_lstm layer, which supports -variable-time length input sequence. The underlying -tensor in this Variable is a matrix with shape -(T X 4D), where T is the total time steps in this -mini-batch, D is the hidden size. -
- size (int) – 4 * hidden size. -
- param_attr (ParamAttr|None) –
The parameter attribute for the learnable -hidden-hidden weights.
--
-
- Weights = {\(W_{ch}, W_{ih}, W_{fh}, W_{oh}\)} -
- The shape is (D x 4D), where D is the hidden -size. -
- - bias_attr (ParamAttr|None) –
The bias attribute for the learnable bias -weights, which contains two parts, input-hidden -bias weights and peephole connections weights if -setting use_peepholes to True.
--
-
- use_peepholes = False -
-
--
-
- Biases = {\(b_c, b_i, b_f, b_o\)}. -
- The shape is (1 x 4D). -
-
-
- use_peepholes = True -
-
--
-
- Biases = { \(b_c, b_i, b_f, b_o, W_{ic}, W_{fc}, W_{oc}\)}. -
- The shape is (1 x 7D). -
- - use_peepholes (bool) – Whether to enable diagonal/peephole connections, -default True. -
- is_reverse (bool) – Whether to compute reversed LSTM, default False. -
- gate_activation (str) – The activation for input gate, forget gate and -output gate. Choices = [“sigmoid”, “tanh”, “relu”, -“identity”], default “sigmoid”. -
- cell_activation (str) – The activation for cell output. Choices = [“sigmoid”, -“tanh”, “relu”, “identity”], default “tanh”. -
- candidate_activation (str) – The activation for candidate hidden state. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], -default “tanh”. -
- dtype (str) – Data type. Choices = [“float32”, “float64”], default “float32”. -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- init – the initial variable for Memory +
- init – Variable +
- pre_mem – the memory variable in previous time step +
- pre_mem – Variable +
- mem – the memory variable in current time step +
- mem – Variable
Parameters: -
+
Returns: The hidden state, and cell state of LSTM. The shape of both is (T x D), and lod is the same with the input.
-Return type: tuple
-Examples
-+-hidden_dim = 512 -forward_proj = fluid.layers.fc(input=input_seq, size=hidden_dim * 4, - act=None, bias_attr=None) -forward, _ = fluid.layers.dynamic_lstm( - input=forward_proj, size=hidden_dim * 4, use_peepholes=False) -
StaticRNNMemoryLink class is used to create a link between two +memory cells of a StaticRNN.
-
+class
paddle.v2.fluid.layers.WhileGuard(while_op)
+ -
+class
paddle.v2.fluid.layers.While(cond, name=None)
+ -
-
paddle.v2.fluid.layers.dynamic_lstmp(input, size, proj_size, param_attr=None, bias_attr=None, use_peepholes=True, is_reverse=False, gate_activation='sigmoid', cell_activation='tanh', candidate_activation='tanh', proj_activation='tanh', dtype='float32', name=None)
- Dynamic LSTMP Layer
-LSTMP (LSTM with recurrent projection) layer has a separate projection -layer after the LSTM layer, projecting the original hidden state to a -lower-dimensional one, which is proposed to reduce the number of total -parameters and furthermore computational complexity for the LSTM, -espeacially for the case that the size of output units is relative -large (https://research.google.com/pubs/archive/43905.pdf).
-The formula is as follows:
--\[ \begin{align}\begin{aligned}i_t & = \sigma(W_{ix}x_{t} + W_{ir}r_{t-1} + W_{ic}c_{t-1} + b_i)\\f_t & = \sigma(W_{fx}x_{t} + W_{fr}r_{t-1} + W_{fc}c_{t-1} + b_f)\\\tilde{c_t} & = act_g(W_{cx}x_t + W_{cr}r_{t-1} + b_c)\\o_t & = \sigma(W_{ox}x_{t} + W_{or}r_{t-1} + W_{oc}c_t + b_o)\\c_t & = f_t \odot c_{t-1} + i_t \odot \tilde{c_t}\\h_t & = o_t \odot act_h(c_t)\\r_t & = \overline{act_h}(W_{rh}h_t)\end{aligned}\end{align} \]-In the above formula:
--
-
- \(W\): Denotes weight matrices (e.g. \(W_{xi}\) is the matrix of weights from the input gate to the input). -
- \(W_{ic}\), \(W_{fc}\), \(W_{oc}\): Diagonal weight matrices for peephole connections. In our implementation, we use vectors to reprenset these diagonal weight matrices. -
- \(b\): Denotes bias vectors (e.g. \(b_i\) is the input gate bias vector). -
- \(\sigma\): The activation, such as logistic sigmoid function. -
- \(i, f, o\) and \(c\): The input gate, forget gate, output gate, and cell activation vectors, respectively, all of which have the same size as the cell output activation vector \(h\). -
- \(h\): The hidden state. -
- \(r\): The recurrent projection of the hidden state. -
- \(\tilde{c_t}\): The candidate hidden state, whose computation is based on the current input and previous hidden state. -
- \(\odot\): The element-wise product of the vectors. -
- \(act_g\) and \(act_h\): The cell input and cell output activation functions and tanh is usually used for them. -
- \(\overline{act_h}\): The activation function for the projection output, usually using identity or same as \(act_h\). -
Set use_peepholes to False to disable peephole connection. The formula -is omitted here, please refer to the paper -http://www.bioinf.jku.at/publications/older/2604.pdf for details.
-Note that these \(W_{xi}x_{t}, W_{xf}x_{t}, W_{xc}x_{t}, W_{xo}x_{t}\) -operations on the input \(x_{t}\) are NOT included in this operator. -Users can choose to use fully-connected layer before LSTMP layer.
+paddle.v2.fluid.layers.lod_rank_table(x, level=0) +LoD Rank Table Operator. Given an input variable x and a level number +of LoD, this layer creates a LodRankTable object. A LoDRankTable object +contains a list of bi-element tuples. Each tuple consists of an index and +a length, both of which are int type. Refering to specified level of LoD, +the index is the sequence index number and the length representes the +sequence length. Please note that the list is ranked in descending order by +the length. The following is an example:
++
++x is a LoDTensor: + x.lod = [[0, 2, 3], + [0, 5, 6, 7]] + x.data = [a, b, c, d, e, f, g] + +1. set level to 0: + Create lod rank table: + lod_rank_table_obj = lod_rank_table(x, level=0) + + Get: + lod_rank_table_obj.items() = [(0, 2), (1, 1)] + +2. set level to 1: + Create lod rank table: + lod_rank_table_obj = lod_rank_table(x, level=1) + + Get: + lod_rank_table_obj.items() = [(0, 5), (1, 1), (2, 1)] +
Parameters: -
-
- input (Variable) – The input of dynamic_lstmp layer, which supports -variable-time length input sequence. The underlying -tensor in this Variable is a matrix with shape -(T X 4D), where T is the total time steps in this -mini-batch, D is the hidden size. -
- size (int) – 4 * hidden size. -
- proj_size (int) – The size of projection output. -
- param_attr (ParamAttr|None) –
The parameter attribute for the learnable -hidden-hidden weight and projection weight.
--
-
- Hidden-hidden weight = {\(W_{ch}, W_{ih}, W_{fh}, W_{oh}\)}. -
- The shape of hidden-hidden weight is (P x 4D), -where P is the projection size and D the hidden -size. -
- Projection weight = {\(W_{rh}\)}. -
- The shape of projection weight is (D x P). -
- - bias_attr (ParamAttr|None) –
The bias attribute for the learnable bias -weights, which contains two parts, input-hidden -bias weights and peephole connections weights if -setting use_peepholes to True.
--
-
- use_peepholes = False -
-
--
-
- Biases = {\(b_c, b_i, b_f, b_o\)}. -
- The shape is (1 x 4D). -
-
-
- use_peepholes = True -
-
--
-
- Biases = { \(b_c, b_i, b_f, b_o, W_{ic}, W_{fc}, W_{oc}\)}. -
- The shape is (1 x 7D). -
- - use_peepholes (bool) – Whether to enable diagonal/peephole connections, -default True. -
- is_reverse (bool) – Whether to compute reversed LSTM, default False. -
- gate_activation (str) – The activation for input gate, forget gate and -output gate. Choices = [“sigmoid”, “tanh”, “relu”, -“identity”], default “sigmoid”. -
- cell_activation (str) – The activation for cell output. Choices = [“sigmoid”, -“tanh”, “relu”, “identity”], default “tanh”. -
- candidate_activation (str) – The activation for candidate hidden state. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], -default “tanh”. -
- proj_activation (str) – The activation for projection output. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], -default “tanh”. -
- dtype (str) – Data type. Choices = [“float32”, “float64”], default “float32”. -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x (Variable) – Input variable, a LoDTensor based which to create the lod +rank table. +
- level (int) – Specify the LoD level, on which to create the lod rank +table.
Returns: The projection of hidden state, and cell state of LSTMP. The shape of projection is (T x P), for the cell state which is (T x D), and both LoD is the same with the input.
+Returns: The created LoDRankTable object.
Return type: tuple
+Return type: Variable
Examples
--hidden_dim, proj_dim = 512, 256 -fc_out = fluid.layers.fc(input=input_seq, size=hidden_dim * 4, - act=None, bias_attr=None) -proj_out, _ = fluid.layers.dynamic_lstmp(input=fc_out, - size=hidden_dim * 4, - proj_size=proj_dim, - use_peepholes=False, - is_reverse=True, - cell_activation="tanh", - proj_activation="tanh") +
x = fluid.layers.data(name='x', shape=[10], + dtype='float32', lod_level=1) +out = layers.lod_rank_table(x=x, level=0)
-dynamic_gru¶
++max_sequence_len¶
-
-
paddle.v2.fluid.layers.dynamic_gru(input, size, param_attr=None, bias_attr=None, is_reverse=False, gate_activation='sigmoid', candidate_activation='tanh', h_0=None)
- Dynamic GRU Layer
-Refer to Empirical Evaluation of Gated Recurrent Neural Networks on -Sequence Modeling
-The formula is as follows:
--\[ \begin{align}\begin{aligned}u_t & = act_g(W_{ux}x_{t} + W_{uh}h_{t-1} + b_u)\\r_t & = act_g(W_{rx}x_{t} + W_{rh}h_{t-1} + b_r)\\\tilde{h_t} & = act_c(W_{cx}x_{t} + W_{ch}(r_t \odot h_{t-1}) + b_c)\\h_t & = (1-u_t) \odot h_{t-1} + u_t \odot \tilde{h_t}\end{aligned}\end{align} \]-The \(\odot\) is the element-wise product of the vectors. \(act_g\) -is the update gate and reset gate activation function and \(sigmoid\) -is usually used for it. \(act_c\) is the activation function for -candidate hidden state and \(tanh\) is usually used for it.
-Note that these \(W_{ux}x_{t}, W_{rx}x_{t}, W_{cx}x_{t}\) operations on -the input \(x_{t}\) are NOT included in this operator. Users can choose -to use fully-connect layer before GRU layer.
+paddle.v2.fluid.layers.max_sequence_len(rank_table) +Max Sequence Len Operator. Given a LoDRankTable object, this layer +returns the max length of a batch of sequences. In fact, a LoDRankTable +object contains a list of tuples(<sequence index, sequence length>) and +the list is already sorted by sequence length in descending order, so the +operator just returns the sequence length of the first tuple element.
- Parameters: -
-
- input (Variable) – The input of dynamic_gru layer, which supports -variable-time length input sequence. The underlying tensor in this -Variable is a matrix with shape \((T \times 3D)\), where -\(T\) is the total time steps in this mini-batch, \(D\) -is the hidden size. -
- size (int) – The dimension of the gru cell. -
- param_attr (ParamAttr|None) –
The parameter attribute for the learnable -hidden-hidden weight matrix. Note:
--
-
- The shape of the weight matrix is \((T \times 3D)\), where -\(D\) is the hidden size. -
- All elements in the weight matrix can be divided into two parts. -The first part are weights of the update gate and reset gate with -shape \((D \times 2D)\), and the second part are weights for -candidate hidden state with shape \((D \times D)\). -
- - bias_attr (ParamAttr) – The parameter attribute for learnable the -hidden-hidden bias. -
- is_reverse (bool) – Whether to compute reversed GRU, default
-
False.
- - gate_activation (str) – The activation for update gate and reset gate. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], default “sigmoid”. -
- activation (str) – The activation for candidate hidden state. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], default “tanh”. -
Parameters: rank_table (Variable) – Input variable which is a LoDRankTable object. Returns: The hidden state of GRU. The shape is (T times D), and lod is the same with the input.
-Returns: The max length of sequence. Return type: Variable
-Return type: Variable Examples
--hidden_dim = 512 -x = fluid.layers.fc(input=data, size=hidden_dim * 3) -hidden = fluid.layers.dynamic_gru(input=x, dim=hidden_dim) +
x = fluid.layers.data(name='x', shape=[10], + dtype='float32', lod_level=1) +rank_table = layers.lod_rank_table(x=x, level=0) +max_seq_len = layers.max_sequence_len(rank_table)
-data¶
++topk¶
-
-
paddle.v2.fluid.layers.data(name, shape, append_batch_size=True, dtype='float32', lod_level=0, type=VarType.LOD_TENSOR, stop_gradient=True)
- Data Layer
-This function takes in the input and based on whether data has -to be returned back as a minibatch, it creates the global variable by using -the helper functions. The global variables can be accessed by all the -following operators in the graph.
-All the input variables of this function are passed in as local variables -to the LayerHelper constructor.
+paddle.v2.fluid.layers.topk(input, k) +topk
+This function performs the operation that selects the k entries in the input +vector and outputs their values and indices as vectors. Thus topk_out[j] is +the j-th largest entry in input, and its index is topk_indices[j]
Parameters: -
-
- name (str) – The name/alias of the function -
- shape (list) – Tuple declaring the shape. -
- append_batch_size (bool) – Whether or not to append the data as a batch. -
- dtype (int|float) – The type of data : float32, float_16, int etc -
- type (VarType) – The output type. By default it is LOD_TENSOR. -
- lod_level (int) – The LoD Level. 0 means the input data is not a sequence. -
- main_program (Program) – Name of the main program that calls this -
- startup_program (Program) – Name of the startup program -
- stop_gradient (bool) – A boolean that mentions whether gradient should flow. +
- input (Variable|list) – The input tensor that has all the data. +
- k (int) – The number of top elements that the function will pick.
Returns: The global variable that gives access to the data.
+Returns: -
+
- The variable of type array that contains the k largest entries +
from input.
+
+- Variable: The variable of type array that contains the indices of k +
largest entries from input.
+
+
Return type: Variable
@@ -692,399 +509,388 @@ to the LayerHelper constructor.Examples
--data = fluid.layers.data(name='x', shape=[784], dtype='float32') +
x = fluid.layers.data(name='x', shape=[10]) +k = 5 +array = fluid.layers.topk(x, k)
-mean¶
++lod_tensor_to_array¶
-
-
paddle.v2.fluid.layers.mean(**kwargs)
- Mean Operator.
-Out is a scalar which is the mean of all elements in X.
+paddle.v2.fluid.layers.lod_tensor_to_array(x, table) +Convert a LOD_TENSOR to an LOD_TENSOR_ARRAY.
-- Parameters: x – The input of mean op -Duplicable: False Optional: False -Returns: The output of mean op +Parameters: -
+
- x (Variable|list) – The LOD tensor to be converted to a LOD tensor array. +
- table (ParamAttr|list) – The variable that stores the level of lod +which is ordered by sequence length in +descending order. +
Returns: -
+
- The variable of type array that has been converted from a +
tensor.
+
+
Return type: Variable
+
++BlockGuard¶
+++BlockGuardWithCompletion¶
+-
+
+-StaticRNNMemoryLink¶
+-
+
-dynamic_lstmp¶
+++WhileGuard¶
+-
+
++While¶
+-
+
+lod_rank_table¶
Examples
+++x = fluid.layers.data(name='x', shape=[10]) +table = fluid.layers.lod_rank_table(x, level=0) +array = fluid.layers.lod_tensor_to_array(x, table) +
-mul¶
++-array_to_lod_tensor¶
-
-
paddle.v2.fluid.layers.mul(**kwargs)
- Mul Operator.
-This operator is used to perform matrix multiplication for input $X$ and $Y$.
-The equation is:
-$$Out = X * Y$$
-Both the input $X$ and $Y$ can carry the LoD (Level of Details) information, -or not. But the output only shares the LoD information with input $X$.
+paddle.v2.fluid.layers.array_to_lod_tensor(x, table) +Convert a LoD_Tensor_Aarry to an LoDTensor.
+Parameters: -
-
- x – (Tensor), The first input tensor of mul op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of mul op. -Duplicable: False Optional: False -
- x_num_col_dims (INT) – (int, default 1), The mul_op can take tensors with more than two -dimensions as its inputs. If the input $X$ is a tensor with more -than two dimensions, $X$ will be flattened into a two-dimensional -matrix first. The flattening rule is: the first num_col_dims -will be flattened to form the first dimension of the final matrix -(the height of the matrix), and the rest rank(X) - num_col_dims -dimensions are flattened to form the second dimension of the final -matrix (the width of the matrix). As a result, height of the -flattened matrix is equal to the product of $X$’s first -x_num_col_dims dimensions’ sizes, and width of the flattened -matrix is equal to the product of $X$’s last rank(x) - num_col_dims -dimensions’ size. For example, suppose $X$ is a 6-dimensional -tensor with the shape [2, 3, 4, 5, 6], and x_num_col_dims = 3. -Thus, the flattened matrix will have a shape [2 x 3 x 4, 5 x 6] = -[24, 30]. -
- y_num_col_dims (INT) – (int, default 1), The mul_op can take tensors with more than two, -dimensions as its inputs. If the input $Y$ is a tensor with more -than two dimensions, $Y$ will be flattened into a two-dimensional -matrix first. The attribute y_num_col_dims determines how $Y$ is -flattened. See comments of x_num_col_dims for more details. +
- x (Variable|list) – The lod tensor array to be converted to a tensor. +
- table (ParamAttr|list) – The variable that stores the level of lod +which is ordered by sequence length in +descending order.
Returns: (Tensor), The output tensor of mul op.
+Returns: -
+
- The variable of type tensor that has been converted +
from an array.
+
+
Return type: Variable
Examples
++x = fluid.layers.data(name='x', shape=[10]) +table = fluid.layers.lod_rank_table(x, level=0) +array = fluid.layers.lod_tensor_to_array(x, table) +lod_tensor = fluid.layers.array_to_lod_tensor(array, table) +
-elementwise_add¶
++-increment¶
-
-
paddle.v2.fluid.layers.elementwise_add(**kwargs)
- Limited Elementwise Add Operator.
-The equation is:
-$$Out = X + Y$$
-$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be -smaller than or equal to the dimensions of $X$.
-There are two cases for this operator: -1. The shape of $Y$ is same with $X$; -2. The shape of $Y$ is a subset of $X$.
-For case 2: -$Y$ will be broadcasted to match the shape of $X$ and axis should be -set to index of the start dimension to broadcast $Y$ onto $X$.
--
-
- For example -
- --
shape(X) = (2, 3, 4, 5), shape(Y) = (,) -shape(X) = (2, 3, 4, 5), shape(Y) = (5,) -shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) -shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 -shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 -
-
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) -information. However, the output only shares the LoD information with input $X$.
+paddle.v2.fluid.layers.increment(x, value=1.0, in_place=True) +This function performs an operation that increments each value in the +input \(x\) by an amount: \(value\) as mentioned in the input +parameter. This operation is performed in-place by default.
+Parameters: -
-
- x – (Tensor), The first input tensor of elementwise op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of elementwise op. -Duplicable: False Optional: False -
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. +
- 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.
Returns: The output of elementwise op.
+Returns: -
+
- The tensor variable storing the transformation of +
element-wise increment of each value in the input.
+
+
Return type: Variable
Examples
++data = fluid.layers.data(name='data', shape=[32, 32], dtype='float32') +data = fluid.layers.increment(x=data, value=3.0, in_place=True) +
-elementwise_sub¶
++-array_write¶
-
-
paddle.v2.fluid.layers.elementwise_sub(**kwargs)
- Limited Elementwise Sub Operator.
-The equation is:
-$$Out = X - Y$$
-$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be -smaller than or equal to the dimensions of $X$.
-There are two cases for this operator: -1. The shape of $Y$ is same with $X$; -2. The shape of $Y$ is a subset of $X$.
-For case 2: -$Y$ will be broadcasted to match the shape of $X$ and axis should be -set to index of the start dimension to broadcast $Y$ onto $X$.
--
-
- For example -
- --
shape(X) = (2, 3, 4, 5), shape(Y) = (,) -shape(X) = (2, 3, 4, 5), shape(Y) = (5,) -shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) -shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 -shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 -
-
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) -information. However, the output only shares the LoD information with input $X$.
+paddle.v2.fluid.layers.array_write(x, i, array=None) +This function writes the given input variable to the specified position +indicating by the arrary index to an output LOD_TENSOR_ARRAY. If the +output LOD_TENSOR_ARRAY is not given(None), a new one will be created and +returned.
+Parameters: -
-
- x – (Tensor), The first input tensor of elementwise op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of elementwise op. -Duplicable: False Optional: False -
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. +
- x (Variable|list) – The input tensor from which the data will be read. +
- i (Variable|list) – The index of the output LOD_TENSOR_ARRAY, pointing to +the position to which the input tensor will be +written. +
- array (Variable|list) – The output LOD_TENSOR_ARRAY to which the input +tensor will be written. If this parameter is +NONE, a new LOD_TENSOR_ARRAY will be created and +returned.
Returns: The output of elementwise op.
+Returns: The output LOD_TENSOR_ARRAY where the input tensor is written.
+Return type: Variable
Examples
-elementwise_mul¶
++create_array¶
-
-
paddle.v2.fluid.layers.elementwise_mul(**kwargs)
- Limited Elementwise Mul Operator.
-The equation is:
-$$Out = X odotY$$
-$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be -smaller than or equal to the dimensions of $X$.
-There are two cases for this operator: -1. The shape of $Y$ is same with $X$; -2. The shape of $Y$ is a subset of $X$.
-For case 2: -$Y$ will be broadcasted to match the shape of $X$ and axis should be -set to index of the start dimension to broadcast $Y$ onto $X$.
--
-
- For example -
- +
shape(X) = (2, 3, 4, 5), shape(Y) = (,) -shape(X) = (2, 3, 4, 5), shape(Y) = (5,) -shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) -shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 -shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 +
paddle.v2.fluid.layers.create_array(dtype) +This function creates an array of type \(LOD_TENSOR_ARRAY\) using the +LayerHelper.
++
++ + + Parameters: dtype (int|float) – The data type of the elements in the array. +Returns: The tensor variable storing the elements of data type. +Return type: Variable +Examples
+-data = fluid.layers.create_array(dtype='float32')
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) -information. However, the output only shares the LoD information with input $X$.
+ + ++-less_than¶
+-
+
-
+
paddle.v2.fluid.layers.less_than(x, y, cond=None, **ignored)
+ Less than
+This layer returns the truth value of \(x < y\) elementwise.
+Parameters: -
-
- x – (Tensor), The first input tensor of elementwise op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of elementwise op. -Duplicable: False Optional: False -
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. +
- x (Variable) – First operand of less_than +
- y (Variable) – Second operand of less_than +
- cond (Variable|None) – Optional output variable to store the result of less_than
Returns: The output of elementwise op.
+Returns: The tensor variable storing the output of less_than.
+Return type: Variable
Examples
++less = fluid.layers.less_than(x=label, y=limit) +
-elementwise_div¶
++-array_read¶
-
-
paddle.v2.fluid.layers.elementwise_div(**kwargs)
- Limited Elementwise Div Operator.
-The equation is:
-$$Out = X / Y$$
-$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be -smaller than or equal to the dimensions of $X$.
-There are two cases for this operator: -1. The shape of $Y$ is same with $X$; -2. The shape of $Y$ is a subset of $X$.
-For case 2: -$Y$ will be broadcasted to match the shape of $X$ and axis should be -set to index of the start dimension to broadcast $Y$ onto $X$.
--
-
- For example -
- --
shape(X) = (2, 3, 4, 5), shape(Y) = (,) -shape(X) = (2, 3, 4, 5), shape(Y) = (5,) -shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) -shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 -shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 -
-
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) -information. However, the output only shares the LoD information with input $X$.
+paddle.v2.fluid.layers.array_read(array, i) +This function performs the operation to read the data in as an +LOD_TENSOR_ARRAY. +:param array: The input tensor that will be written to an array. +:type array: Variable|list +:param i: The subscript index in tensor array, that points the
++
place where data will be written to.
+- Parameters: -
-
- x – (Tensor), The first input tensor of elementwise op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of elementwise op. -Duplicable: False Optional: False -
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. -
Returns: The tensor type variable that has the data written to it. Returns: The output of elementwise op.
-Return type: Variable Examples
-dropout¶
+++shrink_memory¶
-
-
paddle.v2.fluid.layers.dropout(x, dropout_prob, is_test=False, seed=None, **kwargs)
- Computes dropout.
-Drop or keep each element of x independently. Dropout is a regularization -technique for reducing overfitting by preventing neuron co-adaption during -training. The dropout operator randomly set (according to the given dropout -probability) the outputs of some units to zero, while others are remain -unchanged.
+paddle.v2.fluid.layers.shrink_memory(x, i, table) +This function creates an operator to shrink_rnn_memory using the RankTable +as mentioned in the input parameter.
+
+-array_length¶
+-
+
-
+
paddle.v2.fluid.layers.array_length(array)
+ This function performs the operation to find the length of the input +LOD_TENSOR_ARRAY.
- Parameters: -
-
- x (variable) – The input tensor. -
- dropout_prob (float) – Probability of setting units to zero. -
- is_test (bool) – A flag indicating whether it is in test phrase or not. -
- seed (int) – A Python integer used to create random seeds. If this -parameter is set to None, a random seed is used. -NOTE: If an integer seed is given, always the same output -units will be dropped. DO NOT use a fixed seed in training. -
Parameters: array (LOD_TENSOR_ARRAY) – The input array that will be used +to compute the length. Returns: A tensor variable.
-Returns: The length of the input LoDTensorArray. Return type: Variable
-Return type: Variable Examples
--x = fluid.layers.data(name="data", shape=[32, 32], dtype="float32") -droped = fluid.layers.dropout(input=x, dropout_rate=0.5) -
-+reshape¶
--
+
-
-
paddle.v2.fluid.layers.reshape(**kwargs)
- Reshape Operator.
-Reshape Input(X) into the shape specified by Attr(shape).
-An example: -Given a 2-D tensor X with 2 rows and 2 columns : [[1, 2], [3, 4]]
-and target shape = [1, 4], the reshape operator will transform -the tensor X into a 2-D tensor: [[1, 2, 3, 4]]
-One dimension in the target shape can be set -1, representing that its -size is unknown. In this case, the real dimension will be infered from -the original shape of Input(X) and other dimensions in the target shape.
-+class
+paddle.v2.fluid.layers.IfElse(cond, name=None) + + + +++DynamicRNN¶
+-
+
-
+class
paddle.v2.fluid.layers.DynamicRNN(name=None)
+
++ConditionalBlock¶
+-
+
-
+class
paddle.v2.fluid.layers.ConditionalBlock(inputs, name=None)
+
+-StaticRNN¶
+-
+
-
+class
paddle.v2.fluid.layers.StaticRNN(name=None)
+ StaticRNN class.
+StaticRNN class is used to create a StaticRNN. The RNN will have its +own parameters like inputs, outputs, memories, status and length.
+-
+
-
+
memory(init=None, shape=None, batch_ref=None, init_value=0.0, init_batch_dim_idx=0, ref_batch_dim_idx=1)
+ - Parameters: -
-
- x – The input tensor of reshape operator. -Duplicable: False Optional: False -
- shape (INTS) – (vector<int>) Target shape of reshape operator. +
- init – boot memory, if not set, a shape, batch_ref must be provided +
- shape – shape of the boot memory +
- batch_ref – batch size reference variable +
- init_value – the init value of boot memory +
- init_batch_dim_idx – the index of batch size in init’s dimension +
- ref_batch_dim_idx – the index of batch size in batch_ref’s dimension
Parameters: -
+
Returns: The output tensor of reshape operator.
-
-
+
--sigmoid¶
--
-
-
-
paddle.v2.fluid.layers.sigmoid(**kwargs)
- Sigmoid Activation Operator
-$$out = frac{1}{1 + e^{-x}}$$
--
- - - Parameters: x – Input of Sigmoid operator -Duplicable: False Optional: False -Returns: Output of Sigmoid operator -
-scale¶
++-reorder_lod_tensor_by_rank¶
-
-
paddle.v2.fluid.layers.scale(**kwargs)
- Scale operator
-$$Out = scale*X$$
+paddle.v2.fluid.layers.reorder_lod_tensor_by_rank(x, rank_table) +ReorderLoDTensorByRankTable operator.
+Input(X) is a batch of sequences. Input(RankTable) stores new orders of the +input sequence batch. The reorder_lod_tensor_by_rank operator reorders the +Input(X) according to the information provided by Input(RankTable).
+For example:
+If the indices stored in the Input(RankTable) are [3, 0, 2, 1], the +Input(X) will be reordered that the fourth sequence in Input(X) will become the +first one, and then followed by the original first, third, and the second one.
+This is: +X = [Seq0, Seq1, Seq2, Seq3]. The indices in RankTable are [3, 0, 2, 1]. +Out = [Seq3, Seq0, Seq2, Seq1] with a new LoD information.
+If the LoD information of Input(X) is empty, this means Input(X) is not sequence +data. This is also identical to a batch of sequences where each sequence has a +fixed length 1. In this case, the reorder_lod_tensor_by_rank operator reorders +each slice of Input(X) along the first axis according to Input(RankTable).
+This is: +X = [Slice0, Slice1, Slice2, Slice3] and its LoD information is empty. The +indices in RankTable are [3, 0, 2, 1]. +Out = [Slice3, Slice0, Slice2, Slice1] with no LoD information is appended.
+NOTE: This operator sorts Input(X) according to a given LoDRankTable which does +not need to be calculated according to Input(X). It can be calculated according +to another different sequence, and then this operator sorts Input(X) according +to the given LoDRankTable.
-Parameters: -
-
- x – (Tensor) Input tensor of scale operator. +
- x – (LoDTensor), the input lod tensor to be reordered according to Input(RankTable). +Duplicable: False Optional: False +
- rank_table – (LoDRankTable), the rank table according to which Input(X) is reordered. Duplicable: False Optional: False -
- scale (FLOAT) – (float, default 1.0)The scaling factor of the scale operator.
Returns: (Tensor) Output tensor of scale operator.
+Returns: (LoDTensor), the reordered lod tensor.
-transpose¶
+++ParallelDo¶
+-
+
-
+class
paddle.v2.fluid.layers.ParallelDo(places, name=None)
+ ParallelDo class.
+ParallelDo class is used to create a ParallelDo.
+
+Print¶
-
-
paddle.v2.fluid.layers.transpose(x, perm, name=None)
- transpose Layer
-Permute the dimensions of input according to perm.
-The i-th dimension of the returned tensor will correspond to the -perm[i]-th dimension of input.
+paddle.v2.fluid.layers.Print(input, first_n=-1, message=None, summarize=-1, print_tensor_name=True, print_tensor_type=True, print_tensor_shape=True, print_tensor_lod=True, print_phase='both') +Print operator
+This creates a print op that will print when a tensor is accessed.
+Wraps the tensor passed in so that whenever that a tensor is accessed, +the message message is printed, along with the current value of the +tensor t.
Parameters: -
-
- input (Variable) – (Tensor), A Tensor. -
- perm (list) – A permutation of the dimensions of input. +
- input (Variable) – A Tensor to print. +
- summarize (int) – Print this number of elements in the tensor, will print +all if left is negative. +
- message (str) – A string message to print as a prefix. +
- first_n (int) – Only log first_n number of times. +
- print_tensor_name (bool) – Print the tensor name. +
- print_tensor_type (bool) – Print the tensor type. +
- print_tensor_shape (bool) – Print the tensor shape. +
- print_tensor_lod (bool) – Print the tensor lod. +
- print_phase (bool) – Which phase to displace, including ‘forward’, +‘backward’ and ‘both’. If set to ‘backward’ or ‘both’, will +print the gradients of input tensor.
Returns: A transposed Tensor.
+Returns: Output tensor, same data with input tensor.
Return type: Variable
@@ -1120,210 +947,260 @@ perm[i]-th dimension of input.Examples
--x = fluid.layers.data(name='x', shape=[5, 10, 15], dtype='float32') -x_transposed = layers.transpose(x, perm=[1, 0, 2]) +
+value = some_layer(...) +Print(value, summarize=10,
++
message=”The content of some_layer: ”)--sigmoid_cross_entropy_with_logits¶
---cast¶
--
-
-
-
paddle.v2.fluid.layers.cast(x, dtype)
- This function takes in the input with input_dtype -and casts it to the output_dtype as the output.
-
-concat¶
++device¶
++-get_places¶
-
-
paddle.v2.fluid.layers.concat(input, axis=0)
- Concat
-This function concatenates the input along the axis mentioned -and returns that as the output.
+paddle.v2.fluid.layers.get_places(device_count=None, device_type=None) +Returns a list of places based on flags. The list will be used for parallel +execution.
-Parameters: -
-
- input (list) – List of tensors to be concatenated -
- axis (int) – Integer axis along which the tensors will be concatenated +
- device_count (INT) – device count +
- device_type (STRING) – device type
Returns: Output variable of the concatenation
-Return type: Variable
+Returns: vector of Place
Examples
-+sums¶
++io¶
++-data¶
-
-
paddle.v2.fluid.layers.sums(input, out=None)
- This function performs the sum operation on the input and returns the -result as the output.
+paddle.v2.fluid.layers.data(name, shape, append_batch_size=True, dtype='float32', lod_level=0, type=VarType.LOD_TENSOR, stop_gradient=True) +Data Layer
+This function takes in the input and based on whether data has +to be returned back as a minibatch, it creates the global variable by using +the helper functions. The global variables can be accessed by all the +following operators in the graph.
+All the input variables of this function are passed in as local variables +to the LayerHelper constructor.
- Parameters: input (Variable|list) – The input tensor that has the elements -that need to be summed up. +Parameters: -
+
- name (str) – The name/alias of the function +
- shape (list) – Tuple declaring the shape. +
- append_batch_size (bool) – Whether or not to append the data as a batch. +
- dtype (int|float) – The type of data : float32, float_16, int etc +
- type (VarType) – The output type. By default it is LOD_TENSOR. +
- lod_level (int) – The LoD Level. 0 means the input data is not a sequence. +
- main_program (Program) – Name of the main program that calls this +
- startup_program (Program) – Name of the startup program +
- stop_gradient (bool) – A boolean that mentions whether gradient should flow. +
Returns: -
-
- The tensor type variable that has the sum of input -
- written to it. -
Returns: The global variable that gives access to the data.
Return type: Variable +Return type: Variable
+Examples
++data = fluid.layers.data(name='x', shape=[784], dtype='float32') +
-linear_chain_crf¶
--
+
-
-
paddle.v2.fluid.layers.linear_chain_crf(input, label, param_attr=None)
- BlockGuardServ class.
+BlockGuardServ class is used to create an op with a block in a program.
+-
+class
paddle.v2.fluid.layers.ListenAndServ(endpoint, fan_in=1, optimizer_mode=True)
+ ListenAndServ class.
+ListenAndServ class is used to wrap listen_and_serv op to create a server +which can receive variables from clients and run a block.
+-
-
paddle.v2.fluid.layers.embedding(input, size, is_sparse=False, padding_idx=None, param_attr=None, dtype='float32')
- Embedding Layer
-This layer is used to lookup embeddings of IDs, provided by
-input, in -a lookup table. The result of this lookup is the embedding of each ID in the -input.All the input variables are passed in as local variables to the LayerHelper -constructor.
+paddle.v2.fluid.layers.Send(endpoints, send_vars, get_vars) +Send layer
-- Parameters: -
-
- input (Variable) – The tensor variable containing the IDs. -
- size (tuple|list) – The shape of the look up table parameter. It should -have two elements which indicate the size of the dictionary of -embeddings and the size of each embedding vector respectively. -
- is_sparse (bool) – The flag indicating whether to use sparse update. -
- padding_idx (int|long|None) – If
None, it makes no effect to lookup. -Otherwise the givenpadding_idxindicates padding the output -with zeros whenever lookup encounters it ininput. If -\(padding_idx < 0\), the padding_idx to use in lookup is -\(size[0] + dim\).
- - param_attr (ParamAttr) – Parameters for this layer -
- dtype (np.dtype|core.DataType|str) – The type of data : float32, float_16, int etc +
- endpoints – comma seperated IP:PORT pairs in the order +of send_vars to send +
- send_vars – vars to send +
- get_vars – vars to get from server after send completes.
Parameters: -
+
Returns: The tensor variable storing the embeddings of the supplied inputs.
-Return type: Variable
-Examples
-+-dict_size = len(dataset.ids) -data = fluid.layers.data(name='ids', shape=[32, 32], dtype='float32') -fc = fluid.layers.embedding(input=data, size=[dict_size, 16]) -
Send variables to the server side, and get vars from server +side when server have finished running server side program.
-
-
paddle.v2.fluid.layers.split_lod_tensor(input, mask, level=0)
- split_lod_tensor
-This function takes in an input that contains the complete lod information, -and takes in a mask which is used to mask certain parts of the input. -The output is the true branch and the false branch with the mask applied to -the input at a certain level in the tensor.
+paddle.v2.fluid.layers.fc(input, size, num_flatten_dims=1, param_attr=None, bias_attr=None, act=None, name=None) +Fully Connected Layer
+The fully connected layer can take multiple tensors as its inputs. It +creates a variable (one for each input tensor) called weights for each +input tensor, which represents a fully connected weight matrix from +each input unit to each output unit. The fully connected layer +multiplies each input tensor with its coresponding weight to produce +an output Tensor. If multiple input tensors are given, the results of +multiple multiplications will be sumed up. If bias_attr is not None, +a biases variable will be created and added to the output. Finally, +if activation is not None, it will be applied to the output as well.
+This process can be formulated as follows:
++\[Out = Act({\sum_{i=0}^{N-1}W_iX_i + b})\]+In the above equation:
+-
+
- \(N\): Number of the input. +
- \(X_i\): The input tensor. +
- \(W\): The weights created by this layer. +
- \(b\): The bias parameter created by this layer (if needed). +
- \(Act\): The activation funtion. +
- \(Out\): The output tensor. +
Parameters: -
-
- input (tuple|list|None) – The input tensor that contains complete -lod information needed to construct the output. -
- mask (list) – A bool column vector which masks the input. -
- level (int) – The specific lod level to rank. +
- input (Variable|list) – The input tensor(s) to the fully connected layer. +
- size (int) – The number of output units in the fully connected layer. +
- num_flatten_dims (int) – The fc layer can accept an input tensor with more +than two dimensions. If this happens, the +multidimensional tensor will first be flattened +into a 2-dimensional matrix. The parameter +num_flatten_dims determines how the input tensor +is flattened: the first num_flatten_dims +(inclusive, index starts from 1) dimensions will +be flatten to form the first dimension of the +final matrix (height of the matrix), and the rest +rank(X) - num_flatten_dims dimensions are +flattened to form the second dimension of the +final matrix (width of the matrix). For example, +suppose X is a 6-dimensional tensor with a shape +[2, 3, 4, 5, 6], and num_flatten_dims = 3. Then, +the flattened matrix will have a shape +[2 x 3 x 4, 5 x 6] = [24, 30]. By default, +num_flatten_dims is set to 1. +
- param_attr (ParamAttr|list) – The parameter attribute for learnable +parameters/weights of the fully connected +layer. +
- param_initializer (ParamAttr|list) – The initializer used for the +weight/parameter. If set None, +XavierInitializer() will be used. +
- bias_attr (ParamAttr|list) – The parameter attribute for the bias parameter +for this layer. If set None, no bias will be +added to the output units. +
- bias_initializer (ParamAttr|list) – The initializer used for the bias. +If set None, then ConstantInitializer() +will be used. +
- act (str) – Activation to be applied to the output of the fully connected +layer. +
- name (str) – Name/alias of the fully connected layer.
Returns: The true branch of tensor as per the mask applied to input. -Variable: The false branch of tensor as per the mask applied to input.
+Returns: The output tensor variable.
Return type: Variable
+Return type: Variable
+Raises: ValueError– If rank of the input tensor is less than 2.Examples
--x = layers.data(name='x', shape=[1]) -x.persistable = True - -y = layers.data(name='y', shape=[1]) -y.persistable = True - -out_true, out_false = layers.split_lod_tensor( - input=x, mask=y, level=level) +
data = fluid.layers.data(name="data", shape=[32, 32], dtype="float32") +fc = fluid.layers.fc(input=data, size=1000, act="tanh")
-merge_lod_tensor¶
++embedding¶
-
-
paddle.v2.fluid.layers.merge_lod_tensor(in_true, in_false, x, mask, level=0)
- merge_lod_tensor
-This function takes in an input \(x\), the True branch, the False -branch and a binary \(mask\). Using this information, this function -merges the True and False branches of the tensor into a single Output -at a certain lod level indiacted by \(level\).
+paddle.v2.fluid.layers.embedding(input, size, is_sparse=False, padding_idx=None, param_attr=None, dtype='float32') +Embedding Layer
+This layer is used to lookup embeddings of IDs, provided by
+input, in +a lookup table. The result of this lookup is the embedding of each ID in the +input.All the input variables are passed in as local variables to the LayerHelper +constructor.
Parameters: -
-
- in_true (tuple|list|None) – The True branch to be merged. -
- in_false (tuple|list|None) – The False branch to be merged. -
- x (tuple|list|None) – The input tensor that contains complete -lod information needed to construct the output. -
- mask (list) – A bool column vector which masks the input. -
- level (int) – The specific lod level to rank. +
- input (Variable) – The tensor variable containing the IDs. +
- size (tuple|list) – The shape of the look up table parameter. It should +have two elements which indicate the size of the dictionary of +embeddings and the size of each embedding vector respectively. +
- is_sparse (bool) – The flag indicating whether to use sparse update. +
- padding_idx (int|long|None) – If
None, it makes no effect to lookup. +Otherwise the givenpadding_idxindicates padding the output +with zeros whenever lookup encounters it ininput. If +\(padding_idx < 0\), the padding_idx to use in lookup is +\(size[0] + dim\).
+ - param_attr (ParamAttr) – Parameters for this layer +
- dtype (np.dtype|core.DataType|str) – The type of data : float32, float_16, int etc
Returns: The merged output tensor.
+Returns: The tensor variable storing the embeddings of the supplied inputs.
Return type: Variable
@@ -1332,150 +1209,298 @@ lod information needed to construct the output.Examples
--x = layers.data( - name='x', shape=[1], dtype='float32', stop_gradient=False) -y = layers.data( - name='y', shape=[1], dtype='bool', stop_gradient=False) - -level = 0 - -out_true, out_false = layers.split_lod_tensor( - input=x, mask=y, level=level) -out = layers.merge_lod_tensor( - in_true=out_true, in_false=out_false, mask=y, x=x, level=level) +
dict_size = len(dataset.ids) +data = fluid.layers.data(name='ids', shape=[32, 32], dtype='float32') +fc = fluid.layers.embedding(input=data, size=[dict_size, 16])
--cos_sim¶
--
-
-
-
paddle.v2.fluid.layers.cos_sim(X, Y, **kwargs)
- This function performs the cosine similarity between two tensors -X and Y and returns that as the output.
-
-cross_entropy¶
++dynamic_lstm¶
-
-
paddle.v2.fluid.layers.cross_entropy(input, label, **kwargs)
- Cross Entropy Layer
-This layer computes the cross entropy between input and label. It -supports both standard cross-entropy and soft-label cross-entropy loss -computation.
--
-
-
-
- One-hot cross-entropy: -
soft_label = False, Label[i, 0] indicates the class index for sample i:
--\[Y[i] = -\log(X[i, Label[i]])\]-
-
--
-
- Soft-label cross-entropy: -
soft_label = True, Label[i, j] indicates the soft label of class j -for sample i:
--\[Y[i] = \sum_j{-Label[i, j] * log(X[i, j])}\]-
-
Please make sure that in this case the summation of each row of label -equals one.
-
--
-
- One-hot cross-entropy with vecterized label: -
As a special case of 2), when each row of ‘label’ has only one -non-zero element which is equal to 1, soft-label cross-entropy degenerates -to a one-hot cross-entropy with one-hot label representation.
-
-
-
paddle.v2.fluid.layers.dynamic_lstm(input, size, param_attr=None, bias_attr=None, use_peepholes=True, is_reverse=False, gate_activation='sigmoid', cell_activation='tanh', candidate_activation='tanh', dtype='float32', name=None) +Dynamic LSTM Layer
+The defalut implementation is diagonal/peephole connection +(https://arxiv.org/pdf/1402.1128.pdf), the formula is as follows:
++\[ \begin{align}\begin{aligned}i_t & = \sigma(W_{ix}x_{t} + W_{ih}h_{t-1} + W_{ic}c_{t-1} + b_i)\\f_t & = \sigma(W_{fx}x_{t} + W_{fh}h_{t-1} + W_{fc}c_{t-1} + b_f)\\\tilde{c_t} & = act_g(W_{cx}x_t + W_{ch}h_{t-1} + b_c)\\o_t & = \sigma(W_{ox}x_{t} + W_{oh}h_{t-1} + W_{oc}c_t + b_o)\\c_t & = f_t \odot c_{t-1} + i_t \odot \tilde{c_t}\\h_t & = o_t \odot act_h(c_t)\end{aligned}\end{align} \]+where the \(W\) terms denote weight matrices (e.g. \(W_{xi}\) is +the matrix of weights from the input gate to the input), \(W_{ic}, W_{fc}, W_{oc}\) are diagonal weight matrices for peephole connections. In +our implementation, we use vectors to reprenset these diagonal weight +matrices. The \(b\) terms denote bias vectors (\(b_i\) is the input +gate bias vector), \(\sigma\) is the non-linear activations, such as +logistic sigmoid function, and \(i, f, o\) and \(c\) are the input +gate, forget gate, output gate, and cell activation vectors, respectively, +all of which have the same size as the cell output activation vector \(h\).
+The \(\odot\) is the element-wise product of the vectors. \(act_g\) +and \(act_h\) are the cell input and cell output activation functions +and tanh is usually used for them. \(\tilde{c_t}\) is also called +candidate hidden state, which is computed based on the current input and +the previous hidden state.
+Set use_peepholes to False to disable peephole connection. The formula +is omitted here, please refer to the paper +http://www.bioinf.jku.at/publications/older/2604.pdf for details.
+Note that these \(W_{xi}x_{t}, W_{xf}x_{t}, W_{xc}x_{t}, W_{xo}x_{t}\) +operations on the input \(x_{t}\) are NOT included in this operator. +Users can choose to use fully-connect layer before LSTM layer.
Parameters: -
-
- input (Variable|list) – a 2-D tensor with shape [N x D], where N is the -batch size and D is the number of classes. This -input is a probability computed by the previous -operator, which is almost always the result of -a softmax operator. -
- label (Variable|list) – the ground truth which is a 2-D tensor. When -soft_label is set to False, label is a -tensor<int64> with shape [N x 1]. When -soft_label is set to True, label is a -tensor<float/double> with shape [N x D]. -
- soft_label (bool, via **kwargs) – a flag indicating whether to -interpretate the given labels as soft -labels, default False. +
- input (Variable) – The input of dynamic_lstm layer, which supports +variable-time length input sequence. The underlying +tensor in this Variable is a matrix with shape +(T X 4D), where T is the total time steps in this +mini-batch, D is the hidden size. +
- size (int) – 4 * hidden size. +
- param_attr (ParamAttr|None) –
The parameter attribute for the learnable +hidden-hidden weights.
+-
+
- Weights = {\(W_{ch}, W_{ih}, W_{fh}, W_{oh}\)} +
- The shape is (D x 4D), where D is the hidden +size.
Returns: A 2-D tensor with shape [N x 1], the cross entropy loss.
-Raises: ValueError – 1) the 1st dimension of input and label are not equal. -2) when soft_label == True, and the 2nd dimension of
+ +- bias_attr (ParamAttr|None) –
+The bias attribute for the learnable bias +weights, which contains two parts, input-hidden +bias weights and peephole connections weights if +setting use_peepholes to True.
+-
+
- use_peepholes = False +
-
-input and label are not equal.
--
-
- when soft_label == False, and the 2nd dimension of -label is not 1. +
- Biases = {\(b_c, b_i, b_f, b_o\)}. +
- The shape is (1 x 4D). +
- use_peepholes = True
- Biases = { \(b_c, b_i, b_f, b_o, W_{ic}, W_{fc}, W_{oc}\)}. +
- The shape is (1 x 7D). +
+-
+
-
+
+
+-
+
- use_peepholes (bool) – Whether to enable diagonal/peephole connections, +default True.
+- is_reverse (bool) – Whether to compute reversed LSTM, default False.
+- gate_activation (str) – The activation for input gate, forget gate and +output gate. Choices = [“sigmoid”, “tanh”, “relu”, +“identity”], default “sigmoid”.
+- cell_activation (str) – The activation for cell output. Choices = [“sigmoid”, +“tanh”, “relu”, “identity”], default “tanh”.
+- candidate_activation (str) – The activation for candidate hidden state. +Choices = [“sigmoid”, “tanh”, “relu”, “identity”], +default “tanh”.
+- dtype (str) – Data type. Choices = [“float32”, “float64”], default “float32”.
+- name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically.
+ +Returns: The hidden state, and cell state of LSTM. The shape of both is (T x D), and lod is the same with the input.
+Return type: tuple
Examples
--predict = fluid.layers.fc(input=net, size=classdim, act='softmax') -cost = fluid.layers.cross_entropy(input=predict, label=label) +
hidden_dim = 512 +forward_proj = fluid.layers.fc(input=input_seq, size=hidden_dim * 4, + act=None, bias_attr=None) +forward, _ = fluid.layers.dynamic_lstm( + input=forward_proj, size=hidden_dim * 4, use_peepholes=False)
-+square_error_cost¶
+++dynamic_lstmp¶
-
-
paddle.v2.fluid.layers.square_error_cost(input, label, **kwargs)
- Square error cost layer
-This layer accepts input predictions and target label and returns the -squared error cost.
-For predictions, \(X\), and target labels, \(Y\), the equation is:
+paddle.v2.fluid.layers.dynamic_lstmp(input, size, proj_size, param_attr=None, bias_attr=None, use_peepholes=True, is_reverse=False, gate_activation='sigmoid', cell_activation='tanh', candidate_activation='tanh', proj_activation='tanh', dtype='float32', name=None) +Dynamic LSTMP Layer
+LSTMP (LSTM with recurrent projection) layer has a separate projection +layer after the LSTM layer, projecting the original hidden state to a +lower-dimensional one, which is proposed to reduce the number of total +parameters and furthermore computational complexity for the LSTM, +espeacially for the case that the size of output units is relative +large (https://research.google.com/pubs/archive/43905.pdf).
+The formula is as follows:
-\[Out = (X - Y)^2\]-In the above equation:
+\[ \begin{align}\begin{aligned}i_t & = \sigma(W_{ix}x_{t} + W_{ir}r_{t-1} + W_{ic}c_{t-1} + b_i)\\f_t & = \sigma(W_{fx}x_{t} + W_{fr}r_{t-1} + W_{fc}c_{t-1} + b_f)\\\tilde{c_t} & = act_g(W_{cx}x_t + W_{cr}r_{t-1} + b_c)\\o_t & = \sigma(W_{ox}x_{t} + W_{or}r_{t-1} + W_{oc}c_t + b_o)\\c_t & = f_t \odot c_{t-1} + i_t \odot \tilde{c_t}\\h_t & = o_t \odot act_h(c_t)\\r_t & = \overline{act_h}(W_{rh}h_t)\end{aligned}\end{align} \]
In the above formula:
+-
+
- \(W\): Denotes weight matrices (e.g. \(W_{xi}\) is the matrix of weights from the input gate to the input). +
- \(W_{ic}\), \(W_{fc}\), \(W_{oc}\): Diagonal weight matrices for peephole connections. In our implementation, we use vectors to reprenset these diagonal weight matrices. +
- \(b\): Denotes bias vectors (e.g. \(b_i\) is the input gate bias vector). +
- \(\sigma\): The activation, such as logistic sigmoid function. +
- \(i, f, o\) and \(c\): The input gate, forget gate, output gate, and cell activation vectors, respectively, all of which have the same size as the cell output activation vector \(h\). +
- \(h\): The hidden state. +
- \(r\): The recurrent projection of the hidden state. +
- \(\tilde{c_t}\): The candidate hidden state, whose computation is based on the current input and previous hidden state. +
- \(\odot\): The element-wise product of the vectors. +
- \(act_g\) and \(act_h\): The cell input and cell output activation functions and tanh is usually used for them. +
- \(\overline{act_h}\): The activation function for the projection output, usually using identity or same as \(act_h\). +
Set use_peepholes to False to disable peephole connection. The formula +is omitted here, please refer to the paper +http://www.bioinf.jku.at/publications/older/2604.pdf for details.
+Note that these \(W_{xi}x_{t}, W_{xf}x_{t}, W_{xc}x_{t}, W_{xo}x_{t}\) +operations on the input \(x_{t}\) are NOT included in this operator. +Users can choose to use fully-connected layer before LSTMP layer.
++
++ + + Parameters: -
+
- input (Variable) – The input of dynamic_lstmp layer, which supports +variable-time length input sequence. The underlying +tensor in this Variable is a matrix with shape +(T X 4D), where T is the total time steps in this +mini-batch, D is the hidden size. +
- size (int) – 4 * hidden size. +
- proj_size (int) – The size of projection output. +
- param_attr (ParamAttr|None) –
The parameter attribute for the learnable +hidden-hidden weight and projection weight.
+-
+
- Hidden-hidden weight = {\(W_{ch}, W_{ih}, W_{fh}, W_{oh}\)}. +
- The shape of hidden-hidden weight is (P x 4D), +where P is the projection size and D the hidden +size. +
- Projection weight = {\(W_{rh}\)}. +
- The shape of projection weight is (D x P). +
+ - bias_attr (ParamAttr|None) –
The bias attribute for the learnable bias +weights, which contains two parts, input-hidden +bias weights and peephole connections weights if +setting use_peepholes to True.
+-
+
- use_peepholes = False +
-
-
-
- \(X\): Input predictions, a tensor. -
- \(Y\): Input labels, a tensor. -
- \(Out\): Output value, same shape with \(X\). +
- Biases = {\(b_c, b_i, b_f, b_o\)}. +
- The shape is (1 x 4D). +
- use_peepholes = True +
- Biases = { \(b_c, b_i, b_f, b_o, W_{ic}, W_{fc}, W_{oc}\)}. +
- The shape is (1 x 7D).
- use_peepholes (bool) – Whether to enable diagonal/peephole connections, +default True. +
- is_reverse (bool) – Whether to compute reversed LSTM, default False. +
- gate_activation (str) – The activation for input gate, forget gate and +output gate. Choices = [“sigmoid”, “tanh”, “relu”, +“identity”], default “sigmoid”. +
- cell_activation (str) – The activation for cell output. Choices = [“sigmoid”, +“tanh”, “relu”, “identity”], default “tanh”. +
- candidate_activation (str) – The activation for candidate hidden state. +Choices = [“sigmoid”, “tanh”, “relu”, “identity”], +default “tanh”. +
- proj_activation (str) – The activation for projection output. +Choices = [“sigmoid”, “tanh”, “relu”, “identity”], +default “tanh”. +
- dtype (str) – Data type. Choices = [“float32”, “float64”], default “float32”. +
- name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically. +
+-
+
-
+
+
+ +-
+
Returns: The projection of hidden state, and cell state of LSTMP. The shape of projection is (T x P), for the cell state which is (T x D), and both LoD is the same with the input.
+Return type: tuple
+Examples
++ + ++hidden_dim, proj_dim = 512, 256 +fc_out = fluid.layers.fc(input=input_seq, size=hidden_dim * 4, + act=None, bias_attr=None) +proj_out, _ = fluid.layers.dynamic_lstmp(input=fc_out, + size=hidden_dim * 4, + proj_size=proj_dim, + use_peepholes=False, + is_reverse=True, + cell_activation="tanh", + proj_activation="tanh") +
+dynamic_gru¶
+-
+
-
+
paddle.v2.fluid.layers.dynamic_gru(input, size, param_attr=None, bias_attr=None, is_reverse=False, gate_activation='sigmoid', candidate_activation='tanh', h_0=None)
+ Dynamic GRU Layer
+Refer to Empirical Evaluation of Gated Recurrent Neural Networks on +Sequence Modeling
+The formula is as follows:
++\[ \begin{align}\begin{aligned}u_t & = act_g(W_{ux}x_{t} + W_{uh}h_{t-1} + b_u)\\r_t & = act_g(W_{rx}x_{t} + W_{rh}h_{t-1} + b_r)\\\tilde{h_t} & = act_c(W_{cx}x_{t} + W_{ch}(r_t \odot h_{t-1}) + b_c)\\h_t & = (1-u_t) \odot h_{t-1} + u_t \odot \tilde{h_t}\end{aligned}\end{align} \]+The \(\odot\) is the element-wise product of the vectors. \(act_g\) +is the update gate and reset gate activation function and \(sigmoid\) +is usually used for it. \(act_c\) is the activation function for +candidate hidden state and \(tanh\) is usually used for it.
+Note that these \(W_{ux}x_{t}, W_{rx}x_{t}, W_{cx}x_{t}\) operations on +the input \(x_{t}\) are NOT included in this operator. Users can choose +to use fully-connect layer before GRU layer.
Parameters: -
-
- input (Variable) – Input tensor, has predictions. -
- label (Variable) – Label tensor, has target labels. +
- input (Variable) – The input of dynamic_gru layer, which supports +variable-time length input sequence. The underlying tensor in this +Variable is a matrix with shape \((T \times 3D)\), where +\(T\) is the total time steps in this mini-batch, \(D\) +is the hidden size. +
- size (int) – The dimension of the gru cell. +
- param_attr (ParamAttr|None) –
The parameter attribute for the learnable +hidden-hidden weight matrix. Note:
+-
+
- The shape of the weight matrix is \((T \times 3D)\), where +\(D\) is the hidden size. +
- All elements in the weight matrix can be divided into two parts. +The first part are weights of the update gate and reset gate with +shape \((D \times 2D)\), and the second part are weights for +candidate hidden state with shape \((D \times D)\). +
+ - bias_attr (ParamAttr) – The parameter attribute for learnable the +hidden-hidden bias. +
- is_reverse (bool) – Whether to compute reversed GRU, default
+
False.
+ - gate_activation (str) – The activation for update gate and reset gate. +Choices = [“sigmoid”, “tanh”, “relu”, “identity”], default “sigmoid”. +
- activation (str) – The activation for candidate hidden state. +Choices = [“sigmoid”, “tanh”, “relu”, “identity”], default “tanh”.
Returns: -
-
- The tensor variable storing the element-wise squared error -
difference of input and label.
-
-
Returns: The hidden state of GRU. The shape is (T times D), and lod is the same with the input.
Return type: Variable
@@ -1484,199 +1509,884 @@ squared error cost.Examples
-+y = layers.data(name='y', shape=[1], dtype='float32') -y_predict = layers.data(name='y_predict', shape=[1], dtype='float32') -cost = layers.square_error_cost(input=y_predict, label=y) +
+ + ++hidden_dim = 512 +x = fluid.layers.fc(input=data, size=hidden_dim * 3) +hidden = fluid.layers.dynamic_gru(input=x, dim=hidden_dim) +
++gru_unit¶
+-
+
-
+
paddle.v2.fluid.layers.gru_unit(input, hidden, size, weight=None, bias=None, activation='tanh', gate_activation='sigmoid')
+ GRU unit layer. The equation of a gru step is:
++
++\[ \begin{align}\begin{aligned}u_t & = actGate(xu_{t} + W_u h_{t-1} + b_u)\\r_t & = actGate(xr_{t} + W_r h_{t-1} + b_r)\\m_t & = actNode(xm_t + W_c dot(r_t, h_{t-1}) + b_m)\\h_t & = dot((1-u_t), m_t) + dot(u_t, h_{t-1})\end{aligned}\end{align} \]+The inputs of gru unit includes \(z_t\), \(h_{t-1}\). In terms +of the equation above, the \(z_t\) is split into 3 parts - +\(xu_t\), \(xr_t\) and \(xm_t\). This means that in order to +implement a full GRU unit operator for an input, a fully +connected layer has to be applied, such that \(z_t = W_{fc}x_t\).
+The terms \(u_t\) and \(r_t\) represent the update and reset gates +of the GRU cell. Unlike LSTM, GRU has one lesser gate. However, there is +an intermediate candidate hidden output, which is denoted by \(m_t\). +This layer has three outputs \(h_t\), \(dot(r_t, h_{t-1})\) +and concatenation of \(u_t\), \(r_t\) and \(m_t\).
++
++ + + Parameters: -
+
- input (Variable) – The fc transformed input value of current step. +
- hidden (Variable) – The hidden value of lstm unit from previous step. +
- size (integer) – The input dimension value. +
- weight (ParamAttr) – The weight parameters for gru unit. Default: None +
- bias (ParamAttr) – The bias parameters for gru unit. Default: None +
- activation (string) – The activation type for cell (actNode). +Default: ‘tanh’ +
- gate_activation (string) – The activation type for gates (actGate). +Default: ‘sigmoid’ +
Returns: The hidden value, reset-hidden value and gate values.
+Return type: tuple
+Examples
+++# assuming we have x_t_data and prev_hidden of size=10 +x_t = fluid.layers.fc(input=x_t_data, size=30) +hidden_val, r_h_val, gate_val = fluid.layers.gru_unit(input=x_t, + hidden = prev_hidden) +
++linear_chain_crf¶
+-
+
-
+
paddle.v2.fluid.layers.linear_chain_crf(input, label, param_attr=None)
+
++crf_decoding¶
+-
+
-
+
paddle.v2.fluid.layers.crf_decoding(input, param_attr, label=None)
+
++cos_sim¶
+-
+
-
+
paddle.v2.fluid.layers.cos_sim(X, Y, **kwargs)
+ This function performs the cosine similarity between two tensors +X and Y and returns that as the output.
+
++cross_entropy¶
+-
+
-
+
paddle.v2.fluid.layers.cross_entropy(input, label, **kwargs)
+ Cross Entropy Layer
+This layer computes the cross entropy between input and label. It +supports both standard cross-entropy and soft-label cross-entropy loss +computation.
+-
+
-
+
- One-hot cross-entropy: +
soft_label = False, Label[i, 0] indicates the class index for sample i:
++\[Y[i] = -\log(X[i, Label[i]])\]+
+
+-
+
- Soft-label cross-entropy: +
soft_label = True, Label[i, j] indicates the soft label of class j +for sample i:
++\[Y[i] = \sum_j{-Label[i, j] * log(X[i, j])}\]+
+
Please make sure that in this case the summation of each row of label +equals one.
+
+-
+
- One-hot cross-entropy with vecterized label: +
As a special case of 2), when each row of ‘label’ has only one +non-zero element which is equal to 1, soft-label cross-entropy degenerates +to a one-hot cross-entropy with one-hot label representation.
+
+
+
+
++ + + Parameters: -
+
- input (Variable|list) – a 2-D tensor with shape [N x D], where N is the +batch size and D is the number of classes. This +input is a probability computed by the previous +operator, which is almost always the result of +a softmax operator. +
- label (Variable|list) – the ground truth which is a 2-D tensor. When +soft_label is set to False, label is a +tensor<int64> with shape [N x 1]. When +soft_label is set to True, label is a +tensor<float/double> with shape [N x D]. +
- soft_label (bool, via **kwargs) – a flag indicating whether to +interpretate the given labels as soft +labels, default False. +
Returns: A 2-D tensor with shape [N x 1], the cross entropy loss.
+Raises: ValueError – 1) the 1st dimension of input and label are not equal. +2) when soft_label == True, and the 2nd dimension of
++
+input and label are not equal.
+-
+
- when soft_label == False, and the 2nd dimension of +label is not 1. +
Examples
+++predict = fluid.layers.fc(input=net, size=classdim, act='softmax') +cost = fluid.layers.cross_entropy(input=predict, label=label) +
++square_error_cost¶
+-
+
-
+
paddle.v2.fluid.layers.square_error_cost(input, label, **kwargs)
+ Square error cost layer
+This layer accepts input predictions and target label and returns the +squared error cost.
+For predictions, \(X\), and target labels, \(Y\), the equation is:
++\[Out = (X - Y)^2\]+In the above equation:
++
+-
+
- \(X\): Input predictions, a tensor. +
- \(Y\): Input labels, a tensor. +
- \(Out\): Output value, same shape with \(X\). +
+
++ + + Parameters: -
+
- input (Variable) – Input tensor, has predictions. +
- label (Variable) – Label tensor, has target labels. +
Returns: -
+
- The tensor variable storing the element-wise squared error +
difference of input and label.
+
+
Return type: Variable
+Examples
+++y = layers.data(name='y', shape=[1], dtype='float32') +y_predict = layers.data(name='y_predict', shape=[1], dtype='float32') +cost = layers.square_error_cost(input=y_predict, label=y) +
++accuracy¶
+-
+
-
+
paddle.v2.fluid.layers.accuracy(input, label, k=1, correct=None, total=None, **kwargs)
+ This function computes the accuracy using the input and label. +The output is the top_k inputs and their indices.
+
++chunk_eval¶
+-
+
-
+
paddle.v2.fluid.layers.chunk_eval(input, label, chunk_scheme, num_chunk_types, excluded_chunk_types=None, **kwargs)
+ This function computes and outputs the precision, recall and +F1-score of chunk detection.
+
++sequence_conv¶
+-
+
-
+
paddle.v2.fluid.layers.sequence_conv(input, num_filters, filter_size=3, filter_stride=1, padding=None, bias_attr=None, param_attr=None, act=None)
+ This function creates the op for sequence_conv, using the inputs and +other convolutional configurations for the filters and stride as given +in the input parameters to the function.
+
++conv2d¶
+-
+
-
+
paddle.v2.fluid.layers.conv2d(input, num_filters, filter_size, stride=None, padding=None, groups=None, param_attr=None, bias_attr=None, use_cudnn=True, act=None)
+ Convlution2D Layer
+The convolution2D layer calculates the output based on the input, filter +and strides, paddings, dilations, groups parameters. Input(Input) and +Output(Output) are in NCHW format. Where N is batch size, C is the number of +channels, H is the height of the feature, and W is the width of the feature. +The details of convolution layer, please refer UFLDL’s convolution, . +If bias attribution and activation type are provided, bias is added to the +output of the convolution, and the corresponding activation function is +applied to the final result.
+For each input \(X\), the equation is:
++\[Out = \sigma (W \ast X + b)\]+In the above equation:
+-
+
- \(X\): Input value, a tensor with NCHW format. +
- \(W\): Filter value, a tensor with MCHW format. +
- \(\ast\): Convolution operation. +
- \(b\): Bias value, a 2-D tensor with shape [M, 1]. +
- \(\sigma\): Activation function. +
-
+
- \(Out\): Output value, the shape of \(Out\) and \(X\) may be +
- different. +
+
Example
+-
+
Input:
+Input shape: $(N, C_{in}, H_{in}, W_{in})$
+Filter shape: $(C_{out}, C_{in}, H_f, W_f)$
+
+Output: +Output shape: $(N, C_{out}, H_{out}, W_{out})$
+
+
Where
++\[\]+H_{out}&= frac{(H_{in} + 2 * paddings[0] - (dilations[0] * (H_f - 1) + 1))}{strides[0]} + 1 \ +W_{out}&= frac{(W_{in} + 2 * paddings[1] - (dilations[1] * (W_f - 1) + 1))}{strides[1]} + 1
++
++ + + Parameters: -
+
- input (Variable) – The input image with [N, C, H, W] format. +
- num_filters (int) – The number of filter. It is as same as the output +image channel. +
- filter_size (int|tuple|None) – The filter size. If filter_size is a tuple, +it must contain two integers, (filter_size_H, filter_size_W). +Otherwise, the filter will be a square. +
- stride (int|tuple) – The stride size. If stride is a tuple, it must +contain two integers, (stride_H, stride_W). Otherwise, the +stride_H = stride_W = stride. Default: stride = 1. +
- padding (int|tuple) – The padding size. If padding is a tuple, it must +contain two integers, (padding_H, padding_W). Otherwise, the +padding_H = padding_W = padding. Default: padding = 0. +
- groups (int) – The groups number of the Conv2d Layer. According to grouped +convolution in Alex Krizhevsky’s Deep CNN paper: when group=2, +the first half of the filters is only connected to the first half +of the input channels, while the second half of the filters is only +connected to the second half of the input channels. Default: groups=1 +
- param_attr (ParamAttr) – The parameters to the Conv2d Layer. Default: None +
- bias_attr (ParamAttr) – Bias parameter for the Conv2d layer. Default: None +
- use_cudnn (bool) – Use cudnn kernel or not, it is valid only when the cudnn +library is installed. Default: True +
- act (str) – Activation type. Default: None +
Returns: -
+
- The tensor variable storing the convolution and +
non-linearity activation result.
+
+
Return type: Variable
+Raises:
+ValueError– If the shapes of input, filter_size, stride, padding and +groups mismatch.Examples
+++data = fluid.layers.data( + name='data', shape=[3, 32, 32], dtype='float32') +conv2d = fluid.layers.conv2d( + input=data, num_filters=2, filter_size=3, act="relu") +
++sequence_pool¶
+-
+
-
+
paddle.v2.fluid.layers.sequence_pool(input, pool_type, **kwargs)
+ This function add the operator for sequence pooling. +It pools features of all time-steps of each instance, and is applied +on top of the input using pool_type mentioned in the parameters.
+It supports four pool_type:
+-
+
- average: \(Out[i] = \frac{\sum_i X_i}{N}\) +
- sum: \(Out[i] = \sum_jX_{ij}\) +
- sqrt: \(Out[i] = \frac{\sum_jX_{ij}}{\sqrt{len(X_i)}}\) +
- max: \(Out[i] = max(X_i)\) +
++x is a 1-level LoDTensor: + x.lod = [[0, 2, 5, 7]] + x.data = [1, 3, 2, 4, 6, 5, 1] + x.dims = [7, 1] + +then output is a Tensor: + out.dim = [3, 1] + with condition len(x.lod[-1]) - 1 == out.dims[0] + +for different pool_type: + average: out.data = [2, 4, 3], where 2=(1+3)/2, 4=(2+4+6)/3, 3=(5+1)/2 + sum : out.data = [4, 12, 6], where 4=1+3, 12=2+4+6, 6=5+1 + sqrt : out.data = [2.82, 6.93, 4.24], where 2.82=(1+3)/sqrt(2), + 6.93=(2+4+6)/sqrt(3), 4.24=(5+1)/sqrt(2) + max : out.data = [3, 6, 5], where 3=max(1,3), 6=max(2,4,6), 5=max(5,1) +
+
++ + + Parameters: -
+
- input (variable) – The input variable which is a LoDTensor. +
- pool_type (string) – The pooling type of sequence_pool. +It supports average, sum, sqrt and max. +
Returns: The sequence pooling variable which is a Tensor.
+Examples
+++x = fluid.layers.data(name='x', shape=[7, 1], + dtype='float32', lod_level=1) +avg_x = fluid.layers.sequence_pool(input=x, pool_type='average') +sum_x = fluid.layers.sequence_pool(input=x, pool_type='sum') +sqrt_x = fluid.layers.sequence_pool(input=x, pool_type='sqrt') +max_x = fluid.layers.sequence_pool(input=x, pool_type='max') +
++pool2d¶
+-
+
-
+
paddle.v2.fluid.layers.pool2d(input, pool_size, pool_type, pool_stride=None, pool_padding=None, global_pooling=False, use_cudnn=True, name=None)
+ This function adds the operator for pooling in 2 dimensions, using the +pooling configurations mentioned in input parameters.
+
++batch_norm¶
+-
+
-
+
paddle.v2.fluid.layers.batch_norm(input, act=None, is_test=False, momentum=0.9, epsilon=1e-05, param_attr=None, bias_attr=None, data_layout='NCHW', name=None)
+ This function helps create an operator to implement +the BatchNorm layer using the configurations from the input parameters.
+
++beam_search_decode¶
+-
+
-
+
paddle.v2.fluid.layers.beam_search_decode(ids, scores, name=None)
+
++conv2d_transpose¶
+-
+
-
+
paddle.v2.fluid.layers.conv2d_transpose(input, num_filters, output_size=None, filter_size=None, padding=None, stride=None, dilation=None, param_attr=None, use_cudnn=True, name=None)
+ Convlution2D transpose layer
+The convolution2D transpose layer calculates the output based on the input, +filter, and dilations, strides, paddings. Input(Input) and output(Output) +are in NCHW format. Where N is batch size, C is the number of channels, +H is the height of the feature, and W is the width of the feature. +Parameters(dilations, strides, paddings) are two elements. These two elements +represent height and width, respectively. The details of convolution transpose +layer, please refer to the following explanation and references +therein.
+For each input \(X\), the equation is:
++\[Out = W \ast X\]+In the above equation:
+-
+
- \(X\): Input value, a tensor with NCHW format. +
- \(W\): Filter value, a tensor with MCHW format. +
- \(\ast\) : Convolution transpose operation. +
-
+
- \(Out\): Output value, the shape of \(Out\) and \(X\) may be +
- different. +
+
Example
+-
+
Input:
+Input shape: $(N, C_{in}, H_{in}, W_{in})$
+Filter shape: $(C_{in}, C_{out}, H_f, W_f)$
+
+Output:
+Output shape: $(N, C_{out}, H_{out}, W_{out})$
+
+
Where
++\[\begin{split}H_{out} &= (H_{in} - 1) * strides[0] - 2 * paddings[0] + dilations[0] * (H_f - 1) + 1 \\ +W_{out} &= (W_{in} - 1) * strides[1] - 2 * paddings[1] + dilations[1] * (W_f - 1) + 1\end{split}\]++
++ + + Parameters: -
+
- input (Variable) – The input image with [N, C, H, W] format. +
- num_filters (int) – The number of the filter. It is as same as the output +image channel. +
- output_size (int|tuple|None) – The output image size. If output size is a +tuple, it must contain two integers, (image_H, image_W). This +parameter only works when filter_size is None. +
- filter_size (int|tuple|None) – The filter size. If filter_size is a tuple, +it must contain two integers, (filter_size_H, filter_size_W). +Otherwise, the filter will be a square. None if use output size to +calculate filter_size. +
- padding (int|tuple) – The padding size. If padding is a tuple, it must +contain two integers, (padding_H, padding_W). Otherwise, the +padding_H = padding_W = padding. Default: padding = 0. +
- stride (int|tuple) – The stride size. If stride is a tuple, it must +contain two integers, (stride_H, stride_W). Otherwise, the +stride_H = stride_W = stride. Default: stride = 1. +
- dilation (int|tuple) – The dilation size. If dilation is a tuple, it must +contain two integers, (dilation_H, dilation_W). Otherwise, the +dilation_H = dilation_W = dilation. Default: dilation = 1. +
- param_attr (ParamAttr) – The parameters to the Conv2d_transpose Layer. +Default: None +
- use_cudnn (bool) – Use cudnn kernel or not, it is valid only when the cudnn +library is installed. Default: True +
- name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically. +
Returns: The tensor variable storing the convolution transpose result.
+Return type: Variable
+Raises:
+ValueError– If the shapes of input, filter_size, stride, padding and +groups mismatch.Examples
+++data = fluid.layers.data( + name='data', shape=[3, 32, 32], dtype='float32') +conv2d_transpose = fluid.layers.conv2d_transpose( + input=data, num_filters=2, filter_size=3) +
++sequence_expand¶
+-
+
-
+
paddle.v2.fluid.layers.sequence_expand(x, y, name=None)
+ Sequence Expand Layer. This layer will expand the input variable x +according to LoD information of y. And the following examples will +explain how sequence_expand works:
+++* Case 1 + x is a LoDTensor: + x.lod = [[0, 2, 3], + [0, 1, 3, 4]] + x.data = [a, b, c, d] + x.dims = [4, 1] + + y is a LoDTensor: + y.lod = [[0, 2, 4], + [0, 3, 6, 7, 8]] + + with condition len(y.lod[-1]) - 1 == x.dims[0] + + then output is a 2-level LoDTensor: + out.lod = [[0, 2, 4], + [0, 3, 6, 7, 8]] + out.data = [a, a, a, b, b, b, c, d] + out.dims = [8, 1] + +* Case 2 + x is a Tensor: + x.data = [a, b, c] + x.dims = [3, 1] + + y is a LoDTensor: + y.lod = [[0, 2, 3, 6]] + + with condition len(y.lod[-1]) - 1 == x.dims[0] + + then output is a 1-level LoDTensor: + out.lod = [[0, 2, 3, 6]] + out.data = [a, a, b, c, c, c] + out.dims = [6, 1] +
+
++ + + Parameters: -
+
- x (Variable) – The input variable which is a Tensor or LoDTensor. +
- y (Variable) – The input variable which is a LoDTensor. +
- name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically. +
Returns: The expanded variable which is a LoDTensor.
+Return type: Variable
+Examples
+++x = fluid.layers.data(name='x', shape=[10], dtype='float32') +y = fluid.layers.data(name='y', shape=[10, 20], + dtype='float32', lod_level=1) +out = layers.sequence_expand(x=x, y=y) +
++lstm_unit¶
+-
+
-
+
paddle.v2.fluid.layers.lstm_unit(x_t, hidden_t_prev, cell_t_prev, forget_bias=0.0, param_attr=None, bias_attr=None, name=None)
+ Lstm unit layer. The equation of a lstm step is:
++
++\[ \begin{align}\begin{aligned}i_t & = \sigma(W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i)\\f_t & = \sigma(W_{x_f}x_{t} + W_{h_f}h_{t-1} + b_f)\\c_t & = f_tc_{t-1} + i_t tanh (W_{x_c}x_t + W_{h_c}h_{t-1} + b_c)\\o_t & = \sigma(W_{x_o}x_{t} + W_{h_o}h_{t-1} + b_o)\\h_t & = o_t tanh(c_t)\end{aligned}\end{align} \]+The inputs of lstm unit include \(x_t\), \(h_{t-1}\) and +\(c_{t-1}\). The 2nd dimensions of \(h_{t-1}\) and \(c_{t-1}\) +should be same. The implementation separates the linear transformation and +non-linear transformation apart. Here, we take \(i_t\) as an example. +The linear transformation is applied by calling a fc layer and the +equation is:
++
++\[L_{i_t} = W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i\]+The non-linear transformation is applied by calling lstm_unit_op and the +equation is:
++
++\[i_t = \sigma(L_{i_t})\]+This layer has two outputs including \(h_t\) and \(o_t\).
++
++ + + Parameters: -
+
- x_t (Variable) – The input value of current step, a 2-D tensor with shape +M x N, M for batch size and N for input size. +
- hidden_t_prev (Variable) – The hidden value of lstm unit, a 2-D tensor +with shape M x S, M for batch size and S for size of lstm unit. +
- cell_t_prev (Variable) – The cell value of lstm unit, a 2-D tensor with +shape M x S, M for batch size and S for size of lstm unit. +
- forget_bias (float) – The forget bias of lstm unit. +
- param_attr (ParamAttr) – The attributes of parameter weights, used to set +initializer, name etc. +
- bias_attr (ParamAttr) – The attributes of bias weights, if not False, +bias weights will be created and be set to default value. +
- name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically. +
Returns: The hidden value and cell value of lstm unit.
+Return type: tuple
+Raises:
+ValueError– The ranks of x_t, hidden_t_prev and cell_t_prev +not be 2 or the 1st dimensions of x_t, hidden_t_prev +and cell_t_prev not be the same or the 2nd dimensions of +hidden_t_prev and cell_t_prev not be the same.Examples
+++x_t = fluid.layers.fc(input=x_t_data, size=10) +prev_hidden = fluid.layers.fc(input=prev_hidden_data, size=30) +prev_cell = fluid.layers.fc(input=prev_cell_data, size=30) +hidden_value, cell_value = fluid.layers.lstm_unit(x_t=x_t, + hidden_t_prev=prev_hidden, + cell_t_prev=prev_cell) +
++reduce_sum¶
+-
+
-
+
paddle.v2.fluid.layers.reduce_sum(input, dim=None, keep_dim=False, name=None)
+ Computes the sum of tensor elements over the given dimension.
++
++ + + Parameters: -
+
- input (Variable) – The input variable which is a Tensor or LoDTensor. +
- dim (int|None) – The dimension along which the sum is performed. If
+
None, sum all elements ofinputand return a +Tensor variable with a single element, otherwise must be in the +range \([-rank(input), rank(input))\). If \(dim < 0\), +the dimension to reduce is \(rank + dim\).
+ - keep_dim (bool) – Whether to reserve the reduced dimension in the
+output Tensor. The result tensor will have one fewer dimension
+than the
inputunlesskeep_dimis true.
+ - name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically. +
Returns: The reduced Tensor variable.
+Return type: Variable
+Examples
+++# x is a Tensor variable with following elements: +# [[0.2, 0.3, 0.5, 0.9] +# [0.1, 0.2, 0.6, 0.7]] +# Each example is followed by the correspending output tensor. +fluid.layers.reduce_sum(x) # [3.5] +fluid.layers.reduce_sum(x, dim=0) # [0.3, 0.5, 1.1, 1.6] +fluid.layers.reduce_sum(x, dim=-1) # [1.9, 1.6] +fluid.layers.reduce_sum(x, dim=1, keep_dim=True) # [[1.9], [1.6]] +
+-reduce_mean¶
+-
+
-
+
paddle.v2.fluid.layers.reduce_mean(input, dim=None, keep_dim=False, name=None)
+ Computes the mean of tensor elements over the given dimension.
++
++ + + Parameters: -
+
- input (Variable) – The input variable which is a Tensor or LoDTensor. +
- dim (int|None) – The dimension along which the mean is computed. If
+
None, compute the mean over all elements ofinput+and return a Tensor variable with a single element, otherwise +must be in the range \([-rank(input), rank(input))\). If +\(dim < 0\), the dimension to reduce is \(rank + dim\).
+ - keep_dim (bool) – Whether to reserve the reduced dimension in the
+output Tensor. The result tensor will have one fewer dimension
+than the
inputunlesskeep_dimis true.
+ - name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically. +
Returns: The reduced Tensor variable.
+Return type: Variable
+Examples
+# x is a Tensor variable with following elements: +# [[0.2, 0.3, 0.5, 0.9] +# [0.1, 0.2, 0.6, 0.7]] +# Each example is followed by the correspending output tensor. +fluid.layers.reduce_mean(x) # [0.4375] +fluid.layers.reduce_mean(x, dim=0) # [0.15, 0.25, 0.55, 0.8] +fluid.layers.reduce_mean(x, dim=-1) # [0.475, 0.4] +fluid.layers.reduce_mean(x, dim=1, keep_dim=True) # [[0.475], [0.4]]
--accuracy¶
--
-
-
-
paddle.v2.fluid.layers.accuracy(input, label, k=1, correct=None, total=None, **kwargs)
- This function computes the accuracy using the input and label. -The output is the top_k inputs and their indices.
-
--sequence_conv¶
--
-
-
-
paddle.v2.fluid.layers.sequence_conv(input, num_filters, filter_size=3, filter_stride=1, padding=None, bias_attr=None, param_attr=None, act=None)
- This function creates the op for sequence_conv, using the inputs and -other convolutional configurations for the filters and stride as given -in the input parameters to the function.
-
-conv2d¶
++reduce_max¶
-
-
paddle.v2.fluid.layers.conv2d(input, num_filters, filter_size, stride=None, padding=None, groups=None, param_attr=None, bias_attr=None, use_cudnn=True, act=None)
- Convlution2D Layer
-The convolution2D layer calculates the output based on the input, filter -and strides, paddings, dilations, groups parameters. Input(Input) and -Output(Output) are in NCHW format. Where N is batch size, C is the number of -channels, H is the height of the feature, and W is the width of the feature. -The details of convolution layer, please refer UFLDL’s convolution, . -If bias attribution and activation type are provided, bias is added to the -output of the convolution, and the corresponding activation function is -applied to the final result.
-For each input \(X\), the equation is:
--\[Out = \sigma (W \ast X + b)\]-In the above equation:
--
-
- \(X\): Input value, a tensor with NCHW format. -
- \(W\): Filter value, a tensor with MCHW format. -
- \(\ast\): Convolution operation. -
- \(b\): Bias value, a 2-D tensor with shape [M, 1]. -
- \(\sigma\): Activation function. -
-
-
- \(Out\): Output value, the shape of \(Out\) and \(X\) may be -
- different. -
-
Example
--
-
Input:
-Input shape: $(N, C_{in}, H_{in}, W_{in})$
-Filter shape: $(C_{out}, C_{in}, H_f, W_f)$
-
-Output: -Output shape: $(N, C_{out}, H_{out}, W_{out})$
-
-
Where
--\[\]-H_{out}&= frac{(H_{in} + 2 * paddings[0] - (dilations[0] * (H_f - 1) + 1))}{strides[0]} + 1 \ -W_{out}&= frac{(W_{in} + 2 * paddings[1] - (dilations[1] * (W_f - 1) + 1))}{strides[1]} + 1
+paddle.v2.fluid.layers.reduce_max(input, dim=None, keep_dim=False, name=None) +Computes the maximum of tensor elements over the given dimension.
Parameters: -
-
- input (Variable) – The input image with [N, C, H, W] format. -
- num_filters (int) – The number of filter. It is as same as the output -image channel. -
- filter_size (int|tuple|None) – The filter size. If filter_size is a tuple, -it must contain two integers, (filter_size_H, filter_size_W). -Otherwise, the filter will be a square. -
- stride (int|tuple) – The stride size. If stride is a tuple, it must -contain two integers, (stride_H, stride_W). Otherwise, the -stride_H = stride_W = stride. Default: stride = 1. -
- padding (int|tuple) – The padding size. If padding is a tuple, it must -contain two integers, (padding_H, padding_W). Otherwise, the -padding_H = padding_W = padding. Default: padding = 0. -
- groups (int) – The groups number of the Conv2d Layer. According to grouped -convolution in Alex Krizhevsky’s Deep CNN paper: when group=2, -the first half of the filters is only connected to the first half -of the input channels, while the second half of the filters is only -connected to the second half of the input channels. Default: groups=1 -
- param_attr (ParamAttr) – The parameters to the Conv2d Layer. Default: None -
- bias_attr (ParamAttr) – Bias parameter for the Conv2d layer. Default: None -
- use_cudnn (bool) – Use cudnn kernel or not, it is valid only when the cudnn -library is installed. Default: True -
- act (str) – Activation type. Default: None +
- input (Variable) – The input variable which is a Tensor or LoDTensor. +
- dim (int|None) – The dimension along which the maximum is computed.
+If
None, compute the maximum over all elements of +inputand return a Tensor variable with a single element, +otherwise must be in the range \([-rank(input), rank(input))\). +If \(dim < 0\), the dimension to reduce is \(rank + dim\).
+ - keep_dim (bool) – Whether to reserve the reduced dimension in the
+output Tensor. The result tensor will have one fewer dimension
+than the
inputunlesskeep_dimis true.
+ - name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically.
Returns: -
-
- The tensor variable storing the convolution and -
non-linearity activation result.
-
-
Return type: Variable
+Returns: The reduced Tensor variable.
Raises:
+ValueError– If the shapes of input, filter_size, stride, padding and -groups mismatch.Return type: Variable
Examples
--data = fluid.layers.data( - name='data', shape=[3, 32, 32], dtype='float32') -conv2d = fluid.layers.conv2d( - input=data, num_filters=2, filter_size=3, act="relu") +
# x is a Tensor variable with following elements: +# [[0.2, 0.3, 0.5, 0.9] +# [0.1, 0.2, 0.6, 0.7]] +# Each example is followed by the correspending output tensor. +fluid.layers.reduce_max(x) # [0.9] +fluid.layers.reduce_max(x, dim=0) # [0.2, 0.3, 0.6, 0.9] +fluid.layers.reduce_max(x, dim=-1) # [0.9, 0.7] +fluid.layers.reduce_max(x, dim=1, keep_dim=True) # [[0.9], [0.7]]
-sequence_pool¶
++reduce_min¶
-
-
paddle.v2.fluid.layers.sequence_pool(input, pool_type, **kwargs)
- This function add the operator for sequence pooling. -It pools features of all time-steps of each instance, and is applied -on top of the input using pool_type mentioned in the parameters.
-It supports four pool_type:
--
-
- average: \(Out[i] = \frac{\sum_i X_i}{N}\) -
- sum: \(Out[i] = \sum_jX_{ij}\) -
- sqrt: \(Out[i] = \frac{\sum_jX_{ij}}{\sqrt{len(X_i)}}\) -
- max: \(Out[i] = max(X_i)\) -
+-x is a 1-level LoDTensor: - x.lod = [[0, 2, 5, 7]] - x.data = [1, 3, 2, 4, 6, 5, 1] - x.dims = [7, 1] - -then output is a Tensor: - out.dim = [3, 1] - with condition len(x.lod[-1]) - 1 == out.dims[0] - -for different pool_type: - average: out.data = [2, 4, 3], where 2=(1+3)/2, 4=(2+4+6)/3, 3=(5+1)/2 - sum : out.data = [4, 12, 6], where 4=1+3, 12=2+4+6, 6=5+1 - sqrt : out.data = [2.82, 6.93, 4.24], where 2.82=(1+3)/sqrt(2), - 6.93=(2+4+6)/sqrt(3), 4.24=(5+1)/sqrt(2) - max : out.data = [3, 6, 5], where 3=max(1,3), 6=max(2,4,6), 5=max(5,1) -
paddle.v2.fluid.layers.reduce_min(input, dim=None, keep_dim=False, name=None) +Computes the minimum of tensor elements over the given dimension.
Parameters: -
-
- input (variable) – The input variable which is a LoDTensor. -
- pool_type (string) – The pooling type of sequence_pool. -It supports average, sum, sqrt and max. +
- input (Variable) – The input variable which is a Tensor or LoDTensor. +
- dim (int|None) – The dimension along which the minimum is computed.
+If
None, compute the minimum over all elements of +inputand return a Tensor variable with a single element, +otherwise must be in the range \([-rank(input), rank(input))\). +If \(dim < 0\), the dimension to reduce is \(rank + dim\).
+ - keep_dim (bool) – Whether to reserve the reduced dimension in the
+output Tensor. The result tensor will have one fewer dimension
+than the
inputunlesskeep_dimis true.
+ - name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically.
Returns: The sequence pooling variable which is a Tensor.
+Returns: The reduced Tensor variable.
+Return type: Variable
Examples
-x = fluid.layers.data(name='x', shape=[7, 1], - dtype='float32', lod_level=1) -avg_x = fluid.layers.sequence_pool(input=x, pool_type='average') -sum_x = fluid.layers.sequence_pool(input=x, pool_type='sum') -sqrt_x = fluid.layers.sequence_pool(input=x, pool_type='sqrt') -max_x = fluid.layers.sequence_pool(input=x, pool_type='max') +
# x is a Tensor variable with following elements: +# [[0.2, 0.3, 0.5, 0.9] +# [0.1, 0.2, 0.6, 0.7]] +# Each example is followed by the correspending output tensor. +fluid.layers.reduce_min(x) # [0.1] +fluid.layers.reduce_min(x, dim=0) # [0.1, 0.2, 0.5, 0.7] +fluid.layers.reduce_min(x, dim=-1) # [0.2, 0.1] +fluid.layers.reduce_min(x, dim=1, keep_dim=True) # [[0.2], [0.1]]
+-BlockGuardServ¶
+paddle.v2.fluid.layers.BlockGuardServ(server) +-assign¶
+++ListenAndServ¶
+-
+
+-Send¶
-+split_lod_tensor¶
++nn¶
++fc¶
Examples
---x = fluid.layers.data(name='x', shape=[7, 1], - dtype='float32', lod_level=1) -x_last_step = fluid.layers.sequence_last_step(input=x) -
--pool2d¶
--
-
-
-
paddle.v2.fluid.layers.pool2d(input, pool_size, pool_type, pool_stride=None, pool_padding=None, global_pooling=False, use_cudnn=True, name=None)
- This function adds the operator for pooling in 2 dimensions, using the -pooling configurations mentioned in input parameters.
-
--batch_norm¶
--
-
-
-
paddle.v2.fluid.layers.batch_norm(input, act=None, is_test=False, momentum=0.9, epsilon=1e-05, param_attr=None, bias_attr=None, data_layout='NCHW', name=None)
- This function helps create an operator to implement -the BatchNorm layer using the configurations from the input parameters.
-
--beam_search_decode¶
--
-
-
-
paddle.v2.fluid.layers.beam_search_decode(ids, scores, name=None)
-
-lod_rank_table¶
--
-
-
-
paddle.v2.fluid.layers.lod_rank_table(x, level=0)
- LoD Rank Table Operator. Given an input variable x and a level number -of LoD, this layer creates a LodRankTable object. A LoDRankTable object -contains a list of bi-element tuples. Each tuple consists of an index and -a length, both of which are int type. Refering to specified level of LoD, -the index is the sequence index number and the length representes the -sequence length. Please note that the list is ranked in descending order by -the length. The following is an example:
--
x is a LoDTensor: - x.lod = [[0, 2, 3], - [0, 5, 6, 7]] - x.data = [a, b, c, d, e, f, g] - -1. set level to 0: - Create lod rank table: - lod_rank_table_obj = lod_rank_table(x, level=0) - - Get: - lod_rank_table_obj.items() = [(0, 2), (1, 1)] - -2. set level to 1: - Create lod rank table: - lod_rank_table_obj = lod_rank_table(x, level=1) - - Get: - lod_rank_table_obj.items() = [(0, 5), (1, 1), (2, 1)] + +
Returns: The sequence’s last step variable which is a Tensor. +
Examples
+-x = fluid.layers.data(name='x', shape=[7, 1], + dtype='float32', lod_level=1) +x_last_step = fluid.layers.sequence_last_step(input=x)
-
+class
++IfElse¶
++dropout¶
+-
+
-
+
paddle.v2.fluid.layers.dropout(x, dropout_prob, is_test=False, seed=None, **kwargs)
+ Computes dropout.
+Drop or keep each element of x independently. Dropout is a regularization +technique for reducing overfitting by preventing neuron co-adaption during +training. The dropout operator randomly set (according to the given dropout +probability) the outputs of some units to zero, while others are remain +unchanged.
Parameters: -
-
- x (Variable) – Input variable, a LoDTensor based which to create the lod -rank table. -
- level (int) – Specify the LoD level, on which to create the lod rank -table. +
- x (variable) – The input tensor. +
- dropout_prob (float) – Probability of setting units to zero. +
- is_test (bool) – A flag indicating whether it is in test phrase or not. +
- seed (int) – A Python integer used to create random seeds. If this +parameter is set to None, a random seed is used. +NOTE: If an integer seed is given, always the same output +units will be dropped. DO NOT use a fixed seed in training.
Returns: The created LoDRankTable object.
+Returns: A tensor variable.
Return type: Variable
@@ -1830,74 +2492,116 @@ table.Examples
--x = fluid.layers.data(name='x', shape=[10], - dtype='float32', lod_level=1) -out = layers.lod_rank_table(x=x, level=0) +
x = fluid.layers.data(name="data", shape=[32, 32], dtype="float32") +droped = fluid.layers.dropout(input=x, dropout_rate=0.5)
-max_sequence_len¶
++split¶
-
-
paddle.v2.fluid.layers.max_sequence_len(rank_table)
- Max Sequence Len Operator. Given a LoDRankTable object, this layer -returns the max length of a batch of sequences. In fact, a LoDRankTable -object contains a list of tuples(<sequence index, sequence length>) and -the list is already sorted by sequence length in descending order, so the -operator just returns the sequence length of the first tuple element.
+paddle.v2.fluid.layers.split(input, num_or_sections, dim=-1, name=None) +Split the input tensor into multiple sub-tensors.
- Parameters: rank_table (Variable) – Input variable which is a LoDRankTable object. +Parameters: -
+
- input (Variable) – The input variable which is a Tensor or LoDTensor. +
- num_or_sections (int|list) – If
num_or_sectionsis an integer, +then the integer indicates the number of equal sized sub-tensors +that the tensor will be divided into. Ifnum_or_sections+is a list of integers, the length of list indicates the number of +sub-tensors and the integers indicate the sizes of sub-tensors’ +dimdimension orderly.
+ - dim (int) – The dimension along which to split. If \(dim < 0\), the +dimension to split along is \(rank(input) + dim\). +
- name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically. +
Returns: The max length of sequence. +Returns: The list of segmented tensor variables.
+Return type: Variable +Return type: List
+Examples
--x = fluid.layers.data(name='x', shape=[10], - dtype='float32', lod_level=1) -rank_table = layers.lod_rank_table(x=x, level=0) -max_seq_len = layers.max_sequence_len(rank_table) +
# x is a Tensor variable with shape [3, 9, 5]: +x0, x1, x2 = fluid.layers.split(x, num_or_sections=3, dim=1) +x0.shape # [3, 3, 5] +x1.shape # [3, 3, 5] +x2.shape # [3, 3, 5] +x0, x1, x2 = fluid.layers.split(x, num_or_sections=[2, 3, 4], dim=1) +x0.shape # [3, 2, 5] +x1.shape # [3, 3, 5] +x2.shape # [3, 4, 5]
-topk¶
++ctc_greedy_decoder¶
-
-
paddle.v2.fluid.layers.topk(input, k)
- topk
-This function performs the operation that selects the k entries in the input -vector and outputs their values and indices as vectors. Thus topk_out[j] is -the j-th largest entry in input, and its index is topk_indices[j]
+paddle.v2.fluid.layers.ctc_greedy_decoder(input, blank, name=None) +This op is used to decode sequences by greedy policy by below steps: +1. Get the indexes of max value for each row in input. a.k.a.
++
+numpy.argmax(input, axis=0).-
+
- For each sequence in result of step1, merge repeated tokens between two +blanks and delete all blanks. +
A simple example as below:
++Given: + +input.data = [[0.6, 0.1, 0.3, 0.1], + [0.3, 0.2, 0.4, 0.1], + [0.1, 0.5, 0.1, 0.3], + [0.5, 0.1, 0.3, 0.1], + + [0.5, 0.1, 0.3, 0.1], + [0.2, 0.2, 0.2, 0.4], + [0.2, 0.2, 0.1, 0.5], + [0.5, 0.1, 0.3, 0.1]] + +input.lod = [[0, 4, 8]] + +Then: + +output.data = [[2], + [1], + [3]] + +output.lod = [[0, 2, 3]] +
Parameters: -
-
- input (Variable|list) – The input tensor that has all the data. -
- k (int) – The number of top elements that the function will pick. +
- input (Variable) – (LoDTensor<float>), the probabilities of +variable-length sequences, which is a 2-D Tensor with +LoD information. It’s shape is [Lp, num_classes + 1], +where Lp is the sum of all input sequences’ length and +num_classes is the true number of classes. (not +including the blank label). +
- blank (int) – the blank label index of Connectionist Temporal +Classification (CTC) loss, which is in thehalf-opened +interval [0, num_classes + 1).
Returns: -
-
- The variable of type array that contains the k largest entries -
from input.
-
-- Variable: The variable of type array that contains the indices of k -
largest entries from input.
-
-
Returns: CTC greedy decode result.
Return type: Variable
@@ -1906,38 +2610,50 @@ the j-th largest entry in input, and its index is topk_indices[j]Examples
--x = fluid.layers.data(name='x', shape=[10]) -k = 5 -array = fluid.layers.topk(x, k) +
x = fluid.layers.data(name='x', shape=[8], dtype='float32') + +cost = fluid.layers.ctc_greedy_decoder(input=x, blank=0)
-lod_tensor_to_array¶
++edit_distance¶
-
-
paddle.v2.fluid.layers.lod_tensor_to_array(x, table)
- Convert a LOD_TENSOR to an LOD_TENSOR_ARRAY.
+paddle.v2.fluid.layers.edit_distance(input, label, normalized=False, ignored_tokens=None, name=None) +EditDistance operator computes the edit distances between a batch of +hypothesis strings and their references. Edit distance, also called +Levenshtein distance, measures how dissimilar two strings are by counting +the minimum number of operations to transform one string into anthor. +Here the operations include insertion, deletion, and substitution.
+For example, given hypothesis string A = “kitten” and reference +B = “sitting”, the edit distance is 3 for A will be transformed into B +at least after two substitutions and one insertion:
+“kitten” -> “sitten” -> “sittin” -> “sitting”
+Input(Hyps) is a LoDTensor consisting of all the hypothesis strings with +the total number denoted by batch_size, and the separation is specified +by the LoD information. And the batch_size reference strings are arranged +in order in the same way in the LoDTensor Input(Refs).
+Output(Out) contains the batch_size results and each stands for the edit +distance for a pair of strings respectively. If Attr(normalized) is true, +the edit distance will be divided by the length of reference string.
Parameters: -
-
- x (Variable|list) – The LOD tensor to be converted to a LOD tensor array. -
- table (ParamAttr|list) – The variable that stores the level of lod -which is ordered by sequence length in -descending order. +
- input (Variable) – The indices for hypothesis strings. +
- label (Variable) – The indices for reference strings. +
- normalized (bool) – Indicated whether to normalize the edit distance by +the length of reference string. +
- ignored_tokens (list of int) – Tokens that should be removed before +calculating edit distance.
Returns: -
-
- The variable of type array that has been converted from a -
tensor.
-
-
Returns: sequence-to-sequence edit distance in shape [batch_size, 1].
Return type: Variable
@@ -1946,38 +2662,42 @@ descending order.Examples
--x = fluid.layers.data(name='x', shape=[10]) -table = fluid.layers.lod_rank_table(x, level=0) -array = fluid.layers.lod_tensor_to_array(x, table) +
x = fluid.layers.data(name='x', shape=[8], dtype='float32') +y = fluid.layers.data(name='y', shape=[7], dtype='float32') + +cost = fluid.layers.edit_distance(input=x,label=y)
-array_to_lod_tensor¶
++l2_normalize¶
-
-
paddle.v2.fluid.layers.array_to_lod_tensor(x, table)
- Convert a LoD_Tensor_Aarry to an LoDTensor.
+paddle.v2.fluid.layers.l2_normalize(x, axis, epsilon=1e-12, name=None) +L2 normalize Layer
+The l2 normalize layer normalizes x along dimension axis using an L2 +norm. For a 1-D tensor (dim is fixed to 0), this layer computes
+output = x / sqrt(max(sum(x**2), epsilon))
+For x with more dimensions, this layer independently normalizes each 1-D +slice along dimension axis.
Parameters: -
-
- x (Variable|list) – The lod tensor array to be converted to a tensor. -
- table (ParamAttr|list) – The variable that stores the level of lod -which is ordered by sequence length in -descending order. +
- x (Variable|list) – The input tensor to l2_normalize layer. +
- axis (int) – Dimension along which to normalize the input. +
- epsilon (float) – A lower bound value for x‘s l2 norm. sqrt(epsilon) will +be used as the divisor if the l2 norm of x is less than +sqrt(epsilon). +
- name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically.
Returns: -
-
- The variable of type tensor that has been converted -
from an array.
-
-
Returns: The output tensor variable.
Return type: Variable
@@ -1986,37 +2706,59 @@ descending order.Examples
--x = fluid.layers.data(name='x', shape=[10]) -table = fluid.layers.lod_rank_table(x, level=0) -array = fluid.layers.lod_tensor_to_array(x, table) -lod_tensor = fluid.layers.array_to_lod_tensor(array, table) +
data = fluid.layers.data(name="data", + shape=(3, 17, 13), + dtype="float32") +normed = fluid.layers.l2_normalize(x=data, axis=1)
-fill_constant¶
++matmul¶
-
-
paddle.v2.fluid.layers.fill_constant(shape, dtype, value, force_cpu=False, out=None)
- fill_constant
-This function creates a tensor with specified shape and dtype, and -initializes it with a constant specifed by value.
-The attribute stop_gradient of the created tensor is set to True.
+paddle.v2.fluid.layers.matmul(x, y, transpose_x=False, transpose_y=False, name=None) +Applies matrix multiplication to two tensors.
+Currently, the input tensors’ rank can be any, but when the rank of any +inputs is bigger than 3, this two inputs’ rank should be equal.
+The actual behavior depends on the shapes of \(x\), \(y\) and the +flag values of
+transpose_x,transpose_y. Specifically:-
+
- If a transpose flag is specified, the last two dimensions of the tensor +are transposed. If the tensor is rank-1 of shape \([D]\), then for +\(x\) it is treated as \([1, D]\) in nontransposed form and as +\([D, 1]\) in transposed form, whereas for \(y\) it is the +opposite: It is treated as \([D, 1]\) in nontransposed form and as +\([1, D]\) in transposed form. +
- After transpose, the two tensors are 2-D or n-D and matrix multiplication
+performs in the following way.
-
+
- If both are 2-D, they are multiplied like conventional matrices. +
- If either is n-D, it is treated as a stack of matrices residing in the +last two dimensions and a batched matrix multiply supporting broadcast +applies on the two tensors. +
+
Also note that if the raw tensor \(x\) or \(y\) is rank-1 and +nontransposed, the prepended or appended dimension \(1\) will be +removed after matrix multiplication.
Parameters: -
-
- shape (tuple|list|None) – Shape of the output tensor. -
- dtype (np.dtype|core.DataType|str) – Data type of the output tensor. -
- value (float) – The constant value used to initialize the output tensor. -
- out (Variable) – The output tensor. +
- x (Variable) – The input variable which is a Tensor or LoDTensor. +
- y (Variable) – The input variable which is a Tensor or LoDTensor. +
- transpose_x (bool) – Whether to transpose \(x\) before multiplication. +
- transpose_y (bool) – Whether to transpose \(y\) before multiplication. +
- name (str|None) – A name for this layer(optional). If set None, the layer +will be named automatically.
Returns: The tensor variable storing the output.
+Returns: The product Tensor variable.
Return type: Variable
@@ -2025,37 +2767,70 @@ initializes it with a constant specifed by value.Examples
--data = fluid.layers.fill_constant(shape=[1], value=0, dtype='int64') +
# Examples to clarify shapes of the inputs and output +# x: [B, ..., M, K], y: [B, ..., K, N] +fluid.layers.matmul(x, y) # out: [B, ..., M, N] + +# x: [B, M, K], y: [B, K, N] +fluid.layers.matmul(x, y) # out: [B, M, N] + +# x: [B, M, K], y: [K, N] +fluid.layers.matmul(x, y) # out: [B, M, N] + +# x: [M, K], y: [K, N] +fluid.layers.matmul(x, y) # out: [M, N] + +# x: [B, M, K], y: [K] +fluid.layers.matmul(x, y) # out: [B, M] + +# x: [K], y: [K] +fluid.layers.matmul(x, y) # out: [1] + +# x: [M], y: [N] +fluid.layers.matmul(x, y, True, True) # out: [M, N]
-fill_constant_batch_size_like¶
++-warpctc¶
-
-
paddle.v2.fluid.layers.fill_constant_batch_size_like(input, shape, dtype, value, input_dim_idx=0, output_dim_idx=0)
- fill_constant_batch_size_like
-This function creates a tensor of specified shape, dtype and batch size, -and initializes this with a constant supplied in value. The batch size is -obtained from the input tensor.
-It also sets stop_gradient to True.
+paddle.v2.fluid.layers.warpctc(input, label, blank=0, norm_by_times=False, **kwargs) +An operator integrating the open source Warp-CTC library +(https://github.com/baidu-research/warp-ctc) +to compute Connectionist Temporal Classification (CTC) loss. +It can be aliased as softmax with CTC, since a native softmax activation is +interated to the Warp-CTC library, to to normlize values for each row of the +input tensor.
Parameters: -
-
- input (Variable) – Tensor whose dimensions will be used to get batch size -
- shape (tuple|list|None) – Shape of output tensor -
- dtype (np.dtype|core.DataType|str) – Data type of output tensor -
- value (float) – Constant value to initialize the output tensor -
- input_dim_idx (int) – Index of input’s batch size dimension -
- output_dim_idx (int) – Index of output’s batch size dimension +
- input (Variable) – (LodTensor, default: LoDTensor<float>), +the unscaled probabilities of variable-length sequences, +which is a 2-D Tensor with LoD information. +It’s shape is [Lp, num_classes + 1], where Lp is the sum of all input +sequences’ length and num_classes is the true number of classes. +(not including the blank label). +
- label (Variable) – (LodTensor, default: LoDTensor<int>), the ground truth +of variable-length sequence, which is a 2-D Tensor with LoD +information. It is of the shape [Lg, 1], where Lg is th sum of +all labels’ length. +
- blank – (int, default: 0), the blank label index of Connectionist +Temporal Classification (CTC) loss, which is in the +half-opened interval [0, num_classes + 1). +
- norm_by_times – (bool, default: false), whether to normalize +
- gradients by the number of time-step, which is also the (the) – +
- length. There is no need to normalize the gradients (sequence's) – +
- warpctc layer was follewed by a mean_op. (if) –
Returns: The tensor variable storing the output
+Returns: The Connectionist Temporal Classification (CTC) loss, +which is a 2-D Tensor of the shape [batch_size, 1].
Return type: Variable
@@ -2064,33 +2839,49 @@ obtained from the input tensor.Examples
--data = fluid.layers.fill_constant_batch_size_like( - input=like, shape=[1], value=0, dtype='int64') -
-ones¶
++sequence_reshape¶
-
-
paddle.v2.fluid.layers.ones(shape, dtype)
- ones
-This function creates a tensor of specified shape and -dtype, and initializes this with 1.
-It also sets stop_gradient to True.
+paddle.v2.fluid.layers.sequence_reshape(input, new_dim) +Sequence Reshape Layer
+This layer will rearrange the input sequences. The new dimension is set by +user. Length of each sequence is computed according to original length, +original dimension and new dimension. The following example will help to +illustrate the function of this layer:
+++x is a LoDTensor: + x.lod = [[0, 2, 6]] + x.data = [[1, 2], [3, 4], + [5, 6], [7, 8], [9, 10], [11, 12]] + x.dims = [6, 2] + +set new_dim = 4 + +then out is a LoDTensor: + out.lod = [[0, 1, 3]] + out.data = [[1, 2, 3, 4], + [5, 6, 7, 8], [9, 10, 11, 12]] + out.dims = [3, 4] +
Currently, only 1-level LoDTensor is supported and please make sure +(original length * original dimension) can be divided by new dimension with +no remainder for each sequence.
Parameters: -
-
- shape (tuple|list|None) – Shape of output tensor -
- dtype (np.dtype|core.DataType|str) – Data type of output tensor +
- input (Variable) – (LodTensor, default: LoDTensor<float>), a 2-D LoDTensor +with shape being [N, M] where M for dimension. +
- new_dim (int) – New dimension which the input LoDTensor is reshaped to.
Returns: The tensor variable storing the output
+Returns: Reshaped LoDTensor according to new dimension.
Return type: Variable
@@ -2099,32 +2890,34 @@ obtained from the input tensor.Examples
--data = fluid.layers.ones(shape=[1], dtype='int64') +
x = fluid.layers.data(name='x', shape=[5, 20], + dtype='float32', lod_level=1) +x_reshaped = layers.sequence_reshape(input=x, new_dim=10)
-zeros¶
++transpose¶
-
-
paddle.v2.fluid.layers.zeros(shape, dtype)
- zeros
-This function creates a tensor of specified shape and -dtype, and initializes this with 0.
-It also sets stop_gradient to True.
+paddle.v2.fluid.layers.transpose(x, perm, name=None) +transpose Layer
+Permute the dimensions of input according to perm.
+The i-th dimension of the returned tensor will correspond to the +perm[i]-th dimension of input.
Parameters: -
-
- shape (tuple|list|None) – Shape of output tensor -
- dtype (np.dtype|core.DataType|str) – Data type of output tensor +
- input (Variable) – (Tensor), A Tensor. +
- perm (list) – A permutation of the dimensions of input.
Returns: The tensor variable storing the output
+Returns: A transposed Tensor.
Return type: Variable
@@ -2133,134 +2926,249 @@ obtained from the input tensor.Examples
--data = fluid.layers.zeros(shape=[1], dtype='int64') +
x = fluid.layers.data(name='x', shape=[5, 10, 15], dtype='float32') +x_transposed = layers.transpose(x, perm=[1, 0, 2])
-increment¶
++im2sequence¶
-
-
paddle.v2.fluid.layers.increment(x, value=1.0, in_place=True)
- This function performs an operation that increments each value in the -input \(x\) by an amount: \(value\) as mentioned in the input -parameter. This operation is performed in-place by default.
+paddle.v2.fluid.layers.im2sequence(input, filter_size=1, stride=1, padding=0, name=None) +Extracts image patches from the input tensor to form a tensor of shape +{input.batch_size * output_height * output_width, filter_size_H * +filter_size_W * input.channels} which is similar with im2col. +This op use filter / kernel to scan images and convert these images to +sequences. After expanding, the number of time step are +output_height * output_width for an image, in which output_height and +output_width are calculated by below equation:
++\[output\_size = 1 + (2 * padding + img\_size - block\_size + stride - 1) / stride\]+And the dimension of each time step is block_y * block_x * input.channels.
-Parameters: -
-
- 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. +
- input (Variable) – The input should be a tensor in NCHW format. +
- filter_size (int|tuple|None) – The filter size. If filter_size is a tuple, +it must contain two integers, (filter_size_H, filter_size_W). +Otherwise, the filter will be a square. +
- stride (int|tuple) – The stride size. If stride is a tuple, it must +contain two integers, (stride_H, stride_W). Otherwise, the +stride_H = stride_W = stride. Default: stride = 1. +
- padding (int|tuple) – The padding size. If padding is a tuple, it can +contain two integers like (padding_H, padding_W) which means +padding_up = padding_down = padding_H and +padding_left = padding_right = padding_W. Or it can use +(padding_up, padding_left, padding_down, padding_right) to indicate +paddings of four direction. Otherwise, a scalar padding means +padding_up = padding_down = padding_left = padding_right = padding +Default: padding = 0. +
- name (int) – The name of this layer. It is optional.
Returns: -
-
- The tensor variable storing the transformation of -
element-wise increment of each value in the input.
-
-
Returns: The output is a LoDTensor with shape +{input.batch_size * output_height * output_width, +filter_size_H * filter_size_W * input.channels}. +If we regard output as a matrix, each row of this matrix is +a step of a sequence.
Return type: Variable
+Return type: output
Examples
--data = fluid.layers.data(name='data', shape=[32, 32], dtype='float32') -data = fluid.layers.increment(x=data, value=3.0, in_place=True) +
Examples:
+As an example:
++
++Given: + +x = [[[[ 6. 2. 1.] + [ 8. 3. 5.] + [ 0. 2. 6.]] + + [[ 2. 4. 4.] + [ 6. 3. 0.] + [ 6. 4. 7.]]] + + [[[ 6. 7. 1.] + [ 5. 7. 9.] + [ 2. 4. 8.]] + + [[ 1. 2. 1.] + [ 1. 3. 5.] + [ 9. 0. 8.]]]] + +x.dims = {2, 2, 3, 3} + +And: + +filter = [2, 2] +stride = [1, 1] +padding = [0, 0] + +Then: + +output.data = [[ 6. 2. 8. 3. 2. 4. 6. 3.] + [ 2. 1. 3. 5. 4. 4. 3. 0.] + [ 8. 3. 0. 2. 6. 3. 6. 4.] + [ 3. 5. 2. 6. 3. 0. 4. 7.] + [ 6. 7. 5. 7. 1. 2. 1. 3.] + [ 7. 1. 7. 9. 2. 1. 3. 5.] + [ 5. 7. 2. 4. 1. 3. 9. 0.] + [ 7. 9. 4. 8. 3. 5. 0. 8.]] + +output.dims = {8, 9} + +output.lod = [[0, 4, 8]] +The simple usage is:
++output = fluid.layers.im2sequence( + input=layer, stride=[1, 1], filter_size=[2, 2])
-array_write¶
++-nce¶
-
-
paddle.v2.fluid.layers.array_write(x, i, array=None)
- This function writes the given input variable to the specified position -indicating by the arrary index to an output LOD_TENSOR_ARRAY. If the -output LOD_TENSOR_ARRAY is not given(None), a new one will be created and -returned.
+paddle.v2.fluid.layers.nce(input, label, num_total_classes, sample_weight=None, param_attr=None, bias_attr=None, num_neg_samples=None) +Compute and return the noise-contrastive estimation training loss. +See [Noise-contrastive estimation: A new estimation principle for unnormalized statistical models](http://www.jmlr.org/proceedings/papers/v9/gutmann10a/gutmann10a.pdf). +By default this operator uses a uniform distribution for sampling.
-Parameters: -
-
- x (Variable|list) – The input tensor from which the data will be read. -
- i (Variable|list) – The index of the output LOD_TENSOR_ARRAY, pointing to -the position to which the input tensor will be -written. -
- array (Variable|list) – The output LOD_TENSOR_ARRAY to which the input -tensor will be written. If this parameter is -NONE, a new LOD_TENSOR_ARRAY will be created and -returned. +
- input – (Tensor) A tensor of shape [batch_size, dim]. +Duplicable: False Optional: False +
- label – (Tensor) A tensor of shape [batch_size, num_true_class]. ‘num_true_class’ is the number of target classes in each sample.The number of target classes per sample should be same. If you have a variable number of target classes, you can pad them out to a constant number by either repeating them or by padding with an otherwise unused class.) +Duplicable: False Optional: False +
- weight – (Tensor) A tensor of shape [num_class, dim]. ‘num_class’ is the total number of class. +Duplicable: False Optional: False +
- bias – (Tensor) A tensor of shape [num_class, 1]. ‘num_class’ is the total number of class. It is a dispensable input. +Duplicable: False Optional: True +
- sample_weight – (Tensor) A tensor of shape [batch_size, 1] storing a weight for each sample. And it is a dispensable input. The default value of sample is 1. +Duplicable: False Optional: True +
- num_total_classes (INT) – Total number of classes in all samples. +
- num_neg_samples (INT) – The number of negative classes. The default value is 10. +
- custom_neg_classes (INTS) – This attribute only be used in unitest. Classes in this list wiil be used as negative classes for every samples. Under normal conditions, user should avoid setting this attribute.
Returns: The output LOD_TENSOR_ARRAY where the input tensor is written.
-Return type: Variable
+Returns: (Tensor) A tensor of shape [batch_size, 1]. Cost of samples.
Examples
-create_array¶
+++beam_search¶
-
-
paddle.v2.fluid.layers.create_array(dtype)
- This function creates an array of type \(LOD_TENSOR_ARRAY\) using the -LayerHelper.
+paddle.v2.fluid.layers.beam_search(pre_ids, ids, scores, beam_size, end_id, level=0) +This function implements the beam search algorithm.
+
+row_conv¶
+-
+
-
+
paddle.v2.fluid.layers.row_conv(input, future_context_size, param_attr=None, act=None)
+ Row Conv Operator. This layer will apply lookahead convolution to +input. The input variable should be a 2D LoDTensor with shape [T, D]. +Parameters with shape [future_context_size + 1, D] will be created. The math +equation of row convolution is as follows:
++\[Out_{i} = \sum_{j = i} ^ {i + \tau} X_{j} \odot W_{i - j}\]+In the above equation:
+-
+
- \(Out_{i}\): The i-th row of output variable with shape [1, D]. +
- \(\tau\): Future context size. +
- \(X_{j}\): The j-th row of input variable with shape [1, D]. +
- \(W_{i-j}\): The (i-j)-th row of parameters with shape [1, D]. +
More details about row_conv please refer to the paper (http://www.cs.cmu.edu/~dyogatam/papers/wang+etal.iclrworkshop2016.pdf) and +the design document (https://github.com/PaddlePaddle/Paddle/issues/2228#issuecomment-303903645).
- Parameters: dtype (int|float) – The data type of the elements in the array. +Parameters: -
+
- input (Variable) – Input variable, a 2D LoDTensor with shape [T, D]. +
- future_context_size (int) – Future context size. Please note, the shape +of convolution kernel is [future_context_size + 1, D]. +
- param_attr (ParamAttr) – Attributes of parameters, including +name, initializer etc. +
- act (str) – Non-linear activation to be applied to output variable. +
Returns: The tensor variable storing the elements of data type. +Returns: The output tensor with same shape as input tensor.
+Return type: Variable +Return type: Variable
+Examples
--data = fluid.layers.create_array(dtype='float32') +
x = fluid.layers.data(name='x', shape=[16], + dtype='float32', lod_level=1) +out = fluid.layers.row_conv(input=x, future_context_size=2)
-less_than¶
++multiplex¶
-
-
paddle.v2.fluid.layers.less_than(x, y, cond=None, **ignored)
- Less than
-This layer returns the truth value of \(x < y\) elementwise.
+paddle.v2.fluid.layers.multiplex(inputs, index) +Multiplex Layer
+Referring to the given index variable, this layer selects rows from the +input variables to construct a multiplex variable. Assuming that there are +\(m\) input variables and \(I_i\) represents the i-th input +variable and \(i\) is in [0, \(m\)). All input variables are +tensors with same shape [\(d_0\), \(d_1\), ..., \(d_R\)]. +Please note that rank of the input tensor should be at least 2. Each input +variable will be treated as a 2-D matrix with shape [\(M\), \(N\)] +where \(M\) for \(d_0\) and \(N\) for \(d_1\) * \(d_2\) +* ... * \(d_R\). Let \(I_i[j]\) be the j-th row of the i-th input +variable. The given index variable should be a 2-D tensor with shape +[\(M\), 1]. Let ID[i] be the i-th index value of the index variable. +Then the output variable will be a tensor with shape [\(d_0\), +\(d_1\), ..., \(d_R\)]. If we treat the output tensor as a 2-D +matrix with shape [\(M\), \(N\)] and let \(O[i]\) be the i-th +row of the matrix, then O[i] is equal to \(I_{ID[i]}[i]\).
Parameters: -
-
- x (Variable) – First operand of less_than -
- y (Variable) – Second operand of less_than -
- cond (Variable|None) – Optional output variable to store the result of less_than +
- inputs (list) – A list of variables to gather from. All variables have the +same shape and the rank is at least 2. +
- index (Variable) – Tensor<int32>, index variable which is a 2-D tensor +with shape [M, 1] where M is the batch size.
Returns: The tensor variable storing the output of less_than.
+Returns: Multiplex variable gathered from input variables.
Return type: Variable
@@ -2269,721 +3177,664 @@ LayerHelper.Examples
--less = fluid.layers.less_than(x=label, y=limit) +
x1 = fluid.layers.data(name='x1', shape=[4], dtype='float32') +x2 = fluid.layers.data(name='x2', shape=[4], dtype='float32') +index = fluid.layers.data(name='index', shape=[1], dtype='int32') +out = fluid.layers.multiplex(inputs=[x1, x2], index=index)
-+array_read¶
++ops¶
++-mean¶
-
-
paddle.v2.fluid.layers.array_read(array, i)
- This function performs the operation to read the data in as an -LOD_TENSOR_ARRAY. -:param array: The input tensor that will be written to an array. -:type array: Variable|list -:param i: The subscript index in tensor array, that points the
--
+place where data will be written to.paddle.v2.fluid.layers.mean(**kwargs) +Mean Operator.
+Out is a scalar which is the mean of all elements in X.
-- Returns: The tensor type variable that has the data written to it. +Parameters: x – The input of mean op +Duplicable: False Optional: False Return type: Variable +Returns: The output of mean op Examples
-
--shrink_memory¶
--
-
-
-
paddle.v2.fluid.layers.shrink_memory(x, i, table)
- This function creates an operator to shrink_rnn_memory using the RankTable -as mentioned in the input parameter.
-array_length¶
++-mul¶
-
-
paddle.v2.fluid.layers.array_length(array)
- This function performs the operation to find the length of the input -LOD_TENSOR_ARRAY.
+paddle.v2.fluid.layers.mul(**kwargs) +Mul Operator.
+This operator is used to perform matrix multiplication for input $X$ and $Y$.
+The equation is:
+$$Out = X * Y$$
+Both the input $X$ and $Y$ can carry the LoD (Level of Details) information, +or not. But the output only shares the LoD information with input $X$.
-- Parameters: array (LOD_TENSOR_ARRAY) – The input array that will be used -to compute the length. -Returns: The length of the input LoDTensorArray. +Parameters: -
+
- x – (Tensor), The first input tensor of mul op. +Duplicable: False Optional: False +
- y – (Tensor), The second input tensor of mul op. +Duplicable: False Optional: False +
- x_num_col_dims (INT) – (int, default 1), The mul_op can take tensors with more than two +dimensions as its inputs. If the input $X$ is a tensor with more +than two dimensions, $X$ will be flattened into a two-dimensional +matrix first. The flattening rule is: the first num_col_dims +will be flattened to form the first dimension of the final matrix +(the height of the matrix), and the rest rank(X) - num_col_dims +dimensions are flattened to form the second dimension of the final +matrix (the width of the matrix). As a result, height of the +flattened matrix is equal to the product of $X$’s first +x_num_col_dims dimensions’ sizes, and width of the flattened +matrix is equal to the product of $X$’s last rank(x) - num_col_dims +dimensions’ size. For example, suppose $X$ is a 6-dimensional +tensor with the shape [2, 3, 4, 5, 6], and x_num_col_dims = 3. +Thus, the flattened matrix will have a shape [2 x 3 x 4, 5 x 6] = +[24, 30]. +
- y_num_col_dims (INT) – (int, default 1), The mul_op can take tensors with more than two, +dimensions as its inputs. If the input $Y$ is a tensor with more +than two dimensions, $Y$ will be flattened into a two-dimensional +matrix first. The attribute y_num_col_dims determines how $Y$ is +flattened. See comments of x_num_col_dims for more details. +
Return type: Variable +Returns: (Tensor), The output tensor of mul op.
+Examples
-conv2d_transpose¶
++-reshape¶
-
-
paddle.v2.fluid.layers.conv2d_transpose(input, num_filters, output_size=None, filter_size=None, padding=None, stride=None, dilation=None, param_attr=None, use_cudnn=True, name=None)
- Convlution2D transpose layer
-The convolution2D transpose layer calculates the output based on the input, -filter, and dilations, strides, paddings. Input(Input) and output(Output) -are in NCHW format. Where N is batch size, C is the number of channels, -H is the height of the feature, and W is the width of the feature. -Parameters(dilations, strides, paddings) are two elements. These two elements -represent height and width, respectively. The details of convolution transpose -layer, please refer to the following explanation and references -therein.
-For each input \(X\), the equation is:
--\[Out = W \ast X\]-In the above equation:
--
-
- \(X\): Input value, a tensor with NCHW format. -
- \(W\): Filter value, a tensor with MCHW format. -
- \(\ast\) : Convolution transpose operation. -
-
-
- \(Out\): Output value, the shape of \(Out\) and \(X\) may be -
- different. -
-
Example
--
-
Input:
-Input shape: $(N, C_{in}, H_{in}, W_{in})$
-Filter shape: $(C_{in}, C_{out}, H_f, W_f)$
-
-Output:
-Output shape: $(N, C_{out}, H_{out}, W_{out})$
-
-
Where
--\[\begin{split}H_{out} &= (H_{in} - 1) * strides[0] - 2 * paddings[0] + dilations[0] * (H_f - 1) + 1 \\ -W_{out} &= (W_{in} - 1) * strides[1] - 2 * paddings[1] + dilations[1] * (W_f - 1) + 1\end{split}\]+paddle.v2.fluid.layers.reshape(**kwargs) +Reshape Operator.
+Reshape Input(X) into the shape specified by Attr(shape).
+An example: +Given a 2-D tensor X with 2 rows and 2 columns : [[1, 2], [3, 4]]
+and target shape = [1, 4], the reshape operator will transform +the tensor X into a 2-D tensor: [[1, 2, 3, 4]]
+One dimension in the target shape can be set -1, representing that its +size is unknown. In this case, the real dimension will be infered from +the original shape of Input(X) and other dimensions in the target shape.
-Parameters: -
-
- input (Variable) – The input image with [N, C, H, W] format. -
- num_filters (int) – The number of the filter. It is as same as the output -image channel. -
- output_size (int|tuple|None) – The output image size. If output size is a -tuple, it must contain two integers, (image_H, image_W). This -parameter only works when filter_size is None. -
- filter_size (int|tuple|None) – The filter size. If filter_size is a tuple, -it must contain two integers, (filter_size_H, filter_size_W). -Otherwise, the filter will be a square. None if use output size to -calculate filter_size. -
- padding (int|tuple) – The padding size. If padding is a tuple, it must -contain two integers, (padding_H, padding_W). Otherwise, the -padding_H = padding_W = padding. Default: padding = 0. -
- stride (int|tuple) – The stride size. If stride is a tuple, it must -contain two integers, (stride_H, stride_W). Otherwise, the -stride_H = stride_W = stride. Default: stride = 1. -
- dilation (int|tuple) – The dilation size. If dilation is a tuple, it must -contain two integers, (dilation_H, dilation_W). Otherwise, the -dilation_H = dilation_W = dilation. Default: dilation = 1. -
- param_attr (ParamAttr) – The parameters to the Conv2d_transpose Layer. -Default: None -
- use_cudnn (bool) – Use cudnn kernel or not, it is valid only when the cudnn -library is installed. Default: True -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x – The input tensor of reshape operator. +Duplicable: False Optional: False +
- shape (INTS) – (vector<int>) Target shape of reshape operator.
Returns: The tensor variable storing the convolution transpose result.
-Return type: Variable
-Raises:
+ValueError– If the shapes of input, filter_size, stride, padding and -groups mismatch.Returns: The output tensor of reshape operator.
Examples
---data = fluid.layers.data( - name='data', shape=[3, 32, 32], dtype='float32') -conv2d_transpose = fluid.layers.conv2d_transpose( - input=data, num_filters=2, filter_size=3) -
-sequence_expand¶
--
-
-
-
paddle.v2.fluid.layers.sequence_expand(x, y, name=None)
- Sequence Expand Layer. This layer will expand the input variable x -according to LoD information of y. And the following examples will -explain how sequence_expand works:
-+* Case 1 - x is a LoDTensor: - x.lod = [[0, 2, 3], - [0, 1, 3, 4]] - x.data = [a, b, c, d] - x.dims = [4, 1] - - y is a LoDTensor: - y.lod = [[0, 2, 4], - [0, 3, 6, 7, 8]] - - with condition len(y.lod[-1]) - 1 == x.dims[0] - - then output is a 2-level LoDTensor: - out.lod = [[0, 2, 4], - [0, 3, 6, 7, 8]] - out.data = [a, a, a, b, b, b, c, d] - out.dims = [8, 1] - -* Case 2 - x is a Tensor: - x.data = [a, b, c] - x.dims = [3, 1] - - y is a LoDTensor: - y.lod = [[0, 2, 3, 6]] - - with condition len(y.lod[-1]) - 1 == x.dims[0] + - then output is a 1-level LoDTensor: - out.lod = [[0, 2, 3, 6]] - out.data = [a, a, b, c, c, c] - out.dims = [6, 1] -
+-scale¶
+-
+
-
+
paddle.v2.fluid.layers.scale(**kwargs)
+ Scale operator
+$$Out = scale*X$$
-Parameters: -
-
- x (Variable) – The input variable which is a Tensor or LoDTensor. -
- y (Variable) – The input variable which is a LoDTensor. -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x – (Tensor) Input tensor of scale operator. +Duplicable: False Optional: False +
- scale (FLOAT) – (float, default 1.0)The scaling factor of the scale operator.
Returns: The expanded variable which is a LoDTensor.
-Return type: Variable
+Returns: (Tensor) Output tensor of scale operator.
Examples
--x = fluid.layers.data(name='x', shape=[10], dtype='float32') -y = fluid.layers.data(name='y', shape=[10, 20], - dtype='float32', lod_level=1) -out = layers.sequence_expand(x=x, y=y) -
-gru_unit¶
++sigmoid_cross_entropy_with_logits¶
-
-
paddle.v2.fluid.layers.gru_unit(input, hidden, size, weight=None, bias=None, activation='tanh', gate_activation='sigmoid')
- GRU unit layer. The equation of a gru step is:
+paddle.v2.fluid.layers.sigmoid_cross_entropy_with_logits(**kwargs) +SigmoidCrossEntropyWithLogits Operator.
+This measures the element-wise probability error in classification tasks +in which each class is independent. This can be thought of as predicting labels +for a data-point, where labels are not mutually exclusive. +For example, a news article can be about politics, technology or sports +at the same time or none of these.
+The logistic loss is given as follows:
-
--\[ \begin{align}\begin{aligned}u_t & = actGate(xu_{t} + W_u h_{t-1} + b_u)\\r_t & = actGate(xr_{t} + W_r h_{t-1} + b_r)\\m_t & = actNode(xm_t + W_c dot(r_t, h_{t-1}) + b_m)\\h_t & = dot((1-u_t), m_t) + dot(u_t, h_{t-1})\end{aligned}\end{align} \]-The inputs of gru unit includes \(z_t\), \(h_{t-1}\). In terms -of the equation above, the \(z_t\) is split into 3 parts - -\(xu_t\), \(xr_t\) and \(xm_t\). This means that in order to -implement a full GRU unit operator for an input, a fully -connected layer has to be applied, such that \(z_t = W_{fc}x_t\).
-The terms \(u_t\) and \(r_t\) represent the update and reset gates -of the GRU cell. Unlike LSTM, GRU has one lesser gate. However, there is -an intermediate candidate hidden output, which is denoted by \(m_t\). -This layer has three outputs \(h_t\), \(dot(r_t, h_{t-1})\) -and concatenation of \(u_t\), \(r_t\) and \(m_t\).
+$$loss = -Labels * log(sigma(X)) - (1 - Labels) * log(1 - sigma(X))$$+We know that $$sigma(X) = (1 / (1 + exp(-X)))$$. By substituting this we get:
++
+$$loss = X - X * Labels + log(1 + exp(-X))$$For stability and to prevent overflow of $$exp(-X)$$ when X < 0, +we reformulate the loss as follows:
++
+$$loss = max(X, 0) - X * Labels + log(1 + exp(-|X|))$$Both the input X and Labels can carry the LoD (Level of Details) information. +However the output only shares the LoD with input X.
-Parameters: -
-
- input (Variable) – The fc transformed input value of current step. -
- hidden (Variable) – The hidden value of lstm unit from previous step. -
- size (integer) – The input dimension value. -
- weight (ParamAttr) – The weight parameters for gru unit. Default: None -
- bias (ParamAttr) – The bias parameters for gru unit. Default: None -
- activation (string) – The activation type for cell (actNode). -Default: ‘tanh’ -
- gate_activation (string) – The activation type for gates (actGate). -Default: ‘sigmoid’ +
- x – (Tensor, default Tensor<float>), a 2-D tensor with shape N x D, where N is the batch size and D is the number of classes. This input is a tensor of logits computed by the previous operator. Logits are unscaled log probabilities given as log(p/(1-p)). +Duplicable: False Optional: False +
- label – (Tensor, default Tensor<float>), a 2-D tensor of the same type and shape as X. This input is a tensor of probabalistic labels for each logit +Duplicable: False Optional: False
Returns: The hidden value, reset-hidden value and gate values.
-Return type: tuple
+Returns: (Tensor, default Tensor<float>), a 2-D tensor with shape N x D of elementwise logistic losses.
Examples
-+# assuming we have x_t_data and prev_hidden of size=10 -x_t = fluid.layers.fc(input=x_t_data, size=30) -hidden_val, r_h_val, gate_val = fluid.layers.gru_unit(input=x_t, - hidden = prev_hidden) + + +
+-elementwise_add¶
+-
+
-
+
paddle.v2.fluid.layers.elementwise_add(**kwargs)
+ Limited Elementwise Add Operator.
+The equation is:
+$$Out = X + Y$$
+$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be +smaller than or equal to the dimensions of $X$.
+There are two cases for this operator: +1. The shape of $Y$ is same with $X$; +2. The shape of $Y$ is a subset of $X$.
+For case 2: +$Y$ will be broadcasted to match the shape of $X$ and axis should be +set to index of the start dimension to broadcast $Y$ onto $X$.
+-
+
- For example +
- +
shape(X) = (2, 3, 4, 5), shape(Y) = (,) +shape(X) = (2, 3, 4, 5), shape(Y) = (5,) +shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) +shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 +shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0
+
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) +information. However, the output only shares the LoD information with input $X$.
++
+ + + Parameters: -
+
- x – (Tensor), The first input tensor of elementwise op. +Duplicable: False Optional: False +
- y – (Tensor), The second input tensor of elementwise op. +Duplicable: False Optional: False +
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. +
Returns: The output of elementwise op.
+
-lstm_unit¶
+++elementwise_div¶
-
-
paddle.v2.fluid.layers.lstm_unit(x_t, hidden_t_prev, cell_t_prev, forget_bias=0.0, param_attr=None, bias_attr=None, name=None)
- Lstm unit layer. The equation of a lstm step is:
--
--\[ \begin{align}\begin{aligned}i_t & = \sigma(W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i)\\f_t & = \sigma(W_{x_f}x_{t} + W_{h_f}h_{t-1} + b_f)\\c_t & = f_tc_{t-1} + i_t tanh (W_{x_c}x_t + W_{h_c}h_{t-1} + b_c)\\o_t & = \sigma(W_{x_o}x_{t} + W_{h_o}h_{t-1} + b_o)\\h_t & = o_t tanh(c_t)\end{aligned}\end{align} \]-The inputs of lstm unit include \(x_t\), \(h_{t-1}\) and -\(c_{t-1}\). The 2nd dimensions of \(h_{t-1}\) and \(c_{t-1}\) -should be same. The implementation separates the linear transformation and -non-linear transformation apart. Here, we take \(i_t\) as an example. -The linear transformation is applied by calling a fc layer and the -equation is:
--
--\[L_{i_t} = W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i\]-The non-linear transformation is applied by calling lstm_unit_op and the -equation is:
--
--\[i_t = \sigma(L_{i_t})\]-This layer has two outputs including \(h_t\) and \(o_t\).
+paddle.v2.fluid.layers.elementwise_div(**kwargs) +Limited Elementwise Div Operator.
+The equation is:
+$$Out = X / Y$$
+$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be +smaller than or equal to the dimensions of $X$.
+There are two cases for this operator: +1. The shape of $Y$ is same with $X$; +2. The shape of $Y$ is a subset of $X$.
+For case 2: +$Y$ will be broadcasted to match the shape of $X$ and axis should be +set to index of the start dimension to broadcast $Y$ onto $X$.
+-
+
- For example +
- ++
shape(X) = (2, 3, 4, 5), shape(Y) = (,) +shape(X) = (2, 3, 4, 5), shape(Y) = (5,) +shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) +shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 +shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 +
+
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) +information. However, the output only shares the LoD information with input $X$.
+Parameters: -
-
- x_t (Variable) – The input value of current step, a 2-D tensor with shape -M x N, M for batch size and N for input size. -
- hidden_t_prev (Variable) – The hidden value of lstm unit, a 2-D tensor -with shape M x S, M for batch size and S for size of lstm unit. -
- cell_t_prev (Variable) – The cell value of lstm unit, a 2-D tensor with -shape M x S, M for batch size and S for size of lstm unit. -
- forget_bias (float) – The forget bias of lstm unit. -
- param_attr (ParamAttr) – The attributes of parameter weights, used to set -initializer, name etc. -
- bias_attr (ParamAttr) – The attributes of bias weights, if not False, -bias weights will be created and be set to default value. -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x – (Tensor), The first input tensor of elementwise op. +Duplicable: False Optional: False +
- y – (Tensor), The second input tensor of elementwise op. +Duplicable: False Optional: False +
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X.
Returns: The hidden value and cell value of lstm unit.
+Returns: The output of elementwise op.
Return type: tuple
+
+elementwise_sub¶
+-
+
-
+
paddle.v2.fluid.layers.elementwise_sub(**kwargs)
+ Limited Elementwise Sub Operator.
+The equation is:
+$$Out = X - Y$$
+$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be +smaller than or equal to the dimensions of $X$.
+There are two cases for this operator: +1. The shape of $Y$ is same with $X$; +2. The shape of $Y$ is a subset of $X$.
+For case 2: +$Y$ will be broadcasted to match the shape of $X$ and axis should be +set to index of the start dimension to broadcast $Y$ onto $X$.
+-
+
- For example +
- ++
shape(X) = (2, 3, 4, 5), shape(Y) = (,) +shape(X) = (2, 3, 4, 5), shape(Y) = (5,) +shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) +shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 +shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 +
+
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) +information. However, the output only shares the LoD information with input $X$.
++
-+ + + Parameters: -
+
- x – (Tensor), The first input tensor of elementwise op. +Duplicable: False Optional: False +
- y – (Tensor), The second input tensor of elementwise op. +Duplicable: False Optional: False +
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. +
Raises:
+ValueError– The ranks of x_t, hidden_t_prev and cell_t_prev -not be 2 or the 1st dimensions of x_t, hidden_t_prev -and cell_t_prev not be the same or the 2nd dimensions of -hidden_t_prev and cell_t_prev not be the same.Returns: The output of elementwise op.
Examples
-+x_t = fluid.layers.fc(input=x_t_data, size=10) -prev_hidden = fluid.layers.fc(input=prev_hidden_data, size=30) -prev_cell = fluid.layers.fc(input=prev_cell_data, size=30) -hidden_value, cell_value = fluid.layers.lstm_unit(x_t=x_t, - hidden_t_prev=prev_hidden, - cell_t_prev=prev_cell) + + +
+-elementwise_mul¶
+-
+
-
+
paddle.v2.fluid.layers.elementwise_mul(**kwargs)
+ Limited Elementwise Mul Operator.
+The equation is:
+$$Out = X odotY$$
+$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be +smaller than or equal to the dimensions of $X$.
+There are two cases for this operator: +1. The shape of $Y$ is same with $X$; +2. The shape of $Y$ is a subset of $X$.
+For case 2: +$Y$ will be broadcasted to match the shape of $X$ and axis should be +set to index of the start dimension to broadcast $Y$ onto $X$.
+-
+
- For example +
- -
shape(X) = (2, 3, 4, 5), shape(Y) = (,) +shape(X) = (2, 3, 4, 5), shape(Y) = (5,) +shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) +shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 +shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0
--sequence_softmax¶
--
-
-
-
paddle.v2.fluid.layers.sequence_softmax(**kwargs)
- Sequence Softmax Operator.
-SequenceSoftmaxOp computes the softmax activation among all time-steps for each -sequence. The dimension of each time-step should be 1. Thus, the shape of -input Tensor can be either [N, 1] or [N], where N is the sum of the length -of all sequences.
-The algorithm works as follows:
--
-for i-th sequence in a mini-batch:$$ -Out(X[lod[i]:lod[i+1]], :) = frac{exp(X[lod[i]:lod[i+1], :])} {sum(exp(X[lod[i]:lod[i+1], :]))} -$$
-For example, for a mini-batch of 3 sequences with variable-length, -each containing 2, 3, 2 time-steps, the lod of which is [0, 2, 5, 7], -then softmax will be computed among X[0:2, :], X[2:5, :], X[5:7, :] -and N turns out to be 7.
+
+
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) +information. However, the output only shares the LoD information with input $X$.
- Parameters: x – (LoDTensor) 1-D or 2-D input LoDTensor with the 2-nd dimension of length 1. -Duplicable: False Optional: False +Parameters: -
+
- x – (Tensor), The first input tensor of elementwise op. +Duplicable: False Optional: False +
- y – (Tensor), The second input tensor of elementwise op. +Duplicable: False Optional: False +
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. +
Returns: (LoDTensor) 1-D or 2-D output LoDTensor with the 2-nd dimension of length 1. +Returns: The output of elementwise op.
+-reduce_sum¶
++-elementwise_max¶
-
-
paddle.v2.fluid.layers.reduce_sum(input, dim=None, keep_dim=False, name=None)
- Computes the sum of tensor elements over the given dimension.
+paddle.v2.fluid.layers.elementwise_max(**kwargs) +Limited Elementwise Max Operator.
+The equation is:
+$$Out = max(X, Y)$$
+$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be +smaller than or equal to the dimensions of $X$.
+There are two cases for this operator: +1. The shape of $Y$ is same with $X$; +2. The shape of $Y$ is a subset of $X$.
+For case 2: +$Y$ will be broadcasted to match the shape of $X$ and axis should be +set to index of the start dimension to broadcast $Y$ onto $X$.
+-
+
- For example +
- ++
shape(X) = (2, 3, 4, 5), shape(Y) = (,) +shape(X) = (2, 3, 4, 5), shape(Y) = (5,) +shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) +shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 +shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 +
+
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) +information. However, the output only shares the LoD information with input $X$.
-Parameters: -
-
- input (Variable) – The input variable which is a Tensor or LoDTensor. -
- dim (int|None) – The dimension along which the sum is performed. If
-
None, sum all elements ofinputand return a -Tensor variable with a single element, otherwise must be in the -range \([-rank(input), rank(input))\). If \(dim < 0\), -the dimension to reduce is \(rank + dim\).
- - keep_dim (bool) – Whether to reserve the reduced dimension in the
-output Tensor. The result tensor will have one fewer dimension
-than the
inputunlesskeep_dimis true.
- - name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x – (Tensor), The first input tensor of elementwise op. +Duplicable: False Optional: False +
- y – (Tensor), The second input tensor of elementwise op. +Duplicable: False Optional: False +
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X.
Returns: The reduced Tensor variable.
-Return type: Variable
+Returns: The output of elementwise op.
Examples
--# x is a Tensor variable with following elements: -# [[0.2, 0.3, 0.5, 0.9] -# [0.1, 0.2, 0.6, 0.7]] -# Each example is followed by the correspending output tensor. -fluid.layers.reduce_sum(x) # [3.5] -fluid.layers.reduce_sum(x, dim=0) # [0.3, 0.5, 1.1, 1.6] -fluid.layers.reduce_sum(x, dim=-1) # [1.9, 1.6] -fluid.layers.reduce_sum(x, dim=1, keep_dim=True) # [[1.9], [1.6]] -
-reduce_mean¶
++-elementwise_min¶
-
-
paddle.v2.fluid.layers.reduce_mean(input, dim=None, keep_dim=False, name=None)
- Computes the mean of tensor elements over the given dimension.
+paddle.v2.fluid.layers.elementwise_min(**kwargs) +Limited Elementwise Max Operator.
+The equation is:
+$$Out = min(X, Y)$$
+$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be +smaller than or equal to the dimensions of $X$.
+There are two cases for this operator: +1. The shape of $Y$ is same with $X$; +2. The shape of $Y$ is a subset of $X$.
+For case 2: +$Y$ will be broadcasted to match the shape of $X$ and axis should be +set to index of the start dimension to broadcast $Y$ onto $X$.
+-
+
- For example +
- ++
shape(X) = (2, 3, 4, 5), shape(Y) = (,) +shape(X) = (2, 3, 4, 5), shape(Y) = (5,) +shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) +shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 +shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 +
+
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) +information. However, the output only shares the LoD information with input $X$.
-Parameters: -
-
- input (Variable) – The input variable which is a Tensor or LoDTensor. -
- dim (int|None) – The dimension along which the mean is computed. If
-
None, compute the mean over all elements ofinput-and return a Tensor variable with a single element, otherwise -must be in the range \([-rank(input), rank(input))\). If -\(dim < 0\), the dimension to reduce is \(rank + dim\).
- - keep_dim (bool) – Whether to reserve the reduced dimension in the
-output Tensor. The result tensor will have one fewer dimension
-than the
inputunlesskeep_dimis true.
- - name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x – (Tensor), The first input tensor of elementwise op. +Duplicable: False Optional: False +
- y – (Tensor), The second input tensor of elementwise op. +Duplicable: False Optional: False +
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X.
Returns: The reduced Tensor variable.
-Return type: Variable
+Returns: The output of elementwise op.
Examples
--# x is a Tensor variable with following elements: -# [[0.2, 0.3, 0.5, 0.9] -# [0.1, 0.2, 0.6, 0.7]] -# Each example is followed by the correspending output tensor. -fluid.layers.reduce_mean(x) # [0.4375] -fluid.layers.reduce_mean(x, dim=0) # [0.15, 0.25, 0.55, 0.8] -fluid.layers.reduce_mean(x, dim=-1) # [0.475, 0.4] -fluid.layers.reduce_mean(x, dim=1, keep_dim=True) # [[0.475], [0.4]] -
-reduce_max¶
++-elementwise_pow¶
-
-
paddle.v2.fluid.layers.reduce_max(input, dim=None, keep_dim=False, name=None)
- Computes the maximum of tensor elements over the given dimension.
+paddle.v2.fluid.layers.elementwise_pow(**kwargs) +Limited Elementwise Pow Operator.
+The equation is:
+$$Out = X ^ Y$$
+$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be +smaller than or equal to the dimensions of $X$.
+There are two cases for this operator: +1. The shape of $Y$ is same with $X$; +2. The shape of $Y$ is a subset of $X$.
+For case 2: +$Y$ will be broadcasted to match the shape of $X$ and axis should be +set to index of the start dimension to broadcast $Y$ onto $X$.
+-
+
- For example +
- ++
shape(X) = (2, 3, 4, 5), shape(Y) = (,) +shape(X) = (2, 3, 4, 5), shape(Y) = (5,) +shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) +shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 +shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 +
+
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) +information. However, the output only shares the LoD information with input $X$.
-Parameters: -
-
- input (Variable) – The input variable which is a Tensor or LoDTensor. -
- dim (int|None) – The dimension along which the maximum is computed.
-If
None, compute the maximum over all elements of -inputand return a Tensor variable with a single element, -otherwise must be in the range \([-rank(input), rank(input))\). -If \(dim < 0\), the dimension to reduce is \(rank + dim\).
- - keep_dim (bool) – Whether to reserve the reduced dimension in the
-output Tensor. The result tensor will have one fewer dimension
-than the
inputunlesskeep_dimis true.
- - name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x – (Tensor), The first input tensor of elementwise op. +Duplicable: False Optional: False +
- y – (Tensor), The second input tensor of elementwise op. +Duplicable: False Optional: False +
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X.
Returns: The reduced Tensor variable.
-Return type: Variable
+Returns: The output of elementwise op.
Examples
--# x is a Tensor variable with following elements: -# [[0.2, 0.3, 0.5, 0.9] -# [0.1, 0.2, 0.6, 0.7]] -# Each example is followed by the correspending output tensor. -fluid.layers.reduce_max(x) # [0.9] -fluid.layers.reduce_max(x, dim=0) # [0.2, 0.3, 0.6, 0.9] -fluid.layers.reduce_max(x, dim=-1) # [0.9, 0.7] -fluid.layers.reduce_max(x, dim=1, keep_dim=True) # [[0.9], [0.7]] -
-reduce_min¶
++-clip¶
-
-
paddle.v2.fluid.layers.reduce_min(input, dim=None, keep_dim=False, name=None)
- Computes the minimum of tensor elements over the given dimension.
+paddle.v2.fluid.layers.clip(**kwargs) +Clip Operator.
+The clip operator limits the value of given input within an interval. The +interval is specified with arguments ‘min’ and ‘max’:
+$$ +Out = min(max(X, min), max) +$$
-Parameters: -
-
- input (Variable) – The input variable which is a Tensor or LoDTensor. -
- dim (int|None) – The dimension along which the minimum is computed.
-If
None, compute the minimum over all elements of -inputand return a Tensor variable with a single element, -otherwise must be in the range \([-rank(input), rank(input))\). -If \(dim < 0\), the dimension to reduce is \(rank + dim\).
- - keep_dim (bool) – Whether to reserve the reduced dimension in the
-output Tensor. The result tensor will have one fewer dimension
-than the
inputunlesskeep_dimis true.
- - name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x – (Tensor)The input of clip op.The number of dimensions must be between [1, 9]. +Duplicable: False Optional: False +
- min (FLOAT) – (float)Minimum value, under which element is replaced by min. +
- max (FLOAT) – (float)Maximum value, above which element is replaced by max
Returns: The reduced Tensor variable.
-Return type: Variable
+Returns: (Tensor)The output of clip op with shape as input(X)
Examples
--# x is a Tensor variable with following elements: -# [[0.2, 0.3, 0.5, 0.9] -# [0.1, 0.2, 0.6, 0.7]] -# Each example is followed by the correspending output tensor. -fluid.layers.reduce_min(x) # [0.1] -fluid.layers.reduce_min(x, dim=0) # [0.1, 0.2, 0.5, 0.7] -fluid.layers.reduce_min(x, dim=-1) # [0.2, 0.1] -fluid.layers.reduce_min(x, dim=1, keep_dim=True) # [[0.2], [0.1]] -
-split¶
++-clip_by_norm¶
-
-
paddle.v2.fluid.layers.split(input, num_or_sections, dim=-1, name=None)
- Split the input tensor into multiple sub-tensors.
+paddle.v2.fluid.layers.clip_by_norm(**kwargs) +ClipByNorm Operator.
+This operator limits the L2 norm of the input $X$ within $max_norm$. +If the L2 norm of $X$ is less than or equal to $max_norm$, $Out$ will be +the same as $X$. If the L2 norm of $X$ is greater than $max_norm$, $X$ will +be linearly scaled to make the L2 norm of $Out$ equal to $max_norm$, as +shown in the following formula:
+$$ +Out = frac{max_norm * X}{norm(X)}, +$$
+where $norm(X)$ represents the L2 norm of $X$.
-Parameters: -
-
- input (Variable) – The input variable which is a Tensor or LoDTensor. -
- num_or_sections (int|list) – If
num_or_sectionsis an integer, -then the integer indicates the number of equal sized sub-tensors -that the tensor will be divided into. Ifnum_or_sections-is a list of integers, the length of list indicates the number of -sub-tensors and the integers indicate the sizes of sub-tensors’ -dimdimension orderly.
- - dim (int) – The dimension along which to split. If \(dim < 0\), the -dimension to split along is \(rank(input) + dim\). -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x – (Tensor) The input of clip_by_norm op.The number of dimensions must be between [1, 9]. +Duplicable: False Optional: False +
- max_norm (FLOAT) – (float) The maximum norm value.
Returns: The list of segmented tensor variables.
-Return type: List
+Returns: (Tensor) The output of clip_by_norm op with shape as input(X)
Examples
--# x is a Tensor variable with shape [3, 9, 5]: -x0, x1, x2 = fluid.layers.split(x, num_or_sections=3, dim=1) -x0.shape # [3, 3, 5] -x1.shape # [3, 3, 5] -x2.shape # [3, 3, 5] -x0, x1, x2 = fluid.layers.split(x, num_or_sections=[2, 3, 4], dim=1) -x0.shape # [3, 2, 5] -x1.shape # [3, 3, 5] -x2.shape # [3, 4, 5] -
-matmul¶
++sequence_softmax¶
-
-
paddle.v2.fluid.layers.matmul(x, y, transpose_x=False, transpose_y=False, name=None)
- Applies matrix multiplication to two tensors.
-Currently, the input tensors’ rank can be any, but when the rank of any -inputs is bigger than 3, this two inputs’ rank should be equal.
-The actual behavior depends on the shapes of \(x\), \(y\) and the -flag values of
-transpose_x,transpose_y. Specifically:-
-
- If a transpose flag is specified, the last two dimensions of the tensor -are transposed. If the tensor is rank-1 of shape \([D]\), then for -\(x\) it is treated as \([1, D]\) in nontransposed form and as -\([D, 1]\) in transposed form, whereas for \(y\) it is the -opposite: It is treated as \([D, 1]\) in nontransposed form and as -\([1, D]\) in transposed form. -
- After transpose, the two tensors are 2-D or n-D and matrix multiplication
-performs in the following way.
-
-
- If both are 2-D, they are multiplied like conventional matrices. -
- If either is n-D, it is treated as a stack of matrices residing in the -last two dimensions and a batched matrix multiply supporting broadcast -applies on the two tensors. -
-
Also note that if the raw tensor \(x\) or \(y\) is rank-1 and -nontransposed, the prepended or appended dimension \(1\) will be -removed after matrix multiplication.
+paddle.v2.fluid.layers.sequence_softmax(**kwargs) +Sequence Softmax Operator.
+SequenceSoftmaxOp computes the softmax activation among all time-steps for each +sequence. The dimension of each time-step should be 1. Thus, the shape of +input Tensor can be either [N, 1] or [N], where N is the sum of the length +of all sequences.
+The algorithm works as follows:
++
+for i-th sequence in a mini-batch:$$ +Out(X[lod[i]:lod[i+1]], :) = frac{exp(X[lod[i]:lod[i+1], :])} {sum(exp(X[lod[i]:lod[i+1], :]))} +$$
+For example, for a mini-batch of 3 sequences with variable-length, +each containing 2, 3, 2 time-steps, the lod of which is [0, 2, 5, 7], +then softmax will be computed among X[0:2, :], X[2:5, :], X[5:7, :] +and N turns out to be 7.
-- Parameters: -
-
- x (Variable) – The input variable which is a Tensor or LoDTensor. -
- y (Variable) – The input variable which is a Tensor or LoDTensor. -
- transpose_x (bool) – Whether to transpose \(x\) before multiplication. -
- transpose_y (bool) – Whether to transpose \(y\) before multiplication. -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. -
Returns: The product Tensor variable.
-Parameters: x – (LoDTensor) 1-D or 2-D input LoDTensor with the 2-nd dimension of length 1. +Duplicable: False Optional: False Return type: Variable
-Returns: (LoDTensor) 1-D or 2-D output LoDTensor with the 2-nd dimension of length 1. Examples
-+# Examples to clarify shapes of the inputs and output -# x: [B, ..., M, K], y: [B, ..., K, N] -fluid.layers.matmul(x, y) # out: [B, ..., M, N] - -# x: [B, M, K], y: [B, K, N] -fluid.layers.matmul(x, y) # out: [B, M, N] - -# x: [B, M, K], y: [K, N] -fluid.layers.matmul(x, y) # out: [B, M, N] - -# x: [M, K], y: [K, N] -fluid.layers.matmul(x, y) # out: [M, N] - -# x: [B, M, K], y: [K] -fluid.layers.matmul(x, y) # out: [B, M] - -# x: [K], y: [K] -fluid.layers.matmul(x, y) # out: [1] + -# x: [M], y: [N] -fluid.layers.matmul(x, y, True, True) # out: [M, N] -
+sigmoid¶
+-
+
-
+
paddle.v2.fluid.layers.sigmoid(**kwargs)
+ Sigmoid Activation Operator
+$$out = frac{1}{1 + e^{-x}}$$
++
+ + + Parameters: x – Input of Sigmoid operator +Duplicable: False Optional: False +Returns: Output of Sigmoid operator +
-logsigmoid¶
+logsigmoid¶
-
paddle.v2.fluid.layers.logsigmoid(**kwargs)
@@ -3004,7 +3855,7 @@ Duplicable: False Optional: False
-exp¶
+exp¶
-
paddle.v2.fluid.layers.exp(**kwargs)
@@ -3025,7 +3876,7 @@ Duplicable: False Optional: False
-relu¶
+relu¶
-
paddle.v2.fluid.layers.relu(**kwargs)
@@ -3046,7 +3897,7 @@ Duplicable: False Optional: False
-tanh¶
+tanh¶
-
paddle.v2.fluid.layers.tanh(**kwargs)
@@ -3067,7 +3918,7 @@ Duplicable: False Optional: False
-tanh_shrink¶
+tanh_shrink¶
-
paddle.v2.fluid.layers.tanh_shrink(**kwargs)
@@ -3088,7 +3939,7 @@ Duplicable: False Optional: False
-softshrink¶
+softshrink¶
-
paddle.v2.fluid.layers.softshrink(**kwargs)
@@ -3121,7 +3972,7 @@ Duplicable: False Optional: False
-sqrt¶
+sqrt¶
-
paddle.v2.fluid.layers.sqrt(**kwargs)
@@ -3142,7 +3993,7 @@ Duplicable: False Optional: False
-abs¶
+abs¶
-
paddle.v2.fluid.layers.abs(**kwargs)
@@ -3163,7 +4014,7 @@ Duplicable: False Optional: False
-ceil¶
+ceil¶
-
paddle.v2.fluid.layers.ceil(**kwargs)
@@ -3184,7 +4035,7 @@ Duplicable: False Optional: False
-floor¶
+floor¶
-
paddle.v2.fluid.layers.floor(**kwargs)
@@ -3205,7 +4056,7 @@ Duplicable: False Optional: False
-round¶
+round¶
-
paddle.v2.fluid.layers.round(**kwargs)
@@ -3226,7 +4077,7 @@ Duplicable: False Optional: False
-reciprocal¶
+reciprocal¶
-
paddle.v2.fluid.layers.reciprocal(**kwargs)
@@ -3247,7 +4098,7 @@ Duplicable: False Optional: False
-log¶
+log¶
-
paddle.v2.fluid.layers.log(**kwargs)
@@ -3269,7 +4120,7 @@ Duplicable: False Optional: False
-square¶
+square¶
-
paddle.v2.fluid.layers.square(**kwargs)
@@ -3290,7 +4141,7 @@ Duplicable: False Optional: False
-softplus¶
+softplus¶
-
paddle.v2.fluid.layers.softplus(**kwargs)
@@ -3311,12 +4162,12 @@ Duplicable: False Optional: False
-+softsign¶
+softsign¶
-
paddle.v2.fluid.layers.softsign(**kwargs) Softsign Activation Operator.
-$$out = frac{x}{1 + |x|}$$
+$$out = frac{x}{1 + |x|}$$
@@ -3332,7 +4183,7 @@ Duplicable: False Optional: False -brelu¶
+brelu¶
-
paddle.v2.fluid.layers.brelu(**kwargs)
@@ -3359,7 +4210,7 @@ Duplicable: False Optional: False
-leaky_relu¶
+leaky_relu¶
-
paddle.v2.fluid.layers.leaky_relu(**kwargs)
@@ -3385,7 +4236,7 @@ Duplicable: False Optional: False
-soft_relu¶
+soft_relu¶
-
paddle.v2.fluid.layers.soft_relu(**kwargs)
@@ -3411,7 +4262,7 @@ Duplicable: False Optional: False
-elu¶
+elu¶
-
paddle.v2.fluid.layers.elu(**kwargs)
@@ -3439,7 +4290,7 @@ Duplicable: False Optional: False
-
+stanh¶
+-
+
-
+
paddle.v2.fluid.layers.stanh(**kwargs)
+ STanh Activation Operator.
+$$out = b * frac{e^{a * x} - e^{-a * x}}{e^{a * x} + e^{-a * x}}$$
++
++ + + Parameters: -
+
- x – Input of STanh operator +Duplicable: False Optional: False +
- scale_a (FLOAT) – The scale parameter of a for the input +
- scale_b (FLOAT) – The scale parameter of b for the input +
Returns: Output of STanh operator
+
-hard_shrink¶
+hard_shrink¶
-
paddle.v2.fluid.layers.hard_shrink(**kwargs)
@@ -3524,7 +4402,7 @@ Duplicable: False Optional: False
-thresholded_relu¶
+thresholded_relu¶
-
paddle.v2.fluid.layers.thresholded_relu(**kwargs)
@@ -3556,7 +4434,7 @@ Duplicable: False Optional: False
-hard_sigmoid¶
+hard_sigmoid¶
-
paddle.v2.fluid.layers.hard_sigmoid(**kwargs)
@@ -3588,7 +4466,7 @@ Duplicable: False Optional: False
--swish¶
+swish¶
-
paddle.v2.fluid.layers.swish(**kwargs)
@@ -3613,169 +4491,143 @@ Duplicable: False Optional: False
-+im2sequence¶
++tensor¶
+++create_tensor¶
-
-
paddle.v2.fluid.layers.im2sequence(input, filter_size=1, stride=1, padding=0, name=None)
- Extracts image patches from the input tensor to form a tensor of shape -{input.batch_size * output_height * output_width, filter_size_H * -filter_size_W * input.channels} which is similar with im2col. -This op use filter / kernel to scan images and convert these images to -sequences. After expanding, the number of time step are -output_height * output_width for an image, in which output_height and -output_width are calculated by below equation:
--\[output\_size = 1 + (2 * padding + img\_size - block\_size + stride - 1) / stride\]-And the dimension of each time step is block_y * block_x * input.channels.
+paddle.v2.fluid.layers.create_tensor(dtype, name=None) +
+create_parameter¶
+-
+
-
+
paddle.v2.fluid.layers.create_parameter(shape, dtype, attr=None, is_bias=False, default_initializer=None)
+ Create a parameter +:param shape: shape of the parameter +:type shape: list[int] +:param dtype: element type of the parameter +:type dtype: string +:param attr: attributes of the parameter +:type attr: ParamAttr +:param is_bias: This can affect which default initializer is chosen
++
when default_initializer is None. If is_bias, +initializer.Constant(0.0) will be used. Otherwise, +Xavier() will be used.
-- Parameters: -
-
- input (Variable) – The input should be a tensor in NCHW format. -
- filter_size (int|tuple|None) – The filter size. If filter_size is a tuple, -it must contain two integers, (filter_size_H, filter_size_W). -Otherwise, the filter will be a square. -
- stride (int|tuple) – The stride size. If stride is a tuple, it must -contain two integers, (stride_H, stride_W). Otherwise, the -stride_H = stride_W = stride. Default: stride = 1. -
- padding (int|tuple) – The padding size. If padding is a tuple, it can -contain two integers like (padding_H, padding_W) which means -padding_up = padding_down = padding_H and -padding_left = padding_right = padding_W. Or it can use -(padding_up, padding_left, padding_down, padding_right) to indicate -paddings of four direction. Otherwise, a scalar padding means -padding_up = padding_down = padding_left = padding_right = padding -Default: padding = 0. -
- name (int) – The name of this layer. It is optional. -
Parameters: default_initializer (Initializer) – initializer for the parameter Returns: The output is a LoDTensor with shape -{input.batch_size * output_height * output_width, -filter_size_H * filter_size_W * input.channels}. -If we regard output as a matrix, each row of this matrix is -a step of a sequence.
-Returns: the created parameter Return type: output
-Return type: Parameter Examples:
-As an example:
--
-+Given: - -x = [[[[ 6. 2. 1.] - [ 8. 3. 5.] - [ 0. 2. 6.]] - - [[ 2. 4. 4.] - [ 6. 3. 0.] - [ 6. 4. 7.]]] - - [[[ 6. 7. 1.] - [ 5. 7. 9.] - [ 2. 4. 8.]] - - [[ 1. 2. 1.] - [ 1. 3. 5.] - [ 9. 0. 8.]]]] - -x.dims = {2, 2, 3, 3} - -And: - -filter = [2, 2] -stride = [1, 1] -padding = [0, 0] - -Then: - -output.data = [[ 6. 2. 8. 3. 2. 4. 6. 3.] - [ 2. 1. 3. 5. 4. 4. 3. 0.] - [ 8. 3. 0. 2. 6. 3. 6. 4.] - [ 3. 5. 2. 6. 3. 0. 4. 7.] - [ 6. 7. 5. 7. 1. 2. 1. 3.] - [ 7. 1. 7. 9. 2. 1. 3. 5.] - [ 5. 7. 2. 4. 1. 3. 9. 0.] - [ 7. 9. 4. 8. 3. 5. 0. 8.]] + -output.dims = {8, 9} ++create_global_var¶
+-
+
-
+
paddle.v2.fluid.layers.create_global_var(shape, value, dtype, persistable=False, name=None)
+
The simple usage is:
-++output = fluid.layers.im2sequence( - input=layer, stride=[1, 1], filter_size=[2, 2]) -
+-cast¶
+-
+
-
+
paddle.v2.fluid.layers.cast(x, dtype)
+ This function takes in the input with input_dtype +and casts it to the output_dtype as the output.
+
+-concat¶
+-
+
-
+
paddle.v2.fluid.layers.concat(input, axis=0)
+ Concat
+This function concatenates the input along the axis mentioned +and returns that as the output.
++
++ + + Parameters: -
+
- input (list) – List of tensors to be concatenated +
- axis (int) – Integer axis along which the tensors will be concatenated +
Returns: Output variable of the concatenation
+Return type: Variable
+Examples
--edit_distance¶
--ctc_greedy_decoder¶
++sums¶
-
-
paddle.v2.fluid.layers.ctc_greedy_decoder(input, blank, name=None)
- This op is used to decode sequences by greedy policy by below steps: -1. Get the indexes of max value for each row in input. a.k.a.
--
-numpy.argmax(input, axis=0).-
-
- For each sequence in result of step1, merge repeated tokens between two -blanks and delete all blanks. -
A simple example as below:
-+Given: - -input.data = [[0.6, 0.1, 0.3, 0.1], - [0.3, 0.2, 0.4, 0.1], - [0.1, 0.5, 0.1, 0.3], - [0.5, 0.1, 0.3, 0.1], - - [0.5, 0.1, 0.3, 0.1], - [0.2, 0.2, 0.2, 0.4], - [0.2, 0.2, 0.1, 0.5], - [0.5, 0.1, 0.3, 0.1]] - -input.lod = [[0, 4, 8]] - -Then: - -output.data = [[2], - [1], - [3]] +
paddle.v2.fluid.layers.sums(input, out=None) +This function performs the sum operation on the input and returns the +result as the output.
++
++ + + Parameters: input (Variable|list) – The input tensor that has the elements +that need to be summed up. +Returns: -
+
- The tensor type variable that has the sum of input +
- written to it. +
Return type: Variable +Examples
++-assign¶
+-
+
-
+
paddle.v2.fluid.layers.assign(input, output)
+ Assign
+This function copies the input Variable to the output Variable.
Parameters: -
-
- input (Variable) – (LoDTensor<float>), the probabilities of -variable-length sequences, which is a 2-D Tensor with -LoD information. It’s shape is [Lp, num_classes + 1], -where Lp is the sum of all input sequences’ length and -num_classes is the true number of classes. (not -including the blank label). -
- blank (int) – the blank label index of Connectionist Temporal -Classification (CTC) loss, which is in thehalf-opened -interval [0, num_classes + 1). +
- input (Variable|numpy.ndarray) – The source variable +
- output (Variable) – The destination variable
Returns: CTC greedy decode result.
+Returns: The destination variable that was supplied as the output.
Return type: Variable
@@ -3784,41 +4636,34 @@ interval [0, num_classes + 1).Examples
--x = fluid.layers.data(name='x', shape=[8], dtype='float32') - -cost = fluid.layers.ctc_greedy_decoder(input=x, blank=0) -
-l2_normalize¶
++fill_constant_batch_size_like¶
-
-
paddle.v2.fluid.layers.l2_normalize(x, axis, epsilon=1e-12, name=None)
- L2 normalize Layer
-The l2 normalize layer normalizes x along dimension axis using an L2 -norm. For a 1-D tensor (dim is fixed to 0), this layer computes
-output = x / sqrt(max(sum(x**2), epsilon))
-For x with more dimensions, this layer independently normalizes each 1-D -slice along dimension axis.
+paddle.v2.fluid.layers.fill_constant_batch_size_like(input, shape, dtype, value, input_dim_idx=0, output_dim_idx=0) +fill_constant_batch_size_like
+This function creates a tensor of specified shape, dtype and batch size, +and initializes this with a constant supplied in value. The batch size is +obtained from the input tensor.
+It also sets stop_gradient to True.
Parameters: -
-
- x (Variable|list) – The input tensor to l2_normalize layer. -
- axis (int) – Dimension along which to normalize the input. -
- epsilon (float) – A lower bound value for x‘s l2 norm. sqrt(epsilon) will -be used as the divisor if the l2 norm of x is less than -sqrt(epsilon). -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- input (Variable) – Tensor whose dimensions will be used to get batch size +
- shape (tuple|list|None) – Shape of output tensor +
- dtype (np.dtype|core.DataType|str) – Data type of output tensor +
- value (float) – Constant value to initialize the output tensor +
- input_dim_idx (int) – Index of input’s batch size dimension +
- output_dim_idx (int) – Index of output’s batch size dimension
Returns: The output tensor variable.
+Returns: The tensor variable storing the output
Return type: Variable
@@ -3827,55 +4672,35 @@ will be named automatically.Examples
--data = fluid.layers.data(name="data", - shape=(3, 17, 13), - dtype="float32") -normed = fluid.layers.l2_normalize(x=data, axis=1) +
data = fluid.layers.fill_constant_batch_size_like( + input=like, shape=[1], value=0, dtype='int64')
-sequence_reshape¶
++fill_constant¶
-
-
paddle.v2.fluid.layers.sequence_reshape(input, new_dim)
- Sequence Reshape Layer
-This layer will rearrange the input sequences. The new dimension is set by -user. Length of each sequence is computed according to original length, -original dimension and new dimension. The following example will help to -illustrate the function of this layer:
---x is a LoDTensor: - x.lod = [[0, 2, 6]] - x.data = [[1, 2], [3, 4], - [5, 6], [7, 8], [9, 10], [11, 12]] - x.dims = [6, 2] - -set new_dim = 4 - -then out is a LoDTensor: - out.lod = [[0, 1, 3]] - out.data = [[1, 2, 3, 4], - [5, 6, 7, 8], [9, 10, 11, 12]] - out.dims = [3, 4] -
Currently, only 1-level LoDTensor is supported and please make sure -(original length * original dimension) can be divided by new dimension with -no remainder for each sequence.
+paddle.v2.fluid.layers.fill_constant(shape, dtype, value, force_cpu=False, out=None) +fill_constant
+This function creates a tensor with specified shape and dtype, and +initializes it with a constant specifed by value.
+The attribute stop_gradient of the created tensor is set to True.
Parameters: -
-
- input (Variable) – (LodTensor, default: LoDTensor<float>), a 2-D LoDTensor -with shape being [N, M] where M for dimension. -
- new_dim (int) – New dimension which the input LoDTensor is reshaped to. +
- shape (tuple|list|None) – Shape of the output tensor. +
- dtype (np.dtype|core.DataType|str) – Data type of the output tensor. +
- value (float) – The constant value used to initialize the output tensor. +
- out (Variable) – The output tensor.
Returns: Reshaped LoDTensor according to new dimension.
+Returns: The tensor variable storing the output.
Return type: Variable
@@ -3884,49 +4709,32 @@ with shape being [N, M] where M for dimension.Examples
--x = fluid.layers.data(name='x', shape=[5, 20], - dtype='float32', lod_level=1) -x_reshaped = layers.sequence_reshape(input=x, new_dim=10) +
data = fluid.layers.fill_constant(shape=[1], value=0, dtype='int64')
-row_conv¶
++ones¶
-
-
paddle.v2.fluid.layers.row_conv(input, future_context_size, param_attr=None, act=None)
- Row Conv Operator. This layer will apply lookahead convolution to -input. The input variable should be a 2D LoDTensor with shape [T, D]. -Parameters with shape [future_context_size + 1, D] will be created. The math -equation of row convolution is as follows:
--\[Out_{i} = \sum_{j = i} ^ {i + \tau} X_{j} \odot W_{i - j}\]-In the above equation:
--
-
- \(Out_{i}\): The i-th row of output variable with shape [1, D]. -
- \(\tau\): Future context size. -
- \(X_{j}\): The j-th row of input variable with shape [1, D]. -
- \(W_{i-j}\): The (i-j)-th row of parameters with shape [1, D]. -
More details about row_conv please refer to the paper (http://www.cs.cmu.edu/~dyogatam/papers/wang+etal.iclrworkshop2016.pdf) and -the design document (https://github.com/PaddlePaddle/Paddle/issues/2228#issuecomment-303903645).
+paddle.v2.fluid.layers.ones(shape, dtype) +ones
+This function creates a tensor of specified shape and +dtype, and initializes this with 1.
+It also sets stop_gradient to True.
Parameters: -
-
- input (Variable) – Input variable, a 2D LoDTensor with shape [T, D]. -
- future_context_size (int) – Future context size. Please note, the shape -of convolution kernel is [future_context_size + 1, D]. -
- param_attr (ParamAttr) – Attributes of parameters, including -name, initializer etc. -
- act (str) – Non-linear activation to be applied to output variable. +
- shape (tuple|list|None) – Shape of output tensor +
- dtype (np.dtype|core.DataType|str) – Data type of output tensor
Returns: The output tensor with same shape as input tensor.
+Returns: The tensor variable storing the output
Return type: Variable
@@ -3935,48 +4743,32 @@ name, initializer etc.Examples
--x = fluid.layers.data(name='x', shape=[16], - dtype='float32', lod_level=1) -out = fluid.layers.row_conv(input=x, future_context_size=2) +
data = fluid.layers.ones(shape=[1], dtype='int64')
-multiplex¶
++@@ -4003,7 +4793,7 @@ with shape [M, 1] where M is the batch size. @@ -217,7 +217,7 @@zeros¶
-
-
paddle.v2.fluid.layers.multiplex(inputs, index)
- Multiplex Layer
-Referring to the given index variable, this layer selects rows from the -input variables to construct a multiplex variable. Assuming that there are -\(m\) input variables and \(I_i\) represents the i-th input -variable and \(i\) is in [0, \(m\)). All input variables are -tensors with same shape [\(d_0\), \(d_1\), ..., \(d_R\)]. -Please note that rank of the input tensor should be at least 2. Each input -variable will be treated as a 2-D matrix with shape [\(M\), \(N\)] -where \(M\) for \(d_0\) and \(N\) for \(d_1\) * \(d_2\) -* ... * \(d_R\). Let \(I_i[j]\) be the j-th row of the i-th input -variable. The given index variable should be a 2-D tensor with shape -[\(M\), 1]. Let ID[i] be the i-th index value of the index variable. -Then the output variable will be a tensor with shape [\(d_0\), -\(d_1\), ..., \(d_R\)]. If we treat the output tensor as a 2-D -matrix with shape [\(M\), \(N\)] and let \(O[i]\) be the i-th -row of the matrix, then O[i] is equal to \(I_{ID[i]}[i]\).
+paddle.v2.fluid.layers.zeros(shape, dtype) +zeros
+This function creates a tensor of specified shape and +dtype, and initializes this with 0.
+It also sets stop_gradient to True.
Parameters: -
-
- inputs (list) – A list of variables to gather from. All variables have the -same shape and the rank is at least 2. -
- index (Variable) – Tensor<int32>, index variable which is a 2-D tensor -with shape [M, 1] where M is the batch size. +
- shape (tuple|list|None) – Shape of output tensor +
- dtype (np.dtype|core.DataType|str) – Data type of output tensor
Returns: Multiplex variable gathered from input variables.
+Returns: The tensor variable storing the output
Return type: Variable
@@ -3985,14 +4777,12 @@ with shape [M, 1] where M is the batch size.Examples
-x1 = fluid.layers.data(name='x1', shape=[4], dtype='float32') -x2 = fluid.layers.data(name='x2', shape=[4], dtype='float32') -index = fluid.layers.data(name='index', shape=[1], dtype='int32') -out = fluid.layers.multiplex(inputs=[x1, x2], index=index) +
+data = fluid.layers.zeros(shape=[1], dtype='int64')
-Nets¶
+nets¶
-simple_img_conv_pool¶
-
@@ -225,15 +225,6 @@
paddle.v2.fluid.nets.simple_img_conv_pool(input, num_filters, filter_size, pool_size, pool_stride, act, param_attr=None, pool_type='max', use_cudnn=True)-img_conv_group¶
--
-
-
-
paddle.v2.fluid.nets.img_conv_group(input, conv_num_filter, pool_size, conv_padding=1, conv_filter_size=3, conv_act=None, param_attr=None, conv_with_batchnorm=False, conv_batchnorm_drop_rate=0.0, pool_stride=1, pool_type=None, use_cudnn=True)
- Image Convolution Group, Used for vgg net.
-
@@ -217,113 +217,58 @@sequence_conv_pool¶
@@ -361,10 +352,10 @@ parameters. diff --git a/develop/doc/api/v2/fluid/optimizer.html b/develop/doc/api/v2/fluid/optimizer.html index bd08781ae05949ff792804cb657eaa262d981f9b..e75efcc1cbead5875c5abfd96f0b85bd7566189e 100644 --- a/develop/doc/api/v2/fluid/optimizer.html +++ b/develop/doc/api/v2/fluid/optimizer.html @@ -8,7 +8,7 @@ -Optimizer — PaddlePaddle documentation +optimizer — PaddlePaddle documentation @@ -34,8 +34,8 @@ - - + + @@ -162,17 +162,17 @@- Training and Inference
- Fluid
@@ -207,7 +207,7 @@- Fluid >
-- Optimizer
+- optimizer
-Optimizer¶
--Optimizer¶
--
-
-
-class
paddle.v2.fluid.optimizer.Optimizer(learning_rate, global_step=None, regularization=None)
- Optimizer Base class.
-Define the common interface of an optimizer. -User should not use this class directly, -but need to use one of it’s implementation.
+optimizer¶
++-SGD¶
-
-
global_learning_rate
- get global decayed learning rate -:return:
-
-
-
-
-
create_optimization_pass(parameters_and_grads, loss, startup_program=None)
- Add optimization operators to update gradients to variables.
--
-- - - Parameters: -
-
- loss – the target that this optimization is for. -
- parameters_and_grads – a list of (variable, gradient) pair to update. -
Returns: a list of operators that will complete one step of -optimization. This will include parameter update ops, global step -update ops and any other custom ops required by subclasses to manage -their internal state. -:param startup_program:
-Return type: return_op_list
-
-
-
-
-
minimize(loss, startup_program=None, parameter_list=None, no_grad_set=None)
- Add operations to minimize loss by updating parameter_list.
-This method combines interface append_backward() and -create_optimization_pass() into one.
-
paddle.v2.fluid.optimizer.SGD+alias of
SGDOptimizer--SGDOptimizer¶
--
-
-
-class
paddle.v2.fluid.optimizer.SGDOptimizer(learning_rate, **kwargs)
- Simple SGD optimizer without any state.
-
-MomentumOptimizer¶
--
+
-
-class
paddle.v2.fluid.optimizer.MomentumOptimizer(learning_rate, momentum, use_nesterov=False, **kwargs)
- Simple Momentum optimizer with velocity state
+paddle.v2.fluid.optimizer.Momentum+alias of
MomentumOptimizer-
-class
paddle.v2.fluid.optimizer.AdagradOptimizer(learning_rate, epsilon=1e-06, **kwargs)
- Simple Adagrad optimizer with moment state
+paddle.v2.fluid.optimizer.Adagrad+alias of
AdagradOptimizer-
-class
paddle.v2.fluid.optimizer.AdamOptimizer(learning_rate=0.001, beta1=0.9, beta2=0.999, epsilon=1e-08, **kwargs)
- Implements the Adam Optimizer
+paddle.v2.fluid.optimizer.Adam+alias of
AdamOptimizer-
-class
paddle.v2.fluid.optimizer.AdamaxOptimizer(learning_rate=0.001, beta1=0.9, beta2=0.999, epsilon=1e-08, **kwargs)
- Implements the Adamax Optimizer
+paddle.v2.fluid.optimizer.Adamax+alias of
AdamaxOptimizer-
-class
paddle.v2.fluid.optimizer.DecayedAdagradOptimizer(learning_rate, decay=0.95, epsilon=1e-06, **kwargs)
- Simple Decayed Adagrad optimizer with moment state
+paddle.v2.fluid.optimizer.DecayedAdagrad+alias of
DecayedAdagradOptimizer- Training and Inference
- Fluid @@ -207,7 +207,7 @@
- Fluid > -
- ParamAttr +
- param_attr
-
+class
paddle.v2.fluid.param_attr.ParamAttr(name=None, initializer=None, learning_rate=1.0, regularizer=None, trainable=True, gradient_clip=None)
+ -
+class
paddle.v2.fluid.param_attr.WeightNormParamAttr(dim=None, **kwargs)
+ Used for weight normalization. Any field in ParamAttr can also be set here. +Besides, an extra field dim can be set to indicate the dimension except +which to normalize.
+- Training and Inference
- Fluid @@ -207,7 +207,7 @@
- Fluid > -
- Profiler +
- profiler
-
paddle.v2.fluid.profiler.cuda_profiler(*args, **kwds)
@@ -249,6 +249,53 @@ to “Compute Command Line Profiler User Guide”.
-
+
paddle.v2.fluid.profiler.reset_profiler()
+ The profiler clear interface. +reset_profiler will clear the previous time record.
+-
+
paddle.v2.fluid.profiler.profiler(*args, **kwds)
+ The profiler interface. +Different from cuda_profiler, this profiler can be used to profile both CPU +and GPU program. By defalut, it records the CPU and GPU operator kernels, +if you want to profile other program, you can refer the profiling tutorial +to add more records.
++
++ + + Parameters: -
+
- state (string) – The profiling state, which should be ‘CPU’ or ‘GPU’, +telling the profiler to use CPU timer or GPU timer for profiling. +Although users may have already specified the execution place +(CPUPlace/CUDAPlace) in the begining, for flexibility the profiler +would not inherit this place. +
- sorted_key (string) – If None, the profiling results will be printed +in the order of first end time of events. Otherwise, the profiling +results will be sorted by the this flag. This flag should be one +of ‘calls’, ‘total’, ‘max’, ‘min’ or ‘ave’. +The calls means sorting by the number of calls. +The total means sorting by the total execution time. +The max means sorting by the maximum execution time. +The min means sorting by the minimum execution time. +The ave means sorting by the average execution time. +
- Training and Inference
- Fluid @@ -207,7 +207,7 @@
- Fluid > -
- Regularizer +
- regularizer
-
-class
paddle.v2.fluid.regularizer.WeightDecayRegularizer
- Base class for weight decay regularizers
-Defines the common interface of weight-decay regularizers. -Weight-decay regularizers are added only during the backward -pass for faster regularization. They add operations to the network -that correspond to gradient of the regularization function. -Users should not use this class directly, but need to use one -of its implementations
+paddle.v2.fluid.regularizer.append_regularization_ops(parameters_and_grads, regularization=None) +Create and add backward regularization Operators
+Creates and adds backward regularization operators in the BlockDesc. +This will add gradients of the regularizer function to the gradients +of the parameters and return these modified gradients. This is the +same as implementing weight decay in optimizers for regularization.
++
+ + + Parameters: -
+
- parameters_and_grads – A list of (parameters, gradients) pairs +that need to be regularized. +
- regularization – A global regularizer. If the parameter is not +set. It will be applied with regularizer. +
Returns: list of (parameters, gradients) pair with the regularized gradient
+Raises:
+Exception– Unknown regularization type-
-class
paddle.v2.fluid.regularizer.L2DecayRegularizer(regularization_coeff=0.0)
- Implements the L2 Weight Decay Regularization
+paddle.v2.fluid.regularizer.L1Decay+alias of
L1DecayRegularizer-
-class
paddle.v2.fluid.regularizer.L1DecayRegularizer(regularization_coeff=0.0)¶
- Implements the L1 Weight Decay Regularization
++@@ -260,10 +278,10 @@ of its implementations diff --git a/develop/doc/api/v2/model_configs.html b/develop/doc/api/v2/model_configs.html index 8042041351cfc77bc5e545c2b9efb4f6d25ed3e0..29362778f3f0f4b5cf278989a26ae1497948da50 100644 --- a/develop/doc/api/v2/model_configs.html +++ b/develop/doc/api/v2/model_configs.html @@ -162,17 +162,17 @@L2Decay¶
+-
+
-
+
paddle.v2.fluid.regularizer.L2Decay
+ alias of
L2DecayRegularizer
- Training and Inference
- Fluid
diff --git a/develop/doc/api/v2/run_logic.html b/develop/doc/api/v2/run_logic.html index f665bbb1b13c11c7734de71c45a132ce79869ac4..b1d3598a5ace6339f1f70f9dedb9a9562759cc1e 100644 --- a/develop/doc/api/v2/run_logic.html +++ b/develop/doc/api/v2/run_logic.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/api.html b/develop/doc/design/api.html index 1903a81076d2c3b91c4eb6b40a3c719a31602e23..e1d938acfc4d148b9d334d14310af9d67d715de1 100644 --- a/develop/doc/design/api.html +++ b/develop/doc/design/api.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/auto_gradient_check.html b/develop/doc/design/auto_gradient_check.html index aa76c77cdf7bd169c6c77605679adaa4102dd4c7..aa07af71658b274a9d8e2b0a403454d234d810ed 100644 --- a/develop/doc/design/auto_gradient_check.html +++ b/develop/doc/design/auto_gradient_check.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/backward.html b/develop/doc/design/backward.html index 1c9615f3df51cd5c325ad6dccffa4e485a100863..9b6f06dc4272cb4c8b473806a2de11cd7268d3fc 100644 --- a/develop/doc/design/backward.html +++ b/develop/doc/design/backward.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/block.html b/develop/doc/design/block.html index 0070b5ba1fc93f35ef4c3123d7ddd8742c480e93..3ef21489916d360d32762566dc131396b3b19607 100644 --- a/develop/doc/design/block.html +++ b/develop/doc/design/block.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/build_system/README.html b/develop/doc/design/build_system/README.html index fadbf9cc9ec07479d3d87f634722868e501999c6..c94834d70eda38483ffc241a9a3e0a9508123448 100644 --- a/develop/doc/design/build_system/README.html +++ b/develop/doc/design/build_system/README.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/cluster_train/README.html b/develop/doc/design/cluster_train/README.html index 42c30b6e085cb04e18c6fa4d2363bc1de765def3..2f0b21457c9a9d229d53a2a339ced0d66a75efd3 100644 --- a/develop/doc/design/cluster_train/README.html +++ b/develop/doc/design/cluster_train/README.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/cluster_train/checkpointing.html b/develop/doc/design/cluster_train/checkpointing.html index a2a5de6e58075ced932b130f1969c9b4e085fe5b..d1efe9f2bedb138f7ae0dccf1af8a3898b7ad6e6 100644 --- a/develop/doc/design/cluster_train/checkpointing.html +++ b/develop/doc/design/cluster_train/checkpointing.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/cluster_train/data_dispatch.html b/develop/doc/design/cluster_train/data_dispatch.html index fa2104ab47ac3a5f9f35a1e82d9f5ebf1feb7372..a1bce4063f48cd217feb9ecec42b03dcee5b1a62 100644 --- a/develop/doc/design/cluster_train/data_dispatch.html +++ b/develop/doc/design/cluster_train/data_dispatch.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/cluster_train/large_model_dist_train.html b/develop/doc/design/cluster_train/large_model_dist_train.html index 31e0b43a169c6235820f99ac46c1d4c256fb21f1..488d8d92615bbac274599fcd43ded66280ddac0b 100644 --- a/develop/doc/design/cluster_train/large_model_dist_train.html +++ b/develop/doc/design/cluster_train/large_model_dist_train.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/cluster_train/master_server.html b/develop/doc/design/cluster_train/master_server.html index 19384b28aae534787bac031cce3625c05189032b..6e1e19b8a9c1c89d411feed14b70253e0c2f0576 100644 --- a/develop/doc/design/cluster_train/master_server.html +++ b/develop/doc/design/cluster_train/master_server.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/cluster_train/pserver_client.html b/develop/doc/design/cluster_train/pserver_client.html index e8925ce93d0c32445ee9d7fe99f9efc07ef0d1ae..d084c79c11eee37d912f588d15cf8f76b5f1ee6f 100644 --- a/develop/doc/design/cluster_train/pserver_client.html +++ b/develop/doc/design/cluster_train/pserver_client.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/cluster_train/remote_parameter_updater.html b/develop/doc/design/cluster_train/remote_parameter_updater.html index 65c936c03d389af23a97f8360136511d27f6fe5e..552a0cd15de30f7c6596054d225a4f0eb20fdd9b 100644 --- a/develop/doc/design/cluster_train/remote_parameter_updater.html +++ b/develop/doc/design/cluster_train/remote_parameter_updater.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/cluster_train/save_model.html b/develop/doc/design/cluster_train/save_model.html index 124b51953a848f0f1be7d075e0cb93c7ea9b3772..dbccb108f37901272305f71aea11f44d84eac159 100644 --- a/develop/doc/design/cluster_train/save_model.html +++ b/develop/doc/design/cluster_train/save_model.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/cluster_train/submit-job.html b/develop/doc/design/cluster_train/submit-job.html index 1bce5e75acce5cbcdbac9de8b65d2a91d457cf23..b3c2147c7d8b4a0771c144c3e840fee59411703a 100644 --- a/develop/doc/design/cluster_train/submit-job.html +++ b/develop/doc/design/cluster_train/submit-job.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/concurrent_programming.html b/develop/doc/design/concurrent_programming.html index e77e6cb89c5ee9bc7721be5a98cd8390423ec0df..3da59a42bb7c620971a054b4a88e4b56780b276e 100644 --- a/develop/doc/design/concurrent_programming.html +++ b/develop/doc/design/concurrent_programming.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/csp.html b/develop/doc/design/csp.html index 7c8095fd15766754bb62ed4422715806227949be..d9de1420057935e17bc081833bdf5d27911a74bd 100644 --- a/develop/doc/design/csp.html +++ b/develop/doc/design/csp.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/dist_refactor/distributed_architecture.html b/develop/doc/design/dist_refactor/distributed_architecture.html index 00403656b031a901e4e486cc57865f39df8bd081..e51c7abd9ff61e4015bc476e3348088f7589797f 100644 --- a/develop/doc/design/dist_refactor/distributed_architecture.html +++ b/develop/doc/design/dist_refactor/distributed_architecture.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/dist_refactor/multi_cpu.html b/develop/doc/design/dist_refactor/multi_cpu.html index fe645c62b7d5ea3cbe2cd12acdb91859e5149333..56d38fabaf4dbdb82f084d9a1f0a004062243720 100644 --- a/develop/doc/design/dist_refactor/multi_cpu.html +++ b/develop/doc/design/dist_refactor/multi_cpu.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/dist_refactor/parameter_server.html b/develop/doc/design/dist_refactor/parameter_server.html index 11ba8e09ef61bd2eecaf27e34902fb7cfc139b85..f6b922d1678b09306174ca6bdc15c95ccd8e7a07 100644 --- a/develop/doc/design/dist_refactor/parameter_server.html +++ b/develop/doc/design/dist_refactor/parameter_server.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/error_clip.html b/develop/doc/design/error_clip.html index e61047938662dd980993ac8670dd4c31caabdf94..4b4a87a95ee0d81254af86d3a9581b0b87db218c 100644 --- a/develop/doc/design/error_clip.html +++ b/develop/doc/design/error_clip.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/evaluator.html b/develop/doc/design/evaluator.html index fb66dc3ca7cae9f0f1b4951c1a84a8b09a6b4f8e..969be56148894fd20cf7bc38b949dfbe8752ab70 100644 --- a/develop/doc/design/evaluator.html +++ b/develop/doc/design/evaluator.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/executor.html b/develop/doc/design/executor.html index 7cd5028f8c405fcde6066a5c4e377e1c2b20157d..44665f60f76a293da79d7ad542031a9c80da3345 100644 --- a/develop/doc/design/executor.html +++ b/develop/doc/design/executor.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/file_manager/README.html b/develop/doc/design/file_manager/README.html index 4f3387c367be851529f554f5754086cf72aacea8..e1ab469c7d089ef0acb3e6028e021f83afbed50d 100644 --- a/develop/doc/design/file_manager/README.html +++ b/develop/doc/design/file_manager/README.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/file_manager/pfs/pfsclient.html b/develop/doc/design/file_manager/pfs/pfsclient.html index 322bfb080c3b50e2b150f25e886d431ceb5789e0..6b78bbc3a8c26884fcef289e351a92530d055e21 100644 --- a/develop/doc/design/file_manager/pfs/pfsclient.html +++ b/develop/doc/design/file_manager/pfs/pfsclient.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/float16.html b/develop/doc/design/float16.html index 9c3e0184ecd2cf511c9e92846ec5ce50345ecb8a..f487c93f74f9ad7929d74a762ab3d46628ad396c 100644 --- a/develop/doc/design/float16.html +++ b/develop/doc/design/float16.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/fluid.html b/develop/doc/design/fluid.html index 67f5df790bb75cc5315ec6ae06d110496249cce4..3f116ee9d2a2881fb2fdd23cd39b49ace2185dfa 100644 --- a/develop/doc/design/fluid.html +++ b/develop/doc/design/fluid.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/fluid_compiler.html b/develop/doc/design/fluid_compiler.html index b6f0139d18cf64171655c78d86307f7868acece9..0ab78210a8dcb58c880be0e255c7b316213f392e 100644 --- a/develop/doc/design/fluid_compiler.html +++ b/develop/doc/design/fluid_compiler.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/functions_operators_layers.html b/develop/doc/design/functions_operators_layers.html index 13dc2a2f515ce59522664e5e5e08f468bdf4f622..008db0fa351adaa0223175b4db59050b128eda3e 100644 --- a/develop/doc/design/functions_operators_layers.html +++ b/develop/doc/design/functions_operators_layers.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/gan_api.html b/develop/doc/design/gan_api.html index 89f158bd6c600f58acccd09ec0da922bb71e956b..70db6a0d96cc8a2bf3046dfd8fca36b9d2e4dbf8 100644 --- a/develop/doc/design/gan_api.html +++ b/develop/doc/design/gan_api.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/graph.html b/develop/doc/design/graph.html index 2a203313050604e163c8c8ade94f7affdc07a8f2..02219f039bc8f908eac9f1b5f58609dfb33eab13 100644 --- a/develop/doc/design/graph.html +++ b/develop/doc/design/graph.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/graph_survey.html b/develop/doc/design/graph_survey.html index 14583b98236bd296fa95943d49afba8fdd598656..a250ded9cfd85adde967e33734ef5e952a7f3fe6 100644 --- a/develop/doc/design/graph_survey.html +++ b/develop/doc/design/graph_survey.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/if_else_op.html b/develop/doc/design/if_else_op.html index 50d60c82b1feddb68adffa2e5f4c69e40c2a260d..95291ac621b8f05af06eda20eff1697b6506e752 100644 --- a/develop/doc/design/if_else_op.html +++ b/develop/doc/design/if_else_op.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/infer_var_type.html b/develop/doc/design/infer_var_type.html index c792683de62f2bb0dd09e99f0e985b30ec675c17..12bccda538e2d5b00c4a1282e2a51b0c8b1b774e 100644 --- a/develop/doc/design/infer_var_type.html +++ b/develop/doc/design/infer_var_type.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/kernel_hint_design.html b/develop/doc/design/kernel_hint_design.html index fa7a059e1b56f1ab14bcab25d227a5ca14d22517..08301df6746604bdd8cd62d47e89c5c599410820 100644 --- a/develop/doc/design/kernel_hint_design.html +++ b/develop/doc/design/kernel_hint_design.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/memory_optimization.html b/develop/doc/design/memory_optimization.html index fbc08771f0e676b99fefac0af14d285a9717aa3f..089e7c57a333317d1514a9188c8ff0d6e93723ca 100644 --- a/develop/doc/design/memory_optimization.html +++ b/develop/doc/design/memory_optimization.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/mkl/mkl_packed.html b/develop/doc/design/mkl/mkl_packed.html index 16f449849ae4c860a188abe8c92c1b27976f7684..e14967111712a5c9e01f7bcd663ed19d33676d21 100644 --- a/develop/doc/design/mkl/mkl_packed.html +++ b/develop/doc/design/mkl/mkl_packed.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/mkl/mkldnn.html b/develop/doc/design/mkl/mkldnn.html index fc716cc79a8b0f2872518a5497bf91909fc02d02..ef6c603c3f219821d1c67a36718756f48d0fa770 100644 --- a/develop/doc/design/mkl/mkldnn.html +++ b/develop/doc/design/mkl/mkldnn.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/mkl/mkldnn_fluid.html b/develop/doc/design/mkl/mkldnn_fluid.html index b5cb8b62275f77057f7211e84e33e52a56f329c8..a89b3e0202bf90362be47b171b67859e63ddb8cc 100644 --- a/develop/doc/design/mkl/mkldnn_fluid.html +++ b/develop/doc/design/mkl/mkldnn_fluid.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/model_format.html b/develop/doc/design/model_format.html index 5a61ed9f2e63a1cfa0963562800d2950c38b50ad..7d0d84bfefac222356e4d4c3275f1db5bc15e20d 100644 --- a/develop/doc/design/model_format.html +++ b/develop/doc/design/model_format.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/multi_language_interface/00.why_plain_c.html b/develop/doc/design/multi_language_interface/00.why_plain_c.html index 8f7341ada277f2b58242496b8e1c7ab0d3058ef6..f5a7a3cb65d1cba31b1122df4f57070c7c916ff0 100644 --- a/develop/doc/design/multi_language_interface/00.why_plain_c.html +++ b/develop/doc/design/multi_language_interface/00.why_plain_c.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/multi_language_interface/01.inference_implementation.html b/develop/doc/design/multi_language_interface/01.inference_implementation.html index 46714697c21e019feb301fce934e992d578a84ee..ff33b526eaa3f5653cdc1bd58c1472858ed368af 100644 --- a/develop/doc/design/multi_language_interface/01.inference_implementation.html +++ b/develop/doc/design/multi_language_interface/01.inference_implementation.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/operator_kernel_type.html b/develop/doc/design/operator_kernel_type.html index fce8eb0420eb34ecb16a1b1103270873cb9bbfcf..fcd4c1c0115ee2c2bc33c48376ebe468caf3b5a7 100644 --- a/develop/doc/design/operator_kernel_type.html +++ b/develop/doc/design/operator_kernel_type.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/ops/rnn.html b/develop/doc/design/ops/rnn.html index 809cb86708dd12eaa8e429daffc318277dbba254..f90a72a099479207b0c2abcfda0eed2e55684f04 100644 --- a/develop/doc/design/ops/rnn.html +++ b/develop/doc/design/ops/rnn.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/ops/sequence_decoder.html b/develop/doc/design/ops/sequence_decoder.html index 1cf28a854ea13f2e429203eda204352a9c87a42f..8720cd8af59e2b81d6bb3b2909522ad013ad72dc 100644 --- a/develop/doc/design/ops/sequence_decoder.html +++ b/develop/doc/design/ops/sequence_decoder.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/optimizer.html b/develop/doc/design/optimizer.html index b3c9d4b83de40cd73e68b4557f692dba2b9b02e0..494fab91383a0772940a06527d431e04ffef18b1 100644 --- a/develop/doc/design/optimizer.html +++ b/develop/doc/design/optimizer.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/paddle_nccl.html b/develop/doc/design/paddle_nccl.html index c921b6924499e87a526cd93c827d764e793e4a44..01c7658852f99b5fdd11bcfc0ef36465f3110df0 100644 --- a/develop/doc/design/paddle_nccl.html +++ b/develop/doc/design/paddle_nccl.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/parameter_average.html b/develop/doc/design/parameter_average.html index 85b9e457e478a41e020c7285c74b2b9cde41e88d..89d2afc7a7f21858f4931eb0b63f64285b6433c6 100644 --- a/develop/doc/design/parameter_average.html +++ b/develop/doc/design/parameter_average.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/parameters_in_cpp.html b/develop/doc/design/parameters_in_cpp.html index 3fe816902de042a8d506feb8d4314fe255f1ec18..840f4f869e2597cb49c13db5307f7a763b0f6c4e 100644 --- a/develop/doc/design/parameters_in_cpp.html +++ b/develop/doc/design/parameters_in_cpp.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/profiler.html b/develop/doc/design/profiler.html index b6c3a68d4b810374ca7a4a344402930bb0501512..acb0ce9134e97413c7559eeef7c9442ec8b2992f 100644 --- a/develop/doc/design/profiler.html +++ b/develop/doc/design/profiler.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/program.html b/develop/doc/design/program.html index 488f6270ad3cee079b3e777d7471577d2312f638..71156defab2825244013ec36046281ff60dabcd4 100644 --- a/develop/doc/design/program.html +++ b/develop/doc/design/program.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/prune.html b/develop/doc/design/prune.html index a585945e2a4eca3bee93dbb255c16b3b4330840f..7193b7ca0d6f076029647d08de78f0a1c63c8377 100644 --- a/develop/doc/design/prune.html +++ b/develop/doc/design/prune.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/python_api.html b/develop/doc/design/python_api.html index 13317ceb34af6946cbe692794b47f69944460d68..558f7afe3dfe01cfeffd683188eae3d4cbdce176 100644 --- a/develop/doc/design/python_api.html +++ b/develop/doc/design/python_api.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/reader/README.html b/develop/doc/design/reader/README.html index c2c90422850fca9ce44b06ffa15af093dacb2ae0..cb7145dd2ef6071daa709e125c5ed2bd4676b73e 100644 --- a/develop/doc/design/reader/README.html +++ b/develop/doc/design/reader/README.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/refactorization.html b/develop/doc/design/refactorization.html index ba67452b562807f5d1a30648bb47f8dade6c2693..03ec56b1d4cb98f815786f49d4da56f48c312878 100644 --- a/develop/doc/design/refactorization.html +++ b/develop/doc/design/refactorization.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/register_grad_op.html b/develop/doc/design/register_grad_op.html index cb52e5904a93a67063d792e0a866eb09a9fe150d..e4f7e347514e85a190583359921bd5209de33bc0 100644 --- a/develop/doc/design/register_grad_op.html +++ b/develop/doc/design/register_grad_op.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/regularization.html b/develop/doc/design/regularization.html index 72f1790c02761946f17140899660092b93ca71ee..71f341842ae9bd02fe38e8f0d1a44a911d8a4253 100644 --- a/develop/doc/design/regularization.html +++ b/develop/doc/design/regularization.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/releasing_process.html b/develop/doc/design/releasing_process.html index c80000df72b21934769beb4d597926161f6a0cd0..a71443ad7ba4206d807c8e7bd9163039b8b7cadf 100644 --- a/develop/doc/design/releasing_process.html +++ b/develop/doc/design/releasing_process.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/scope.html b/develop/doc/design/scope.html index c794bd4af3ef04c9418c29fe91f4945eb7e18e75..09d3639c7a6157086d071f1d0b764a35dddc0a4f 100644 --- a/develop/doc/design/scope.html +++ b/develop/doc/design/scope.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/selected_rows.html b/develop/doc/design/selected_rows.html index 2fe39da2cff2a9976e4e8db57b069641d3036cc0..5158cf8d2f96d9380dfb703947e87f80d89cb369 100644 --- a/develop/doc/design/selected_rows.html +++ b/develop/doc/design/selected_rows.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/simple_op_design.html b/develop/doc/design/simple_op_design.html index 7c28b0d50176bc8e826c43e20d7b832461181d03..b412b82f3a528b9a78ac3573a696b3cc06b252ac 100644 --- a/develop/doc/design/simple_op_design.html +++ b/develop/doc/design/simple_op_design.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/speech/deep_speech_2.html b/develop/doc/design/speech/deep_speech_2.html index 20fc13c9fa1f7073f8bb5b382989ba60279da7aa..b63d4024a176a6824191ec4b1f487746bcb9f5f6 100644 --- a/develop/doc/design/speech/deep_speech_2.html +++ b/develop/doc/design/speech/deep_speech_2.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/support_new_device.html b/develop/doc/design/support_new_device.html index ae5d59ee67073eda6c728bd3d0ee2d8c41da19fa..9fc56f2006318c59fc39badd18ae858c44e69f02 100644 --- a/develop/doc/design/support_new_device.html +++ b/develop/doc/design/support_new_device.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/switch_kernel.html b/develop/doc/design/switch_kernel.html index 1afc2ccb54e7d29ec481e8f981ea894fa29af3b6..28881b1c2bdd57ddc859b955893cd661799c2a7e 100644 --- a/develop/doc/design/switch_kernel.html +++ b/develop/doc/design/switch_kernel.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/tensor_array.html b/develop/doc/design/tensor_array.html index 843e8ce810342f60d9f8477b97b620654ac97269..f581f4018f59d5bf26961fac3349bea2ab1417af 100644 --- a/develop/doc/design/tensor_array.html +++ b/develop/doc/design/tensor_array.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/design/var_desc.html b/develop/doc/design/var_desc.html index 62516236a8d988dfddf5e4da517ce11ee6af43c2..52fdc0883f4b2fe45c4d074e63acef37e3ff0ac2 100644 --- a/develop/doc/design/var_desc.html +++ b/develop/doc/design/var_desc.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/genindex.html b/develop/doc/genindex.html index 9e1b8768ce79ebcc82050f00a0a589ea2c170bfb..1bc21c386ebc6440abc7904f5b815bde9ac51855 100644 --- a/develop/doc/genindex.html +++ b/develop/doc/genindex.html @@ -160,17 +160,17 @@- Training and Inference
- Fluid
@@ -217,7 +217,6 @@ B | C | L - | M | P | R | S @@ -243,14 +242,12 @@L
-- + - load_image() (in module paddle.v2.image)
- load_image_bytes() (in module paddle.v2.image) @@ -258,20 +255,8 @@
M
-
-- - P
- -- paddle.v2.image (module) @@ -294,10 +279,6 @@
- Training and Inference
- Fluid
diff --git a/develop/doc/getstarted/build_and_install/build_from_source_en.html b/develop/doc/getstarted/build_and_install/build_from_source_en.html index 9e9f14cae22a865fe4aaa1b906b07af132f8ce24..92192bd976820b25b2d7ac02719509feb8fb04d9 100644 --- a/develop/doc/getstarted/build_and_install/build_from_source_en.html +++ b/develop/doc/getstarted/build_and_install/build_from_source_en.html @@ -162,17 +162,17 @@- - Training and Inference
- Fluid
diff --git a/develop/doc/getstarted/build_and_install/index_en.html b/develop/doc/getstarted/build_and_install/index_en.html index a8551ac0366eb2b4fb4ebae11d7648a0c644572c..300df43fc6f7eadb69f2896f87e873453e371163 100644 --- a/develop/doc/getstarted/build_and_install/index_en.html +++ b/develop/doc/getstarted/build_and_install/index_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/getstarted/build_and_install/pip_install_en.html b/develop/doc/getstarted/build_and_install/pip_install_en.html index af9bd096115a9b8b200c3fff42243c7d2c837be2..bda39adf163127b059a86ee1a309ce33c2263222 100644 --- a/develop/doc/getstarted/build_and_install/pip_install_en.html +++ b/develop/doc/getstarted/build_and_install/pip_install_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/getstarted/index_en.html b/develop/doc/getstarted/index_en.html index 46535ff6f3218266cb507dcc2058b8995b50d3ff..55369cdf90abe905e86c801b3cb0e30ad40e634d 100644 --- a/develop/doc/getstarted/index_en.html +++ b/develop/doc/getstarted/index_en.html @@ -161,17 +161,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/deep_model/rnn/index_en.html b/develop/doc/howto/deep_model/rnn/index_en.html index 7258726fbf9d319878806ac326d3739bf881f530..020d628ab269ce4ccc83c68df56a0bfc87c50dc6 100644 --- a/develop/doc/howto/deep_model/rnn/index_en.html +++ b/develop/doc/howto/deep_model/rnn/index_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/deep_model/rnn/rnn_config_en.html b/develop/doc/howto/deep_model/rnn/rnn_config_en.html index 1c80190ad07b5f00a94dc6c00a700dfde1ad86cb..bf53caf0632a61c7da22cc38fdb586eacb574a2f 100644 --- a/develop/doc/howto/deep_model/rnn/rnn_config_en.html +++ b/develop/doc/howto/deep_model/rnn/rnn_config_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/dev/build_en.html b/develop/doc/howto/dev/build_en.html index 2d133a7aeb8808249606284a7cb3dd72882ce43c..2785b254b18e6626445f31f519280d3919535a16 100644 --- a/develop/doc/howto/dev/build_en.html +++ b/develop/doc/howto/dev/build_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/dev/contribute_to_paddle_en.html b/develop/doc/howto/dev/contribute_to_paddle_en.html index 94e96db96f3e999a5b66bf38a220de4cb5edf51e..6ff377b51f7abc392c8d156b3bee80e988508f36 100644 --- a/develop/doc/howto/dev/contribute_to_paddle_en.html +++ b/develop/doc/howto/dev/contribute_to_paddle_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/dev/new_layer_en.html b/develop/doc/howto/dev/new_layer_en.html index bd94698a902156de4cdd30a142e398952740fa6f..b82b2ea4f18f17c1b176d46d4ec4c86f91cd70ef 100644 --- a/develop/doc/howto/dev/new_layer_en.html +++ b/develop/doc/howto/dev/new_layer_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/dev/new_op_en.html b/develop/doc/howto/dev/new_op_en.html index 444f064aaffb8a0fa1e00a91d37f15a40426f818..fdb00830ee88f459afd186769a39b86497dffab2 100644 --- a/develop/doc/howto/dev/new_op_en.html +++ b/develop/doc/howto/dev/new_op_en.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/dev/new_op_kernel_en.html b/develop/doc/howto/dev/new_op_kernel_en.html index e1501166ab1b5b44c28ded4e70a9bcaea4d58197..42a623487992f39ff339be554c5381ac8ccfbb1f 100644 --- a/develop/doc/howto/dev/new_op_kernel_en.html +++ b/develop/doc/howto/dev/new_op_kernel_en.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/dev/use_eigen_en.html b/develop/doc/howto/dev/use_eigen_en.html index 45a07afe77c9f291f1dbde6cfdd933ff15e9a8bf..8234a2f822275de7b3954e0684e61075ffe9cb66 100644 --- a/develop/doc/howto/dev/use_eigen_en.html +++ b/develop/doc/howto/dev/use_eigen_en.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/dev/write_docs_en.html b/develop/doc/howto/dev/write_docs_en.html index a0ba728807d0879dfae4983fb3c802f96b7903d4..9b7c69c32a1234c9495c9a37339d1ed0ab8f2e6d 100644 --- a/develop/doc/howto/dev/write_docs_en.html +++ b/develop/doc/howto/dev/write_docs_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/index_en.html b/develop/doc/howto/index_en.html index 7de5347cd453675b12bece45eff723d21378b3f9..3659db79058043971bd5f6dc91310c5fe987932b 100644 --- a/develop/doc/howto/index_en.html +++ b/develop/doc/howto/index_en.html @@ -161,17 +161,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/optimization/cpu_profiling.html b/develop/doc/howto/optimization/cpu_profiling.html index f04d192c2559d9e623b08f02cab52bf2a38cbaa5..8696c436445b488eaa21e01022d9d55e93040afe 100644 --- a/develop/doc/howto/optimization/cpu_profiling.html +++ b/develop/doc/howto/optimization/cpu_profiling.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/optimization/gpu_profiling_en.html b/develop/doc/howto/optimization/gpu_profiling_en.html index e0caa3af36fd4a712d9b6f27417b011f283c716c..80d9288f05c2b49d748624a59b54714a00333ab5 100644 --- a/develop/doc/howto/optimization/gpu_profiling_en.html +++ b/develop/doc/howto/optimization/gpu_profiling_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/read_source.html b/develop/doc/howto/read_source.html index 1a3ae0b43a8f2538a8a73513dcc63d78d9fbd29e..8e6ee0d958efe72f24fd1aa1cc3be4ee28d74e42 100644 --- a/develop/doc/howto/read_source.html +++ b/develop/doc/howto/read_source.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cluster/cluster_train_en.html b/develop/doc/howto/usage/cluster/cluster_train_en.html index 9f6574ca3694d94046f35936acd96881b625d505..5fec7df248054ea86b11b29ce32f5fdb41d31dd9 100644 --- a/develop/doc/howto/usage/cluster/cluster_train_en.html +++ b/develop/doc/howto/usage/cluster/cluster_train_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cluster/fabric_en.html b/develop/doc/howto/usage/cluster/fabric_en.html index 292a8d875d1b2affe609173ce5c2322443d8110b..5e960fe3563b32df403eec5459bab3a189f1645e 100644 --- a/develop/doc/howto/usage/cluster/fabric_en.html +++ b/develop/doc/howto/usage/cluster/fabric_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cluster/fluid_cluster_train_en.html b/develop/doc/howto/usage/cluster/fluid_cluster_train_en.html index 15eeb6259841523e02764ae4df9c5edc796baf36..5446faa0c752d80497755325db4fcfee583516df 100644 --- a/develop/doc/howto/usage/cluster/fluid_cluster_train_en.html +++ b/develop/doc/howto/usage/cluster/fluid_cluster_train_en.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cluster/k8s_aws_en.html b/develop/doc/howto/usage/cluster/k8s_aws_en.html index 850e03b53698455b0b370f509359774e98551cae..794cdd6ea0befb415ed354536b947f23dfa7c64c 100644 --- a/develop/doc/howto/usage/cluster/k8s_aws_en.html +++ b/develop/doc/howto/usage/cluster/k8s_aws_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cluster/k8s_en.html b/develop/doc/howto/usage/cluster/k8s_en.html index 4153404fa1d22bfaf0470e48b41cb10ce5b40492..fecc89430dd473cb6f23949b28032a0905cb3765 100644 --- a/develop/doc/howto/usage/cluster/k8s_en.html +++ b/develop/doc/howto/usage/cluster/k8s_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cluster/openmpi_en.html b/develop/doc/howto/usage/cluster/openmpi_en.html index 65c6d010c055a4de249c73bd3748fe523482031b..29289e68ae4bc3bd519571aca41525dceba45ed3 100644 --- a/develop/doc/howto/usage/cluster/openmpi_en.html +++ b/develop/doc/howto/usage/cluster/openmpi_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cluster/src/k8s_data/README.html b/develop/doc/howto/usage/cluster/src/k8s_data/README.html index 10c83c0239d6120e345f7cbf4fcef1af8f599f5e..9c50fa31140b9fa91dc57d28d1dc485db763c985 100644 --- a/develop/doc/howto/usage/cluster/src/k8s_data/README.html +++ b/develop/doc/howto/usage/cluster/src/k8s_data/README.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cluster/src/k8s_train/README.html b/develop/doc/howto/usage/cluster/src/k8s_train/README.html index 0dd51f4333066152a1c44d6be959409a4a6a9f04..ab14b4cb86c3ab1bdfadf94ae251b6743cd98770 100644 --- a/develop/doc/howto/usage/cluster/src/k8s_train/README.html +++ b/develop/doc/howto/usage/cluster/src/k8s_train/README.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cmd_parameter/arguments_en.html b/develop/doc/howto/usage/cmd_parameter/arguments_en.html index 8ffbc2d03e902e2df65c7731b4b637cd47b34ce2..5f09ec2b63632fc6e781b06ea52b27eb0ae2aae9 100644 --- a/develop/doc/howto/usage/cmd_parameter/arguments_en.html +++ b/develop/doc/howto/usage/cmd_parameter/arguments_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cmd_parameter/detail_introduction_en.html b/develop/doc/howto/usage/cmd_parameter/detail_introduction_en.html index 87033573385563e9e8b65de44c580d629030167f..90e7268450ed38150ff99c0a4dccf9ff3b4b1a7f 100644 --- a/develop/doc/howto/usage/cmd_parameter/detail_introduction_en.html +++ b/develop/doc/howto/usage/cmd_parameter/detail_introduction_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cmd_parameter/index_en.html b/develop/doc/howto/usage/cmd_parameter/index_en.html index 49cd7d599a06f0b48a0a83937bfd7e3920c1b269..f47cb2500c32b8adcd54e7071120d7dbe4fa38fd 100644 --- a/develop/doc/howto/usage/cmd_parameter/index_en.html +++ b/develop/doc/howto/usage/cmd_parameter/index_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/howto/usage/cmd_parameter/use_case_en.html b/develop/doc/howto/usage/cmd_parameter/use_case_en.html index bc422f6246668b0036d29dee0c22114a746c7458..93939c03887ba62f2c8b4b7fc9de58e8d6f56931 100644 --- a/develop/doc/howto/usage/cmd_parameter/use_case_en.html +++ b/develop/doc/howto/usage/cmd_parameter/use_case_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/index_en.html b/develop/doc/index_en.html index d95a7cfce9c3f2943f5fbd6c2d7317d7a2071620..e9e89586655f5ba97cf697b720315dec50d74c53 100644 --- a/develop/doc/index_en.html +++ b/develop/doc/index_en.html @@ -160,17 +160,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/mobile/cross_compiling_for_android_en.html b/develop/doc/mobile/cross_compiling_for_android_en.html index 0e3b653435e199335517727aa376399f2e695e97..cd91d89724bd6c99034704d1b1f50949924026a2 100644 --- a/develop/doc/mobile/cross_compiling_for_android_en.html +++ b/develop/doc/mobile/cross_compiling_for_android_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/mobile/cross_compiling_for_ios_en.html b/develop/doc/mobile/cross_compiling_for_ios_en.html index 4f1d77f0c49a39340365e310cf2fc6266c36bfa0..040f1636028dfc8e0977a7841d81af70d93c2fed 100644 --- a/develop/doc/mobile/cross_compiling_for_ios_en.html +++ b/develop/doc/mobile/cross_compiling_for_ios_en.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/mobile/cross_compiling_for_raspberry_en.html b/develop/doc/mobile/cross_compiling_for_raspberry_en.html index da676e7da6d233c3c63e89f76f8d28fdd0331ca9..4372f26f2598e8854fc316d07859bc6f8853b427 100644 --- a/develop/doc/mobile/cross_compiling_for_raspberry_en.html +++ b/develop/doc/mobile/cross_compiling_for_raspberry_en.html @@ -161,17 +161,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc/mobile/index_en.html b/develop/doc/mobile/index_en.html index 991d819535f35303d30b4fc8b1df41e3530157f7..788d4ef52feb4f34e67f3d8b1d34cbfe4de481a0 100644 --- a/develop/doc/mobile/index_en.html +++ b/develop/doc/mobile/index_en.html @@ -34,7 +34,7 @@ - + @@ -161,17 +161,17 @@- Training and Inference
- Fluid
@@ -232,7 +232,7 @@ Next - Previous + Previous diff --git a/develop/doc/objects.inv b/develop/doc/objects.inv index e43f7b4a0aac4d2ab59eac80ce81d1f33c354a31..0a32d308584901424f6662291e5a977a86c48955 100644 Binary files a/develop/doc/objects.inv and b/develop/doc/objects.inv differ diff --git a/develop/doc/py-modindex.html b/develop/doc/py-modindex.html index 21e196d7be88ff18f130af10a437378a3ad598e7..8dc03fc797bcb32717fef9bed0ca245a0423d986 100644 --- a/develop/doc/py-modindex.html +++ b/develop/doc/py-modindex.html @@ -162,17 +162,17 @@- Training and Inference
- Fluid
@@ -229,11 +229,6 @@paddle- - - paddle.v2.fluid.regularizer- diff --git a/develop/doc/search.html b/develop/doc/search.html index 5a57ddffb4bc342d71fa1a78873bbea8533d85ca..248db3bb421454087fe9ada04c14526ed0438fd3 100644 --- a/develop/doc/search.html +++ b/develop/doc/search.html @@ -159,17 +159,17 @@ - Training and Inference
- Fluid
diff --git a/develop/doc/searchindex.js b/develop/doc/searchindex.js index cd5fe608f0d125c578ede21f3d25b99ac340ec1c..470ba3a48412dc907d5c4346d24af090d5fcbcaf 100644 --- a/develop/doc/searchindex.js +++ b/develop/doc/searchindex.js @@ -1 +1 @@ -Search.setIndex({docnames:["api/index_en","api/v2/config/activation","api/v2/config/attr","api/v2/config/evaluators","api/v2/config/layer","api/v2/config/networks","api/v2/config/optimizer","api/v2/config/pooling","api/v2/data","api/v2/data/data_reader","api/v2/data/dataset","api/v2/data/image","api/v2/fluid","api/v2/fluid/data_feeder","api/v2/fluid/evaluator","api/v2/fluid/executor","api/v2/fluid/initializer","api/v2/fluid/io","api/v2/fluid/layers","api/v2/fluid/nets","api/v2/fluid/optimizer","api/v2/fluid/param_attr","api/v2/fluid/profiler","api/v2/fluid/regularizer","api/v2/model_configs","api/v2/run_logic","design/api","design/auto_gradient_check","design/backward","design/block","design/build_system/README","design/cluster_train/README","design/cluster_train/checkpointing","design/cluster_train/data_dispatch","design/cluster_train/large_model_dist_train","design/cluster_train/master_server","design/cluster_train/pserver_client","design/cluster_train/remote_parameter_updater","design/cluster_train/save_model","design/cluster_train/submit-job","design/concurrent_programming","design/csp","design/dist_refactor/distributed_architecture","design/dist_refactor/multi_cpu","design/dist_refactor/parameter_server","design/error_clip","design/evaluator","design/executor","design/file_manager/README","design/file_manager/pfs/pfsclient","design/float16","design/fluid","design/fluid_compiler","design/functions_operators_layers","design/gan_api","design/graph","design/graph_survey","design/if_else_op","design/infer_var_type","design/kernel_hint_design","design/memory_optimization","design/mkl/mkl_packed","design/mkl/mkldnn","design/mkl/mkldnn_fluid","design/model_format","design/multi_language_interface/00.why_plain_c","design/multi_language_interface/01.inference_implementation","design/operator_kernel_type","design/ops/rnn","design/ops/sequence_decoder","design/optimizer","design/paddle_nccl","design/parameter_average","design/parameters_in_cpp","design/profiler","design/program","design/prune","design/python_api","design/reader/README","design/refactorization","design/register_grad_op","design/regularization","design/releasing_process","design/scope","design/selected_rows","design/simple_op_design","design/speech/deep_speech_2","design/support_new_device","design/switch_kernel","design/tensor_array","design/var_desc","getstarted/build_and_install/build_from_source_en","getstarted/build_and_install/docker_install_en","getstarted/build_and_install/index_en","getstarted/build_and_install/pip_install_en","getstarted/index_en","howto/deep_model/rnn/index_en","howto/deep_model/rnn/rnn_config_en","howto/dev/build_en","howto/dev/contribute_to_paddle_en","howto/dev/new_layer_en","howto/dev/new_op_en","howto/dev/new_op_kernel_en","howto/dev/use_eigen_en","howto/dev/write_docs_en","howto/index_en","howto/optimization/cpu_profiling","howto/optimization/gpu_profiling_en","howto/read_source","howto/usage/cluster/cluster_train_en","howto/usage/cluster/fabric_en","howto/usage/cluster/fluid_cluster_train_en","howto/usage/cluster/k8s_aws_en","howto/usage/cluster/k8s_en","howto/usage/cluster/openmpi_en","howto/usage/cluster/src/k8s_data/README","howto/usage/cluster/src/k8s_train/README","howto/usage/cmd_parameter/arguments_en","howto/usage/cmd_parameter/detail_introduction_en","howto/usage/cmd_parameter/index_en","howto/usage/cmd_parameter/use_case_en","index_en","mobile/cross_compiling_for_android_en","mobile/cross_compiling_for_ios_en","mobile/cross_compiling_for_raspberry_en","mobile/index_en","survey/cluster_bootstrapping_tools"],envversion:50,filenames:["api/index_en.rst","api/v2/config/activation.rst","api/v2/config/attr.rst","api/v2/config/evaluators.rst","api/v2/config/layer.rst","api/v2/config/networks.rst","api/v2/config/optimizer.rst","api/v2/config/pooling.rst","api/v2/data.rst","api/v2/data/data_reader.rst","api/v2/data/dataset.rst","api/v2/data/image.rst","api/v2/fluid.rst","api/v2/fluid/data_feeder.rst","api/v2/fluid/evaluator.rst","api/v2/fluid/executor.rst","api/v2/fluid/initializer.rst","api/v2/fluid/io.rst","api/v2/fluid/layers.rst","api/v2/fluid/nets.rst","api/v2/fluid/optimizer.rst","api/v2/fluid/param_attr.rst","api/v2/fluid/profiler.rst","api/v2/fluid/regularizer.rst","api/v2/model_configs.rst","api/v2/run_logic.rst","design/api.md","design/auto_gradient_check.md","design/backward.md","design/block.md","design/build_system/README.md","design/cluster_train/README.md","design/cluster_train/checkpointing.md","design/cluster_train/data_dispatch.md","design/cluster_train/large_model_dist_train.md","design/cluster_train/master_server.md","design/cluster_train/pserver_client.md","design/cluster_train/remote_parameter_updater.md","design/cluster_train/save_model.md","design/cluster_train/submit-job.md","design/concurrent_programming.md","design/csp.md","design/dist_refactor/distributed_architecture.md","design/dist_refactor/multi_cpu.md","design/dist_refactor/parameter_server.md","design/error_clip.md","design/evaluator.md","design/executor.md","design/file_manager/README.md","design/file_manager/pfs/pfsclient.md","design/float16.md","design/fluid.md","design/fluid_compiler.md","design/functions_operators_layers.md","design/gan_api.md","design/graph.md","design/graph_survey.md","design/if_else_op.md","design/infer_var_type.md","design/kernel_hint_design.md","design/memory_optimization.md","design/mkl/mkl_packed.md","design/mkl/mkldnn.md","design/mkl/mkldnn_fluid.md","design/model_format.md","design/multi_language_interface/00.why_plain_c.md","design/multi_language_interface/01.inference_implementation.md","design/operator_kernel_type.md","design/ops/rnn.md","design/ops/sequence_decoder.md","design/optimizer.md","design/paddle_nccl.md","design/parameter_average.md","design/parameters_in_cpp.md","design/profiler.md","design/program.md","design/prune.md","design/python_api.md","design/reader/README.md","design/refactorization.md","design/register_grad_op.md","design/regularization.md","design/releasing_process.md","design/scope.md","design/selected_rows.md","design/simple_op_design.md","design/speech/deep_speech_2.md","design/support_new_device.md","design/switch_kernel.md","design/tensor_array.md","design/var_desc.md","getstarted/build_and_install/build_from_source_en.rst","getstarted/build_and_install/docker_install_en.rst","getstarted/build_and_install/index_en.rst","getstarted/build_and_install/pip_install_en.rst","getstarted/index_en.rst","howto/deep_model/rnn/index_en.rst","howto/deep_model/rnn/rnn_config_en.rst","howto/dev/build_en.md","howto/dev/contribute_to_paddle_en.md","howto/dev/new_layer_en.rst","howto/dev/new_op_en.md","howto/dev/new_op_kernel_en.md","howto/dev/use_eigen_en.md","howto/dev/write_docs_en.rst","howto/index_en.rst","howto/optimization/cpu_profiling.md","howto/optimization/gpu_profiling_en.rst","howto/read_source.md","howto/usage/cluster/cluster_train_en.md","howto/usage/cluster/fabric_en.md","howto/usage/cluster/fluid_cluster_train_en.md","howto/usage/cluster/k8s_aws_en.md","howto/usage/cluster/k8s_en.md","howto/usage/cluster/openmpi_en.md","howto/usage/cluster/src/k8s_data/README.md","howto/usage/cluster/src/k8s_train/README.md","howto/usage/cmd_parameter/arguments_en.md","howto/usage/cmd_parameter/detail_introduction_en.md","howto/usage/cmd_parameter/index_en.rst","howto/usage/cmd_parameter/use_case_en.md","index_en.rst","mobile/cross_compiling_for_android_en.md","mobile/cross_compiling_for_ios_en.md","mobile/cross_compiling_for_raspberry_en.md","mobile/index_en.rst","survey/cluster_bootstrapping_tools.md"],objects:{"paddle.v2":{image:[11,1,0,"-"]},"paddle.v2.fluid":{regularizer:[23,1,0,"-"]},"paddle.v2.fluid.evaluator.Evaluator":{metrics:[14,0,1,""],states:[14,0,1,""]},"paddle.v2.fluid.regularizer":{L1DecayRegularizer:[23,2,1,""]},"paddle.v2.image":{batch_images_from_tar:[11,3,1,""],center_crop:[11,3,1,""],left_right_flip:[11,3,1,""],load_and_transform:[11,3,1,""],load_image:[11,3,1,""],load_image_bytes:[11,3,1,""],random_crop:[11,3,1,""],resize_short:[11,3,1,""],simple_transform:[11,3,1,""],to_chw:[11,3,1,""]}},objnames:{"0":["py","attribute","Python attribute"],"1":["py","module","Python module"],"2":["py","class","Python class"],"3":["py","function","Python function"]},objtypes:{"0":"py:attribute","1":"py:module","2":"py:class","3":"py:function"},terms:{"00m":107,"03m":107,"0424m":107,"0473v3":5,"055ee37d":112,"0630u":107,"06u":107,"0810u":107,"0957m":107,"0_cudnn5":91,"0_cudnn5_avx_mkl":[92,94],"0_cudnn7_avx_mkl":94,"0ab":4,"0rc":109,"0rc1":82,"0rc2":82,"0x10f256d50":56,"0x7ffe4de00110":56,"100gb":107,"100gi":112,"10g":39,"10m":107,"1150u":107,"11\u5b9e\u73b0\u4e86c":66,"11e6":113,"124n":107,"12gb":60,"13m":113,"1490u":107,"1550u":107,"16u":107,"173n":107,"1770u":107,"18ad":112,"18e457ce3d362ff5f3febf8e7f85ffec852f70f3b629add10aed84f930a68750":113,"197u":107,"1gb":107,"1st":18,"210u":107,"211839e770f7b538e2d8":5,"215n":107,"228u":107,"2520u":107,"2680u":107,"279n":107,"27m":107,"285m":107,"2863m":107,"28m":107,"2977m":107,"2cbf7385":112,"2nd":18,"302n":107,"30u":107,"328n":107,"32u":107,"32x32":10,"331n":107,"3320u":107,"365e":112,"36u":107,"3710m":107,"3768m":107,"387u":107,"38u":107,"3920u":107,"39u":107,"3rd":123,"4035m":107,"4090u":107,"4096mb":118,"4279m":107,"43u":107,"448a5b355b84":113,"4560u":107,"4563m":107,"45u":107,"4650u":107,"4726m":107,"473m":113,"4gb":118,"50bd":112,"50gi":112,"514u":107,"525n":107,"526u":107,"536u":107,"5460u":107,"5470u":107,"54u":107,"5690m":107,"573u":107,"578n":107,"5798m":107,"586u":107,"58s":113,"5969m":107,"5_cudnn5_avx_mkl":94,"5_cudnn5_avx_openbla":[94,95],"6080u":107,"6140u":107,"6305m":107,"639u":107,"64m":64,"655u":107,"6780u":107,"6810u":107,"682u":107,"6970u":107,"6ce9":112,"704u":107,"7090u":107,"72u":107,"73u":107,"75u":107,"760u":107,"767u":107,"783n":107,"784u":107,"78m":107,"7eamaa":10,"7kb":113,"8250u":107,"8300u":107,"830n":107,"849m":107,"861u":107,"8661m":107,"892m":107,"901n":107,"90u":107,"918u":107,"9247m":107,"924n":107,"9261m":107,"9330m":107,"94u":107,"9530m":107,"983m":107,"988u":107,"997u":107,"99u":107,"9a235":123,"9f18":113,"\u4e00\u4e2a\u5178\u578b\u7684chunk\u5982\u4e0b\u6240\u793a":48,"\u4e00\u4e2a\u795e\u7ecf\u7f51\u7edc\u7684\u6a21\u578b\u7531\u5927\u91cf\u7684\u53c2\u6570\u7ec4\u6210":32,"\u4e00\u4e2achunk\u7531\u6240\u5728\u7684\u6587\u4ef6\u504f\u79fb":48,"\u4e00\u4e2aposix\u517c\u5bb9\u7684\u6587\u4ef6\u7cfb\u7edf":48,"\u4e00\u822c\u4e0d\u5141\u8bb8\u518d\u4ece":82,"\u4e00\u822c\u7531mkl":62,"\u4e0a\u4f20\u5230cloud\u6216\u8005\u4e0b\u8f7d\u5230\u672c\u5730\u7684\u65f6\u95f4\u53ef\u80fd\u6bd4\u8f83\u957f":48,"\u4e0a\u4f20\u65b9\u6cd5":82,"\u4e0a\u6ce8\u518c\u4e00\u4e0b":48,"\u4e0a\u8ff0paddlepaddl":82,"\u4e0b\u4e00\u4e2awheel\u5305\u9700\u8981\u66f4\u65b0\u7248\u672c\u53f7\u624d\u53ef\u4ee5\u4e0a\u4f20":82,"\u4e0b\u5b58\u653e\u516c\u5171\u6570\u636e\u96c6\u5408":33,"\u4e0b\u62c9\u6846\u4e2d\u627e\u5230\u751f\u6210\u76843\u4e2a\u4e8c\u8fdb\u5236\u6587\u4ef6":82,"\u4e0b\u8f7d":48,"\u4e0b\u8f7d\u5230\u672c\u5730":48,"\u4e0b\u8f7d\u5f97\u5230":82,"\u4e0b\u9762\u5206\u522b\u4ecb\u7ecd\u67d0\u4e00\u7c7b\u6587\u4ef6\u7684\u5b9e\u73b0\u65b9\u5f0f":66,"\u4e0d\u4e00\u81f4\u7684\u7531pfsclient\u4e0b\u8f7d\u6216\u8005\u4f20\u8f93chunk\u5b8c\u6210":48,"\u4e0d\u4f7f\u7528\u9759\u6001\u5e93":65,"\u4e0d\u4f7f\u7528c":65,"\u4e0d\u4f7f\u7528swig":65,"\u4e0d\u53ef\u4ee5\u66f4\u6539":82,"\u4e0d\u540c":62,"\u4e0d\u540c\u7248\u672c\u7684\u7f16\u8bd1\u5668\u4e4b\u95f4":65,"\u4e0d\u540c\u8bed\u8a00\u7684\u63a5\u53e3\u9002\u5e94\u4e0d\u540c\u8bed\u8a00\u7684\u7279\u6027":65,"\u4e0d\u5728":66,"\u4e0d\u5bb9\u6613\u51fa\u9519":48,"\u4e0d\u5d4c\u5165\u5176\u4ed6\u8bed\u8a00\u89e3\u91ca\u5668":65,"\u4e0d\u5d4c\u5165python\u89e3\u91ca\u5668":65,"\u4e0d\u663e\u793a\u7684\u5199\u6bcf\u4e2a\u7c7b\u5177\u4f53\u5305\u542b\u4ec0\u4e48":65,"\u4e0d\u7528mount\u7684\u65b9\u5f0f\u6765\u8bbf\u95ee\u6570\u636e":33,"\u4e0e":62,"\u4e0e\u4e4b\u76f8\u5bf9\u7684\u662flocal":48,"\u4e0e\u5176\u4ed6\u7b2c\u4e09\u65b9\u5e93\u4e00\u6837":62,"\u4e0e\u529f\u80fd\u5206\u652f\u4e0d\u540c\u7684\u662f":82,"\u4e0e\u53ef\u80fd\u6709\u7684":82,"\u4e0ebatch":61,"\u4e14\u589e\u52a0\u4e00\u4e2a\u7b2c\u4e09\u65b9\u8bed\u8a00":65,"\u4e14\u8c03\u7528\u65f6\u4e0d\u80fd\u629b\u51fa\u5f02\u5e38\u6216\u51fa\u73b0\u8fd0\u884c\u65f6\u9519\u8bef":66,"\u4e14c99\u652f\u6301bool\u7c7b\u578b\u548c\u5b9a\u957f\u6574\u6570":65,"\u4e14c99\u76f8\u5bf9\u4e8ec11\u4f7f\u7528\u66f4\u52a0\u5e7f\u6cdb":65,"\u4e25\u683c\u7684\u547d\u540d\u89c4\u8303pep":82,"\u4e2a\u6027\u5316\u63a8\u8350":82,"\u4e2d":[61,62,65,66],"\u4e2d\u4f1a\u63d0\u4f9b\u4e00\u4e9b\u5fc5\u8981\u7684\u63a5\u53e3\u548c\u51fd\u6570":62,"\u4e2d\u5199\u5165json\u5185\u5bb9":32,"\u4e2d\u5b8c\u5168\u4e00\u81f4":65,"\u4e2d\u5b9e\u73b0\u4e86\u4e00\u4e2amerge\u7684\u65b9\u6cd5":62,"\u4e2d\u5b9e\u73b0\u7684\u7ed3\u6784\u4f53":66,"\u4e2d\u5bf9\u5e94\u7684layer\u5904":61,"\u4e2d\u5f15\u5165\u7684":61,"\u4e2d\u63d0\u4f9b\u4e00\u4e2a\u4e0emkl\u6709\u5173\u7684\u603b\u5f00\u5173":62,"\u4e2d\u6839\u636e":61,"\u4e2d\u6dfb\u52a0":61,"\u4e2d\u6dfb\u52a0\u4e00\u4e2a":62,"\u4e2d\u7684\u7248\u672c\u4fe1\u606f":82,"\u4e2d\u8fd0\u884c\u4efb\u52a1\u7684\u89d2\u5ea6":33,"\u4e3a":[61,62],"\u4e3a\u4e86\u5c3d\u53ef\u80fd\u5c11\u7684\u5728\u7236\u7c7blayer\u4e2d\u6dfb\u52a0\u53d8\u91cf\u6216\u8005\u51fd\u6570":62,"\u4e3a\u4e86\u5e94\u5bf9\u4ee5\u4e0a\u7684\u95ee\u9898":48,"\u4e3a\u4e86\u66b4\u9732\u7684\u63a5\u53e3\u5c3d\u91cf\u7b80\u5355":66,"\u4e3a\u4e86\u66f4\u597d\u7684\u7b26\u5408paddlepaddle\u7684\u4ee3\u7801\u98ce\u683c":62,"\u4e3a\u4e86\u6700\u5927\u7a0b\u5ea6\u51cf\u5c11\u591a\u6b21\u8c03\u7528":61,"\u4e3a\u4e86\u8fdb\u4e00\u6b65\u63d0\u5347paddlepaddle\u5728\u57fa\u672c\u6570\u5b66\u8fd0\u7b97\u7684\u8ba1\u7b97\u901f\u5ea6":62,"\u4e3b\u8981\u529f\u80fd\u5305\u62ec":48,"\u4e3b\u8981\u5305\u62ec":62,"\u4e3b\u8981\u5305\u62ec\u4e86\u6df1\u5ea6\u5b66\u4e60\u76f8\u5173\u7684\u6570\u5b66\u539f\u8bed\u4e0e\u64cd\u4f5c":62,"\u4e3b\u8981\u9488\u5bf9paddlepaddle\u5728\u91cd\u6784\u4e4b\u524d\u7684\u4ee3\u7801\u6846\u67b6\u4ee5\u53cav1\u7684api":62,"\u4e4b\u5916\u7684\u6240\u6709\u5934\u6587\u4ef6":66,"\u4e5f\u4e0d\u4f7f\u7528\u5176\u4ed6\u52a8\u6001\u5e93":65,"\u4e5f\u4e0d\u5e94\u8be5\u62a5\u9519":66,"\u4e5f\u4e0d\u751f\u6210":66,"\u4e5f\u53ef\u4ee5\u4f7f\u7528\u8fd9\u4e9b\u955c\u50cf":82,"\u4e5f\u5c31\u662f\u8bf4\u8f93\u51fa\u7684\u7ed3\u679c\u4e0d\u4f1a\u5728\u539f\u6765\u7684\u6570\u636e\u4e0a\u7d2f\u52a0":62,"\u4e66\u5199":65,"\u4eba\u8138\u8bc6\u522b":33,"\u4ec5\u4ec5\u4f7f\u7528":65,"\u4ec5\u4f1a\u5728\u652f\u6301avx2\u6307\u4ee4\u96c6\u53ca\u4ee5\u4e0a\u7684\u673a\u5668\u624d\u4f7f\u7528mkl":62,"\u4ece":82,"\u4ece\u78c1\u76d8\u6587\u4ef6\u4e2d\u52a0\u8f7duuid\u6587\u4ef6\u540d\u7684\u68c0\u67e5\u70b9\u5feb\u7167\u6587\u4ef6":32,"\u4ece\u800c\u907f\u514d\u4e86packing\u5197\u4f59":61,"\u4eceetcd\u4e2d\u8bfb\u53d6\u8282\u70b9":32,"\u4ed6\u4e3b\u8981\u5305\u542b\u4e86\u5b9e\u9645\u66b4\u9732\u7684\u7c7b\u578b\u7ed3\u6784":66,"\u4ed6\u662f\u5c06":66,"\u4ed6\u7684\u76ee\u6807\u662f\u4f7f\u7528c":65,"\u4ee3\u7801\u751f\u6210\u7684\u7b26\u53f7\u53ef\u80fd\u4e0d\u4e00\u81f4":65,"\u4ee3\u8868\u8fd9\u4e2alayer\u662f\u7528\u4e8e\u8dd1\u5728mkl":62,"\u4ee3\u8868\u8fd9\u4e2ashard\u7684\u6700\u5927index":33,"\u4ee3\u8868shard\u7684index":33,"\u4ee5\u4e0a\u4ee3\u7801\u7684reader\u8f93\u51fa\u7684data":33,"\u4ee5\u4e0a\u547d\u4ee4\u4f1a\u5728\u5f53\u524d\u76ee\u5f55\u4e0b\u751f\u6210100\u4e2a\u6587\u4ef6":33,"\u4ee5\u4e0b":33,"\u4ee5\u4e0b\u7b80\u79f0rnn":61,"\u4ee5\u4fbf\u6211\u4eec\u53ef\u4ee5\u628a\u66f4\u591a\u7684\u7cbe\u529b\u653e\u5230\u903b\u8f91\u672c\u8eab\u4e0a":48,"\u4ee5\u53ca":61,"\u4ee5\u53canumpi":33,"\u4ee5\u6b64\u8fbe\u5230\u6700\u597d\u7684\u6027\u80fd":62,"\u4ee5\u793a\u533a\u5206":[61,62],"\u4efb\u610f\u65f6\u523b\u53ea\u53ef\u80fd\u540c\u65f6\u6709\u4e00\u53f0\u670d\u52a1\u5668\u6545\u969c":32,"\u4f18\u5316\u524d":61,"\u4f18\u5316\u540e":61,"\u4f1a\u4ee5":[61,62],"\u4f1a\u4f7f\u7528\u76f8\u540c\u7684\u539f\u6570\u636e":61,"\u4f1a\u5148\u4e34\u65f6\u4fdd\u5b58\u5728":62,"\u4f1a\u5728":62,"\u4f1a\u5728\u7f16\u8bd1paddlepaddle\u7684\u65f6\u5019\u4e0b\u8f7d\u5e76\u7f16\u8bd1mkl":62,"\u4f1a\u5bfc\u81f4\u4e0d\u540c\u7248\u672cpython\u5728\u4e00\u4e2a\u8fdb\u7a0b\u91cc\u7684bug":65,"\u4f1a\u5f15\u5165":62,"\u4f1a\u628acpu\u7684buffer\u5bf9\u9f50\u4e3a4096":62,"\u4f1a\u6dfb\u52a0\u76f8\u5e94\u7684\u811a\u672c\u5728":62,"\u4f1a\u6dfb\u52a0\u76f8\u5e94\u7684\u811a\u672c\u7528\u4e8e\u6d4b\u8bd5\u548c\u5bf9\u6bd4\u5728\u4f7f\u7528mkl":61,"\u4f1a\u76f4\u63a5\u62a5\u9519\u9000\u51fa":65,"\u4f1a\u81ea\u52a8\u4f7f\u7528mklml\u5e93\u4f5c\u4e3apaddlepaddle\u7684cblas\u548clapack\u5e93":62,"\u4f1a\u81ea\u52a8\u6839\u636e\u786c\u4ef6\u914d\u7f6e":62,"\u4f1a\u88abpickle\u5e8f\u5217\u5316\u6210\u5b57\u7b26\u4e32":33,"\u4f20\u5165":33,"\u4f46":66,"\u4f46\u4e0d\u66b4\u9732":66,"\u4f46\u5e76\u6ca1\u6709\u7ecf\u8fc7\u56de\u5f52\u6d4b\u8bd5":82,"\u4f46\u6240\u6709fork\u7684\u7248\u672c\u5e93\u7684\u6240\u6709\u5206\u652f\u90fd\u76f8\u5f53\u4e8e\u7279\u6027\u5206\u652f":82,"\u4f46\u662f\u53c8\u8fc7\u4e8e\u7410\u788e":66,"\u4f46\u662f\u5728mkl":62,"\u4f46\u662f\u5728paddlepaddle\u4e2d":62,"\u4f46\u662f\u6574\u4e2a\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u4e0d\u9700\u8981\u4efb\u4f55\u8f6c\u6362":62,"\u4f46\u662f\u6ce8\u610f\u7684\u662f":62,"\u4f46\u662f\u89e3\u91ca\u6027\u8bed\u8a00":65,"\u4f5c\u4e3a\u53e6\u4e00\u4e2a\u7b2c\u4e09\u65b9\u5e93\u96c6\u6210\u8fdbpaddlepaddl":62,"\u4f5c\u4e3a\u5b58\u50a8\u7cfb\u7edf":33,"\u4f5c\u4e3a\u7c7b\u53e5\u67c4":65,"\u4f7f\u7528":[62,66,82],"\u4f7f\u7528\u4e0b\u9762\u547d\u4ee4":33,"\u4f7f\u7528\u52a8\u6001\u5e93":65,"\u4f7f\u7528\u540c\u6837\u7684\u8bad\u7ec3\u6570\u636eblock":32,"\u4f7f\u7528\u667a\u80fd\u6307\u9488\u7684\u539f\u56e0\u662f":66,"\u4f7f\u7528\u7684\u53c2\u6570\u4e0epaddlepaddle\u7533\u8bf7\u7684buffer\u5171\u7528\u4e00\u5757\u5185\u5b58":62,"\u4f7f\u7528\u76f8\u5bf9\u8def\u5f84\u7684\u5f15\u7528\u65b9\u5f0f":66,"\u4f7f\u7528\u8fd9\u4e2a\u795e\u7ecf\u7f51\u7edc\u53ef\u4ee5\u5b8c\u6210\u5bf9\u65b0\u6570\u636e\u7684\u9884\u6d4b":32,"\u4f7f\u7528\u9759\u6001\u5e93\u548c\u52a8\u6001\u5e93\u96be\u5ea6\u5dee\u4e0d\u591a":65,"\u4f7f\u7528c":66,"\u4f7f\u7528c99\u505a\u63a5\u53e3":65,"\u4f7f\u7528c99\u800c\u4e0d\u4f7f\u7528c11\u7684\u539f\u56e0\u662f":65,"\u4f7f\u7528c99\u800c\u4e0d\u4f7f\u7528c89":65,"\u4f7f\u7528regress":82,"\u4f7f\u7528swig\u53ea\u652f\u6301cpython\u89e3\u91ca\u5668":65,"\u4f7f\u7528swig\u9700\u8981\u591a\u8bed\u8a00\u7ed1\u5b9a\u7684\u5f00\u53d1\u4eba\u5458\u719f\u7ec3\u638c\u63e1swig\u914d\u7f6e":65,"\u4f7f\u7528void":65,"\u4f8b\u5982":[33,65,66,82],"\u4f8b\u5982\u5728deepspeech2":61,"\u4f8b\u5982\u5bf9\u4e8ejava\u6216\u8005python":65,"\u4f8b\u5982\u5bf9\u4e8ejava\u6765\u8bf4":65,"\u4f8b\u5982\u5bf9\u4e8epython":65,"\u4f8b\u5982c":65,"\u4f8b\u5982java\u4e0epython\u7684\u9519\u8bef\u5904\u7406\u662f\u76f4\u63a5\u6254\u51fa\u6765except":65,"\u4f8b\u5982python\u53ef\u4ee5\u4f7f\u7528":65,"\u4f8b\u5982python\u7684":65,"\u4f8b\u5982rnn":61,"\u4f9d\u6b21\u7c7b\u63a8":82,"\u4fbf\u662f\u5c06\u9759\u6001\u5e93\u52a0\u5165jvm\u4e2d":65,"\u4fee\u590d\u6240\u6709bug\u540e":82,"\u4fee\u590ddocker\u7f16\u8bd1\u955c\u50cf\u95ee\u9898":82,"\u4fee\u6539":[62,82],"\u4fee\u6539\u6210":82,"\u505a\u53ea\u8bfb\u6302\u8f7d":33,"\u505a\u5982\u4e0b\u51e0\u4e2a\u64cd\u4f5c":82,"\u505a\u63a5\u53e3":65,"\u505c\u6b62\u4fdd\u5b58\u68c0\u67e5\u70b9\u7684\u7ebf\u7a0b":32,"\u5145\u5206\u53d1\u6325\u82f1\u7279\u5c14\u5e73\u53f0\u7684\u4f18\u52bf":61,"\u5145\u5206\u5c55\u73b0\u82f1\u7279\u5c14\u5e73\u53f0\u7684\u4f18\u52bf":62,"\u5148\u5b8c\u6210\u5bf9\u6743\u91cd\u7684packing\u64cd\u4f5c":61,"\u5148\u5b9e\u73b0\u6a21\u578b\u63a8\u65ad\u7684api":66,"\u5171\u4eab\u5185\u5b58":62,"\u5171\u4eab\u540c\u4e00\u4e2a\u6743\u91cd":61,"\u5176\u4e2d":[65,82],"\u5176\u4ed6\u51fd\u6570\u5747\u8fd4\u56de":66,"\u5176\u4ed6\u7528\u6237\u7684fork\u7248\u672c\u5e93\u5e76\u4e0d\u9700\u8981\u4e25\u683c\u9075\u5b88":82,"\u5176\u8f6c\u6362\u6b21\u6570\u51cf\u5c11\u81f3":61,"\u5177\u4f53\u4f7f\u7528\u65b9\u6cd5\u4e3a":66,"\u5177\u4f53\u539f\u56e0\u53c2\u8003":66,"\u5177\u4f53\u53ef\u4ee5\u53c2\u8003mkl":62,"\u5177\u4f53\u5b9e\u73b0\u65b9\u5f0f\u6bd4\u5982":[61,62],"\u5177\u4f53\u7684\u5b8c\u6210\u72b6\u6001\u53ef\u4ee5\u53c2\u89c1":62,"\u5177\u4f53\u8bf7\u53c2\u8003":66,"\u5185\u90e8\u5b58\u50a8":62,"\u5185\u90e8\u9a71\u52a8python\u89e3\u91ca\u5668\u8fdb\u884c\u6a21\u578b\u914d\u7f6e\u89e3\u6790\u548c\u6570\u636e\u8bfb\u53d6":65,"\u518d\u5728\u6bcf\u4e00\u4e2aapi\u4e2d\u81ea\u5df1\u68c0\u67e5\u7c7b\u578b":65,"\u518d\u57fa\u4e8e":82,"\u518d\u628a\u5df2\u8f6c\u6362\u4e3apacked\u683c\u5f0f\u7684\u6570\u636e\u4f20\u9012\u7ed9\u90a3\u4e9b\u590d\u7528\u540c\u4e00\u6570\u636e\u7684gemm":61,"\u5199\u4ee3\u7801":65,"\u5199\u5165\u5feb\u7167\u6570\u636e":32,"\u51fd\u6570":[61,62],"\u51fd\u6570\u5373\u53ef\u5b8c\u6210\u8f6c\u6362":33,"\u51fd\u6570\u540d\u4e3a":66,"\u51fd\u6570\u547d\u540d":65,"\u5206\u522b\u4ee3\u8868\u8f93\u5165\u6570\u636e":62,"\u5206\u522b\u5bf9\u5e94capi":82,"\u5206\u5e03\u5f0f\u5b58\u50a8\u670d\u52a1":32,"\u5206\u652f":82,"\u5206\u652f\u4e00\u65e6\u5efa\u7acb":82,"\u5206\u652f\u4e2d":82,"\u5206\u652f\u4e3a\u5f00\u53d1":82,"\u5206\u652f\u4e3a\u6bcf\u4e00\u6b21release\u65f6\u5efa\u7acb\u7684\u4e34\u65f6\u5206\u652f":82,"\u5206\u652f\u4e3a\u7a33\u5b9a":82,"\u5206\u652f\u529f\u80fd\u7684\u5c01\u95ed":82,"\u5206\u652f\u5408\u5165":82,"\u5206\u652f\u5408\u5165master\u5206\u652f":82,"\u5206\u652f\u540c\u6b65\u4e3b\u7248\u672c\u5e93\u7684":82,"\u5206\u652f\u540d\u4e3a":82,"\u5206\u652f\u5b58\u5728\u7684\u65f6\u5019":82,"\u5206\u652f\u6d3e\u751f\u51fa\u65b0\u7684\u5206\u652f":82,"\u5206\u652f\u7684\u7248\u672c\u90fd\u662f\u7ecf\u8fc7\u5355\u5143\u6d4b\u8bd5\u548c\u56de\u5f52\u6d4b\u8bd5\u7684\u7248\u672c":82,"\u5206\u652f\u7684\u7248\u672c\u90fd\u7ecf\u8fc7\u5355\u5143\u6d4b\u8bd5":82,"\u5206\u7247":32,"\u5219\u4f7f\u7528\u542f\u52a8\u53c2\u6570\u5b9a\u4e49\u7684\u521d\u59cb\u5316\u65b9\u6cd5\u521d\u59cb\u5316\u53c2\u6570":32,"\u5219\u5ffd\u7565":32,"\u5219\u628a\u53e6\u4e00\u4e2a\u6162\u901f\u7684kill\u6389":32,"\u5219\u76f4\u63a5\u5f15\u5165\u53e6\u4e00\u79cd\u7c7b\u578b\u7684\u5934\u6587\u4ef6":66,"\u5219\u9700\u8981\u56de\u6eda\u5230\u4e0a\u4e00\u4e2a\u68c0\u67e5\u70b9":32,"\u521b\u5efa":62,"\u5220\u9664\u78c1\u76d8\u76ee\u5f55\u4e2d\u4e0d\u662f\u5f53\u524duuid\u7684\u5feb\u7167\u6587\u4ef6":32,"\u5230":32,"\u5230\u7b2c\u4e8c\u6b65":82,"\u524d\u540e\u7684\u7f51\u7edc\u6027\u80fd":61,"\u529f\u80fd":48,"\u529f\u80fd\u7684\u6b63\u786e\u6027\u5305\u62ec\u9a8c\u8bc1paddlepaddle\u76ee\u524d\u7684":82,"\u52a8\u6001\u5e93":65,"\u5305\u542b\u4e86\u67d0\u79cd\u7c7b\u578b\u7684\u7c7b\u578b\u5b9a\u4e49\u548c\u66b4\u9732\u7684\u5168\u90e8\u51fd\u6570":66,"\u5305\u62ec":[33,61,62],"\u5305\u62ec\u6743\u91cdw\u548c\u504f\u7f6eb":32,"\u5305\u62ecmkl":62,"\u534f\u540c\u5b8c\u6210releas":82,"\u5355\u4e2a\u503c":33,"\u5355\u70b9\u6545\u969c":32,"\u5373":66,"\u5373\u4f7f\u7528":66,"\u5373\u4f7f\u7528\u6237\u76f4\u63a5\u5f15\u7528\u67d0\u79cd\u7c7b\u578b\u7684\u5934\u6587\u4ef6":66,"\u5373\u4f7fc":66,"\u5373\u4f8b\u5982":66,"\u5373\u4fbfpaddl":66,"\u5373\u5b8c\u6210\u67d0\u4e00\u4e2a\u4efb\u52a1\u7684\u6700\u5c11\u51fd\u6570":66,"\u5373\u66b4\u9732":66,"\u5373\u8868\u793a\u4e0d\u9700\u8981\u8f6c\u6362":62,"\u5373\u8fd9\u4e2a\u52a8\u6001\u5e93\u662f\u4e0d\u4f9d\u8d56\u4e8e\u5176\u4ed6\u4efb\u4f55\u6587\u4ef6\u7684":65,"\u539f\u6765\u7684\u65b9\u6848":62,"\u53c2\u6570":65,"\u53c2\u8003":[48,65],"\u53c2\u8003\u4e0b\u56fe":82,"\u53c8\u53ef\u4ee5\u907f\u514d\u4e0d\u5fc5\u8981\u7684\u8f6c\u6362":62,"\u53cc\u5411\u9a8c\u8bc1":48,"\u53d1\u578b\u7248":82,"\u53d1\u5e03\u5230dockerhub":82,"\u53d1\u5e03docker\u955c\u50cf\u53ea\u9700\u8981\u5bf9\u81ea\u52a8push\u7684\u955c\u50cf\u6253\u4e0a":82,"\u53d8\u91cf\u6765\u533a\u5206layer\u7684\u5c5e\u6027":62,"\u53ea\u5bf9\u7279\u6b8a\u5728\u7ebf\u7cfb\u7edf\u8003\u8651\u4e24\u53f0\u4ee5\u4e0a\u540c\u65f6\u6545\u969c\u7684\u5bb9\u707e":32,"\u53ea\u66b4\u9732\u6982\u5ff5\u7684\u63a5\u53e3":66,"\u53ea\u80fd\u8c03\u7528paddle\u7684\u52a8\u6001\u5e93":65,"\u53ea\u9700\u8981\u6062\u590d\u8fd9\u53f0\u8282\u70b9":32,"\u53ef\u4ee5":82,"\u53ef\u4ee5\u51cf\u5c0f\u7cfb\u7edf\u590d\u6742\u6027":32,"\u53ef\u4ee5\u5728\u4efb\u4f55\u673a\u5668\u4e0a\u6267\u884c\u7684":65,"\u53ef\u4ee5\u5728\u6b64\u9875\u9762\u7684":82,"\u53ef\u4ee5\u628a\u672c\u5730\u7684\u6570\u636e\u4e0a\u4f20\u5230\u5b58\u50a8\u96c6\u7fa4\u4e2d":33,"\u53ef\u4ee5\u6709\u6548\u7684\u907f\u514dparamet":32,"\u53ef\u4ee5\u7528":48,"\u53ef\u4ee5\u7528\u4ee5\u4e0b\u6307\u4ee4":33,"\u53ef\u4ee5\u7ee7\u7eed\u5728\u81ea\u5df1\u7684\u529f\u80fd\u5206\u652f\u63d0\u4ea4\u4ee3\u7801":82,"\u53ef\u4ee5\u901a\u8fc7\u9636\u6bb5\u6027\u7684\u4fdd\u5b58\u6bcf\u4e2aparamet":32,"\u53ef\u80fd\u4f1a\u9020\u6210\u7f51\u7edc\u62e5\u585e":32,"\u53f3\u4fa7\u7684":82,"\u5404\u6b21\u524d\u5411\u4e4b\u95f4\u4e5f\u90fd\u4f7f\u7528\u4e86\u76f8\u540c\u7684\u6743\u91cd":61,"\u540c\u4e00\u6b21\u524d\u5411":61,"\u540c\u65f6":[61,62],"\u540c\u65f6\u4f1a\u5f00\u542fintel":62,"\u540c\u65f6\u518d\u5c06":82,"\u540c\u65f6\u53c8\u5c3d\u53ef\u80fd\u5c11\u7684\u727a\u7272mkl":62,"\u540c\u65f6\u63d0\u8d77":82,"\u540c\u65f6\u6570\u636e\u683c\u5f0f\u5c31\u662f":62,"\u540d\u5b57\u4fee\u9970":65,"\u540e\u5411":61,"\u540e\u5411\u65f6\u590d\u7528\u5df2\u7ecf\u8f6c\u6362\u8fc7\u7684\u6743\u91cd":61,"\u5411\u6307\u5b9a\u7684\u76ee\u5f55\u4e2d\u4e00\u4e2a\u65b0\u7684\u6587\u4ef6":32,"\u5411paddlepaddle\u7684\u4e3b\u7248\u672c\u5e93\u63d0\u4ea4":82,"\u5426\u5219\u5f97\u628apaddle\u9759\u6001\u5e93\u94fe\u63a5\u5230\u89e3\u91ca\u5668\u91cc":65,"\u542f\u52a8\u4e00\u4e2a\u65b0\u7684\u7ebf\u7a0b\u5f00\u59cb\u4fdd\u5b58\u68c0\u67e5\u70b9":32,"\u548c":[33,61,62,65,66,82],"\u548c\u672a\u6765\u53ef\u80fd\u8fd8\u4f1a\u7528\u5230":62,"\u548c\u79bb\u7ebf\u6570\u636e\u7684\u65b9\u5f0f":33,"\u54ea\u4e2atrainer\u5148\u5b8c\u6210block\u7684\u8bad\u7ec3":32,"\u56e0\u4e3a\u8fd9\u6837\u505a\u4e5f\u6ca1\u6cd5\u4fdd\u8bc1\u6d88\u9664\u968f\u673a\u6027":32,"\u56e0\u4e3aswig\u5728\u7b2c\u4e09\u65b9\u8bed\u8a00\u4e2d\u66b4\u9732\u7684\u51fd\u6570\u540d":65,"\u56e0\u6b64":61,"\u56fe\u50cf\u5206\u7c7b":82,"\u5728":[61,62,66,82],"\u5728\u4e00\u4e2a\u4e0d\u53ef\u4e2d\u65ad\u5e76\u7f3a\u5c11\u5907\u4efd\u7684\u8bad\u7ec3\u4efb\u52a1\u4e2d":32,"\u5728\u4e0a\u56fe\u4e2d\u663e\u793a\u4e86\u5728\u4e00\u4e2a\u5b9e\u9645\u751f\u4ea7\u73af\u5883\u4e2d\u7684\u5e94\u7528":33,"\u5728\u4f7f\u7528twine\u4e0a\u4f20\u4e4b\u524d":82,"\u5728\u51fa\u73b0\u5355\u70b9\u6545\u969c\u65f6":32,"\u5728\u5b9e\u73b0\u6bcf\u4e2a\u5b50\u7c7b\u7684\u65f6\u5019\u5c31\u4e0d\u9700\u8981\u5173\u5fc3\u5206\u652f\u7684\u4e8b\u60c5\u4e86":62,"\u5728\u5b9e\u73b0\u8fc7\u7a0b\u4e2d":66,"\u5728\u5bf9\u5e94\u7684":61,"\u5728\u5c42\u521d\u59cb\u5316\u7684\u65f6\u5019":61,"\u5728\u5f00\u59cb\u8bad\u7ec3\u4e4b\u524d":33,"\u5728\u5f02\u6784\u96c6\u7fa4\u4e2d":32,"\u5728\u5f15\u5165\u5176\u4ed6\u7c7b\u578b\u7684\u5934\u6587\u4ef6\u65f6":66,"\u5728\u5feb\u7167\u5199\u5165\u5b8c\u6210\u540e":32,"\u5728\u60a8\u7684\u5b9e\u9645\u73af\u5883\u4e2d":32,"\u5728\u6709\u666e\u901a\u7684cpu":62,"\u5728\u672c\u6587\u6863\u4e2d":48,"\u5728\u673a\u7fa4\u4e0a\u8fd0\u884c\u8f6c\u6362\u7a0b\u5e8f":33,"\u5728\u6837\u4f8b\u4e2d":66,"\u5728\u7528\u6237\u4f7f\u7528c":66,"\u5728\u7b2c\u4e8c\u4e2atab":82,"\u5728\u7ebf\u6a21\u578b\u9884\u6d4b\u670d\u52a1":33,"\u5728\u8bad\u7ec3\u7ed3\u675f\u7684\u65f6\u5019\u518d\u4fdd\u5b58\u4e3apaddlepaddle\u7684\u683c\u5f0f":62,"\u5728\u8bc4\u5ba1\u8fc7\u7a0b\u4e2d":82,"\u5728\u8fd9\u4e2a":82,"\u5728\u8fd9\u4e2a\u52a8\u6001\u5e93\u4e2d\u4e0d\u5d4c\u5165\u4efb\u4f55\u5176\u4ed6\u8bed\u8a00\u7684\u89e3\u91ca\u5668":65,"\u5728\u8fd9\u4e2a\u9636\u6bb5\u7684\u4ee3\u7801\u6b63\u5728\u7ecf\u5386\u56de\u5f52\u6d4b\u8bd5":82,"\u5728\u8fd9\u4e9b\u5934\u6587\u4ef6\u4e2d":66,"\u5728\u8fd9\u4e9b\u6587\u4ef6\u4e2d":66,"\u5728\u91cd\u6784\u524d\u7684paddlepaddle\u4e2d":62,"\u5728\u95ee\u9898\u672c\u8eab\u7684\u8ba1\u7b97\u91cf\u6bd4\u8f83\u5c0f\u7684\u65f6\u5019":61,"\u5728batch":61,"\u5728c":65,"\u5728c\u7684\u5934\u6587\u4ef6":65,"\u5728packing\u4e0a\u7684\u8017\u65f6":61,"\u5728paddle\u4e4b\u4e0a\u8fd0\u884c\u7684\u6df1\u5ea6\u5b66\u4e60\u8bad\u7ec3\u8f93\u51fa\u7684\u6a21\u578b\u4f1a\u63d0\u4f9b\u7ed9\u5728\u7ebf\u4eba\u8138\u8bc6\u522b\u7684\u5e94\u7528\u4f7f\u7528":33,"\u5728paramet":32,"\u5728rnn\u7684\u60c5\u51b5\u4e0b":61,"\u5747\u4f1a\u88ab\u5b89\u88c5\u5230includ":66,"\u5747\u662f\u5728":66,"\u57fa\u4e8e\u7c98\u6027\u4f1a\u8bdd\u7684\u8d1f\u8f7d\u5747\u8861\u529f\u80fd":48,"\u5916\u90e8\u5b58\u50a8":62,"\u591a\u4e2a\u503c":33,"\u591a\u4e2aparamet":32,"\u591a\u6b21\u8c03\u7528":61,"\u5927\u591a\u6570\u8bed\u8a00\u90fd\u652f\u6301\u4f7f\u7528c\u8bed\u8a00api":65,"\u5982\u56fe\u4e2dtrainer":32,"\u5982\u679c\u4e0a\u9762\u4e24\u6b65\u51fa\u73b0\u9519\u8bef":32,"\u5982\u679c\u4e0d\u9700\u8981\u5916\u90e8\u5b58\u50a8\u7528\u4e8e\u8f6c\u6362":62,"\u5982\u679c\u4f7f\u7528swig\u6211\u4eec\u9700\u8981\u5c06\u5728interface\u6587\u4ef6\u91cc":65,"\u5982\u679c\u5728\u4f7f\u7528mkl":62,"\u5982\u679c\u5931\u8d25":82,"\u5982\u679c\u5b58\u5728\u6570\u636e\u6392\u5217\u683c\u5f0f\u4e0d\u4e00\u6837\u7684\u60c5\u51b5\u65f6":62,"\u5982\u679c\u5b58\u5728\u67d0\u4e9btrainer\u6267\u884c\u901f\u5ea6\u8fc7\u6162\u4f1a\u5f71\u54cd\u6574\u4f53\u96c6\u7fa4\u7684\u901f\u5ea6":32,"\u5982\u679c\u5df2\u7ecf\u6b63\u5728\u6267\u884c\u4fdd\u5b58\u68c0\u67e5\u70b9\u7684\u7ebf\u7a0b":32,"\u5982\u679c\u662f\u5176\u5b83\u7c7b\u578b":33,"\u5982\u679c\u6709bugfix\u7684\u884c\u4e3a":82,"\u5982\u679c\u67d0\u4e00\u4e2a\u7c7b\u578b\u9700\u8981\u5f15\u7528\u53e6\u4e00\u4e2a\u7c7b\u578b":66,"\u5982\u679c\u67d0\u4e00\u4e2apaddl":66,"\u5982\u679c\u67d0\u4e00\u4e2apaddle\u6982\u5ff5\u5fc5\u987b\u8981\u66b4\u9732":66,"\u5982\u679c\u6ee1\u8db3\u6761\u4ef6":32,"\u5982\u679c\u7528\u6237\u8981\u628apaddle\u7684\u9759\u6001\u5e93":65,"\u5982\u679c\u8981\u4e0a\u4f20gpu\u7248\u672c\u7684\u5305":82,"\u5982\u679c\u8c03\u7528\u9759\u6001\u5e93\u53ea\u80fd\u5c06\u9759\u6001\u5e93\u4e0e\u89e3\u91ca\u5668\u94fe\u63a5":65,"\u5982\u679c\u9700\u8981\u624b\u52a8\u7f16\u8bd1":82,"\u5982\u679cmkl":62,"\u5982\u679cparamet":32,"\u5b50\u7c7b\u53ea\u9700\u8981\u4f7f\u7528\u5b9a\u4e49\u597d\u7684\u63a5\u53e3":62,"\u5b57\u6bb5\u8bbe\u4e3a":82,"\u5b57\u7b26\u4e32":33,"\u5b58\u50a8":33,"\u5b66\u4e60\u6210\u672c\u9ad8":65,"\u5b83\u4eec\u4e3b\u8981\u662f\u7528\u4e8e":62,"\u5b83\u4eec\u7684\u6587\u4ef6\u540d\u662f":33,"\u5b83\u53ea\u4f1a\u5305\u62ec\u751f\u6210\u597d\u7684\u52a8\u6001\u5e93\u548c\u5934\u6587\u4ef6":62,"\u5b83\u8d1f\u8d23\u51b3\u5b9a\u7f16\u8bd1\u65f6\u662f\u5426\u4f7f\u7528mklml\u548cmkl":62,"\u5b89\u88c5\u540e\u7684\u76ee\u5f55\u7ed3\u6784\u4e3a":66,"\u5b8c\u6210\u4e00\u4e2a\u4f20\u8f93\u52a8\u4f5c\u5b8c\u6210\u7684\u65f6\u95f4\u4e5f\u6bd4\u8f83\u77ed":48,"\u5b8c\u6210\u5e38\u7528layer\u7684mkl":62,"\u5b8c\u6210\u5e38\u89c1\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edcvgg":62,"\u5b8c\u6210\u6570\u636e\u7684\u9884\u5904\u7406":33,"\u5b8c\u6210\u81ea\u52a8\u5316\u4e8c\u8fdb\u5236\u7f16\u8bd1":82,"\u5b9a\u4e49":62,"\u5b9a\u4e49\u4e00\u4e9b\u9664\u4e86layer\u548cmemory\u76f8\u5173\u7684\u7c7b\u548c\u51fd\u6570":62,"\u5b9e\u73b0\u5177\u4f53\u7684\u51fd\u6570\u529f\u80fd\u5373\u53ef":62,"\u5b9e\u73b0\u7b80\u5355":65,"\u5bf9\u4e8e\u4e0d\u540c\u8bed\u8a00":65,"\u5bf9\u4e8e\u540c\u4e00\u6bb5c":65,"\u5bf9\u4e8e\u540c\u6837\u8bbe\u7f6e\u7684\u7f51\u7edc\u6a21\u578b":61,"\u5bf9\u4e8e\u591a\u8bed\u8a00\u63a5\u53e3":65,"\u5bf9\u4e8e\u5927\u591a\u6570\u8bed\u8a00":65,"\u5bf9\u4e8e\u5e8f\u5217\u957f\u5ea6":61,"\u5bf9\u4e8e\u6709\u53c2\u6570\u7684\u5c42":62,"\u5bf9\u4e8e\u6bcf\u4e00\u4e2a\u65b0\u52a0\u7684rnn":61,"\u5bf9\u4e8e\u6bcf\u79cd\u7c7b\u578b":66,"\u5bf9\u4e8e\u6bcf\u79cdc":66,"\u5bf9\u65b0\u7684\u6743\u91cd\u8fdb\u884c\u8f6c\u6362\u7528\u4e8e\u4e0b\u6b21\u8fed\u4ee3":61,"\u5bf9\u6bd4":65,"\u5bf9\u6bd4\u4f18\u5316\u540elayer\u4e0e\u76f8\u5bf9\u5e94\u7684paddlepaddle\u539f\u6709lay":61,"\u5bf9\u6bd4\u4f18\u5316\u540elayer\u81ea\u8eab":61,"\u5bf9\u8f93\u5165\u53c2\u6570\u7684\u5b89\u5168\u6027\u8fdb\u884c\u4e86\u5fc5\u8981\u7684\u5224\u65ad":66,"\u5bf9\u8fd9\u4e2a\u7248\u672c\u7684\u63d0\u4ea4":82,"\u5bfb\u627e\u6709\u6ca1\u6709\u5176\u4ed6\u53ef\u4ee5\u4f18\u5316\u7684\u53ef\u80fd":62,"\u5bfc\u51fa\u8fd9\u4e9b\u63a5\u53e3":66,"\u5c06":82,"\u5c06\u4e00\u4e2a\u795e\u7ecf\u7f51\u7edc\u53c2\u6570\u62c6\u5206\u6210\u591a\u4efd":32,"\u5c06\u5927\u91cf\u7684":65,"\u5c06\u65b0\u5206\u652f\u7684\u7248\u672c\u6253\u4e0atag":82,"\u5c06master\u5206\u652f\u7684\u5408\u5165commit\u6253\u4e0atag":82,"\u5c0f\u4e8e\u67d0\u4e2a\u6bd4\u8f83\u5c0f\u7684\u9608\u503c\u8ba4\u4e3a\u901a\u8fc7":62,"\u5c31\u9700\u8981\u5bf9\u8fd9\u4e2a\u7b2c\u4e09\u65b9\u8bed\u8a00\u589e\u52a0\u4e00\u4e9b\u5b9a\u4e49":65,"\u5de5\u5177\u4e0a\u4f20\u5373\u53ef":82,"\u5e73\u5747\u6545\u969c\u4fee\u590d\u65f6\u95f4":32,"\u5e73\u5747\u6545\u969c\u7387":32,"\u5e76\u4e14\u4f1a\u5199\u597d":62,"\u5e76\u4e14\u4f7f\u7528":66,"\u5e76\u4e14\u53ea\u9700\u8981\u5728\u5fc5\u8981\u7684\u65f6\u5019\u8f6c\u6362\u8fd9\u79cd\u683c\u5f0f":62,"\u5e76\u4e14\u5728\u5e38\u89c1\u7684\u5e73\u53f0\u4e0a":65,"\u5e76\u4e14\u5f53\u7f16\u8bd1\u65f6":61,"\u5e76\u4e14\u628a\u7cfb\u7edf\u751f\u6210\u7684ca":48,"\u5e76\u4e14\u628a\u7ed3\u679c\u8fd4\u56depfsclient\u7aef":48,"\u5e76\u4e14\u8ba9\u63a5\u53e3\u8131\u79bb\u5b9e\u73b0\u7ec6\u8282":65,"\u5e76\u4e14\u8f93\u5165\u8f93\u51fa\u90fd\u662f\u5171\u7528\u4e00\u5757\u5185\u5b58":62,"\u5e76\u5220\u9664":82,"\u5e76\u5220\u9664\u66f4\u65e9\u7684\u5feb\u7167":32,"\u5e76\u52a0\u8f7d\u5176\u4e2d\u7684\u53c2\u6570":32,"\u5e76\u53d1\u5e03\u5230pypi":82,"\u5e76\u5728\u6bcf\u6b21\u6743\u91cd\u66f4\u65b0\u540e":61,"\u5e76\u5728\u96c6\u7fa4\u4e2d\u8fd0\u884c\u591a\u4e2a\u5206\u5e03\u5f0f\u6570\u636e\u5904\u7406\u4efb\u52a1":33,"\u5e76\u5c06":82,"\u5e76\u5c06c":66,"\u5e76\u628a\u5feb\u7167\u4fdd\u5b58\u5230\u8fd9\u4e2a\u76ee\u5f55\u4e0b":32,"\u5e76\u628a\u7ed3\u679c\u653e\u5230\u5f53\u524d\u5c42\u7684":62,"\u5e76\u6ca1\u6709paddle\u7279\u522b\u9700\u8981\u7684\u7279\u6027":65,"\u5e76\u6dfb\u52a0\u5934\u6587\u4ef6":61,"\u5e76\u88ab\u5b58\u50a8\u5728\u8bf8\u5982hadoop":33,"\u5e76\u9002\u5e94github\u7684\u7279\u6027\u505a\u4e86\u4e00\u4e9b\u533a\u522b":82,"\u5e76\u91cd\u65b0\u6253\u5305wheel\u5305":82,"\u5efa\u8bae":82,"\u5f00\u53d1\u4e86\u6a21\u578b\u9884\u6d4b\u7684\u6837\u4f8b\u4ee3\u7801":66,"\u5f00\u53d1\u8005\u4fee\u6539\u81ea\u5df1\u7684\u4ee3\u7801":82,"\u5f00\u53d1\u8005fork\u7684\u7248\u672c\u5e93\u4e2d":82,"\u5f00\u53d1\u8005fork\u7684\u7248\u672c\u5e93\u4f7f\u7528":82,"\u5f00\u5934":[61,62],"\u5f00\u59cb\u63d0\u4f9b\u670d\u52a1":32,"\u5f15\u5165\u4e86\u4ee5\u4e0b\u56db\u4e2aapi":61,"\u5f15\u5165\u4e86\u7c7b\u578b\u7684\u5934\u6587\u4ef6":66,"\u5f39\u51fa\u4e0b\u9762\u7684\u9009\u62e9\u6846":82,"\u5f53\u529f\u80fd\u5206\u652f\u5f00\u53d1\u5b8c\u6bd5\u540e":82,"\u5f53\u53ea\u505a\u63a8\u65ad":61,"\u5f53\u5f00\u542f":62,"\u5f53\u6253\u5f00":62,"\u5f53\u6570\u636e\u683c\u5f0f\u4e0epaddlepaddle\u9ed8\u8ba4\u7684":62,"\u5f53\u7136\u8fd9\u4e24\u8005\u4e5f\u53ef\u4ee5\u76f8\u7b49":62,"\u5f53\u7528\u6237\u4f7f\u7528\u5b8c\u8fd9\u4e2a\u53c2\u6570\u540e":66,"\u5f53\u7f51\u7edc\u51fa\u73b0\u5206\u652f\u4e14\u5728":62,"\u5f53destination\u6587\u4ef6\u4e0d\u5b58\u5728\u6216\u8005\u5927\u5c0f\u548csource\u6587\u4ef6\u4e0d\u4e00\u81f4\u65f6":48,"\u5f88\u96be\u4fdd\u8bc1\u591a\u8bed\u8a00\u4ee3\u7801\u98ce\u683c\u7684\u4e00\u81f4\u6027":65,"\u5f97\u4f7f\u7528":65,"\u5fc5\u8981":66,"\u5fc5\u987b\u5206\u522b\u4e0e":62,"\u60c5\u611f\u5206\u6790":82,"\u6211\u4eec\u4e5f\u53ef\u4ee5\u786e\u5b9a\u6bcf\u4e00\u4e2a\u53c2\u6570\u7684\u7c7b\u578b":66,"\u6211\u4eec\u4e5f\u5c06mklml\u5373":62,"\u6211\u4eec\u4f1a\u4fdd\u8bc1":62,"\u6211\u4eec\u4f1a\u5728\u7f51\u7edc\u8bad\u7ec3\u4e4b\u524d\u628a\u683c\u5f0f\u8f6c\u6362\u4e3amkl":62,"\u6211\u4eec\u4f1a\u5bf9\u6bd4\u5982\u4e0b2\u4e2a\u65b9\u9762":61,"\u6211\u4eec\u4f1a\u628amkl":62,"\u6211\u4eec\u4f1a\u6dfb\u52a0":[61,62],"\u6211\u4eec\u4f7f\u7528\u52a8\u6001\u5e93\u6765\u5206\u53d1paddl":65,"\u6211\u4eec\u51b3\u5b9a\u4f7f\u7528\u5df2\u6709\u7684":62,"\u6211\u4eec\u53ef\u4ee5\u5148\u5b8c\u6210\u5bf9\u539f\u6570\u636e\u7684packing\u64cd\u4f5c":61,"\u6211\u4eec\u603b\u7ed3\u51fa\u4e00\u4e9b\u7279\u522b\u9700\u8981\u6ce8\u610f\u7684\u70b9":62,"\u6211\u4eec\u63d0\u4f9b\u4e24\u4e2a\u8f6c\u6362\u65b9\u5f0f":33,"\u6211\u4eec\u63d0\u51fa\u4e86chunk\u7684\u6982\u5ff5":48,"\u6211\u4eec\u6700\u7ec8\u7684\u52a8\u6001\u5e93\u4e2d\u4e0d\u5d4c\u5165python\u6216\u8005\u5176\u4ed6\u4efb\u4f55\u8bed\u8a00\u7684\u89e3\u91ca\u5668":65,"\u6211\u4eec\u8ba1\u5212\u5c06":61,"\u6211\u4eec\u8ba1\u5212\u5c06\u82f1\u7279\u5c14\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\u6570\u5b66\u5e93":62,"\u6211\u4eec\u8bbe\u8ba1\u8bf4\u660e\u4e86\u540d\u4e3afilemanager\u7cfb\u7edf":48,"\u6211\u4eec\u9009\u62e9":33,"\u6211\u4eec\u901a\u8fc7\u4f7f\u7528\u65b0\u5f15\u5165\u7684gemm":61,"\u6211\u4eec\u90fd\u63d0\u4f9bpython\u7684\u8f6c\u6362\u5e93":33,"\u6216\u8005":[62,65,66],"\u6216\u8005\u5c06\u8fd9\u53f0\u8282\u70b9\u8fc1\u79fb\u5230\u53e6\u4e00\u4e2a\u8282\u70b9\u5e76\u542f\u52a8\u5373\u53ef\u6062\u590d\u8bad\u7ec3\u4efb\u52a1":32,"\u6216\u8005\u7528tuple\u8868\u793a\u7684\u591a\u4e2a\u503c":33,"\u6216\u8005\u7531\u5b83\u4eec\u7ec4\u6210\u7684list":33,"\u6216activ":62,"\u6240\u4ee5":[62,82],"\u6240\u4ee5\u4e00\u4e2a\u7248\u672c\u53f7\u7684wheel\u5305\u53d1\u5e03\u4e4b\u540e":82,"\u6240\u4ee5\u4e0d\u5b58\u5728\u8fd9\u4e2a\u95ee\u9898":62,"\u6240\u4ee5\u5728":62,"\u6240\u4ee5\u5728\u5199\u5165\u5feb\u7167\u7684\u8fc7\u7a0b\u4e2d":32,"\u6240\u4ee5\u6211\u4eec\u5b9a\u4e49\u4e86\u4e00\u4e2a":62,"\u6240\u4ee5\u6574\u4f53\u4e0a":62,"\u6240\u4ee5\u6dfb\u52a0\u4e86\u5bf9\u5e94\u7684":62,"\u6240\u4ee5\u7528\u6237\u9700\u8981\u9996\u5148\u5728":48,"\u6240\u4ee5\u9700\u8981\u5f15\u5165\u4e00\u4e2a\u8f6c\u6362\u65b9\u6cd5":62,"\u6240\u6709\u4e0e\u7c7b\u578b\u76f8\u5173\u7684\u51fd\u6570":66,"\u6240\u6709\u5916\u90e8\u7684\u8f6c\u6362\u5de5\u4f5c\u90fd\u4f1a\u5728reset\u7cfb\u5217\u51fd\u6570\u4e2d\u90fd\u51c6\u5907\u597d":62,"\u6240\u6709\u7684":61,"\u6240\u6709\u7684\u63a5\u53e3\u5747\u4e3ac\u63a5\u53e3":66,"\u6240\u6709\u76f8\u5173\u7684":61,"\u6240\u6709\u7c7b\u578b\u540d\u4e3a":66,"\u6240\u6709mkl":62,"\u624b\u5199\u591a\u8bed\u8a00\u7ed1\u5b9a":65,"\u624d\u80fd\u66f4\u597d\u7684\u53d1\u6325mkl":62,"\u6253\u5f00\u8fd9\u4e2a\u7f16\u8bd1\u9009\u9879":66,"\u6267\u884c":82,"\u628a":33,"\u628a\u4e4b\u524d\u793a\u4f8b\u4e2d\u8f6c\u6362\u5b8c\u6bd5\u7684random":33,"\u6307\u6df1\u5ea6\u5b66\u4e60\u8bad\u7ec3\u4e4b\u540e\u5f97\u5230\u7684\u6240\u6709\u53c2\u6570":32,"\u6309\u94ae":82,"\u63a5\u53e3":[65,66],"\u63a5\u53e3\u5c42\u505a\u8fc7\u591a\u5c01\u88c5":66,"\u63a5\u53e3\u662f":33,"\u63a5\u6536\u5904\u7406pfsclient\u7aef\u7684\u6587\u4ef6\u7ba1\u7406\u8bf7\u6c42":48,"\u63a7\u5236\u662f\u5426\u4f7f\u7528mkl":62,"\u63a7\u5236\u662f\u5426\u4f7f\u7528mklml\u5e93":62,"\u63a7\u5236\u7528\u6237\u6743\u9650":33,"\u63d0\u4f9b\u4e03\u5c42\u534f\u8bae\u7684\u53cd\u5411\u4ee3\u7406":48,"\u63d0\u4f9b\u5e38\u7528\u7684\u547d\u4ee4\u884c\u7ba1\u7406\u547d\u4ee4\u7ba1\u7406\u6587\u4ef6\u548c\u76ee\u5f55":48,"\u63d0\u4f9b\u7528\u6237\u7ba1\u7406\u6587\u4ef6\u7684\u547d\u4ee4":48,"\u63d0\u4f9b\u7ed9paddle\u4f5c\u4e3a\u8bad\u7ec3\u6570\u636e":33,"\u652f\u6301\u5927\u6587\u4ef6\u7684\u65ad\u70b9\u4e0a\u4f20":48,"\u6570\u636e":48,"\u6570\u636e\u8bfb\u53d6\u5747\u4ea4\u7531\u5176\u4ed6\u8bed\u8a00\u5b8c\u6210":65,"\u6570\u636e\u957f\u5ea6\u53ca\u6821\u9a8c\u503c\u7ec4\u6210":48,"\u6570\u636e\u96c6\u9700\u8981\u9884\u5148\u88ab\u8f6c\u6362\u6210paddlepaddle\u5206\u5e03\u5f0f\u8bad\u7ec3\u4f7f\u7528\u7684\u5b58\u50a8\u683c":33,"\u6570\u636e\u9884\u5904\u7406\u4efb\u52a1":33,"\u6587\u4ef6":65,"\u6587\u4ef6\u4f20\u8f93\u7684\u7684\u5173\u952e\u5728\u4e8e\u9700\u8981pfsclient\u7aef\u5bf9\u6bd4source\u548cdestination\u7684\u6587\u4ef6chunks\u7684checksum\u662f\u5426\u4fdd\u6301\u4e00\u81f4":48,"\u6587\u4ef6\u5185\u5bb9\u4e3a":65,"\u6587\u4ef6\u540d\u4e3a\u6b64uuid":32,"\u6587\u4ef6\u5bf9\u5e94\u7684data":33,"\u6587\u4ef6\u7684\u4e0a\u4f20\u548c\u4e0b\u8f7d\u90fd\u662f\u901a\u8fc7\u5bf9chunk\u7684\u64cd\u4f5c\u6765\u5b9e\u73b0\u7684":48,"\u65b0\u624b\u5165\u95e8\u7ae0\u8282":82,"\u65b0\u7248\u672c":62,"\u65b9\u4fbf\u6d4b\u8bd5\u4eba\u5458\u6d4b\u8bd5paddlepaddle\u7684\u884c\u4e3a":82,"\u65b9\u4fbf\u7528\u6237\u4e0a\u4f20\u81ea\u5df1\u7684\u8bad\u7ec3\u6570\u636e\u4ee5\u8fdb\u884c\u5206\u5e03\u5f0f\u8bad\u7ec3":48,"\u65b9\u4fbf\u7528\u6237\u5728python\u7aef\u9009\u62e9\u662f\u5426\u542f\u7528\u8fd9\u4e2a\u529f\u80fd":61,"\u65b9\u4fbf\u7528\u6237\u9009\u62e9\u4f7f\u7528mkl":62,"\u65b9\u5f0f\u7c7b\u4f3c\u4e8e":62,"\u65e0\u6cd5\u505a\u5230\u5bf9\u4e8e\u5404\u79cd\u8bed\u8a00\u9519\u8bef\u5904\u7406\u65b9\u5f0f\u7684\u9002\u914d":65,"\u65e0\u8bba\u5728\u672c\u5730\u8fd8\u662f\u5728\u4e91\u7aef":33,"\u65e0\u8bba\u662f\u4ece":33,"\u65e0\u8bba\u662f\u5728\u672c\u5730\u6216\u662f\u4e91\u7aef\u8f6c\u6362":33,"\u65e0\u8bba\u662f\u91cd\u6784\u524d\u7684layer\u8fd8\u662f\u91cd\u6784\u540e\u7684op":62,"\u65f6":[32,61,62],"\u65f6\u4e00\u8d77\u66f4\u65b0":62,"\u662f":[48,62],"\u662f\u4e00\u4e2a\u591a\u8bed\u8a00\u63a5\u53e3\u7684\u4ee3\u7801\u751f\u6210\u5668":65,"\u662f\u4e00\u4e2a\u7c7b\u578b\u7684\u6807\u5fd7":66,"\u662f\u4e0d\u5e38\u89c1\u7684\u505a\u6cd5":65,"\u662f\u5404\u4e2a\u5b9e\u73b0\u4e2d\u5171\u4eab\u7684\u5934\u6587\u4ef6":66,"\u662f\u5426\u6253\u5f00":61,"\u662f\u56e0\u4e3ac99\u652f\u6301":65,"\u662f\u5bf9\u7528\u6237\u6587\u4ef6\u5b58\u50a8\u7a7a\u95f4\u7684\u62bd\u8c61":48,"\u662f\u6307":66,"\u662f\u7528\u6237\u4f7f\u7528c":66,"\u662fc":66,"\u663e\u5f97\u76f8\u5bf9\u6765\u8bf4\u8f83\u4e3a\u8017\u65f6":61,"\u6682\u65f6\u4e0d\u8003\u8651\u591a\u4e2aparamet":32,"\u66b4\u9732\u8fd9\u4e2a\u6982\u5ff5\u5fc5\u8981\u51fd\u6570":66,"\u6700\u540e\u5220\u9664":82,"\u6700\u5e38\u89c1\u7684\u9519\u8bef\u5904\u7406\u65b9\u5f0f\u662fexcept":65,"\u6709\u6548\u63d0\u5347paddlepaddle\u5728\u82f1\u7279\u5c14\u67b6\u6784\u4e0a\u7684\u6027\u80fd":[61,62],"\u6709\u6807\u51c6\u7684":65,"\u6709\u7684\u65f6\u5019":65,"\u672c\u5217\u8868\u8bf4\u660epaddlepaddle\u53d1\u7248\u4e4b\u524d\u9700\u8981\u6d4b\u8bd5\u7684\u529f\u80fd\u70b9":82,"\u672c\u6587\u6863\u63cf\u8ff0paddl":66,"\u673a\u5668\u7ffb\u8bd1":82,"\u6765\u4fdd\u8bc1\u8bad\u7ec3\u8fc7\u7a0b\u53ef\u4ee5\u4ece\u4e2d\u95f4\u72b6\u6001\u91cd\u65b0\u542f\u52a8":32,"\u6765\u51b3\u5b9a\u662f\u5426\u5f00\u542fmkl":61,"\u6765\u5b9e\u73b0":62,"\u6765\u786e\u4fdd\u628a":65,"\u6765\u8868\u793apaddle\u5185\u90e8\u7c7b":65,"\u6765\u8bbf\u95ee\u7528\u6237\u81ea\u5df1\u7684\u6570\u636e":33,"\u6765\u8fdb\u884c\u8ba8\u8bba":66,"\u67e5\u770blatest":82,"\u6807\u51c6\u8868\u793apaddlepaddle\u7248\u672c\u53f7":82,"\u683c\u5f0f\u4e0d\u5339\u914d\u65f6":62,"\u68c0\u67e5\u70b9\u4fdd\u5b58\u7a0b\u5e8f\u6d41\u7a0b":32,"\u6a21\u578b\u53c2\u6570\u68c0\u67e5\u70b9\u901a\u8fc7\u5b9a\u671f\u5411\u78c1\u76d8\u4e0a\u4fdd\u5b58\u4e00\u4efd\u5b58\u50a8\u5728paramet":32,"\u6a21\u578b\u6570\u636e\u68c0\u67e5\u70b9\u7684\u5b9e\u73b0":32,"\u6a21\u578b\u914d\u7f6e\u89e3\u6790":65,"\u6b21\u8fed\u4ee3\u6267\u884c\u7684\u8f6c\u6362\u6b21\u6570\u4e3a":61,"\u6b64\u65f6\u6bcf\u4e2a\u5c0f\u5206\u652f\u7684":62,"\u6b64\u65f6master\u5c06\u8d1f\u8d23\u542f\u52a8\u4e00\u4e2a\u65b0\u7684train":32,"\u6bcf\u4e00\u4e2a":82,"\u6bcf\u4e00\u4e2a\u6587\u4ef6\u662f\u6570\u636e\u96c6\u7684\u4e00\u4e2ashard":33,"\u6bcf\u4e2a":62,"\u6bcf\u4e2a\u503c\u7684\u7c7b\u578b\u53ef\u4ee5\u662f\u6574\u5f62":33,"\u6bcf\u4e2a\u6d4b\u8bd5\u4f1a\u5bf9\u6bd4paddlepaddle\u4e2dcpu\u7b97\u51fa\u7684\u7ed3\u679c\u4e0emkl":62,"\u6bcf\u4e2adata":33,"\u6bcf\u4e2amkldnnlayer\u90fd\u5305\u542b\u7528\u4e8e\u5185\u90e8\u5b58\u50a8\u548c\u5916\u90e8\u5b58\u50a8\u7684\u4e00\u7cfb\u5217mkldnnmatrix":62,"\u6bcf\u4e2aparamet":32,"\u6bcf\u4e2ashard\u5206\u522b\u5b58\u50a8\u5728\u5176\u4e2d\u4e00\u53f0paramet":32,"\u6bcf\u6b21\u8c03\u7528\u65f6\u5bf9\u539f\u6570\u636e\u7684\u91cd\u590dpacking\u4fbf\u6210\u4e3a\u4e86\u5197\u4f59":61,"\u6bcf\u6b21\u8f93\u51fa\u4e00\u4e2adata":33,"\u6bcf\u969410\u5206\u949f":32,"\u6bd4\u5982":[33,62],"\u6bd4\u5982\u53ef\u80fd\u4f1a\u7528openmp\u6539\u8fdbsgd\u7684\u66f4\u65b0\u6027\u80fd":62,"\u6bd4\u5982\u5c06":82,"\u6bd4\u5982\u6bcf\u969410\u5206\u949f\u6700\u65b0\u7684\u5feb\u7167":32,"\u6bd4\u5982\u6d41\u5f0f\u6570\u636e\u5904\u7406":33,"\u6bd4\u5982imagenet\u8fd9\u4e2a\u6570\u636e\u96c6\u53ef\u80fd\u88ab\u5206\u62101000\u4e2ashard":33,"\u6ca1\u6709\u5fc5\u8981\u5728\u6bcf\u6b21\u524d\u5411\u4e2d\u6bcf\u4e2a\u65f6\u95f4\u6b65\u7684\u8ba1\u7b97\u65f6\u5bf9\u6743\u91cd\u8fdb\u884c\u91cd\u590d\u7684packing\u64cd\u4f5c":61,"\u6ce8":[32,82],"\u6ce8\u518clayer\u7684\u65f6\u5019\u4fdd\u8bc1":[61,62],"\u6ce8\u610f":62,"\u6d4b\u8bd5\u5206\u4e3a\u6bcf\u4e2alayer":62,"\u6d4b\u8bd5\u672c\u6b21release\u7684\u6b63\u786e\u6027":82,"\u6d4b\u8bd5\u7684\u6027\u80fd\u5bf9\u6bd4\u7ed3\u679c\u4f1a\u5728":62,"\u6d6e\u70b9\u578b\u6570\u636e":33,"\u6df1\u5165paddlepaddl":62,"\u6dfb\u52a0":61,"\u6dfb\u52a0\u7684\u76f8\u5173\u6587\u4ef6\u548c\u76ee\u5f55\u7ed3\u6784\u5982\u4e0b":[61,62],"\u6fc0\u6d3b\u51fd\u6570\u662f\u72ec\u7acb\u4e8e":62,"\u70b9\u51fb":82,"\u7136\u540e\u5728\u524d\u5411":61,"\u7136\u540e\u5728etcd\u7684":32,"\u7136\u540e\u5c31\u53ef\u4ee5\u5e76\u53d1\u5199\u5165\u591a\u4e2achunk":48,"\u7136\u540e\u624d\u80fd\u4f7f\u7528pfsclient":48,"\u7136\u540e\u6309\u7167\u4e0a\u8ff0\u7684\u65b9\u6cd5":82,"\u7136\u540e\u70b9\u51fb":82,"\u7248\u672c\u5206\u652f":82,"\u7248\u672c\u53f7":82,"\u7248\u672c\u53f7\u5bf9\u5e94\u7684tag\u5373\u53ef":82,"\u7248\u672c\u53f7rc":82,"\u7248\u672cfork\u51fa\u81ea\u5df1\u7684\u529f\u80fd\u5206\u652f":82,"\u7279\u6709\u7684\u8bbe\u5907id":62,"\u73b0\u9636\u6bb5\u7684\u4f18\u5316\u4e3b\u8981\u9488\u5bf9":61,"\u73b0\u9636\u6bb5paddle\u6709\u4e00\u4e2a\u95ee\u9898\u662f":65,"\u751f\u4ea7\u73af\u5883\u4e2d\u7684\u8bad\u7ec3\u6570\u636e\u96c6\u901a\u5e38\u4f53\u79ef\u5f88\u5927":33,"\u751f\u4ea7\u73af\u5883\u7684\u65e5\u5fd7\u6570\u636e\u4f1a\u901a\u8fc7\u5b9e\u65f6\u6d41\u7684\u65b9\u5f0f":33,"\u751f\u6210\u5404\u79cd\u8bed\u8a00\u7684\u7ed1\u5b9a\u4ee3\u7801":65,"\u751f\u6210\u6587\u6863":65,"\u751f\u6210\u7684":33,"\u751f\u6210\u7ed9\u5b9a":33,"\u751f\u6210api\u6587\u6863":65,"\u751f\u6210pfsclient\u548cpfsserver\u7684\u6846\u67b6\u90e8\u5206":48,"\u7528":48,"\u7528\u4e8e\u6d4b\u8bd5\u548c\u5bf9\u6bd4\u5728\u4f7f\u7528mkl":62,"\u7528\u4e8e\u7ba1\u7406mkl":62,"\u7528\u4e8e\u9009\u62e9\u662f\u5426\u4f7f\u7528\u76f8\u5173\u529f\u80fd":61,"\u7528\u4e8e\u9009\u62e9\u662f\u5426\u4f7f\u7528mkl":62,"\u7528\u4e8emkl":[61,62],"\u7528\u6237\u4e0a\u4f20\u6570\u636e\u540e":33,"\u7528\u6237\u4e5f\u53ef\u4ee5\u4e0a\u4f20label":33,"\u7528\u6237\u53ef\u4ee5\u5b89\u5168\u7684\u91ca\u653e\u67d0\u4e2ac":66,"\u7528\u6237\u53ef\u4ee5\u628a\u81ea\u5df1\u7684\u6570\u636e\u5206\u4eab\u7ed9\u522b\u4eba":33,"\u7528\u6237\u53ef\u4ee5\u76f4\u63a5\u4f7f\u7528\u8fd9\u4e2a\u52a8\u6001\u5e93\u6765\u5f15\u5165paddl":66,"\u7528\u6237\u5728\u672c\u5730\u8f6c\u6362\u597d\u518d\u4e0a\u4f20":33,"\u7528\u6237\u6587\u4ef6\u53ef\u80fd\u662f\u6bd4\u8f83\u5927\u7684":48,"\u7528\u6237\u901a\u8fc7c":66,"\u7531\u4e8e\u5728\u73b0\u6709\u7684\u67d0\u4e9b\u60c5\u51b5\u4e0b":61,"\u7531\u4e8e\u5bf9parameters\u7684\u66f4\u65b0\u9700\u8981\u83b7\u53d6parameters\u5185\u5b58\u7684":32,"\u7531\u4e8e\u96c6\u7fa4\u4e2d\u540c\u65f6\u5b58\u5728\u4e24\u53f0\u673a\u5668\u6545\u969c\u7684\u6982\u7387\u6781\u4f4e":32,"\u7531\u4e8ec":65,"\u7531\u4e8echunk\u6bd4\u8f83\u5c0f":48,"\u7531\u4e8emkl":62,"\u7531\u4e8epypi":82,"\u7531\u5206\u652f\u5904\u7684layer\u8d1f\u8d23\u6c42\u548c":62,"\u7533\u8bf7\u7528\u6237\u7a7a\u95f4":48,"\u7684\u4e00\u4e2a\u5b50\u96c6":62,"\u7684\u4fe1\u606f":62,"\u7684\u5355\u5143\u6d4b\u8bd5\u548c\u7b80\u5355\u7f51\u7edc\u7684\u6574\u4f53\u6d4b\u8bd5":62,"\u7684\u547d\u540d\u98ce\u683c\u5e76\u4e0d\u80fd\u9002\u5e94\u5176\u4ed6\u7b2c\u4e09\u65b9\u8bed\u8a00":65,"\u7684\u57fa\u672c\u903b\u8f91":62,"\u7684\u5934\u6587\u4ef6":65,"\u7684\u5b50\u7c7b\u53ea\u9700\u8981\u4f7f\u7528\u5185\u90e8\u5b58\u50a8\u5c31\u53ef\u4ee5\u4e86":62,"\u7684\u60c5\u51b5\u4e0b":61,"\u7684\u63a5\u53e3\u6837\u5f0f":65,"\u7684\u6570\u636e\u6d41\u56fe":33,"\u7684\u65f6\u5019":62,"\u7684\u683c\u5f0f\u59cb\u7ec8\u662f":62,"\u7684\u683c\u5f0f\u5b58\u50a8":62,"\u7684\u6982\u5ff5":62,"\u7684\u6e90\u7801\u91cc\u4f7f\u7528\u4e86":65,"\u7684\u7248\u672c":82,"\u7684\u7ed3\u679c":61,"\u7684\u7f29\u5199":48,"\u7684\u7f51\u7edc\u6a21\u578b":61,"\u7684\u89c4\u8303":65,"\u7684\u89d2\u5ea6":33,"\u7684\u914d\u7f6e\u5199\u5230\u914d\u7f6e\u6587\u4ef6\u4e2d":33,"\u7684flag":[61,62],"\u7684vanilla":61,"\u76ee\u524d\u53ea\u8003\u8651":62,"\u76ee\u524d\u53ea\u8003\u8651\u52a8\u6001\u6269\u5bb9trainer\u6570\u91cf":32,"\u76ee\u524d\u5728paddlepaddle\u4e2d":62,"\u76ee\u524d\u5728paddlepaddle\u4e2d\u6570\u636e\u90fd\u662f\u4ee5":62,"\u76ee\u524d\u5d4c\u5165python\u89e3\u91ca\u5668":65,"\u76ee\u524d\u6211\u4eec\u7528cephfs\u6765\u642d\u5efa":48,"\u76ee\u524d\u7684\u4f18\u5316":62,"\u76ee\u524dpaddle\u7684\u8fdb\u7a0b\u6a21\u578b\u662fc":65,"\u76ee\u524dpaddlepaddle\u91c7\u7528\u4e86":61,"\u76ee\u5f55\u4e0b":66,"\u76ee\u5f55\u4e0b\u5bf9\u5e94\u7684\u5730\u65b9":62,"\u76f4\u63a5\u4f7f\u7528c\u8bed\u8a00\u7684":65,"\u76f4\u63a5\u5220\u9664\u8fd9\u4e2a\u53c2\u6570\u5373\u53ef":66,"\u76f4\u63a5\u5bfc\u51fa\u5230c\u7684\u63a5\u53e3\u6bd4\u8f83\u56f0\u96be":65,"\u76f8\u5173\u5c42":61,"\u77e9\u9635\u5927\u5c0f\u662f":61,"\u793e\u533a\u53c2\u4e0e\u56f0\u96be":65,"\u793e\u533a\u8d21\u732e\u4ee3\u7801\u5b66\u4e60\u6210\u672c\u9ad8":65,"\u795e\u7ecf\u7f51\u7edc\u4e2d\u7684\u53c2\u6570":32,"\u79bb\u7ebf\u6279\u5904\u7406":33,"\u7b2c\u4e00\u4e2atag\u4e3a":82,"\u7b2c\u4e09\u6b65\u5b8c\u6210\u540e":82,"\u7b2c\u4e8c\u4e2a\u4e3a":82,"\u7b49":[62,66],"\u7b49\u5168\u90e8\u9759\u6001\u5e93\u4e2d\u7684\u76ee\u6807\u6587\u4ef6\u5168\u90e8\u6253\u5305\u540e\u4ea7\u751f\u7684\u6587\u4ef6":66,"\u7b49\u5f85\u7f16\u8bd1\u5b8c\u6210\u540e":82,"\u7b49\u6587\u4ef6":66,"\u7c7b\u4f3c":66,"\u7c7b\u540d\u548cc":65,"\u7c7b\u578b":65,"\u7ed3\u8bba":65,"\u7edf\u4e00\u7528":33,"\u7f16\u8bd1\u5668\u6ca1\u6709":65,"\u7f16\u8bd1\u578b\u8bed\u8a00":65,"\u7f16\u8bd1\u65f6\u4f1a\u628a\u5bf9\u5e94\u7684\u5934\u6587\u4ef6\u548c\u5e93\u653e\u5728":62,"\u7f16\u8bd1\u8fd9\u4e2a\u7248\u672c\u7684docker\u53d1\u884c\u955c\u50cf":82,"\u7f16\u8bd1\u8fd9\u4e2a\u7248\u672c\u7684python":82,"\u7f16\u8bd1c":66,"\u800c\u4e0d\u5fc5\u5728\u610fpaddl":66,"\u800c\u4e0d\u652f\u6301pypy\u89e3\u91ca\u5668":65,"\u800c\u4e0d\u66b4\u9732\u6982\u5ff5\u7684\u5b9e\u73b0":66,"\u800c\u4e14\u5728\u4f20\u8f93\u7684\u8fc7\u7a0b\u4e2d\u4e5f\u53ef\u80fd\u51fa\u73b0\u7f51\u7edc\u4e0d\u7a33\u5b9a\u7684\u60c5\u51b5":48,"\u800c\u51fa\u73b0\u9636\u6bb5\u6027\u7684\u8fd0\u884c\u505c\u6ede":32,"\u800c\u5728cpp\u91cc\u9762\u5b9e\u73b0\u8fd9\u4e2ac\u7684\u63a5\u53e3":65,"\u800c\u591a\u8bed\u8a00\u63a5\u53e3\u9700\u8981\u76f4\u63a5\u8bfb\u53d6\u751f\u6210\u7684\u4e8c\u8fdb\u5236":65,"\u800c\u5bf9\u4e8egolang":65,"\u800c\u5bf9\u4e8egolang\u9519\u8bef\u5904\u7406\u5e94\u8be5\u4f7f\u7528\u8fd4\u56de\u503c":65,"\u800c\u662f\u76f4\u63a5\u4fee\u6539paddl":66,"\u800c\u662f\u76f4\u63a5\u7528api\u7684\u63a5\u53e3\u8fdc\u7a0b\u8bbf\u95ee":33,"\u800cswig\u53ea\u80fd\u7b80\u5355\u7684\u66b4\u9732c":65,"\u81ea\u52a8\u6302\u8f7d\u5206\u5e03\u5f0f\u5b58\u50a8\u76ee\u5f55":32,"\u81f3\u4e8e\u4e3a\u4ec0\u4e48\u9700\u8981c":66,"\u826f\u597d\u7684\u6587\u6863":65,"\u8282\u7701\u4e86\u4e0d\u5fc5\u8981\u7684\u64cd\u4f5c":62,"\u83b7\u53d6\u6700\u65b0\u7684\u68c0\u67e5\u70b9\u7684\u6587\u4ef6uuid":32,"\u867d\u7136\u4e0d\u9f13\u52b1\u8fd9\u6837":66,"\u8868\u793a\u5bf9\u8f93\u5165\u6570\u636e":62,"\u89e3\u91ca\u578b\u8bed\u8a00\u53ea\u80fd\u8c03\u7528\u52a8\u6001\u5e93":65,"\u89e3\u91ca\u6027\u8bed\u8a00\u5b9e\u9645\u8fd0\u884c\u7684\u4e8c\u8fdb\u5236\u662f\u89e3\u91ca\u5668\u672c\u8eab":65,"\u8ba1\u5212\u5728":[61,62],"\u8ba1\u7b97\u8fd9\u4e2a\u6587\u4ef6\u7684md5":32,"\u8ba9paddle\u6838\u5fc3\u4e2d":66,"\u8bad\u7ec3\u4efb\u52a1\u7684\u8fd0\u884c\u53ef\u80fd\u4f1a\u5360\u6ee1trainer\u548cparamet":32,"\u8bad\u7ec3\u548c\u7eaf\u4f7f\u7528":82,"\u8bad\u7ec3\u6a21\u578b\u6b63\u786e\u6027":82,"\u8bb0\u5f55\u4e0b\u6240\u6709\u5931\u8d25\u7684\u4f8b\u5b50":82,"\u8bbe\u7f6e":66,"\u8bc6\u522b\u6570\u5b57":82,"\u8bcd\u5411\u91cf":82,"\u8be5\u6587\u4ef6\u5bf9\u76f8\u5173gemm":61,"\u8be5\u7c7b\u7ee7\u627f\u4e8epaddlepaddle\u7684\u57fa\u7c7b":62,"\u8be6\u7ec6\u8bbe\u8ba1":48,"\u8bed\u610f\u89d2\u8272\u6807\u6ce8":82,"\u8bf4\u660e":32,"\u8bf7\u53c2\u8003":66,"\u8f6c\u6362\u5185\u5b58\u7684\u5de5\u4f5c":62,"\u8f6c\u6362\u5197\u4f59":61,"\u8f6c\u6362\u51fd\u6570":62,"\u8f6c\u6362\u751f\u6210\u7684\u6587\u4ef6\u540d\u4f1a\u662f\u4ee5\u4e0b\u683c\u5f0f":33,"\u8f6c\u6362\u8017\u65f6":61,"\u8f93\u5165\u68af\u5ea6":62,"\u8f93\u51fa\u6570\u636e\u548c\u8f93\u51fa\u68af\u5ea6":62,"\u8f93\u51fa\u6570\u636e\u548c\u8f93\u51fa\u68af\u5ea6\u7684\u8f6c\u6362":62,"\u8fbe\u5230\u5bb9\u707e\u7684\u76ee\u7684":32,"\u8fc7\u7a0b\u4e2d\u6240\u6709\u65f6\u95f4\u6b65":61,"\u8fd1\u671f\u76ee\u6807":62,"\u8fd4\u56de\u7b2c\u4e8c\u6b65":82,"\u8fd8\u662f\u4ece":33,"\u8fd9\u4e00\u5c42\u8fdb\u884c\u5c01\u88c5":66,"\u8fd9\u4e00\u6570\u636e\u683c\u5f0f\u7684\u8f6c\u6362\u64cd\u4f5c":61,"\u8fd9\u4e00\u6982\u5ff5\u4e0d\u518d\u7410\u788e":66,"\u8fd9\u4e09\u4e2a\u5206\u652f":82,"\u8fd9\u4e2a\u51fd\u6570\u672c\u8eab\u4f1a\u5728\u8ba1\u7b97\u524d\u5c06\u539f\u6570\u636e\u8f6c\u6362\u4e3a\u66f4\u9002\u5408\u82f1\u7279\u5c14\u5e73\u53f0\u7684\u5185\u90e8\u683c\u5f0f":61,"\u8fd9\u4e2a\u52a8\u6001\u5e93\u7684\u8fde\u63a5\u53c2\u6570\u4e0epaddle\u7684\u5176\u4ed6\u4e8c\u8fdb\u5236":66,"\u8fd9\u4e2a\u53c2\u6570\u4e5f\u4e0d\u4f1a\u4e00\u5e76\u5220\u9664":66,"\u8fd9\u4e2a\u5934\u6587\u4ef6\u4e0d\u5047\u8bbe\u5176\u4ed6\u6587\u4ef6\u7684\u5f15\u7528\u987a\u5e8f":66,"\u8fd9\u4e2a\u63a5\u53e3\u9700\u8981\u505a\u5230":65,"\u8fd9\u4e2a\u6587\u4ef6\u5177\u6709\u72ec\u7279\u7684\u8bed\u6cd5":65,"\u8fd9\u4e2a\u76ee\u5f55\u4e2d\u9664\u4e86":66,"\u8fd9\u4e2a\u7ed3\u6784\u4f53\u4e2d\u7684\u53e6\u4e00\u4e2a\u9879\u76ee\u662f":66,"\u8fd9\u4e2a\u7ed3\u6784\u4f53\u5305\u542b\u4e24\u4e2a\u9879\u76ee":66,"\u8fd9\u4e2a\u9009\u62e9":[61,62],"\u8fd9\u4e2a\u9759\u6001\u5e93\u5305\u542b\u4e86paddle\u7684\u5168\u90e8\u7b26\u53f7":66,"\u8fd9\u4e2ainstance\u53ef\u4ee5\u662f\u5355\u4e2a\u503c":33,"\u8fd9\u4e9b\u4f1a\u5728":[61,62],"\u8fd9\u4e9b\u51fd\u6570\u4f1a\u6839\u636e\u8f93\u5165\u53c2\u6570\u91cd\u65b0\u8bbe\u7f6e\u5185\u90e8\u548c\u5916\u90e8\u5b58\u50a8":62,"\u8fd9\u4e9b\u5206\u5e03\u5f0f\u5b58\u50a8\u670d\u52a1\u901a\u5e38\u4f1a\u628a\u6570\u636e\u5207\u5272\u6210\u591a\u4e2a\u5206\u7247\u5206\u5e03\u5f0f\u7684\u5b58\u50a8\u5728\u591a\u4e2a\u8282\u70b9\u4e4b\u4e0a":33,"\u8fd9\u4e9b\u955c\u50cf\u4e5f\u53ef\u4ee5\u4ece":82,"\u8fd9\u5bf9\u4e8e\u901a\u5e38\u7684java\u7684\u5f00\u53d1\u8005\u6765\u8bf4":65,"\u8fd9\u662f\u56e0\u4e3a":65,"\u8fd9\u6837":66,"\u8fd9\u6837\u4e0b\u4e00\u4e2acpu":62,"\u8fd9\u6837\u4fdd\u8bc1":82,"\u8fd9\u6837\u5c31\u53ef\u4ee5\u5728\u4e91\u7aef\u6267\u884c\u591a\u79cd\u6570\u636e\u7c7b\u8ba1\u7b97\u4efb\u52a1":33,"\u8fd9\u6837\u5df2\u7ecf\u4f20\u8f93\u6210\u529f\u7684\u90e8\u5206\u5c31\u4e0d\u7528\u91cd\u65b0\u4f20\u8f93\u4e86":48,"\u8fd9\u6837\u5e26\u6765\u7684\u597d\u5904\u5c31\u662f\u4e0d\u9700\u8981\u4e00\u76f4\u6e05\u7a7amemori":62,"\u8fd9\u6837\u65e2\u4f7f\u5f97\u6700\u7ec8\u4fdd\u5b58\u7684\u53c2\u6570\u683c\u5f0f\u4e0epaddlepaddle\u4e00\u81f4":62,"\u8fd9\u90fd\u9700\u8981\u8fd9\u4e2a\u63a5\u53e3\u6309\u7167\u7ea6\u5b9a\u4fd7\u6210\u7684\u89c4\u5219\u6765\u6ce8\u91ca\u5b8c\u5907":65,"\u8fd9\u91cc":62,"\u8fd9\u91cc\u7684dockerimage\u4f5c\u4e3a\u7f16\u8bd1\u73af\u5883\u4ee5\u652f\u6301\u66f4\u591a\u7684linux":82,"\u8fd9\u91cc\u9009\u62e90":82,"\u8fd9\u91cc\u9700\u8981\u7528\u6237\u989d\u5916\u6ce8\u610f":32,"\u8fdb\u4e00\u6b65\u4f18\u5316":62,"\u8fdb\u5165":82,"\u8fdb\u800c\u8fdb\u884c\u4ee3\u7801\u8bc4\u5ba1":82,"\u9009\u62e9\u662f\u5426\u7f16\u8bd1mkl":62,"\u9009\u62e9\u9700\u8981\u53d1\u5e03\u7684\u7248\u672c":82,"\u900f\u4f20\u7528\u6237\u8eab\u4efd\u7684\u529e\u6cd5":48,"\u901a\u5e38":66,"\u901a\u5e38\u5305\u542b\u4e00\u4e2acpu\u7248\u672c\u548c\u4e00\u4e2agpu\u7248\u672c":82,"\u901a\u5e38\u6307\u5c06\u4e00\u4e2a\u6574\u4f53\u62c6\u5206\u6210\u591a\u4efd\u7684\u5176\u4e2d\u7684\u4e00\u4efd":32,"\u901a\u8fc7\u4f7f\u7528\u8fd9\u4e9bapi":61,"\u901a\u8fc7\u6a21\u578b\u63a8\u65adapi\u7684\u5b9e\u73b0\u4f5c\u4e3a\u4e00\u4e2a\u6837\u4f8b":66,"\u903b\u8f91\u5212\u4e0a\u6587\u4ef6\u5206\u5757\u7684\u5355\u4f4d":48,"\u9075\u5faa\u4ee5\u4e0b\u6d41\u7a0b":82,"\u90a3\u4e48":66,"\u90a3\u4e48\u5bf9\u5e94\u7684\u5185\u90e8\u5b58\u50a8\u4e5f\u4f1a\u4e0e\u5b83\u4eec\u5171\u4eab\u5185\u5b58":62,"\u90a3\u4e48\u5c31\u4f1a\u4f7f":62,"\u90fd\u4e0d\u4f1a\u60f3\u8981\u77e5\u9053next":62,"\u90fd\u662f\u4e94\u4f4d\u7684\u6570\u5b57":33,"\u90fd\u662f\u4ee5ext\u5f00\u5934":62,"\u90fd\u662fabi\u8c03\u7528\u6807\u51c6\u7684":65,"\u90fd\u7ee7\u627f\u4e8epaddlepaddle\u7684\u57fa\u7c7b":61,"\u914d\u7f6e\u7684\u65b9\u6cd5\u53c2\u8003":48,"\u91ca\u653e\u5bf9paramters\u5185\u5b58\u7684\u9501\u5b9a":32,"\u91cc\u6240\u6709\u7684\u7b26\u53f7\u90fd\u5199\u5165\u81ea\u5df1\u7684\u7a0b\u5e8f\u7684\u4e8c\u8fdb\u5236\u6587\u4ef6\u91cc":65,"\u91cc\u9009\u62e9\u9700\u8981\u53d1\u5e03\u7684\u5206\u652f":82,"\u91cc\u9762\u6dfb\u52a0":62,"\u91cd\u5199\u7236\u7c7blayer\u7684":62,"\u91cd\u547d\u540d\u6210":65,"\u94fe\u63a5\u5230\u81ea\u5df1\u7684\u7a0b\u5e8f\u91cc":65,"\u9519\u8bef\u5904\u7406":65,"\u9519\u8bef\u5904\u7406\u65b9\u5f0f\u662f\u8fd4\u56de\u503c":65,"\u9519\u8bef\u5904\u7406\u7684\u65b9\u5f0f\u4e5f\u4e0d\u5c3d\u76f8\u540c":65,"\u9664\u6784\u9020\u67d0\u79cd\u7c7b\u578b\u7684\u51fd\u6570":66,"\u96c6\u6210\u5230":61,"\u96c6\u6210\u5230paddlepaddl":62,"\u9700\u8981":33,"\u9700\u8981\u4fee\u6539build":82,"\u9700\u8981\u53ef\u4ee5\u8de8\u5e73\u53f0\u6267\u884c":48,"\u9700\u8981\u5728cmake\u7684\u65f6\u5019":66,"\u9700\u8981\u5c06bugfix\u7684\u5206\u652f\u540c\u65f6merge\u5230":82,"\u9700\u8981\u5f15\u7528":66,"\u9700\u8981\u6709\u7a33\u5b9a\u7684\u5bfc\u51fa\u7b26\u53f7":65,"\u9700\u8981\u6ce8\u610f\u7684\u662f":[62,82],"\u9700\u8981\u7d2f\u52a0\u4e0d\u540clayer\u4f20\u8fc7\u6765\u7684\u68af\u5ea6":62,"\u9700\u8981\u88ab\u66b4\u9732\u5230\u5176\u4ed6\u8bed\u8a00":66,"\u9700\u8981\u91cd\u547d\u540dwheel\u5305\u4e2dplatform\u76f8\u5173\u7684\u540e\u7f00":82,"\u9ed8\u8ba4256k":48,"\u9ed8\u8ba4\u8bbe\u7f6e\u4e3a":61,"abstract":[40,47,51,72,81,83,100,118],"api\u4e2d\u4f7f\u7528":65,"api\u5bfc\u51fa\u7684\u52a8\u6001\u5e93":66,"api\u5bfc\u51fa\u7684\u9759\u6001\u5e93":66,"api\u63a5\u53d7\u7684\u7c7b\u578b\u5168\u662f":66,"api\u63a5\u53e3":48,"api\u63a5\u53e3\u7684\u53c2\u6570\u8f6c\u53d1\u7ed9":66,"api\u65f6":66,"api\u65f6\u6240\u552f\u4e00\u9700\u8981\u5f15\u5165\u7684\u5934\u6587\u4ef6":66,"api\u662f\u591a\u8bed\u8a00api\u7684\u57fa\u7840\u90e8\u5206":66,"api\u66b4\u9732\u7684\u7c7b\u578b":66,"api\u751f\u6210\u7684\u4e8c\u8fdb\u5236\u6587\u4ef6\u4f1a\u88ab\u5b89\u88c5\u5230":66,"api\u7684\u5b9e\u4f8b":66,"api\u7684\u5b9e\u73b0\u7ec6\u8282":66,"api\u7684\u63a5\u53e3":66,"api\u7684\u65f6\u5019\u63a8\u8350paddle\u4e0d\u5d4c\u5165python\u89e3\u91ca\u5668":66,"api\u7684\u7f16\u8bd1\u9009\u9879\u9ed8\u8ba4\u5173\u95ed":66,"api\u76ee\u5f55\u7ed3\u6784\u5982\u4e0a\u56fe\u8868\u6240\u793a":66,"api\u83b7\u5f97\u4e86\u795e\u7ecf\u7f51\u7edc\u7684\u53c2\u6570\u5b9e\u4f8b":66,"apis\u505a\u4e86\u5c01\u88c5":61,"block\u6784\u6210\u4e00\u4e2amodel":32,"book\u4e2d\u6240\u6709\u7ae0\u8282\u529f\u80fd\u7684\u6b63\u786e\u6027":82,"boolean":[17,18,47,49,57,65],"break":[9,30,86,89,90],"bugfix\u5206\u652f\u4e5f\u662f\u5728\u5f00\u53d1\u8005\u81ea\u5df1\u7684fork\u7248\u672c\u5e93\u7ef4\u62a4":82,"bugfix\u5206\u652f\u9700\u8981\u5206\u522b\u7ed9\u4e3b\u7248\u672c\u5e93\u7684":82,"byte":[9,11,48,64],"c99\u662f\u76ee\u524dc\u6700\u5e7f\u6cdb\u7684\u4f7f\u7528\u6807\u51c6":65,"c\u6709\u6807\u51c6\u7684abi":65,"c\u8bed\u8a00\u662f\u6709\u5bfc\u51fa\u7b26\u53f7\u7684\u6807\u51c6\u7684":65,"case":[4,16,18,34,40,42,47,51,60,66,72,76,78,79,97,100,101,107,111,112,119,126],"char":36,"ci\u73af\u5883\u4f7f\u7528":82,"ci\u7f16\u8bd1wheel\u5b8c\u6210\u540e\u4f1a\u81ea\u52a8\u5c06docker\u955c\u50cfpush\u5230dockerhub":82,"class":[1,2,3,4,5,6,7,9,10,14,16,18,20,23,25,26,29,40,41,42,45,46,50,51,52,53,55,56,58,60,65,69,70,74,75,79,80,81,83,84,85,87,89,101,102,103,108,117],"compute\u51fd\u6570":61,"const":[29,34,36,50,52,58,59,73,74,76,80,83,85,87,88,89,100,101,102,103],"core\u4e2d\u7684\u6a21\u578b\u8fd8\u5728\u4f7f\u7528\u8fd9\u4e2a\u53c2\u6570":66,"core\u4e2d\u8fd9\u4e00\u7c7b\u578b\u63a5\u53e3\u7684\u667a\u80fd\u6307\u9488":66,"core\u662f\u5426\u8fd8\u5728\u4f7f\u7528\u8fd9\u4e2a\u5b9e\u4f8b":66,"core\u6982\u5ff5":66,"data\u5230\u5206\u5e03\u5f0f\u5b58\u50a8\u8865\u5145\u8bad\u7ec3\u6570\u636e":33,"default":[2,3,4,5,6,7,9,10,11,14,18,19,22,25,26,29,30,40,45,54,58,64,67,68,76,77,83,84,85,90,91,92,94,98,99,101,102,106,109,110,112,113,118,120,122,123,124,126],"device\u5c31\u80fd\u62ff\u5230\u6b63\u786e\u7684\u6570\u636e":62,"dnn\u4e09\u8005\u5173\u7cfb\u5982\u4e0b\u8868":62,"dnn\u4e2d\u7684":62,"dnn\u4e2d\u7684\u6392\u5217\u65b9\u5f0f\u4e0d\u6b62\u8fd9\u4e00\u79cd":62,"dnn\u4f1a\u4f5c\u4e3a\u7b2c\u4e09\u65b9\u5e93\u96c6\u6210\u8fdbpaddlepaddl":62,"dnn\u4f1a\u7528\u5230":62,"dnn\u5171\u540c\u4f7f\u7528":62,"dnn\u524d\u540e\u7684cnn\u7f51\u7edc\u6027\u80fd":62,"dnn\u5728\u53d1\u5e03":62,"dnn\u5b9e\u73b0":62,"dnn\u5e0c\u671b\u7684\u683c\u5f0f":62,"dnn\u6570\u636e\u7684\u4e0d\u540c\u683c\u5f0f\u4ee5\u53ca\u76f8\u4e92\u4e4b\u95f4\u7684\u8f6c\u6362":62,"dnn\u7684":62,"dnn\u7684\u5e93\u76ee\u524d\u53ea\u6709\u52a8\u6001\u5e93":62,"dnn\u7684\u6027\u80fd":62,"dnn\u7684\u60c5\u51b5\u4e0b":62,"dnn\u7684\u64cd\u4f5c\u90fd\u662f\u76f4\u63a5\u8986\u76d6\u7684\u5f62\u5f0f":62,"dnn\u7684\u6d4b\u8bd5":62,"dnn\u7684\u73af\u5883\u4e0b":62,"dnn\u7684\u76f8\u5173\u529f\u80fd":62,"dnn\u7684\u7ed3\u679c":62,"dnn\u7684\u9ad8\u6027\u80fd\u683c\u5f0f\u4e0epaddlepaddle\u539f\u6709\u7684":62,"dnn\u7684layer":62,"dnn\u7684layers\u90fd\u4f1a\u7ee7\u627f\u4e8e":62,"enum":[34,36,41,67,74,75,84,85,90,102],"export":[51,56,92,104,109],"final":[4,5,18,27,28,42,56,68,69,86,89,100,101],"float":[2,3,4,6,9,18,19,45,50,58,85,87,88,100,101,102,103,107,120],"function":[4,5,9,17,18,19,23,26,28,29,31,35,36,37,39,40,41,42,45,46,50,52,55,58,63,68,69,72,73,74,75,76,78,79,80,81,83,85,89,97,100,101,102,103,106,107,109,111,118,126],"golang\u53ef\u4ee5\u4f7f\u7528":65,"golang\u7684":65,"gpu\u7b49":82,"h\u5e76\u4e0d\u56f0\u96be":65,"images\u6570\u636e\u96c6\u4e0a\u4f20\u5230\u4e91\u7aef\u7684":33,"import":[3,4,26,29,30,40,41,44,52,54,56,57,63,68,69,75,83,86,94,95,97,101,107,109,111,112,124],"ingress\u9700\u8981\u628apfsclient\u7684\u8eab\u4efd\u4fe1\u606f\u4f20\u7ed9pfsserv":48,"instance\u4e0e\u751f\u6210\u6570\u636e\u96c6\u65f6":33,"instance\u5305\u6db5\u4e24\u4e2a\u503c":33,"instance\u662f\u4e00\u6a21\u4e00\u6837\u7684":33,"int":[2,3,4,5,9,10,11,18,19,28,29,34,35,36,39,40,41,43,57,58,61,62,63,65,66,74,75,77,78,84,85,87,89,90,100,102,103,109,120],"interface\u6587\u4ef6\u7684\u5199\u6cd5\u975e\u5e38":65,"layer\u65f6":62,"layer\u7684\u540e\u9762\u63a5\u6709cpu":62,"list\u4f5c\u4e3a\u68c0\u67e5\u5217\u8868":82,"long":[4,5,9,18,41,107],"mkl\u5e93\u7684":61,"mklml\u4ee5\u53camkl":62,"mklml\u53ef\u4ee5\u4e0emkl":62,"mklml\u7684\u5e93\u76ee\u524d\u90fd\u662f\u52a8\u6001\u5e93":62,"mode\u4e0b\u7684\u7ed3\u679c":61,"model\u505a\u5206\u652f\u7ba1\u7406":82,"ndarray\u7c7b\u578b\u7684\u503c\u548c\u6574\u578b\u7684\u503c":33,"new":[4,9,18,27,28,29,30,31,34,35,36,37,38,41,42,45,50,51,60,61,63,67,69,72,77,78,79,81,85,86,89,95,98,99,105,112,113,126],"note\u7684\u4e66\u5199":82,"null":[56,100,118],"op\u7684\u4fe1\u606f":62,"openmp\u7528\u4e8e\u63d0\u9ad8mklml\u7684\u6027\u80fd":62,"org\u76ee\u524d\u9075\u5faa":82,"packed\u4f18\u5316\u540elayer\u7684\u6d4b\u8bd5":61,"packed\u76f8\u5173\u529f\u80fd":61,"paddle\u4e00\u4e2a\u52a8\u6001\u5e93\u53ef\u4ee5\u5728\u4efb\u4f55linux\u7cfb\u7edf\u4e0a\u8fd0\u884c":65,"paddle\u5185\u5d4c\u7684python\u89e3\u91ca\u5668\u548c\u5916\u90e8\u4f7f\u7528\u7684python\u5982\u679c\u7248\u672c\u4e0d\u540c":65,"paddle\u5185\u90e8\u7684\u7c7b\u4e3ac":65,"paddle\u7684\u591a\u8bed\u8a00\u63a5\u53e3\u5b9e\u73b0\u5305\u62ec\u4e00\u4e0b\u51e0\u4e2a\u65b9\u9762":65,"paddle\u7684\u7c7b\u578b\u5168\u90e8\u9000\u5316\u6210":66,"paddle\u7684\u94fe\u63a5\u65b9\u5f0f\u6bd4\u8f83\u590d\u6742":65,"paddle\u7684c":66,"paddle\u8bad\u7ec3\u4efb\u52a1":33,"paddle\u8def\u5f84\u4e0b":66,"paddle\u9700\u8981\u4e00\u4e2a\u591a\u8bed\u8a00\u63a5\u53e3":65,"paddle\u9700\u8981\u66b4\u9732\u7684api\u5f88\u591a":66,"paddle\u9759\u6001\u5e93\u94fe\u63a5\u590d\u6742":65,"paddle_\u7c7b\u578b\u540d":66,"paddle_\u7c7b\u578b\u540d_\u51fd\u6570\u540d":66,"paddlepaddle\u4e2d\u7684cudnn\u90e8\u5206\u4f7f\u7528\u7684\u4e5f\u662f":62,"paddlepaddle\u4f7f\u7528git":82,"paddlepaddle\u5f00\u53d1\u8fc7\u7a0b\u4f7f\u7528":82,"paddlepaddle\u63d0\u4f9b\u4e13\u7528\u7684":33,"paddlepaddle\u6bcf\u6b21\u53d1\u65b0\u7684\u7248\u672c":82,"paddlepaddle\u6bcf\u6b21\u53d1\u7248\u672c\u9996\u5148\u8981\u4fdd\u8bc1paddlepaddl":82,"paddlepaddle\u7684\u4e3b\u7248\u672c\u5e93\u9075\u5faa":82,"paddlepaddle\u7684activation\u4f1a\u76f4\u63a5\u4f7f\u7528":62,"patch\u53f7":82,"patch\u53f7\u52a0\u4e00":82,"pfsclient\u9700\u8981\u548cingress\u4e4b\u95f4\u505a\u53cc\u5411\u9a8c\u8bc1":48,"pfsclient\u9700\u8981\u5728\u4f20\u8f93\u5b8c\u6bd5\u6700\u540e\u4e00\u4e2achunk\u7684\u65f6\u5019\u68c0\u67e5destination\u6587\u4ef6\u7684md5\u503c\u662f\u5426\u548csource\u6587\u4ef6\u4e00\u81f4":48,"pfsserver\u63d0\u4f9brest":48,"public":[10,14,18,29,50,53,58,74,80,83,85,86,87,89,100,101,102,103,109,112,113],"py\u4e2d":82,"pypi\u4e0a\u7684package\u540d\u79f0\u4e3apaddlepaddle\u548cpaddlepaddl":82,"pypi\u4e0d\u652f\u6301\u8986\u76d6\u4e0a\u4f20":82,"reader\u7684\u4f7f\u7528\u65b9\u5f0f\u90fd\u662f\u4e00\u81f4\u7684":33,"reader\u8f93\u51fa\u7684data":33,"resnet\u7684mkl":62,"return":[2,3,4,5,7,9,10,11,14,17,18,19,20,25,26,27,28,29,33,34,36,39,40,46,50,52,53,54,56,58,59,60,63,68,69,70,74,75,76,80,83,85,87,89,97,100,101,102,103,111,112],"rnn\u90e8\u5206\u4e2d":61,"s3\u4e4b\u7c7b\u7684\u5206\u5e03\u5f0f\u5b58\u50a8\u4e4b\u4e0a":33,"server\u4e4b\u4e0a":32,"server\u4e4b\u95f4\u7684\u7f51\u7edc\u5e26\u5bbd":32,"server\u4f1a\u6682\u505c\u53c2\u6570\u66f4\u65b0\u5e76\u7b49\u5f85":32,"server\u4f1a\u83b7\u53d6parameters\u5185\u5b58\u7684":32,"server\u5185\u5b58\u4e2d\u7684\u6a21\u578b\u6570\u636e\u7684\u5b8c\u6574\u955c\u50cf":32,"server\u540c\u6b65\u7684\u4fdd\u5b58\u4e00\u4e2a\u7279\u5b9a\u65f6\u95f4\u70b9\u7684\u5168\u5c40\u68c0\u67e5\u70b9":32,"server\u5728\u96c6\u7fa4\u4e2d\u542f\u52a8\u540e":32,"server\u6545\u969c\u540e\u88abkubernetes\u91cd\u65b0\u542f\u52a8":32,"server\u6b64\u65f6\u8fd8\u9700\u8981\u901a\u8fc7\u7f51\u7edc\u8bbf\u95ee\u5206\u5e03\u5f0f\u5b58\u50a8\u4ee5\u4fdd\u5b58\u5feb\u7167":32,"server\u751f\u6210\u4e00\u4e2auuid":32,"server\u7684\u5355\u70b9\u6216\u591a\u70b9\u540c\u65f6\u6545\u969c":32,"server\u7684\u6570\u636e\u5feb\u7167":32,"server\u7684\u68c0\u67e5\u70b9\u5404\u81ea\u72ec\u7acb\u4fdd\u5b58":32,"server\u7b2c\u4e00\u6b21\u542f\u52a8\u6216\u4efb\u610f\u65f6\u95f4paramet":32,"short":[4,5,50,54,77,83,86,89,101],"static":[25,36,66,83,85,112,123,126],"super":[77,100],"swig\u652f\u6301\u7684\u8bed\u8a00\u6216\u8005\u89e3\u91ca\u5668\u6709\u5c40\u9650":65,"swig\u66b4\u9732\u7684\u63a5\u53e3\u4fdd\u7559\u4e86c":65,"swig\u751f\u6210\u7684\u4ee3\u7801\u4e0d\u80fd\u4fdd\u8bc1\u591a\u8bed\u8a00\u4ee3\u7801\u98ce\u683c\u7684\u4e00\u81f4\u6027":65,"swig\u76f4\u63a5\u8bfb\u53d6c":65,"swig\u9700\u8981\u5199\u4e00\u4e2ainterface\u6587\u4ef6":65,"switch":[29,66,112],"tag\u4e3a":82,"tag\u53ef\u4ee5\u662flatest\u6216latest":82,"tag\u7684\u66f4\u65b0\u65f6\u95f4\u662f\u5426\u5728\u4e0a\u8ff0\u7f16\u8bd1wheel\u5305\u5b8c\u6210\u540e\u662f\u5426\u6700\u65b0":82,"throw":112,"true":[2,3,4,5,6,7,9,11,16,18,19,25,26,28,29,34,51,57,61,70,75,76,77,78,82,85,89,97,100,109,112,118,120],"try":[30,31,34,35,36,51,56,60,63,78,83,86,92,94,98,107,111,123],"type\u5b57\u6bb5\u5747\u4e0d\u5c3d\u76f8\u540c":66,"var":[17,28,29,40,45,52,53,55,57,70,75,76,77,79,83,88,89,104],"void":[29,34,36,47,50,52,53,58,63,64,65,66,75,76,84,85,87,88,100,101,102,103],"wheel\u5305":82,"while":[2,4,9,18,29,38,41,51,56,59,60,69,72,73,78,81,83,87,97,101,103,118],AGE:[112,113],AWS:[9,33,109,115,116],Abs:18,Added:86,And:[3,4,6,9,10,11,18,25,27,34,38,39,47,54,56,67,71,74,78,83,87,97,112,120],But:[4,5,9,18,27,53,59,67,74,83,94,99,109,126],EOS:4,For:[3,4,5,6,9,18,25,26,28,29,35,36,37,39,40,42,45,46,51,52,53,55,58,60,64,67,68,69,72,73,74,75,76,77,78,79,80,81,84,85,86,87,88,90,91,92,95,97,99,100,101,102,103,106,107,111,117,118,120,122,124,126],IDE:[98,123],IDs:[10,18,38,41,69],IPs:109,IRs:42,Into:112,Its:[3,4,52,84,97,101,112],K8s:126,NMS:4,NOT:[18,77,101],Not:[26,31,60,86,94,126],OPs:[42,44,111],One:[3,5,18,25,27,38,64,67,83,86,88,97,100,118],Ops:[79,81,85,101],PFS:48,PRs:104,QoS:113,Such:[58,77,86,89],TLS:[26,48,112],That:[4,9,71,92,118,120],The:[1,2,3,4,5,6,9,10,11,14,17,18,19,22,25,26,27,28,30,31,35,37,38,39,41,42,44,45,46,49,50,52,56,59,60,63,64,66,68,69,71,72,74,75,76,77,78,81,83,84,85,86,87,88,89,90,92,94,97,98,99,100,101,102,103,104,106,107,108,109,110,111,112,113,118,120,122],Their:[4,31],Then:[4,5,18,40,42,53,58,60,71,74,76,91,92,94,97,100,106,107,109,111,112,113,114],There:[3,4,10,18,25,26,29,30,31,36,38,39,41,42,49,50,51,56,60,67,68,69,72,73,74,77,81,83,84,87,101,107,112,122,123],These:[3,11,18,28,29,45,50,55,70,81,84,85,86,91,109,120],Use:[3,9,18,26,43,49,78,79,86,91,100,106,107,112,118,119,124],Used:[5,14,19,79,87],Uses:[60,123],Using:[18,31,51,72,78,79,81,83,93,98,103,113],VMs:98,VPS:112,WITH:124,Will:[9,25],With:[4,5,40,45,51,71,75,86,89],YES:39,Yes:[62,92,98],___fc_layer_0__:112,__align__:50,__cuda_align__:50,__device__:50,__doc__:85,__file__:39,__forceinline__:50,__fp16:50,__global__:50,__gradient_machines__:25,__hadd:50,__half:50,__half_raw:50,__impl__:85,__init__:[45,46,54,60,70,77,89,100,106],__main__:54,__metaclass__:101,__name__:54,__param_conf__:25,__rnn_step__:97,__tmp_params__:25,__va_args__:80,__x:50,_addup_repetitive_outputs_:28,_append_backward_ops_:[28,45],_append_backward_vars_:28,_binari:30,_create_global_var:77,_def:60,_dtype:56,_filer:63,_filter:63,_fwd:63,_fwd_pd:63,_input:63,_librari:30,_link:5,_live_in:60,_live_out:60,_loss:54,_op:[56,101],_output:63,_presucessor:60,_program:60,_proj:4,_remove_no_grad_branch_:28,_reorder_input:63,_source_language_embed:97,_src_input:63,_sucessor:60,_target_language_embed:97,_test:30,_update_op:46,_use:60,_value_index:56,a75:50,a_op:101,a_prev:86,aaaaa:33,aaaaaaaaaaaaa:112,abbrevi:11,abc:4,abi:122,abil:54,abl:[4,26,28,41,42,58,70,74,77,126],about:[5,11,18,29,30,39,44,49,52,60,68,78,83,85,86,87,92,94,95,101,106,107,112,117,118,122,124],abov:[3,4,18,26,28,29,30,31,35,40,41,42,50,51,52,53,55,63,68,69,70,71,72,74,75,77,85,86,88,89,92,93,95,98,99,101,102,103,106,107,111,112,113,122,126],abs:[5,16,27,54],abs_numerical_grad:27,absolut:[92,122,124],acc:42,acceler:[4,32,62,71,72,92,120],accept:[2,4,9,18,26,79,122],access:[4,5,18,26,30,35,38,39,40,42,77,94,97,109],accessmod:112,accessor:77,accord:[3,4,11,18,27,28,36,42,44,55,69,79,89,101,109,117,118,120],accordingli:[3,4,100],account:[79,99,126],accoust:86,accrodingli:34,accumul:[31,36,46,71,72,86],accur:[27,38,74],accuraci:[3,14,46,86,100],achiev:[44,71,72,86,87,107],ack:118,acquir:51,across:[4,9,28,42,68,86],act1:56,act2:56,act:[4,5,18,19,29,42,56,69,77,89,95,97,108,111],act_output:85,act_typ:56,actgat:18,action:112,activ:[5,18,19,24,30,56,60,69,74,77,81,85,95,97,100,103,104,118],activi:5,actnod:18,actual:[4,18,34,45,51,54,56,63,67,72,85,87,88,103],actual_layout:63,adagrad:[20,72,84],adagradoptim:70,adam:[20,26,36,42,54],adamax:20,adapt:[3,6,18,25,74,88],add:[4,5,9,16,18,20,23,25,27,28,29,30,34,38,41,42,44,46,50,53,57,59,70,72,76,77,79,81,83,87,91,98,99,100,101,103,107,108,120,124],add_activ:77,add_bia:77,add_depend:30,add_execut:30,add_input:[68,100],add_memori:68,add_output:68,add_scalar:[29,69,75],add_sum:77,add_test:[30,100],add_two:[29,68],add_unittest_without_exec:100,addattr:[85,101],addbia:100,addcom:[85,101],added:[3,4,14,18,23,25,29,44,45,50,67,71,72,81,99,100,101,102],adding:[81,99],addinput:[85,101],addit:[4,5,28,41,71,74,79,81,89,101,103,111],addition:68,addmemori:63,addop:[53,103],addoutput:101,addprimit:63,addprimitivedesc:63,addr:31,address:[31,36,40,42,92,107,109,110,111,114,118,126],addrow:100,addtolay:4,addtyp:85,adjust:[28,45],admin:126,administr:[38,98,126],adopt:[50,54],advanc:[27,97,107,118],advantag:[27,50,51,72,78,109],adversari:[54,78],advic:107,affect:[4,29],affili:69,afford:35,aforement:30,after:[4,5,10,11,18,28,29,30,35,36,38,42,43,44,45,47,49,50,60,63,71,73,74,77,86,91,92,97,99,100,101,102,103,106,109,110,112,113,118,120,122,123],aftern:74,again:[26,31,72,107],against:112,age:10,agg_level:4,aggreg:[46,71,112],ago:30,ahead:86,aid:107,alex:18,alexnet_pass1:120,alexnet_pass2:120,algo:63,algorithm:[4,6,18,28,35,45,60,63,69,72,81,86,88,97],alia:[1,2,4,18],align:[4,5,9],all:[2,3,4,14,16,18,19,25,26,28,29,30,31,34,36,38,39,40,41,42,43,45,47,49,51,54,55,56,59,60,63,64,66,67,68,69,70,71,72,74,75,77,79,85,86,87,92,97,98,100,103,107,108,109,110,111,112,113,117,118,120,122,123,124,126],all_output_nam:28,alloc:[2,36,39,60,63,87,100,103,108,120],allow:[26,36,40,42,45,51,72,81,99,100,107,112,118],allow_only_one_model_on_one_gpu:[117,118,120],allreduc:71,almost:[18,98,110],along:[4,11,18,19],alpha:[18,30,81],alreadi:[18,30,31,51,63,77,83,92,107,110,112,118,124],alreali:117,also:[4,5,10,18,26,28,29,30,34,37,41,42,50,51,53,54,55,56,59,60,67,68,69,72,73,74,75,76,77,78,81,83,85,86,87,89,90,91,92,94,97,98,100,101,104,107,109,111,113,122,126],altern:[101,106],although:[28,71],altogeth:126,alwai:[4,5,18,25,30,64,84,112,118],amazon:[112,113],amazonaw:112,amazonec2fullaccess:112,amazonelasticfilesystemfullaccess:112,amazonroute53domainsfullaccess:112,amazonroute53fullaccess:112,amazons3fullaccess:112,amazonvpcfullaccess:112,ambigu:[78,86],amd64:112,amd:67,amend:99,amodei:86,among:[18,112],amort:71,amount:[18,107],analys:74,analysi:[74,106,107],analyz:60,ancestor:[75,77],andd:112,andrew:60,android:[124,125],android_abi:122,android_api:122,android_arm_neon:122,android_native_api_level:122,android_standalone_toolchain:122,android_toolchain:122,ani:[4,5,9,18,20,26,30,31,36,38,39,40,41,42,47,50,51,58,60,64,69,71,72,77,78,80,81,86,88,91,98,101,103,107,111,112,114,124],annoi:110,announc:50,anoth:[4,9,25,26,28,29,39,41,51,52,63,69,77,83,85,87,88,98,112,118],anroid_arm_mod:122,ans:112,answer:[40,51,99,112],anymor:71,anyth:[9,69,78,112],anytim:54,anywai:[106,122],apach:62,apart:18,api:[10,14,25,26,28,30,36,37,39,40,46,48,53,54,56,68,73,74,79,82,89,90,91,94,100,101,106,107,109,111,112,121,122,123,124,126],api_shar:30,api_test:30,api_trainer_config_helpers_lay:97,apiserv:112,apivers:[112,113],app:123,appar:28,appear:[40,51,55,87],appel:60,append:[18,25,28,45,46,69,77,78,86,97,99,100,109],append_backward:[20,28,70,106],append_batch_s:18,append_clip_op:45,append_gradient_machin:25,append_op:[45,59,77],append_oper:77,appl:123,appleyard:107,appli:[4,18,54,55,71,74,83,97,100],applic:[22,40,41,50,51,52,55,77,79,99,101,106,107,109,112,113,126],applyl1:34,appoint:101,appreci:[86,99],approach:[4,42,43,44,71,72,73,81,86,122,124,126],approxim:[16,18,72],apt:[92,106],arbitrari:[4,42,64,103],arch:122,archetectur:86,architectur:[50,86,91,109,122,123],archiv:[10,18,65,66],area:54,arg:[3,5,22,28,56,70,85,101,117],arg_nam:4,argmax:18,argu:76,argument:[4,9,11,22,28,29,34,35,42,70,73,76,77,91,97,99,100,118,119],arithmet:50,arm64:[122,123],arm64_standalone_toolchain:122,arm:[50,122,123,124],arm_standalone_toolchain:122,armeabi:122,armv7:[50,123],armv8:50,arn:112,around:[4,38,60,77,112,126],arrai:[2,4,9,11,18,25,36,40,41,55,69,75,77,78,79,89,101],arrang:89,arrari:18,array_to_lod_tensor:60,arrow:54,articl:[52,55,98,109,111,113,124],artifact:[82,94,112],artifici:[16,60],arxiv:[5,16,18,54,86],as_row_vector:4,as_step_input:29,asap:102,asgd:72,ask:[28,31,38,98],asr:86,assgin:60,assign:[3,4,19,28,35,40,43,45,50,52,71,86,103,109,111,112,118,126],assigne:86,assignmemt:60,associ:[73,80,103],assum:[3,4,18,29,42,63,92,97,111,120],assumpt:42,ast:40,astyp:[78,101],asyc:31,async:[31,44,117],async_count:118,async_lagged_grad_discard_ratio:118,async_lagged_ratio_default:[117,118],async_lagged_ratio_min:[117,118],asynchron:[31,41,71,74,109,118],atom:43,att_seq:5,attach:5,attend:5,attended_sequ:5,attenion:5,attent:[4,5,19],attr1:4,attr2:4,attr:[2,4,5,18,29,40,56,59,63,75,76,77,85,97,101],attr_map:85,attrdesc:75,attribu:63,attribut:[4,5,18,24,28,29,44,45,59,75,77,79,83,85,89,100,101],attributemap:101,attrproto:85,attrtyp:[75,85,101],attrvalu:85,auc:[46,117],aucvalidationlay:118,audio:86,augment:86,authent:112,author:[48,86,112],auto:[29,34,43,52,63,65,76,79,83,88,89,91,99,100,101,103,107],autom:[109,112],automat:[4,18,26,28,36,42,44,45,53,70,79,85,86,91,97,99,100,101,106,110,111,112,117,118,123],avail:[31,36,44,50,51,60,94,112,126],averag:[3,4,7,18,25,35,118],average_test_period:[117,118],avg:[107,111],avg_cost:[42,108,111],avg_loss_valu:111,avg_x:18,avgpool:4,avoid:[27,29,31,42,59,63,71,72,73,91,107],avx2:91,avx:[91,92],awai:51,await:113,awar:[26,40,46,52,68,77,98,106,112],awk:114,awni:86,aws:48,aws_account_id:112,awsaccountid:112,awskeymanagementservicepowerus:112,axi:[4,18],axis:4,b363:113,b8561f5c79193550d64fa47418a9e67ebdd71546186e840f88de5026b8097465:113,ba5f:112,back:[4,18,25,28,31,42,50,54,72,92,101],background:[3,4,81,86,109],background_id:[3,4],backpropag:[27,28],backward:[1,4,5,23,27,29,34,36,45,54,61,62,70,72,73,76,80,81,97,100,108,118,120],backward_first:97,backward_op:27,backwardactiv:100,baidu:[51,86,113],bake:42,balanc:[44,71,112,118],bandwidth:[50,71],bare:[111,113,126],barrier:[109,118],barrierstatset:107,basci:56,base:[3,4,7,9,10,14,16,18,20,23,26,35,45,46,50,51,58,63,67,70,71,72,74,79,80,81,87,89,97,98,100,101,106,107,108,112,118,122,123],baseactiv:5,baseerrorclipattr:45,baseev:25,baselin:86,basematrix:100,basenam:3,basepoolingtyp:[4,5],basestr:[2,3,4,5,7,25],bash:[91,92,98,109,112,113],basic:[4,25,42,56,63,74,75,79,80,86,89,100],batch:[4,5,9,11,14,18,19,25,26,29,31,33,34,41,42,46,47,51,54,57,67,68,69,71,72,86,89,99,100,111,112,113,118],batch_id:[25,54],batch_im:54,batch_images_from_tar:11,batch_label:54,batch_norm:[54,86],batch_norm_lay:5,batch_norm_typ:4,batch_read:[33,78],batch_siz:[9,18,42,54,61,69,111],batch_szi:54,batch_z:54,batchnorm:[18,54,86],batchsiz:[4,100],bazel:30,bbbbb:33,bbox:3,bcd:4,bcebo:10,bcm2708:124,bdist_wheel:82,beacus:56,beam:[4,97,118],beam_gen:[4,97],beam_search:[25,69,97],beam_siz:[4,69,97,117,118,120],becaus:[3,4,10,26,29,30,31,36,50,69,73,77,78,81,83,84,88,89,90,97,98,100,103,106,111,112,120,122],becom:[43,44,83,87,107],been:[4,5,18,28,30,35,41,51,98,99],befor:[4,5,18,28,31,38,41,45,49,52,55,67,72,73,74,78,81,91,92,99,101,102,106,112,122,123,126],begin:[3,4,14,18,28,34,36,46,49,55,69,71,100,109],beginiter:[25,26],beginn:97,beginpass:[25,26],begintrain:26,behavior:[18,107],behind:[51,89,111],being:[18,28,38,45,51,76,78,103,106],belong:[3,4,42,83],below:[18,29,31,36,42,44,50,51,64,73,78,81,89,90,91,94,97,100,107,109,112,122,123],benchmark:[64,86],benefit:[5,38,39,69],bengio:16,besid:[4,10,42,60,67,71,94],best:[30,63,92,98,118],besteffort:113,beta1:[6,20],beta2:[6,20],beta:[18,54],better:[5,30,51,60,63,69,88,112,123,126],between:[3,4,11,18,25,28,30,31,36,42,44,50,51,63,66,71,73,80,83,92,101,103,112],bgr:11,bi_gru:5,bi_lstm:5,bia:[4,5,18,69,77,97,100],bias:[4,18,100],bias_attr:[4,5,18,77,97],bias_initi:18,bias_param_attr:5,biases_:100,biasparameter_:100,biassiz:100,bidi:113,bidirect:[4,5,86,97],big:[40,44,60,107,126],bigger:[18,31],bilinear:4,bilinear_interpol:4,bilinearfwdbwd:107,bin:[92,109,112,113],binari:[3,4,9,18,30,39,42,50,52,54,64,91,93,94,98,106,107,112],bind:[40,41,50,53,83,87],bioinf:18,bit:50,bitcod:123,bla:[92,123],black:[54,123],blank:[4,18,112,123],block0:60,block1:60,block2:60,block:[4,28,32,34,36,40,41,42,43,44,45,46,47,51,58,60,67,68,70,87,90,100,101,103,107,118],block_expand:86,block_i:[4,18],block_id:[40,47],block_x:[4,18],blockdesc:[29,55,77,79],blockdescbind:58,blockingcount:43,blueprint:69,bn_bias_attr:5,bn_layer_attr:5,bn_param_attr:5,book:[10,79,86,97,104,108],book_distribut:111,bool:[2,3,4,5,6,7,9,11,18,25,29,50,57,59,61,62,63,76,77,84,85,89,90,100,102,118,120],boost:[67,86,87,102],boot:[4,97,126],boot_bia:4,boot_bias_active_typ:4,boot_lay:97,boot_stat:89,boot_with_const_id:4,bootstrapp:126,borrow:[54,89],bos_id:[4,97],both:[1,2,4,5,11,18,26,29,30,31,38,42,44,50,51,54,58,60,67,69,71,74,76,84,86,87,97,100,101,103,104,107,109,111,112,123],bottl:71,bottleneck:[74,107],bottom:[25,86],bound:[4,18,60],boundari:42,boundri:3,box:[4,54,107],brace:[29,55],brain:38,branch:[4,18,26,29,30,42,51,57,75,82,94,99,101,104],breadth:118,break_if:89,brief:[30,36,50,87,103],briefli:107,bring:[51,60,103],broadcast:[18,31,71,79,126],broken:99,browser:[92,104,106,112],bsd:[41,71,98],bsp:41,bucket_nam:112,buddy_alloc:99,buf:34,buf_siz:[9,42,111],buffer:[9,34,41,63,64,72,78,83,108,118],buffer_s:9,buffered_read:78,bufsiz:9,bug:[99,112],build:[4,10,30,39,42,55,56,60,62,72,81,82,85,86,92,94,95,99,101,106,109,111,112,115,116,118,125],build_android:122,build_dict:10,build_model:54,builder:99,buildtool:82,built:[30,40,42,50,52,60,67,71,85,86,89,91,94,98,106,122,124,126],bulk:41,bunch:[64,107,109],button:[99,104,112],c11:65,c703c041:99,c99:66,c99e:112,cach:[50,91],cacul:[5,46,109],caff:[29,51],caffe2:[29,40,41,51],calcul:[3,4,5,14,18,27,28,31,36,43,46,50,60,92,97,100,107,109,118,120],calcut:60,calendar:74,call:[3,4,5,9,18,25,26,27,28,29,34,35,36,37,39,40,41,42,45,52,54,55,60,68,69,70,74,77,79,80,83,85,87,89,95,97,98,100,101,102,103,106,107,109,111,112,118],callabl:[2,4,9,10],callback:[45,100],caller:[27,106,112],can:[2,3,4,5,9,10,11,18,19,22,25,26,27,28,29,30,31,34,35,38,39,40,41,42,44,45,47,50,51,52,53,54,55,56,58,59,60,63,67,68,69,70,71,72,74,75,76,77,78,79,80,81,85,87,88,89,90,91,92,93,94,95,97,98,99,100,101,102,103,104,106,107,109,110,111,112,113,114,117,118,120,122,123,124,126],cancel:38,candid:[4,18,69,86],candidate_activ:18,cannot:[79,83,88,89,100,101],cantain:56,capabl:[50,73,79],capac:[81,112],capi:[65,91],capi_prvi:66,caption:69,captur:[4,110],card:[71,109,111],care:[5,39,60,78,86,87,92,117,118,126],carefulli:[86,118],caret:25,carpedm20:54,carri:18,cast:[50,88],cast_to_op_attr:85,cat:[9,11,92,114],categor:101,categori:[4,10,31],categorig:10,categoryfil:113,caus:[31,49,94,101],caution:[112,113],cbla:61,cc_:30,cc_binari:30,cc_test:30,cclient:37,cde:4,cdn:10,cduadevicecontext:[67,87],ceil:4,ceil_mod:4,cell:[4,5,18],cell_activ:18,cell_t_prev:18,cell_valu:18,center:11,center_crop:11,cento:[91,94,95,126],central:[81,111],ceph:[9,33,113],cephf:[33,39,48],cer:86,certain:[18,59,67,70,74,83,87,102,117],certif:[26,48,112],cffi:65,cfg:[60,113],cgo:65,ch1:41,chain:[9,28,55,100],challeng:[4,31,51,57,87],chan:41,chanc:[26,50,100],chang:[4,10,30,35,39,42,51,63,73,75,78,80,82,83,86,91,97,99,100,101,103,107,109,111,112,118,122],changes:63,channel:[4,5,11,18,40,107],channel_shar:4,chapter:[68,69,86,111],chapter_data:68,chapter_out:68,charact:86,characterist:120,check:[9,17,28,29,30,45,63,76,79,91,94,99,101,104,109,112,118,120],check_align:9,check_attr:85,check_eq:100,check_grad:[27,101],check_l:100,check_output:101,check_sparse_distribution_batch:[117,118],check_sparse_distribution_in_pserv:[117,118],check_sparse_distribution_ratio:[117,118],check_sparse_distribution_unbalance_degre:[117,118],check_styl:99,checker:79,checkgrad:118,checkgrad_ep:118,checkmark:126,checkout:99,checkpoint:[44,76],checksum:48,child:29,china:92,chines:104,chip:51,chmod:112,choic:[18,30,51,92],choos:[18,59,91,92,93,98,102,118,122],chosen:[54,67],chunk:[35,48],chunk_schem:3,chunktyp:3,chw:11,circl:55,circular:41,circumst:87,claim:112,claimnam:112,clang:[50,65,99,122],clarifi:[3,18],clariti:69,classdim:18,classic:[4,60,86],classif:[4,16,18,55,120],classifi:[4,54],classification_error_evalu:3,classification_evalu:3,claster:112,clean:[29,30,47,73,79,91,98,99],clear:[3,30,69,73,83,88,124],clearer:[73,77],clearli:83,cli:112,click:[94,99,104,106,107,112],client:[34,37,79],clip:[2,5,118],clip_op:45,clip_op_desc:45,clock:4,clone:[4,91,98,99,104,106,122,124],close:[78,99],cloud:[30,31,39,48,49,79,126],cloud_read:9,cludform:112,cluster:[9,25,26,29,31,36,42,86,113,117,118],cluster_test_fil:109,cluster_train:110,cluster_train_fil:109,cluster_train_v2:[110,114],cm469:112,cmake:[66,91,98,99,100,101,104,106,107,111,122,123],cmake_build_typ:[106,122,123,124],cmake_c:123,cmake_install_prefix:122,cmake_system_nam:[122,123,124],cmakelist:[30,61,62,100],cmatrix:[65,66],cmd:113,cmu:18,cname:112,cnn:[4,18,113],coars:53,code:[4,9,26,28,30,38,41,42,44,47,50,53,54,55,59,64,67,70,72,73,74,76,78,79,80,81,85,89,91,92,94,95,97,98,100,101,102,103,104,105,107,109,111,112,113],codebas:[79,99],coeff:4,collabor:31,collect:[4,10,25,74],collectbia:100,color:11,colour:10,column:[3,4,18,55,78,100,106],column_evalu:3,com:[4,5,10,18,30,54,82,91,92,98,99,104,106,108,112,113,122,124,126],combin:[3,4,5,9,20,25,60,70,79,83],come:[42,46,60,75,86,89],comma:[22,25,36,111,118],command:[9,22,30,34,39,49,91,92,94,95,98,99,100,101,104,105,106,107,110,111,112,113,114,115,116,122,123,124],commandlin:107,comment:[18,30,56,85,86,99,101],commit:[30,99],common:[11,16,20,23,33,81,87,97,100,117],commonli:[49,81,97,106,107,120,124],commun:[31,36,37,41,42,44,71,99,100,109,111,112],compani:51,compar:[27,30,40,79,98,100,101],comparison:[30,51,102],compat:[19,50,53,71],compil:[4,30,42,51,56,58,60,67,71,80,84,85,90,98,100,104,109],complaint:30,complet:[4,5,10,18,20,25,28,29,31,35,36,45,48,55,64,67,79,93,100,101,103,106,111,112,113,126],complex:[5,18,38,41,60,69,79,97,107],complic:[4,42,53,78,88,89],compon:[41,42,56,86,89,90,100,102],compos:[9,19,26,41,53,56,68,77,79],composenotalign:9,composit:53,compress:35,compris:28,compromis:98,comput:[4,5,18,19,22,26,27,31,41,42,44,47,50,51,52,56,60,64,67,70,71,72,74,80,83,86,87,88,90,92,97,98,99,100,101,102,106,107,108,109,111,112,120,122,123,124],computation:[4,97],computationgraph:56,con:71,concat:[54,97],concaten:[4,5,18,54,68,89],concentr:79,concept:[3,26,40,41,51,53,54,56,63,68,69,72,73,75,83,89,90,97],conceptu:[41,47,51,54,56],concern:[26,41,46,123],concis:[54,89],conckerneltrac:22,conclud:101,concret:[79,87,101],concurr:[31,38,44,74,109],concurrentremoteparameterupdat:118,cond:[18,29,51,57,75],condit:[4,18,35,42,51,57,63,86,97,113],condtion:54,conduct:107,conf:[4,110],conf_paddle_gradient_num:112,conf_paddle_n:112,conf_paddle_port:112,conf_paddle_ports_num:112,conf_paddle_ports_num_spars:112,confer:16,confid:4,confidence_threshold:4,config:[2,4,22,33,49,69,100,112,113,117,118,126],config_:[34,118],config_arg:[117,118,120],config_bas:[3,4,25],config_lay:100,config_len:36,config_pars:[61,62,100],config_proto:36,configmap:42,configur:[0,4,18,25,28,34,36,38,39,42,44,51,56,59,77,86,87,88,95,96,98,99,100,101,103,107,109,111,118,122,124,126],confirm:49,conflict:[83,99],confus:[11,54,59],congest:118,conll:10,connect:[5,18,39,40,42,44,86,100,109,111,112,113,126],connectionist:[4,18],consequ:[4,5],consid:[3,4,16,28,76,87,98,107,120,126],consider:[4,5,67,86],consist:[3,4,10,11,18,35,41,52,64,75,78,79,80,85,86,90,101],consol:[107,112],consolid:[29,104],constant:[4,16,18,56,58,59,67,100,101],constantiniti:18,constraint:83,construct:[3,18,26,47,56,60,68,77,79,83,85,97,102],constructbackwardgraph:55,constructoptimizationgraph:55,constructor:[18,45,50,74,77,79,83,85,100,101],consum:[31,106],consumpt:60,contact:38,contain:[3,4,5,7,9,10,11,18,25,26,28,29,35,47,54,56,63,64,67,73,74,77,79,80,83,84,85,86,89,90,93,94,97,98,101,109,111,112,114,123],container:109,containerport:112,content:[36,49,64,69,104,113],content_dir:104,content_len:36,context:[4,5,10,18,19,45,63,83,84,87,97,101,103,108],context_attr:5,context_len:[4,5],context_proj_layer_nam:5,context_proj_param_attr:5,context_project:5,context_start:[4,5],contin:112,continu:[3,28,31,64,86,109,118,122],contrast:4,contrib:81,contribut:[81,86,98,105],contributor:79,control:[2,29,40,41,112,113,118,126],controlflowgraph:60,conv2d:[54,102],conv:[5,18,54,63,88],conv_act:[5,19],conv_batchnorm_drop_r:[5,19],conv_bias_attr:5,conv_filter_s:[5,19],conv_fwd:63,conv_layer_attr:5,conv_num_filt:[5,19],conv_op:4,conv_pad:[5,19],conv_param_attr:5,conv_pool_2:42,conv_strid:5,conv_with_batchnorm:[5,19],conveni:[26,28,56,70,85,86],convent:[18,28,36,99,101],converg:110,convers:[50,51],convert:[10,18,33,42,43,44,50,51,52,63,78,80,86,111],convlay:4,convlut:86,convlution2d:18,convolut:[4,5,9,18,19,54,67,77,87],convolution2d:18,convolution_algorithm_opt:63,convoper:4,convproject:4,convtranslay:4,convtransproject:4,cool:99,cooper:86,coordin:[31,36],copi:[25,26,35,38,49,55,68,69,71,72,89,91,99,109,112,114],copy_from:45,copyvariablewithtensor:88,core:[2,14,18,28,56,59,66,72,73,89,98,108,118],coreo:[112,126],corespond:18,corner:79,corpu:[10,86],correct:[4,18,27,28,50,71,100,101,102,112],correctli:[3,9,28,50,54,100],corresond:50,correspend:18,correspoind:26,correspond:[4,18,19,23,26,28,29,30,45,50,56,57,63,67,68,69,77,79,80,81,85,87,100,101,102,103,106,123],correspondingli:123,corss_entropi:26,cortex:50,cos:[4,85],cosin:[4,18,85],cosineop:85,cosineopproto:85,cosineopprotomak:85,cost:[18,25,26,28,42,55,70,71,75,76,88,108,111,118],cost_id:4,cost_np:76,could:[4,9,25,26,27,35,40,41,42,43,44,50,51,52,68,70,72,73,75,77,78,80,98,102,106,107,110,112,122],count:[3,31,39,46,76,78,86,107,109,111,113,118,120],counter:[22,31,35,43,55],cours:[3,39,67,98],covari:4,cover:[51,86,103],cp27:94,cp27m:[82,94],cp27mu:[82,94],cpp:[27,34,53,61,62,65,66,73,79,90,100,102,107],cprofil:106,cprofilev:106,cpu:[2,4,27,39,50,59,67,72,73,74,79,81,82,87,88,91,92,98,101,102,103,106,107,108,113,118],cpu_avx_mkl:[92,94],cpu_avx_openbla:[94,95],cpu_kernel:59,cpu_noavx_openbla:94,cpu_ns_:74,cpu_per_pserv:42,cpu_per_train:42,cpudevicecontext:[67,87,101,102],cpuelapsedu:74,cpuengin:62,cpuinfo:92,cpuplac:[42,59,63,67,87,88,101,102,103,108,111],cpusparsematrix:66,crash:[31,107,110,118],creat:[2,9,14,18,25,26,27,29,31,36,40,43,45,46,47,48,49,50,51,53,54,55,63,67,68,70,71,72,73,77,80,81,86,92,95,98,99,100,101,104,109,111,114,118,122,126],create_backward_pass:70,create_bias_paramet:100,create_block:77,create_doc_str:85,create_input_paramet:100,create_local_scop:47,create_oper:53,create_optimization_pass:[20,70],create_paramet:77,create_python_ops_creatation_funct:85,create_rnn:29,create_rnn_op:68,create_st:14,create_tmp_var:77,create_tmp_vari:77,create_var:77,create_whileloop:89,creategradientoper:80,creatememori:63,createop:85,createoper:29,createprimitivedesc:63,createstack:112,createvari:29,creation:[53,112],creationd:112,creator:[9,10,33,79,80],creator_:80,credenti:49,crf:[87,88],critic:[54,106],crlf:99,crop:[11,87],crop_grad:87,crop_siz:11,crope:11,cropgradkernel:87,cropkernel:87,cross:[4,18,77,101],cross_compil:124,cross_entropi:[4,26,42,54,60,88],cross_entropy_with_selfnorm:4,crt:48,csc:100,csr:100,csv:22,ctc:[3,18],ctc_error_evalu:86,ctc_evalu:3,ctest:[91,98,101],ctor:77,ctrl:[98,110],ctx:[63,88,101,103],cubla:[67,102],cublas_handle_:87,cublashandle_t:87,cuda7:[94,95],cuda8:[91,92,94],cuda:[22,30,52,67,74,79,87,92,94,98,101,102,107,109,111,118],cuda_context:52,cuda_dir:[117,118],cuda_fp16:50,cuda_profil:22,cuda_so:92,cudaconfigurecal:107,cudadevicecontext:[52,67,87,101],cudadevicegetattribut:107,cudaelapsedu:74,cudaevent_t:74,cudaeventcr:107,cudaeventcreatewithflag:107,cudafre:107,cudagetdevic:107,cudagetdevicecount:107,cudagetdeviceproperti:107,cudagetlasterror:107,cudahostalloc:107,cudalaunch:107,cudamalloc:107,cudamemcpi:107,cudaplac:[67,87,88,102],cudaprofilerstart:107,cudaprofilerstop:107,cudaruntimegetvers:107,cudasetdevic:107,cudasetupargu:107,cudastream_t:87,cudastreamcr:107,cudastreamcreatewithflag:107,cudastreamsynchron:107,cudeviceget:107,cudevicegetattribut:107,cudevicegetcount:107,cudevicegetnam:107,cudevicetotalmem:107,cudnn:[4,7,18,30,59,63,67,87,88,102,118],cudnn_batch_norm:4,cudnn_conv:4,cudnn_conv_workspace_limit_in_mb:[117,118],cudnn_convt:4,cudnn_dir:[117,118],cudnn_kernel:59,cudnnavginclpadpool:4,cudnnavgpool:4,cudnnconvopkernel:102,cudnnv5:91,cudrivergetvers:107,cuinit:107,cumtim:106,cumul:4,cur_mem:69,curl:112,curli:[29,55],current:[4,18,28,29,30,31,34,36,40,44,46,51,59,67,68,69,72,73,74,77,83,88,89,92,94,97,98,100,104,110,111,112,118,123],current_block:[75,77],current_endpoint:111,current_oper:75,current_word:97,curv:26,custom:[20,26,39,50,54,69,72,79,86,100,112],custom_batch_read:78,cut:[9,89],cut_lin:9,cutoff:10,cv2:11,cxx:123,cxx_compil:[122,123,124],cxx_flag:123,cxxabi_1:94,cycl:31,cyclic:4,cython:65,d3e0:112,d_b0:54,d_b1:54,d_b2:54,d_block:54,d_f:54,d_g:54,d_h0:54,d_h0_bn:54,d_h0_relu:54,d_h1:54,d_h1_bn:54,d_h1_relu:54,d_h2:54,d_loss:54,d_loss_fak:54,d_loss_real:54,d_optim:54,d_step:54,d_t:54,d_w0:54,d_w1:54,d_w2:54,daili:99,dandroid_abi:122,dandroid_arm_mod:122,dandroid_arm_neon:122,dandroid_standalone_toolchain:122,dangl:98,dario:86,darwin:112,dash:54,dat:33,data:[0,3,10,11,14,25,26,27,29,33,34,35,41,44,46,48,50,51,54,55,56,58,59,60,63,64,67,68,69,70,71,72,73,75,77,79,81,83,84,85,86,87,89,90,95,97,100,101,103,107,108,109,111,114,115,117,118,120],data_fil:11,data_i:54,data_lay:34,data_layout:18,data_layout_:88,data_read:[9,78],data_reader_creator_random_imag:78,data_shar:89,data_typ:[9,10,64,84,86,88,90,95,97,102],data_type_:[59,67,88],data_x:54,databas:10,datacent:[33,49],datacenter1:33,datacenter2:33,datacenter_1:33,datacenter_2:33,datacenter_nam:33,datadim:4,datafeed:[12,108,111],dataflow:56,dataflow_analysi:60,datalayout:88,dataparallel:42,dataprovider_convert:86,datasci:4,dataset:[0,18,33,39,42,72,78,86,95,97,106,111,118],dataset_nam:11,datatransform:88,datatyp:[10,14,18,59,63,84,86,88,90],date:109,dcgan:54,dcmake_install_prefix:[122,123,124],dcmake_system_nam:[122,123,124],dcuda_arch_nam:91,dcudnn_root:91,ddim:[67,87,103],dead:31,deal:[28,126],debug:[27,28,42,49,51,77,92,99,106],debug_str:56,decai:[6,20,23],decar:9,decayr:34,decent:35,decid:[26,38,54,64,72,80,81,84],declar:[18,29,54,68],decod:[4,5,18,86,97],decoder_boot:97,decoder_dim:69,decoder_group_nam:97,decoder_input:[69,97],decoder_mem:[69,97],decoder_prev:5,decoder_s:97,decoder_st:[5,97],deconv:[4,54],deconvolut:4,decor:[9,100],decrement:43,decrementcount:43,decrypt:112,deduc:79,deep:[4,16,18,28,38,41,47,54,55,60,62,74,79,81,86,87,92,107,111,123],deeper:[52,92],deepspeech2:61,def:[4,9,26,27,28,33,39,45,46,53,54,56,59,60,68,69,70,77,78,85,89,97,100,101],def_block:54,defalut:[18,118,120],default_block:54,default_devic:120,default_main_program:[14,108,111],default_param_attr:77,default_st:89,default_startup_program:[14,108,111],default_valu:120,defaultdict:60,defaultinfervartyp:58,defect:73,defer:38,defin:[4,5,9,16,20,23,25,26,28,29,30,31,38,40,43,44,45,50,51,52,53,54,56,59,60,67,68,71,75,77,78,79,83,85,87,89,97,100,103,106,108,109,110,111,118],definit:[28,29,31,35,42,47,52,59,75,80,85,89,92,101,106,108],definiton:53,degener:18,degre:4,deivc:102,delai:[72,87,103,118],delet:[18,39,48,99],deletestack:112,delimit:3,deliv:126,delta:[4,27],delv:[4,16],demand:[31,87],demo:[4,10,79,110,113,115],demolish:113,demonstr:[97,103],denot:[18,101,120],dens:[4,9,36,37,84,86,100,112],dense_arrai:9,dense_vector:[9,95],dense_vector_sequ:9,densescann:86,dep:30,depart:86,depend:[18,29,30,31,39,42,44,56,71,76,84,92,98,101,109,120,122,123,124,126],dependent_var:76,deploi:[4,110,120,126],deploy:[56,64,79,109,110,112,123,126],deprec:[4,86],depth:[29,51,86],dequeu:44,deriv:[1,26,42,45,57,70,122],desc:[29,45,63,64,77,85,89],desc_:29,descend:[18,89],descent:[4,31,72,109],descproto:64,describ:[26,28,29,30,35,40,42,47,52,59,63,64,68,69,73,75,77,79,84,85,88,90,100,101,102,103,112,113],describestack:112,describestackev:112,describestackresourc:112,descripotor:63,descript:[3,29,30,58,62,64,67,80,84,86,88,90,91,94,99,101,109,112,119],descriptor:[41,63,88],deseri:[25,64,73],deserializ:79,desgin:55,design:[4,9,16,18,28,34,59,60,65,72,74,81,101,126],desir:[9,31,42,72,112,113],destin:[36,49],destroi:[29,47],destruct:83,destructor:[74,100],det_output:3,detail:[2,3,4,5,6,18,27,28,35,39,42,44,49,51,54,56,60,63,64,67,68,74,77,81,83,87,88,89,90,91,95,97,98,100,101,102,103,104,106,107,110,111,112,113,119,120,124,126],detect:[58,91,99,122],detection_evalu:3,detection_output:3,determin:[4,9,18,29,42,60,67,79,100],dev:[79,92,98,106,122,126],dev_ctx:[29,63,74],devel:82,develop:[28,30,51,58,73,74,77,80,82,86,92,94,99,103,104,106,108,109,117,118,123,124],deverlop:118,deviat:[2,16],devic:[2,40,42,46,50,56,62,63,67,71,73,74,79,88,92,101,103,108,118,123],device_:74,device_context:[63,101],devicecontext:[29,67,74,101,102],deviceid:[62,120],deviceid_:62,deviceplac:87,devid:[4,118],devot:86,devtools2:91,dhcp:126,diagnos:110,diagon:18,diagram:[68,109],diamond:54,dic:11,dict:[3,10,25,28,77,109,114],dict_fil:3,dict_siz:[10,18,34,69],dictionari:[3,4,10,18,25,26,27,77,120],did:[73,92],diff_mat:27,differ:[3,4,18,25,28,29,30,31,36,38,42,43,44,45,46,47,50,51,54,56,57,60,63,67,69,71,72,74,76,80,83,86,88,89,90,92,97,100,101,102,103,106,110,112,113,118,123],differenti:53,difficult:[3,27,51,98],difficulti:16,dig:[92,107,112],digit:[4,109],digraph:56,dilat:[4,18,63],dilation_h:18,dilation_i:4,dilation_w:18,dim0:101,dim1:101,dim:[4,9,18,19,34,63,64,68,79,84,87,90,100,101,103],dim_:[87,103],dimens:[1,4,5,7,9,18,19,54,79,84,86,87,89,100,101,103,120],dimension:[4,18,97,100,103],dimes:4,dios_arch:123,dios_enable_bitcod:123,dios_platform:123,dios_use_veclib_for_bla:123,dir:122,direcit:86,direct:[4,5,11,18,51,60,72,86,106],directli:[5,16,20,23,30,37,39,42,50,59,73,85,88,89,91,93,110,113,122],director:101,directori:[4,30,33,38,48,49,87,91,92,98,103,104,107,109,110,113,114,118,122,123,124],disabl:[18,74],disadvantag:[72,77],discard:[9,31,35,69,118],discount:4,discov:31,discoveri:112,discrep:107,discret:4,discrim:54,discuss:[26,29,35,36,37,42,63,86],disk:[64,98,113],dispatch:[42,73,109,110,118],displai:[39,49,99],dist:[82,91],dist_train:[26,39],distanc:[3,4],distinguish:[30,110],distribut:[4,16,29,35,36,37,38,40,41,46,52,71,79,86,90,94,105,110,113,115,116,126],distribute_test:[117,118],distributedli:[42,100],distributetranspil:111,disucss:26,div:18,divid:[4,6,18,28,46,85,90,106,117],divisor:18,diy_beam_search_prob_so:[117,118],django:104,dnn:[63,86,91],dns:112,do_forward_backward:78,doc:[9,56,68,89,101,103,104,109],doc_cn:104,dockefil:98,docker:[82,91,93,99,104,109,112,115,116,126],docker_build:26,docker_clust:[110,114],docker_push:26,dockerfil:[98,122,124],dockerhub:92,document:[4,5,18,27,42,48,55,68,69,74,79,86,91,98,99,101,102,103,105,109,111,120,123],doe:[5,31,35,36,38,39,40,42,44,47,50,56,60,68,73,77,79,80,81,94,98,100,101,103,107,108],doesn:[2,4,9,26,29,40,41,78,92,98,99,106,107,113,122,123],doing:[34,38,42,55,107],domain:112,don:[5,26,30,53,55,60,78,86,91,92,98,99,101,104,112],done:[3,4,5,28,30,31,35,36,42,43,58,60,64,72,80,81,86,99,106,107,112],dot:[4,5,19,101,118],dot_period:[118,120],dotmuloper:4,dotmulproject:4,doubl:[18,42,50,55,74,88,91,101,102,118],down:[86,107],download:[10,30,31,34,38,48,91,92,94,109,123,126],dozen:30,draw:69,drive:83,driver:[92,109,111],drop:[4,5,18,19,69],drop_rat:2,drope:18,dropout:[2,5,19,100],dropout_prob:18,dropout_r:[4,18,19],drpi_arm_neon:124,drpi_toolchain:124,drwxr:113,ds2:86,dst:[36,63],dst_primitive_desc:63,dtoh:107,dtype:[14,18,41,42,56,77,108,111],due:[35,38,54,60,69,77,106],dummi:[25,35],dump:64,duplic:[18,44],durat:[35,107],dure:[4,5,18,23,28,29,31,35,38,39,46,51,60,71,72,74,77,79,86,90,100,101,112,117,118,126],duse_eigen_for_bla:122,dwith_c_api:[66,122,123,124],dwith_distribut:111,dwith_doc:111,dwith_gpu:[91,111,124],dwith_profil:107,dwith_python:[66,111,124],dwith_swig_pi:[66,111,122,123,124],dwith_test:[91,101,123],dwith_tim:107,dynam:[18,36,66,68,77,78,91,107,118],dynamic_cast:100,dynamic_recurrent_op:89,dyogatam:18,e2e:126,each:[3,4,5,7,9,10,14,18,19,25,27,28,30,31,34,35,36,38,39,40,41,42,45,46,47,52,55,58,60,63,67,68,69,71,73,74,76,77,78,79,80,83,84,85,86,87,88,89,90,97,100,102,106,109,110,111,112,118,120,126],each_feature_vector:1,each_time_step_output:1,eager:51,earli:[50,52,99,101],eas:[9,58,101],easi:[27,28,69,72,78,79,81,99,100,110],easier:[26,44,50,51,78,89,98,99,100],easili:[26,54,71,74,78,80,83,87],echo:92,edg:[11,60],edit:[3,41,92,98,112],editor:[77,98],edu:[10,18,112,113],eeoi3ezpr86c:112,effect:[4,18,25,91,112,118,123],effici:[4,42,64,78,86,87,97,98,100],effort:[42,86],efg:4,efs:112,efs_dns_nam:112,efsvol:112,egd:60,eigen:[50,67,72,79,81,87,101,122,123],eigen_device_:87,eigen_test:103,eigen_use_gpu:101,eigenmatrix:103,eigenscalar:103,eigentensor:103,eigenvector:103,either:[4,5,9,18,25,26,42,54,57,58,68,72,81,93,107,123],elabor:86,elb:112,elbapis:112,electr:60,electron:113,elem_dim:4,elememt:4,element:[3,4,5,9,11,18,25,27,35,41,44,56,69,79,101,103],element_typ:[36,102],elementari:79,elementwis:[18,19],elif:[26,85],els:[26,34,39,41,42,44,45,51,54,57,58,59,60,83,85,91,92,98,100,101],elsewher:74,emac:98,email:99,emailweixu:30,emb1:34,emb2:34,emb:113,embed:[26,29,34,44,58,69,84,89,97,109],embedding_lay:34,embedding_nam:[4,97],embedding_s:[4,97],emphas:107,empir:[4,18],emplace_back:100,emploi:[28,45,85,97],empti:[3,9,28,31,69,101],emul:50,enabl:[2,4,18,29,30,35,40,44,45,56,74,98,99,107,109,112,118,123],enable_grad_shar:[117,118],enable_parallel_vector:118,enableonstart:22,enc_proj:[5,97],enc_seq:5,enc_vec:97,encapsul:36,encod:[5,35,69,97],encoded_proj:[5,97],encoded_sequ:[5,97],encoded_vector:97,encoder_ctx:69,encoder_ctx_expand:69,encoder_ctx_proj:69,encoder_dim:69,encoder_last:4,encoder_out_seq:69,encoder_s:97,encount:[18,34],encourag:[42,47],encrypt:112,encrypt_decrypt:112,end2end:126,end:[3,4,18,25,28,29,42,45,52,56,60,69,73,74,78,83,86,94,97,98,99,118],end_pass:26,end_po:4,endforwardbackward:25,endian:64,endif:[67,74],enditer:[25,26],endpass:[25,26],endpoint:[9,33,112],endtrain:26,enforc:123,engin:[39,62,63,86,107],english:[4,86,104],enjoi:92,enough:[28,29,59,60,67,98],enqueu:44,ensembl:5,ensur:[31,63,71,83,92,94,98,100,123],enter:[29,47],enterpris:79,entir:[4,5,36,38,101],entiti:[3,29,83],entranc:47,entri:[9,18,35,39,58,98,99,100,112,122],entropi:[4,18,77],entry_point:39,enumer:[1,67],env:[104,106,112],environ:[26,42,91,94,98,99,106,107,109,112,113,117,118,123],environmenterror:109,eos_id:[4,97],epoch:54,epol:41,epsilon:[4,6,18,20],equal:[4,5,18,19,31,89,101,102,118],equat:[3,4,5,6,18,60,101],equip:97,equival:[26,29,40,45,51,57,85,126],erlang:41,error:[2,3,4,5,18,26,27,35,49,50,51,63,83,86,100,101,110,112,118],error_clip:45,error_clip_callback:45,error_clipping_threshold:2,errorclipbyvalu:45,espeaci:18,especi:[4,5,62,98],essenc:[26,28],essenti:[4,26,47,50,103],establish:40,estim:[4,26,44,72],eta:113,etal:18,etc:[3,9,18,29,41,42,46,63,71,72,78,83,86,91,102,109,112,117,120,126],etcd:[9,25,31,35,36,38],etcd_addr:36,etcd_endpoint:9,eth0:112,etyp:41,euclidean:4,eval:[3,14,29,46,54,79],eval_program:[14,46],eval_result:46,evalu:[4,12,18,24,25,38,56,76,86,107,108],evaluate_difficult:3,even:[26,50,71,77,78,98,99,107,118,123],evenli:[36,112],event:113,event_:74,event_block:74,event_handl:[25,26],eventkind:74,eventlist:74,eventu:[42,89],everi:[3,4,5,9,14,26,31,35,36,38,45,46,55,56,58,60,63,67,68,71,77,83,85,88,97,99,100,101,103,108,109,114,118],everyon:99,everyth:[42,44,54,122],everywher:98,evid:73,evolv:51,exactli:[4,5,112],exampl:[3,4,5,9,10,11,18,19,25,29,39,42,44,46,49,51,52,53,54,55,56,58,60,63,67,68,69,73,74,75,77,78,79,80,81,84,87,88,89,97,98,99,100,101,102,103,106,107,108,109,111,112,113,117,118,120,124],example_read:9,exceed:4,except:[4,10,38,40,51,55,74,86,89,120],excess:60,exchang:73,exclud:4,exclude_mod:4,exclude_param:25,excluded_chunk_typ:3,exconv:4,exconvt:4,exdb:10,exe:[42,108,111],exec:118,execut:[4,30,31,35,39,40,41,42,46,47,52,54,56,60,63,71,74,80,90,98,100,106,107,112],executioncontext:[63,88,101,102,103],executor:[12,14,40,42,46,50,51,52,54,70,75,77,88,90,106,108,111],exist:[26,29,31,49,51,69,77,78,80,85,87,89,94,98,100,103,112,118,123],exit:[36,49,111,113,118],expand:[18,69,92,100],expand_a:4,expand_level:4,expandconvlay:4,expans:4,expect:[4,88,107],expected_desc:63,expected_kernel_kei:88,experi:[64,86,120],experienc:99,expert:30,expir:31,explain:[3,18,31,40,51,53,55,99,102,106,109,111],explan:[4,18,27,39,40,42,83,88],explicit:[74,89,100,102],explicitli:[26,42,47,101,103,123],explod:45,explor:[4,69,81],expon:4,exponenti:[1,18],expos:[28,37,41,63,64,87,89,112],express:[26,44,46,56,60,101,112],extend:[3,72,89],extens:[38,44,69,101,122],extent:66,extern:[30,62,65,66,79,86],external_librari:30,extingrad_:62,extinval_:62,extoutgrad_:62,extoutval_:62,extra:[2,4,5,42,81,87,126],extra_lay:25,extraattr:[2,120],extraattribut:4,extraattributenon:4,extract:[3,4,18,51,73,86,101,112],extralayerattribut:[2,5],extralayeroutput:5,extrem:[4,40,51,107],f120da72:113,f7e3:112,fa0wx:113,fabric:109,face:[30,81],fact:[18,40,51,71,75,77],factor:[2,6,18],factor_s:4,factori:65,fail:[31,35,69,101,113,118,120],failur:[31,36,101],fake:54,fake_imag:78,faked_imag:54,fall:[50,76],falloc:48,fals:[2,3,4,5,6,9,18,19,20,27,28,29,51,57,59,61,68,75,76,78,84,90,95,97,100,101,109,113,118,120],false_block:[29,57,75],false_label:78,false_neg:46,false_posit:46,false_read:78,fan_in:16,fan_out:16,far:[45,89],fashion:42,fast:[4,35,51,107],faster:[4,5,18,23,31,51,92,97,107],fastest:51,fastli:99,fat:123,father:28,fault:[25,35,79,91],favorit:98,fbd1f2bb71f4:113,fc1:[56,100,120],fc1_bia:56,fc1_weight:56,fc2:[56,120],fc3:[56,120],fc4:120,fc8a365:112,fc8a:112,fc_act:5,fc_attr:5,fc_bias_attr:5,fc_layer:[77,85,100,120],fc_layer_nam:5,fc_mat:25,fc_op:85,fc_out:[18,29],fc_output:85,fc_param_attr:5,fc_without_b:29,fclayer:100,fcop:53,feasibl:72,featur:[1,4,9,10,18,28,42,50,56,71,74,86,99,118],feed:[5,25,26,42,55,68,81,108,111],feed_dict:54,feed_list:[108,111],feeder:[9,42,108,111],feedforward:16,feel:99,fetch:[10,31,34,42,76,97,100,108],fetch_list:[42,77,108,111],fetch_op:76,few:[30,31,41,42,60,72,78,84,86,98],fewer:[4,18,41,77],fft:86,fg0:4,field1:25,field2:25,field:[4,25,29,56,58,64,76,77,80,84,85,107,112],fifth:55,figur:[26,30,42,44,54,62,68,74,77,86,97,100,107],file:[3,4,9,10,11,22,25,26,28,30,31,33,35,36,38,39,41,48,49,51,52,56,64,66,78,79,86,87,90,92,94,95,97,98,99,100,101,102,103,108,109,110,111,114,118,122,123,124,126],file_typ:9,filelist:86,filenam:[11,33,77,106],fileoffset:48,filesystem:[38,39,42,48,112],fill:[4,31,35,67,77,112],fill_zero_grad:79,fill_zeros_like_op:28,filter:[4,5,18,45,63],filter_s:[4,5,18,19],filter_size_h:18,filter_size_i:4,filter_size_w:18,filter_strid:18,find:[4,18,29,31,38,41,50,56,63,69,83,88,91,94,107,111,114,122,123],find_var:27,findmemori:63,findop:29,findprimit:63,findprimitivedesc:63,findvar:[29,83],fine:[2,35,53],fingerprint:112,finish:[31,35,38,39,47,60,71,85,91,109,110,112,113],finit:100,finnal:92,first:[4,18,25,26,28,29,31,35,38,39,40,42,47,49,51,54,55,56,63,68,69,75,76,77,79,84,85,86,87,89,91,97,98,99,100,101,103,107,111,112,118,120,126],first_seq:97,firstli:[3,4,102],firstn:9,firstseen:113,fit:[10,50,59,60,64,69,79,111],fit_a_lin:111,five:[75,107],fix:[2,4,18,42,60,65,77,86,99,106],flag:[4,10,18,61,62,74,99,101,104,118],flatten0:56,flatten:[18,56,75,77,103],flexibl:[4,5,26,36,42,51,55,59,68,69,72,78,87,89,97],flip:11,flist:109,fliud:40,float16:41,float16_t:50,float32:[9,18,42,50,53,54,77,78,101,108,111],float64:18,float_16:18,float_to_half_rn:50,floor:4,flow:[18,29,40,41,68,74,82],fluid:[0,14,16,17,18,19,20,22,23,28,42,44,47,67,74,77,87,88,102,106],fluid_cuda_create_tensor:52,fluid_cuda_mult:52,fluid_cuda_read:52,fly:28,fnt03:112,focu:[41,56,106,107],focus:101,folder:[30,33,39,49,111,112],follow:[3,4,5,6,9,11,18,19,25,26,27,28,29,30,31,35,39,40,41,42,44,47,50,51,52,53,54,55,56,57,58,60,63,67,68,69,71,72,74,75,76,77,78,79,80,81,83,84,85,86,87,88,89,91,92,94,95,97,98,99,100,101,103,104,106,107,108,111,112,113,114,115,116,120,122,123,124,126],footprint:52,forbid:26,forc:[71,77,88],force_cpu:[18,59],force_cudnn:59,force_load:65,forest:29,forget:[6,18,26],forget_bia:18,fork:[4,99],form:[4,5,18,41,46,94,107],formal:88,format:[3,9,10,11,18,22,25,27,35,42,50,51,67,69,86,89,95,99,100,101,103,109,112,118],former:[26,30,51,60,72],formul:18,formula:[4,5,6,18,27,60],formular:4,forth:54,forward:[1,4,5,18,27,28,29,34,36,45,51,54,61,62,63,64,70,73,75,78,79,80,81,84,97,100,120],forward_infer:63,forward_list:74,forward_op:27,forward_proj:18,forward_train:63,forwardactiv:100,forwardbackward:25,found:[50,75,81,83,97,102,109,111,124],four:[3,18,46,51,55,63,67],foward:76,fp16:[50,79,90],fp32:[67,79,88,90],fp64:[67,90],fpga:[67,108],fpgadevicecontext:87,fpgaengin:62,fpgaplac:[67,87],frac:18,frame:[3,47,79,86,89],framework:[26,28,29,41,45,46,50,51,56,67,71,72,74,75,79,81,83,85,87,99,100,101,106,108,109,123],free:[10,52,87,99,126],freememoryop:52,frequenc:[10,86,107],frequent:[35,78,79,81,87,110,122,123],fresh:38,friend:83,friendli:54,from:[3,4,5,9,10,11,16,18,25,27,28,29,30,31,33,34,35,36,40,41,42,44,45,46,49,50,51,53,54,55,56,57,59,60,63,68,69,70,71,73,75,77,78,79,80,83,86,87,88,89,92,94,97,98,99,100,101,102,103,106,107,109,111,112,113,118,120,122,123,126],from_no_sequ:4,from_sequ:4,from_tar:25,fromfil:78,front:[56,60],fuction:22,fulfil:107,full:[4,18,31,38,68,71,72,97,100,102,126],full_matrix_project:[5,97],fulli:[18,42,44,86,100,107,111,126],fullmatrixproject:4,fullsiz:34,fully_matrix_project:5,fullyconnect:[56,77],fullyconnectedlay:100,func:[9,35,40,52,80],funciton:[5,18],functor:[53,56],fundament:[41,44,50,79],funtion:18,further:[4,85,126],furthermor:18,futur:[4,18,38,42,50,60,68,79,122],future_context_s:18,fvs:85,fwd_desc:63,fwd_op:80,fwd_primit:63,fwd_primitive_desc:63,fwd_var:45,g_b0:54,g_b1:54,g_b2:54,g_block:54,g_command_config_arg:[61,62],g_h0:54,g_h0_bn:54,g_h0_relu:54,g_h1:54,g_h1_bn:54,g_h1_relu:54,g_h2:54,g_im:54,g_loss:54,g_optim:54,g_program:77,g_state:74,g_step:54,g_w0:54,g_w1:54,g_w2:54,gain:4,gan:26,gangliao:30,gap:118,gate:[4,5,18,19],gate_act:[4,5],gate_activ:18,gate_attr:4,gate_bias_attr:4,gate_param_attr:4,gate_recurr:4,gate_v:18,gatedrecurrentlay:61,gather:[4,18,28,60,71,73,100,101],gauss:2,gaussian:16,gaussian_normal_random:54,gcc:[50,52,65,79,91,98,106,122,124],gcc_3:94,gcreators_:85,gemm:61,gemmconvkernel:102,gen:4,gender:10,gendrated_id:69,gener:[3,4,5,9,14,25,26,27,28,29,30,31,33,35,36,38,40,42,51,53,58,60,63,67,71,72,75,76,77,78,79,80,84,85,86,87,89,99,101,107,112,114,118,120,122,124],generated_id:69,generated_scor:69,generated_word_embed:4,generatedinput:[4,97],gereat:3,get:[3,4,10,18,20,25,27,28,29,30,31,35,36,38,39,48,51,54,56,59,60,61,62,63,67,68,69,74,77,79,80,83,85,88,89,91,92,94,97,99,100,101,102,106,107,109,110,112,114,121],get_all_op_proto:85,get_block:77,get_config_arg:120,get_data:113,get_dict:10,get_dim:27,get_embed:10,get_float_el:27,get_grad:25,get_grad_op_desc:28,get_input_lay:100,get_lin:9,get_movie_title_dict:10,get_numeric_gradi:27,get_numerical_gradi:27,get_output:27,get_program:60,get_pserver_program:111,get_shap:25,get_startup_program:111,get_support:94,get_symbol:56,get_tensor:27,get_trainer_program:111,get_vari:29,get_word_dict:10,get_worker_addr:40,getactualkerneltyp:59,getattr:45,getbatchs:100,getdeviceid:102,geteigendevic:103,getengin:63,getenv:[26,39,109],getexpectedkerneltyp:[59,63,88],getinfervartyp:58,getinput:100,getinputgrad:100,getinputvalu:100,getkerneltyp:50,getkerneltypeforvar:88,getlayeroutput:25,getlibrari:63,getmat:34,getoptconfig:34,getoutputgrad:100,getoutputvalu:100,getparam:34,getparameterconfig:34,getparameterptr:100,getparameterspars:34,getparametersremot:34,getplac:[63,87,101,102,103],getsiz:100,gettask:35,gettempl:112,gettensor:88,gettranspos:100,getw:100,getweight:100,getwgrad:100,gist:5,git:[82,91,98,99,104,122,124],github:[5,18,30,54,67,82,91,98,99,104,106,108,122,124],give:[18,31,68,77,79,88,98,99,100,107,112],given:[4,9,18,19,25,28,36,38,41,44,45,51,53,54,69,78,81,89,100,118],glibc:[94,122,124],glibc_2:94,glibcxx_3:94,glide:30,global:[2,18,20,26,29,30,31,52,56,59,73,74,79,83,85,87,88,98,107,112,118],global_block:77,global_learning_r:[2,20],global_pool:18,global_step:20,globalstat:107,globalstatinfo:107,glog:99,glog_v:99,glog_vmodul:99,glorot10a:16,glorot:16,gnueabihf:124,go_librari:30,go_test:30,goal:[41,44,50,55,71,79,86,107],gob:35,godep:30,godoc:65,goe:[5,31,51,57,83,108],going:[28,53,72,106,109,126],golang:30,good:[41,54,72,77,78,81,106,107,111,126],googl:[18,26,74,79,99,106,109,122],googleapi:112,googlenet:62,goroutin:[40,41],got:[59,83],gpg2:112,gpg:112,gprotos_:85,gpu:[2,4,7,27,39,41,46,50,60,67,71,72,73,74,79,81,82,87,88,91,94,95,98,102,103,105,108,109,111,126],gpu_id:[118,120],gpu_per_train:42,gpudevic:87,gpugpu_id:117,gpukernel:79,gpustarttimestamp:22,grab:31,grad:[27,28,36,45,62,77,84,118],grad_info_map:28,grad_n:45,grad_nam:45,grad_op:45,grad_op_class:79,grad_op_desc:45,grad_op_maker_:80,grad_op_typ:[79,80],grad_op_type_:80,grad_s_block:28,grad_share_block_num:[117,118],grad_to_var:[28,45],grad_var_nam:27,gradient:[2,3,4,6,16,18,20,23,25,31,35,41,43,45,55,58,70,71,72,73,77,79,84,101,106,109,111,118],gradient_clipping_threshold:2,gradient_evalu:3,gradient_flat:27,gradient_machin:[25,66],gradientmachin:[25,66,73],gradientmachine_:34,gradopdescmak:[58,80],gradopdescmakerbas:80,gradopmak:80,gradual:107,grai:11,grain:53,gram:86,grandient:25,grant:112,graph:[4,18,25,28,29,30,31,40,41,42,43,44,46,51,54,68,71,72,75,103],great:[44,86,126],greater:[4,45,72,109],greaterthan:85,greedi:[18,86],green:[40,54],grep:[92,114],gridsize3d:22,groudtruth:97,ground:[3,4,18],group:[5,18,19,35,56,63,87,101,126],group_input1:97,group_input2:97,group_input:97,grouplen:10,grow:99,grpc:126,gru:[4,18,69,86,97],gru_bias_attr:5,gru_decod:97,gru_decoder_with_attent:97,gru_layer_attr:5,gru_memori:5,gru_out:69,gru_param_attr:5,gru_step:[69,97],gru_step_lay:5,grumemori:[5,97],gserver:[4,61,62,100],gsizex:107,gtx:60,guarante:[63,77,100],guard:34,guest:[94,98],gui:[106,107],guid:[22,48,60,79,97,99,100,107,111,112,113,123],gzip:[35,113],h0_bn:54,h_0:18,h_f:18,h_prev:29,had:98,hadoop:26,half:[4,18,19,50,112],half_to_float:50,hand:[60,79,86,87,103,109,111],handi:30,handl:[9,26,28,39,40,42,56,60,63,67,73,78,83,87,89,102,108],handler:[25,29],hannun:86,happen:[18,35,85],hard:[42,51,69,86,89,98,112],hardshrink:18,hardsigmoid:18,hardwar:[51,52,87,98,102,107],has:[3,4,5,10,18,19,22,26,27,28,29,30,31,35,36,38,41,42,44,45,46,50,51,54,56,60,64,67,69,71,74,75,79,84,85,87,88,91,97,98,99,100,107,108,112,113,123,126],has_kei:[25,28,45],has_selected_colum:4,has_var_recurs:28,hasdependentvar:76,hash:[67,71],hasn:51,have:[4,5,9,18,19,26,27,28,29,30,31,35,36,38,39,41,42,44,45,47,50,51,52,53,54,55,59,60,63,64,67,68,69,71,72,73,74,75,77,78,79,80,83,84,86,87,88,90,91,92,97,98,100,101,102,107,109,112,118,120,123,124,126],haven:[51,98],hdf:[9,33],head:[19,99,101,109,114],header:[36,64,66,79,87,100,102,122,123,124],headip:114,heard:98,heavi:110,height:[4,9,11,18,29,65,78,100,101],height_:84,held:31,hello:26,help:[4,18,29,49,51,56,63,69,78,79,89,98,99,106,110],helper:[18,42,63,80,89,100],henc:[42,72,77,80,81,83],here:[2,3,4,5,9,18,19,26,30,31,37,41,44,45,47,49,51,55,56,63,67,68,78,81,85,91,92,94,97,99,101,102,104,109,110,111,112,113,117,120,123,124,126],heterogen:[42,44,74],heurist:[4,44,69,118],hidden:[4,5,18,70,77,97,112],hidden_dim:18,hidden_out:29,hidden_s:5,hidden_t_prev:18,hidden_v:18,hidden_valu:18,hierarch:[4,75,77,79,97],hierarchi:79,high:[2,16,50,71,86,87,100,109,126],higher:[53,68,89,99],highest:[9,29],highli:[10,86,89,97,120],him:26,hint:[59,106],histor:[53,102],histori:6,hl_get_sync_flag:100,hold:[26,28,31,35,37,41,50,54,56,58,60,83,85,87,88,103,111,112],holder_:[87,103],home:[33,49,92,106,112,113,114],honor:35,hook:2,hookattr:2,hookattribut:2,horizont:[4,11],host:[30,39,74,112,113,122,123,124],host_c:[122,123,124],hostfil:114,hostnam:112,hostpath:113,hostport:112,hot:18,hour:98,hourli:99,hous:[10,95],how:[2,4,18,26,29,31,35,40,41,42,47,49,51,53,56,59,63,68,69,73,74,81,85,88,97,98,102,106,109,111,112,113,118,121,124],howev:[4,5,18,27,28,38,41,42,47,51,60,67,72,73,77,78,80,81,84,85,86,87,88,97,112,117,118],howto:109,hpp:[50,65],html:[10,16],htod:107,http:[4,5,10,16,18,30,39,54,82,91,92,98,99,104,106,108,112,113,122,124,126],hub:82,huber:4,huge:72,human:[4,16,74,86],hundr:102,hwc:11,hyper:[4,54,100],hyperparamet:[4,81],hyperplan:9,i1117:107,i386:123,iOS:[124,125],iamfullaccess:112,iamusersshkei:112,icc:52,iclrworkshop2016:18,icml:86,ics:10,id_input:3,id_rsa:114,idea:[30,41,51,52,72,78,81,106,111],ideal:[42,88],ident:[4,18,80,101,112],identifi:[4,57,67,100],identityoffsetproject:4,identityproject:4,ids:[3,4,18,69,100],idx:[35,54,60,100],ies:49,if_else_op:28,ifdef:[67,74],ifels:[29,75],ifelseop:75,ignor:[4,18,118],iii:86,illustr:[3,18,31,36,42,53,68,97,100,107],im2col:18,im_siz:54,imag:[7,8,9,10,18,19,26,42,51,54,55,69,70,75,78,86,91,98,99,112,115,116,120,126],image_a:78,image_b:78,image_conv_lay:86,image_fil:78,image_h:18,image_lay:78,image_nam:26,image_path:78,image_reader_cr:78,image_w:18,imagenet:[4,16,33],imagepullpolici:112,imageri:4,images_reader_cr:78,imagin:55,img2label:11,img:[4,5],img_conv_lay:5,img_pool_lay:5,imgsiz:107,imgsizei:107,imgsizex:107,imikolov:109,immedi:[60,63,72,81,91,112],immutable_paramet:26,imper:40,imperfect:79,implement:[4,5,9,16,18,19,20,23,29,35,36,37,38,39,40,41,42,44,51,53,56,57,58,60,63,65,66,67,69,73,76,83,85,86,87,88,89,97],implemet:34,impli:30,implicitli:40,imposs:[69,126],impractic:88,improv:[4,43,44,60,79,86,106,107,112],in_fals:18,in_plac:18,in_tru:18,inarg:34,inbound:112,includ:[3,4,5,10,11,18,20,26,29,30,36,39,41,50,51,54,56,60,65,66,68,69,74,75,77,79,85,91,94,97,98,100,101,106,107,109,112,113,118,122,123,124],inclus:[18,69],incom:[40,59],incorpor:4,incorrect:4,increas:[31,35,50,109,118],increment:[46,55,60,118],incupd:100,inde:[9,41],independ:[4,18,27,28,36,43,83,87,126],index:[3,4,7,9,10,18,25,27,28,29,31,35,40,75,77,89,102,112],indexslot:4,indiact:18,indic:[3,4,18,28,29,36,47,54,68,75,80,84,87,89,110,112,122],indice_map:89,indices_map:89,individu:[31,71,112],industri:[31,64,126],ineffici:[73,88],infer:[0,11,18,26,28,29,31,46,51,57,58,59,60,61,65,67,76,77,79,84,86,88,95,123,124],infer_shap:77,infer_var_type_:58,inferenc:123,inferer:86,inferfac:58,inferior:38,infernec:124,infershap:[29,77,79,101,103],infershapecontext:[101,103],infervartypefn:58,info:[3,4,10,50,68,100,110,126],infom:4,inform:[4,10,18,25,29,39,49,56,59,60,63,64,67,68,71,77,81,83,84,99,100,101,103,106,107,112,118,122],infrastructur:[51,112],ingor:118,ingrad_:62,ingredi:[41,86],inherit:[29,70,79,87,101],ininst:26,init:[2,16,25,29,43,54,62,68,69,95,100,109,112,120],init_attr:77,init_from_tar:25,init_model_path:[117,118,120],initi:[2,4,5,10,12,18,25,28,30,35,40,42,43,44,46,55,68,71,72,77,81,85,89,95,97,100,101,108,118],initial_max:2,initial_mean:[2,4],initial_min:2,initial_std:[2,4],initialize_op_attr:77,initrd:126,inlcud:5,inlin:[87,102,103,112],inner:[4,100],inner_param_attr:5,inproj_attr:4,inproj_bias_attr:4,inproj_param_attr:4,input0:103,input1:[4,5,103],input2:4,input:[1,3,4,5,7,9,11,17,18,19,25,27,28,29,34,38,40,42,43,44,45,46,50,51,52,53,54,55,56,58,59,60,62,63,67,68,69,72,73,76,77,78,79,80,83,85,86,87,88,89,95,97,99,100,101,102,103,108,111,114,120],input_conf:4,input_data:100,input_data_target:100,input_dim_idx:18,input_dtyp:18,input_featur:1,input_hassub_sequence_data:100,input_id:4,input_imag:5,input_index:100,input_label:100,input_lay:100,input_loc:4,input_nam:26,input_proj_bias_attr:5,input_proj_layer_attr:5,input_seg:89,input_seq:[4,18],input_sequence_data:100,input_sequence_label:100,input_sparse_float_value_data:100,input_sparse_non_value_data:100,input_t:100,input_to_check:27,input_valu:27,input_var:[27,77],inputbuff:34,inputdef:100,inputgradi:80,inputlayers_:100,inputs_to_check:27,inputsizechang:63,inputtyp:9,insert:[28,45,52,71,76,79,80,99],insid:[3,5,28,31,42,44,45,46,59,63,73,74,78,79,80,92,112],inspir:74,instal:[4,18,39,62,82,91,92,98,99,104,106,109,113],install_android:122,instanc:[4,18,27,29,31,33,37,40,42,43,45,47,52,57,63,68,69,72,77,79,80,97,100,103,107,111,118],instance_ip:112,instanti:[31,47,108],instead:[4,5,7,28,30,34,39,40,41,42,50,51,55,56,86,98,99],instrins:50,instruct:[29,55,92,98,107,122],int16:90,int32:[18,67,75,89,90,118],int64:[18,42,48,67,84,88,90],int64_t:74,int8:67,integ:[3,4,9,18,35,39,40,50,65,69,100],integer_valu:9,integer_value_sequ:[9,69,86,97],integr:[3,91,126],intel:[51,67,87,102],intellig:[16,60],inteloptimizedpaddl:62,intend:91,intens:86,inter:[4,42],interact:[4,42,92,112],intercept:4,interchang:[55,79],interconnect:71,interest:[40,50,71,107],interfac:[0,2,4,5,16,20,22,23,25,29,35,39,49,56,71,73,79,80,86,87,91,101,103,112,123,126],intergr:4,intermedi:[18,42,49,52,54,60,70,86,98,122,124],intern:[4,5,16,20,25,50,86,106,109,110,112],internel:62,internet:[30,31,126],interpret:[3,18,47,51,52,90,91,107],interv:18,inth:103,intrins:[40,47,50,124],introduc:[4,11,29,31,54,61,64,81,83,85,101,106,109,113],introductori:98,intuit:[38,79],inval_:62,invalid:[78,83],invent:51,invoc:[30,53,79],invok:[4,14,25,28,42,45,73,77,79,80,85,88,98,99,107,112],involv:[69,101],iob:3,ioe:3,ios:123,ios_arch:123,ios_deployment_target:123,ios_development_root:123,ios_enable_bitcod:123,ios_platform:123,ios_sdk_root:123,ios_use_veclib_for_bla:123,ipad:123,iphon:123,ips:112,ipt:[4,77,85,97],ipx:126,ipython:26,is_color:11,is_cpu_plac:63,is_loc:25,is_mkldnn_librari:63,is_revers:18,is_seq:[4,97],is_spars:18,is_stat:2,is_target:76,is_tensor:85,is_test:[18,63],is_traget:76,is_train:11,isinst:45,ismkldnnkernel:63,isn:107,isspars:100,issu:[18,30,54,86,92,94,98,99,107],issuecom:18,istag:82,item:[4,9,18,25,38,50,78,95,126],iter:[4,5,6,9,25,26,31,42,51,52,60,63,72,74,78,86,89],iter_multiple_input_and_param:77,its:[4,5,16,18,23,25,26,27,28,29,31,35,40,41,44,45,46,51,52,54,55,56,58,60,64,68,69,71,72,73,76,77,79,80,83,84,85,87,88,94,100,101,102,103,107,109,112,118],itself:[28,31,38,52,63,72,83],ivs:85,java:[29,65,75,79],jeremi:107,jian:16,jku:18,job:[10,28,38,40,42,45,79,92,109,111,117,118,120],job_desc:42,job_dispatch_packag:110,job_id:10,job_nam:[39,112],job_namespac:112,job_path:112,job_workspac:110,jobdesc:42,jobnam:42,jobpath:112,jobport0:112,jobport1:112,jobport2:112,jobport3:112,jobserv:39,join:31,jointli:5,journei:92,jpg:11,json:[56,86,112,113],jth:5,judg:4,juditski:72,jupyt:[39,92],just:[1,3,4,5,10,18,30,35,36,40,42,51,52,54,58,63,72,73,77,78,79,80,81,83,84,91,94,98,99,110,112,120,122],jx4xr:112,jypyt:26,k8s:[40,126],k8s_data:112,k8s_job:26,k8s_token:26,k8s_train:112,k8s_user:26,kafka:33,kaim:16,kcpu:74,kcuda:74,kcudnn:102,kdisabl:74,kebilinearinterpbw:107,kebilinearinterpfw:107,keep:[4,9,11,16,18,31,41,51,52,55,69,72,77,83,85,91,99,126],keep_dim:18,keep_top_k:4,kei:[10,11,19,22,25,27,28,29,31,33,35,48,50,59,63,79,80,85,86,89,98,99,101,107],kenlm:86,kept:[4,60,77],kera:81,kernel0:102,kernel1:102,kernel:[4,18,27,41,50,52,59,62,72,74,81,84,86,87,88,101,103,107],kernel_hint:59,kernel_type_for_var:88,kerneltyp:[59,63],key1:118,key2:118,key_pair_nam:112,keyid:112,keymetadata:112,keypair:112,keyserv:112,keystat:112,keyusag:112,keyword:77,kforcecpu:59,kill:[31,112],kind:[26,27,31,37,42,45,52,55,59,63,70,71,74,87,88,90,92,102,112,113],kind_:74,kmark:74,kmkldnn:102,kms:112,knchw8c:67,knchw:67,knhwc:67,know:[26,35,40,60,64,99,100,106,107,109,112,122],knowledg:86,known:[28,29,41,51,53,68],kplain:102,kpoprang:74,kpushrang:74,kqueue:41,kriz:10,krizhevski:18,kselectedrow:84,ksimonyan:5,kstate:74,kube_cluster_tl:26,kube_ctrl_start_job:26,kube_get_workers_addr:40,kube_list_containers_in_job_and_return_current_containers_rank:26,kubeconfig:112,kubectl:[110,113,114],kuberent:[31,112],kubernet:[26,31,40,42,79,109,115,116,126],kubernetes_service_host:26,kusecudnn:59,kusemkldnn:59,kvp:22,kwarg:[5,6,9,14,18,20,46,56,77,85],kwd:22,l1_rate:2,l1_regularization_op:81,l2_rate:2,l2_regularization_op:81,l2_sim:4,l93:34,label:[3,4,9,10,11,18,25,42,46,51,54,55,56,70,75,78,86,88,108,111,113],label_dim:4,label_fil:78,label_lay:78,label_path:78,lag:118,lambda:[18,40,45],lambdacost:4,lambdarank:4,lan:109,languag:[4,10,19,40,41,51,55,60,74,79,83,86,120],larg:[7,10,18,42,44,45,60,64,72,86,99],larger:[2,3,4,60],larger_than:[29,57,75],largest:18,last:[3,4,5,18,28,45,60,68,74,75,97,118],last_seq:69,last_time_step_output:4,lastseen:113,latenc:[4,50,86,110,112],latent:4,later:[30,79,81,86,87,91,94,101,103,112],latest:[4,29,30,31,38,82,91,92,94,104,113,122,123],latter:[72,89,106],launch:[63,112,118],launcher:26,layer1:[4,5],layer2:4,layer3:4,layer:[2,3,5,7,9,12,16,24,25,28,29,34,40,42,44,51,54,55,57,70,72,75,78,79,81,85,86,87,89,95,97,105,108,111,117,118],layer_0:100,layer_attr:[4,97,120],layer_help:59,layer_num:120,layer_typ:[4,61,62],layerbas:100,layerconfig:100,layergradutil:100,layerhelp:[18,59,77],layermap:100,layeroutout:4,layeroutput:5,layout:[11,63,88],layout_:[59,67],layouttyp:59,lazi:[72,81],lbl:3,lead:[60,67,107],leaki:54,leakyrelu:18,learing_r:70,learn:[2,3,4,5,6,10,19,20,26,28,36,38,41,42,44,47,54,55,60,62,69,71,72,74,78,79,81,87,92,97,98,100,101,104,107,111,123],learnabl:[18,19,25],learning_r:[2,20,36,42,108,111],leas:31,least:[3,18,31,94,122],leav:[29,112],lectur:60,lecun:10,left:[4,29,103,123],left_right_flip:11,legaci:92,legal:85,len:[4,18,36,40,48,51,77,95,100],length:[4,5,9,10,11,18,36,50,61,64,68,69,79,86,89,97,113,118],leran:60,less:[4,18,26,45,111,126],less_than:[26,60],lesser:18,let02:113,let:[3,4,18,26,29,38,40,52,53,55,59,63,67,68,69,70,80,87,88,101,106,111,112],level:[2,4,16,18,50,53,56,64,68,69,74,87,89,90,99,110,118,122],lgtest:30,lgtest_main:30,lib64:[92,118],lib:[66,91,92,106,109,122,123,124],libapi:30,libari:66,libc:94,libcuda:92,libgcc_:94,libgoogl:106,libiomp5:62,libmkldnn:62,libmklml_intel:62,libnvidia:92,libpaddl:[65,66,79,106],libpaddle_capi:66,libpaddle_gserv:66,libpaddle_math:66,libpython2:91,librari:[4,18,30,37,41,42,62,63,66,71,86,88,91,94,101,102,109,111,118,123,124],library_:67,library_typ:102,library_type_:88,librarydevicecontext:67,librarytyp:[88,102],libstdc:94,licens:[62,71],life:31,lifecycl:[74,126],lifetim:[83,94],lightweight:53,like:[3,4,9,10,18,28,29,30,31,34,39,40,41,47,51,52,53,54,55,56,58,63,67,71,72,77,78,79,80,81,83,84,86,88,89,91,94,97,98,99,106,107,108,109,112,117,120,122,123,124,126],limit:[4,9,18,51,60,64,69,79,81,107,118],linaro:124,line:[3,9,22,30,34,39,41,49,55,72,75,77,79,81,98,99,105,106,107,110,111,112,120],line_break:9,linear:[4,18,19,69,95],lineno:106,link1:50,link2:50,link:[4,5,30,48,49,83,94,101,112,126],linux:[9,41,48,92,94,98,99,109,112,124],linux_x86_64:[82,94],lipo:123,list:[3,4,5,9,11,14,18,20,22,25,26,28,29,30,35,39,40,47,49,51,54,67,70,73,74,77,80,83,89,95,97,98,100,101,106,109,111,112,118,120,124],listdir:109,listen:[31,40,42,109,111,118],listen_and_do:40,listenanddo:40,littl:[36,59,64,118],live:[101,108],live_in:60,live_out:60,load:[11,26,31,42,54,71,77,91,101,112,118],load_and_transform:11,load_imag:11,load_image_byt:11,load_missing_parameter_strategi:[117,118,120],load_mnist:54,load_persist:111,loadsave_parameters_in_pserv:[34,117,118],loc:[3,16],local:[2,18,25,27,29,31,37,38,41,55,60,68,75,77,79,91,92,98,99,106,110,113,117,118],local_scop:27,localhost:[92,104],localpath:49,locat:[4,18,25,30,51,67,74,87,89,97,100,109,124],lock:[30,31,35,36],lod:[18,41,64,68,84,89,90],lod_desc:[84,90],lod_expand:69,lod_level:[18,77,84,90],lod_rank_table_obj:18,lod_tensor:[18,68,84,90],lod_tensor_aarri:18,lod_tensor_arrai:18,lodrankt:18,lodtensor:[18,19,41,58,64,79,90],lodtensorarrai:18,lodtensordesc:[64,84],log:[35,42,49,54,94,100,109,110,112,113,114,118,124],log_barrier_abstract:118,log_barrier_lowest_nod:[117,118],log_barrier_show_log:[117,118],log_clip:[117,118],log_error_clip:[117,118],log_period:[113,118,120],log_period_serv:[117,118],logarithm:[1,18],logic:[38,42,44,45,54,58,70,71,73,83,89,101],login:[94,114],logist:18,logit:[54,88],longer:[31,42,60],look:[3,18,29,39,40,51,52,55,72,77,80,81,86,108,109,111,112,113,117],lookahead:[4,18,86],lookup:[18,58,69,108],lookup_t:60,loop:[27,29,51,60,74,78,83],loop_var:89,loss:[4,18,20,28,42,54,56,70,72,81,86,100],lot:[42,67,69,72,77,81,87,109,117,126],low:[4,16,70,71,86,87,89],low_rnn:68,lower:[4,18,50,68,69,99,110],lower_level_rnn:68,lowest:118,lpaddle_capi_shar:66,lpaddle_capi_whol:66,lrelu:54,lstm:[4,18,97,113],lstm_bias_attr:5,lstm_cell_attr:5,lstm_group:5,lstm_layer_attr:5,lstm_step:5,lstm_unit_op:18,lstmemori:[5,97],lstmemory_group:4,lstmlayer:61,lstmp:18,ltr:4,luckili:60,mac:[66,98,99,122],machin:[5,10,25,42,44,51,54,60,62,71,72,81,91,94,98,100,112,113,114,117,118,120,124,126],machine_transl:97,maco:[94,95,98,123],macro:[53,67,80,101],made:[31,36,51,97],mai:[4,5,18,27,29,42,46,50,52,59,60,63,71,74,78,79,83,86,88,91,92,103,104,107,109,112,124],main:[18,40,41,45,51,52,56,71,75,79,94,106,109,112],main_program:[14,18,28,46],mainli:[37,60,67,87,91,101,118],mainlin:94,maintain:[4,29,35,72,77,79,112],majel:30,major:[42,50,88,122],make:[3,4,18,26,28,29,30,31,35,36,38,41,42,43,50,51,55,68,69,72,73,77,78,79,81,86,89,91,98,99,100,101,102,106,107,109,111,112,122,123,124,126],make_chan:41,make_ddim:103,make_function_oper:53,make_vari:85,maker:[79,80],malloc:[87,100],man:48,manag:[20,25,31,36,37,40,41,42,49,74,83,87,94,104],mandarin:[4,86],mandatori:123,mani:[5,11,28,30,35,40,41,51,54,59,60,69,73,74,77,79,80,83,84,85,88,89,98,118],manili:56,manipul:[51,77,80,110,123],manner:[4,72,81,86,87],mantain:60,manual:[42,70,72,80,110,122,123,126],manufactur:51,manylinux1:94,manylinux1_x86_64:[82,94],manylinux:82,map:[3,4,9,19,25,26,29,35,45,63,67,77,80,83,85,87,89,98,118,126],map_fn:89,map_read:9,mapper:9,mapreduc:[26,109],margin:18,mark:[28,44,54,55,68,69,74,83,97,106,126],marker:74,market:50,mask:[2,4,18],master:[26,38,79,82,118,124],mastermind:30,mat:[65,66],mat_cache_row:34,mat_norm:34,mat_normal_shar:34,mat_param_attr:5,mat_sparse_row:34,mat_sparse_row_auto_grow:34,mat_sparse_row_id:34,mat_sparse_row_prefetch:34,mat_sparse_row_prefetch_full_s:34,mat_value_shar:34,match:[18,30,50,94,107],matchbox:126,math:[5,18,62,65,79,99,100,101,107],mathemat:81,matirx:4,matmul:[29,56,68,89,101],matric:[18,97,100],matrix:[3,4,5,9,18,19,25,34,65,66,100,101,117,120,123],matrixptr:100,matrixtyp:66,mattyp:34,matur:109,max:[2,4,9,10,18,19,27,43,45,60,77,107,118,120],max_diff:27,max_id:[4,25],max_job_id:10,max_length:[4,69,97],max_movie_id:10,max_relative_error:[27,101],max_seq_len:18,max_sort_s:4,max_user_id:10,max_x:18,maxframe_evalu:3,maxid:3,maxid_evalu:3,maxim:[4,45],maximum:[3,4,10,18,29,36,97,101,107,118],maxinum:7,maxoutfunctor:87,maxpool:4,mayb:[29,63,101],mchw:18,md5:[10,32],mean:[2,3,4,5,6,7,9,11,16,25,28,30,42,43,45,56,69,76,78,83,86,88,92,97,98,101,106,107,108,111,112,118,120,126],mean_var_nam:4,meant:89,measur:[46,107],mechan:[4,5,19,28,37,46,63,77,80,97,102,112],mem:[4,29,39,69],mem_per_pserv:42,mem_per_train:42,member:[4,10,26,45,55,56,67,73,77,83,101],memcpi:[73,107],memor:4,memori:[5,28,29,34,35,39,50,52,62,63,64,67,69,72,74,79,88,97,98,99,100,103,107,108,113,118,120],memory_boot:5,memory_nam:4,memory_optim:60,memory_test:98,memory_threshold_on_load_data:118,memoryalloc:87,memorydesc:63,mention:[18,28,30,35,42,44,51,68,71,72,74,98],mere:5,merg:[4,14,18,36,38,43,46,62,68,71,73,99,104,118,123],messag:[29,40,41,47,51,52,55,64,74,75,76,77,79,80,84,90,99,113,118],metaclass:101,metadata:[48,112,113],metal:[111,126],metaphor:55,metaplotlib:26,method:[4,6,16,20,25,27,29,38,40,42,43,45,50,54,55,56,59,70,71,77,78,79,83,84,88,89,91,92,94,100,101,103,104,106,107,118,120],methodolog:72,metric:[14,46,74],microarchitectur:50,might:[4,29,30,40,41,51,60,75,86,98,99,100,106,112,122],mileag:107,million:[10,120],min:[2,4,18,43,45,77,107,112,120],min_block:29,min_count:44,min_desc:29,min_word_freq:10,mind:106,mini:[4,9,14,18,25,29,31,41,46,47,51,57,68],mini_batch:78,minibatch:[4,18,29,46,55,57,75],minim:[20,29,42,44,45,51,54,70,79,108,111,118,122,123,124],minimum:[4,18,86,111,123],minimun:118,minsizerel:[122,123,124],minu:80,minus_grad:80,minusgradop:80,minusop:80,minusopgradmak:80,minusopprotoandcheckermak:80,minut:[31,38,92,98,112],mirror:[30,92],mislead:36,mismatch:18,miss:[54,118],mistak:51,misus:102,mit:112,mix:[5,74,89,97],mixed_lay:5,mixed_layer_attr:5,mixedlayertyp:4,mixtur:106,mkdir:[49,91,104,112,114],mkl:[63,79,87,88,91,92,102],mkl_packed_:61,mkldnn:[4,62,67,88],mkldnn_:62,mkldnn_batch_norm:4,mkldnnactiv:62,mkldnnbase:62,mkldnnlayer:62,mkldnnmatrix:62,mkldnnstream:62,mkldnntester:62,mklml:62,mklpack:61,mklpackedgatedrecurrentlay:61,mklpackedgemm:61,mklpackedlstmlay:61,mklpackedrecurrentlay:61,mlp:56,mlr:16,mnist:[33,42,54,55,75,78,79,106],mnist_random_image_batch_read:78,mnist_train:78,mnist_train_batch_read:78,mobil:[50,51,60,79,104,121],mod:109,mode:[4,22,25,50,61,71,99,118],model:[0,4,5,10,18,19,25,28,29,31,32,40,42,44,45,46,55,60,61,70,71,72,79,81,86,88,89,95,99,100,104,105,109,111,112,118],model_list:[118,120],model_path:120,modelparallel:42,modern:60,modif:[86,93],modifi:[4,42,50,56,81,97,100,101,109,110,112],modul:[5,10,25,42,53,54,69,86,89,101,106],modular:69,modulo:4,moment:[20,106],momentum:[2,18,20,83,102],momentumop:106,mon:113,monitor:[41,74],mono:4,month:30,more:[3,4,5,9,18,26,27,28,30,31,35,38,39,41,42,44,49,50,51,52,53,55,59,60,63,67,68,69,70,74,77,78,79,81,86,87,89,91,92,95,97,98,100,101,102,103,104,106,107,108,109,111,113,120,124,126],most:[4,9,25,26,28,30,38,41,42,52,55,56,67,69,72,74,78,81,86,87,94,97,100,106,107,108,117,126],mostli:[50,126],motiv:79,mount:[39,92,109,112,113],mountpath:[112,113],move:[4,31,35,49,51,72,92,107,112,126],movement:107,movi:10,movidiu:51,movie_categori:10,movie_info:10,movie_review:10,movieinfo:10,moving_average_fract:4,mpi:[41,71,114],mpirun:114,mse:[51,55,70,75],msra:16,much:[4,18,31,51,63,70,78,81,89,107],mul:[53,60,77,100,101],mul_grad:101,mul_op:[18,101],mul_ratio:4,mul_result:77,mulgradkernel:101,mulkernel:101,mulop:[53,101],mulopgrad:101,mulopmak:101,mult:[40,52],multi:[4,19,46,71,73,88,100,106,110,111,117,118,126],multi_binary_label_cross_entropi:4,multidimension:18,multigradientmachin:73,multinomi:4,multip:19,multipl:[3,4,5,9,14,18,19,25,26,27,35,36,38,40,41,42,44,46,51,52,53,59,71,74,79,86,88,90,97,100,101,106,109,112,118,120],multiple_input:77,multiple_param_attr:77,multipli:[3,4,18,40,100],multiprocess:9,must:[1,3,4,5,9,11,18,28,36,45,60,63,64,67,74,76,77,78,79,85,90,97,100,101,103,109,111,112,118,120,122,124],mutabl:[87,103],mutable_data:[63,87,101,103],mutex:41,mutuable_data:[87,103],mxnet:[29,40,41,51],my_cluster_nam:112,my_external_dns_nam:112,my_lib:109,myerrorclip:45,myfil:9,mypaddl:113,naiv:40,name:[2,3,4,5,7,11,14,18,22,25,26,27,28,29,31,33,34,36,39,40,42,46,50,53,56,59,62,63,64,66,67,69,74,75,77,79,82,84,85,89,90,92,94,95,97,98,100,101,102,107,108,109,111,113,115,116,118,120,123,126],name_:74,name_prefix:33,namespac:[29,57,65,77,100,101,113],nativ:[4,50,99],natur:[18,35,38,44,69,89,120],navig:104,ncall:106,nccl1:71,nccl2:71,ncclinit:71,nchw8:88,nchw8c:88,nchw:[4,18,62,67],ndarrai:[11,25,33],ndcg:4,ndcg_num:4,ndk:122,nearest:50,nearli:27,necess:89,necessari:[4,28,29,36,38,45,46,60,64,69,73,77,85,89,100,114],necessarili:[40,100],neck:71,need:[3,4,5,9,16,18,19,20,23,26,27,28,30,34,35,36,38,39,41,42,44,45,46,49,51,52,53,54,59,60,63,67,69,70,71,72,73,74,76,77,79,80,81,83,84,85,86,87,89,91,92,93,94,95,97,100,101,102,103,104,109,111,112,113,117,118,120,122,123,124,126],neg:[3,4,18],neg_distribut:4,neg_overlap:4,neg_pos_ratio:4,neglect:4,neighberhood:71,neither:4,neon:[50,122,124],ner:3,nervana:51,nessesari:86,nest:[4,9,28,29,74,75,90],net:[4,5,12,18,29,54,68,83],netop:[29,79],network:[2,3,4,9,16,18,19,23,24,25,26,27,28,29,31,34,42,44,46,54,56,60,61,62,68,70,72,74,77,78,81,83,85,86,87,88,90,95,100,101,103,107,109,111,118,126],network_config:120,networkadministr:112,neural:[4,5,9,16,18,25,26,28,29,31,42,56,60,61,62,68,72,81,83,87,88,90,95,103,107,109,111,118],neuralnetwork:73,neuron:[18,100],never:[9,60,78,83,112,113],new_block_idx:77,new_dim:18,new_op_and_kernel:102,new_op_desc:45,new_scop:88,new_stat:68,newblock:77,newbuff:63,newer:122,newest:36,newli:[50,123,126],newop:29,newopdesc:77,newprogram:77,newscop:88,newvardesc:77,next:[4,10,28,31,37,41,45,69,71,89,97,100,101,106,107,112,113,118],nextlay:62,nfs4:112,nfs:112,nfsver:112,ngram:10,nic:[117,118],nil:35,nine:10,nlp:4,nltk:10,nms_threshold:4,nms_top_k:4,nnz:100,no_grad_dict:28,no_grad_set:[20,27,28,101],no_gradi:28,no_sequ:4,node1ip:114,node2ip:114,node3ip:114,node:[4,30,38,40,42,44,56,60,69,71,79,98,100,109,110,111,112,113,114,118,126],node_0:112,node_1:112,node_2:112,node_id:109,nodeattr:56,nodeentri:56,nodefil:110,nodesep:56,nohup:109,nois:[4,31,54,109],noisi:[4,54],non:[4,18,31,50,51,84,100,101,112,118],none:[2,3,4,5,6,7,11,14,16,18,19,20,25,26,27,28,29,45,46,54,56,57,68,69,70,75,77,85,89,97,108,111],noneedtran:63,nonlinear:[16,100],nontranspos:18,nor:[40,98],norm:[5,18,54,67],norm_by_tim:4,normal:[4,5,10,16,18,72,86,97,100,109,113,118],notat:[4,60],note:[2,4,5,7,11,14,18,25,26,28,29,34,35,39,60,64,67,71,78,79,87,88,91,92,101,103,104,107,109,111,112,118,120,123],notebook:[39,92],notest_dist_fit_a_lin:111,noteworthi:51,noth:[1,25,59,77,83,98,118],notic:[4,45,51,71,80,97,99,100],notif:99,notimplementederror:45,notin:88,notingradi:101,notion:89,notori:27,now:[9,28,30,31,44,54,64,67,72,79,80,81,83,111,112,118,123],np_arrai:9,nproc:98,nullptr:[63,74,80,83,100],num:[4,5,109,118],num_channel:[4,5],num_chunk_typ:3,num_class:[4,5,18,56],num_col_dim:18,num_filt:[4,5,18,19],num_flatten_dim:18,num_gradient_serv:[109,117,118],num_head:19,num_hidden:56,num_input:99,num_neg_sampl:4,num_or_sect:18,num_parameter_serv:26,num_pass:[25,113,117,118,120],num_per_batch:11,num_pserv:42,num_repeat:4,num_result:3,num_results_per_sampl:4,num_row:84,num_shard:33,num_step:89,num_train:42,number:[3,4,5,9,10,11,18,19,29,31,33,44,46,60,72,74,78,79,85,89,98,100,106,109,111,112,118],numchunktyp:3,numdevices_:120,numer:[4,101],numeric_grad:27,numerical_grad:27,numlogicaldevices_:120,numofallsampl:3,numofwrongpredict:3,numpi:[2,9,11,18,25,33,50,54,77,78,91,101],numreal:34,numsampl:107,numtagtyp:3,numtimeout:35,nv_:30,nv_gpu:98,nv_librari:30,nv_test:30,nvcc:[30,50,52],nvidia:[50,67,71,87,92,98,107,118],nvlink:71,nvprof:74,obei:3,object:[2,4,5,9,18,25,26,34,42,45,46,54,56,60,65,70,74,77,79,81,83,103,107],observ:[4,100,107],obtain:[4,18,38,72,87],obvious:[30,67,106],occup:60,occupi:[50,74],occur:[10,25,60],occurr:29,oct:113,odd:4,odoti:18,off:[66,91,92,98,109,111,122,123,124,126],offer:[29,79,85],offici:[4,30,99,104,112,122],offlin:[31,33,126],offset:[4,18,34],often:[4,34,56,60,67,99,106,109],ograd:100,old:[27,36,38,69,79,118],older:[18,51,122],omega:81,omit:18,omp_num_thread:106,ompi_comm_world_rank:109,onc:[4,31,35,40,42,44,46,51,55,72,99,100,104,112],one:[1,3,4,5,7,9,16,18,20,23,25,26,27,28,29,31,34,35,36,38,39,40,42,45,46,47,50,51,52,53,54,56,58,59,63,64,67,68,69,70,71,72,73,75,76,77,78,79,80,83,84,86,87,88,89,92,98,99,100,101,102,108,110,111,112,113,114,118,120,126],onehotcrossentropyopkernel:101,ones:[53,54,79,99],onli:[3,4,5,7,11,18,23,25,26,27,28,30,34,35,36,37,38,39,40,42,44,45,46,47,49,50,51,54,55,60,63,68,69,70,71,73,74,77,79,84,85,86,87,88,89,91,93,94,97,98,100,101,102,103,104,107,111,112,113,117,118,120,123,126],onlin:[4,6,31,33,60,78],only_cpu:27,onnx:51,onto:[18,42,44,112,114],op1:[60,88],op1_2_op2:88,op1_to_op2:88,op2:[60,88],op3:60,op_:101,op_check:101,op_class:[79,85],op_desc:[45,58,76],op_info:108,op_kei:63,op_maker_class:[79,85],op_proto:85,op_registri:108,op_siz:45,op_test:101,op_typ:[79,101,102],op_unique_kei:63,opattrcheck:101,opcreat:85,opdesc:[29,45,55,75,76,77,79,80,85,90],opdescbind:[58,80],opdescbuild:29,open:[4,11,18,26,33,51,54,62,78,99,106,109,112],openbla:[91,92,122],opencv:11,openmp:106,openmpi:109,opensourc:71,oper:[4,5,9,11,16,18,20,23,27,29,40,41,42,43,44,46,47,50,51,52,54,55,56,58,59,68,69,70,71,74,76,81,83,86,87,88,90,97,99,100,102,103,107,108,112,118,122],operand:[18,50],operator_grad:27,operator_list:74,operatorbas:[29,53,79,80,85,101],operatorwithkernel:[88,101],opinfo:[58,79,80],opinfomak:58,opinfomap:80,opkernel:103,opkernelbas:102,opkernelkei:79,opkerneltyp:[67,88,102],opmak:85,opposit:18,opproto:101,opprotoandcheckermak:[80,101],opprotomak:[85,101],opregist:85,opregistri:85,ops:[20,27,28,29,30,40,52,55,56,72,75,76,77,79,87,101,126],ops_:29,ops_test:30,opt:[26,70,76,85,91],opt_op_list:70,optest:101,optestmeta:101,optim:[2,12,24,25,27,28,42,43,44,52,54,71,72,73,75,79,81,84,86,100,106,107,108,109,111,122,124],optimis:70,optimize_op:111,optimize_op_attr:77,optimizer_op:111,option:[3,4,14,18,22,26,30,42,54,59,64,75,76,77,79,84,85,86,90,98,100,106,109,110,111,120,122,123,126],optmization_op_list:70,opts_np:76,optyp:[58,85],opwithkernel:84,order:[4,5,9,11,18,25,28,55,64,74,78,81,89,91,100,106,109,112,113,118,126],ordereddict:25,orderli:18,oregon:112,org:[3,4,5,10,16,18,33,48,54],organ:[3,4],orient:85,origin:[4,5,9,10,18,27,50,54,83,89,99,103],other:[3,4,5,9,18,20,29,31,36,40,49,50,51,52,58,60,63,67,68,72,76,81,83,85,86,87,88,91,97,98,99,106,108,109,111,112,113,120,122,123,124,126],otherchunktyp:3,otherwis:[4,9,10,11,18,25,26,28,31,36,38,54,58,63,78,86,97,99,110,120],our:[18,26,28,30,41,42,44,54,58,60,67,71,72,83,89,91,94,97,98,99,100,106,111,112,113,122],out:[4,18,25,26,29,30,35,38,42,45,51,56,60,63,68,69,77,88,95,97,101,103,106,107,112,113,114,118],out_dir:112,out_fals:18,out_left:4,out_mem:97,out_memori:5,out_right:4,out_size_i:4,out_size_x:4,out_tru:18,outer:4,outgrad_:62,outlier:4,outlin:119,outout_lay:25,outout_layer1:25,outout_layer2:25,output:[1,2,3,5,7,9,18,19,22,25,26,27,28,29,33,38,40,44,45,49,52,53,54,55,56,57,58,60,63,64,68,69,72,75,76,77,78,79,80,83,84,85,87,88,89,91,97,98,99,100,101,102,103,106,107,109,113,118,120,122],output_:[4,62,100],output_all_step:68,output_arg_nam:45,output_dim_idx:18,output_dtyp:18,output_fil:22,output_height:18,output_id:4,output_lay:[25,95],output_max_index:7,output_mem:[4,97],output_mod:22,output_nam:27,output_num:68,output_path:33,output_s:18,output_seg:89,output_width:18,outputbuff:34,outputgradi:80,outputh:4,outputw:4,outsid:[4,5,42,83],outupt:89,outv:100,outval_:62,over:[4,5,18,25,26,51,60,71,72,89,99,100,107],overal:[54,72,74,99,126],overfit:[18,81],overhead:107,overlap:[3,4,100],overlap_threshold:[3,4],overload:[50,59],overrid:[29,31,49,63,87,100,101,103],overview:[35,36,37,87],overwhelm:99,overwrit:[49,109],own:[4,28,36,38,45,47,56,58,70,71,72,81,85,101,109,110,112,122,123],owner:[98,99],pack:[89,122],packag:[9,10,35,39,40,53,62,82,91,92,99,101,106,112],pad:[5,18,63,86],pad_c:4,pad_h:4,pad_w:4,padding_attr:4,padding_down:18,padding_h:18,padding_i:4,padding_idx:18,padding_left:18,padding_right:18,padding_up:18,padding_w:18,padding_x:4,paddl:[1,2,3,4,5,6,7,9,10,11,14,16,17,18,19,20,22,23,25,26,29,30,31,33,39,42,49,52,53,54,57,61,62,63,64,65,66,68,69,73,75,79,81,82,85,86,87,89,91,92,94,95,97,98,99,100,101,102,104,106,107,108,109,110,111,112,113,114,118,120,122,126],paddle_begin_init_param:36,paddle_dir:101,paddle_element_typ:36,paddle_element_type_float32:36,paddle_element_type_float64:36,paddle_element_type_int32:36,paddle_element_type_int64:36,paddle_element_type_uint32:36,paddle_element_type_uint64:36,paddle_enforc:[29,63],paddle_enforce_eq:[101,103],paddle_error:[65,66],paddle_exampl:39,paddle_finish_init_param:36,paddle_get_param:36,paddle_gradi:36,paddle_init_num_gradient_serv:109,paddle_init_param:36,paddle_init_port:109,paddle_init_ports_num:109,paddle_init_ports_num_for_spars:109,paddle_init_pserv:109,paddle_init_trainer_count:109,paddle_init_trainer_id:109,paddle_init_use_gpu:109,paddle_job:39,paddle_manylinux_devel:91,paddle_matrix:[65,66],paddle_matrix_cr:66,paddle_matrix_get_shap:65,paddle_matrix_shap:65,paddle_new_etcd_pserver_cli:36,paddle_new_pserver_cli:36,paddle_on_cloud:39,paddle_output:113,paddle_paramet:36,paddle_pserver2:110,paddle_pserver_cli:36,paddle_pserver_client_releas:36,paddle_save_model:36,paddle_send_grad:36,paddle_train:[66,82,110],paddle_with_cuda:74,paddle_with_mkldnn:67,paddlepaddl:[4,5,9,10,11,18,25,30,31,33,36,37,38,39,40,42,48,49,53,54,55,57,59,64,68,69,70,73,74,77,78,79,83,89,90,93,95,97,98,99,100,101,102,106,107,109,110,114,115,116,125,126],paddlepaddle_gpu:94,paddlepaddlebook:92,paddlepaddlehub:92,page:[99,112],pair:[3,19,20,22,28,29,42,55,70,74,79],pairwis:4,pakcag:30,paper:[4,16,18,54,86],para:34,paradigm:[40,47,79],paragraph:68,paragraph_data:68,paragraph_out:68,parallel:[40,41,42,44,71,74,79,88,98,107,109,112,113,118,120],parallel_for:40,parallel_nn:[2,117,118],paralleldo:43,parallelfor:40,paralleliz:86,param:[2,4,5,9,18,20,27,29,36,73,77,87,103],param_attr:[4,5,18,19,34,77,97],param_config_proto:36,param_initi:18,paramattr:[2,4,12,18,97],paramet:[3,5,6,7,9,10,11,14,17,18,19,20,22,24,27,28,29,30,32,34,38,40,42,43,45,47,49,51,52,54,55,56,58,64,68,70,71,75,78,83,85,86,89,91,95,99,100,101,102,103,105,108,110,120,123],parameter_block_s:[117,118],parameter_block_size_for_spars:[117,118],parameter_learning_r:2,parameter_list:[20,28,70],parameter_nam:[25,26],parameter_serv:26,parameter_valu:34,parameterattribut:[2,4,5,34],parameterclient_:34,parametermap:100,parametermutex_:34,parameters_:100,parameters_and_grad:[20,70],parameterserver2:34,parameterset:26,parameterupdat:73,parameterupdater_:34,parametr:4,params_grad:[70,111],paramt:112,paraspars:100,parent:[29,40,75,77,79,100],parent_:[29,83],parent_idx:77,parenthes:79,pars:[9,10,30,42,56,98,112,120],parser:9,part:[3,4,18,19,28,29,38,42,51,63,64,75,77,86,87,97,100,106,107,109,111,126],parti:[98,107,122,123,124],partial:[4,25],partial_sum:4,particip:101,particular:[55,64,79,86,88,107],particularli:16,partit:[31,33,42,44,79,109,112],pass:[4,14,18,23,25,28,29,31,41,45,46,51,54,60,64,70,72,73,76,77,78,79,81,83,86,89,99,100,107,109,110,111,112,113,118],pass_gener:4,pass_id:[25,42,111],pass_idx:78,pass_num:111,passtyp:100,password:114,past:[26,92,95,112],patch:[18,48],path:[3,9,10,11,25,31,35,36,39,60,69,78,86,91,92,109,112,113,118,120,122,123,124],path_to_paddlepaddle_working_directori:104,pattern:[10,31,65,72,81,112],paus:[31,38],pcie:71,pdf:[5,18],peephol:18,peer:71,pem:[26,33,112],pend:[31,35],peopl:98,pep425tag:94,pep8:99,per:[3,4,10,11,18,31,36,71,72,78,81,101,118],percal:106,perf_test:106,perfectli:86,perfom:[118,120],perform:[4,5,16,18,27,36,41,42,46,50,51,54,60,71,73,74,78,79,81,86,87,88,97,98,100,101,105,109,111,117,122,123,124],perftool:[74,106],period:[31,38,118],perm:18,permiss:112,permut:18,peroid:[4,11],persist:[18,47,84,86,90,112],persistentvolum:112,persistentvolumeclaim:112,person:[3,26,59],perspect:[79,107],perturb:[27,100],pex:126,pfs:[33,49],pfsclient:33,pfspath:49,pgp:112,phase:[63,69,71,72,78,80,86,126],philosophi:[72,81],photo:54,phrase:18,physic:[123,126],pick:[18,112,123],pickl:[109,114],pictur:71,piec:[5,40,74,103,111],pil:[11,109],pillow:39,ping:99,pip:[82,91,93,95,99,104,106],pipe:9,pipelin:[46,86],piperead:9,pivot:63,pixel:[4,9,10,42],place:[18,28,29,31,38,42,44,47,59,63,71,79,88,100,103,107,108,111],place_:[59,67,87,88],place_typ:102,placehold:[54,87,103],placement:44,plain:[3,4,9,39,64,66,67],plan:[31,40,63,79,86,100,122],platform:[29,52,63,67,74,87,88,94,99,101,102,103,108,112,122,123,124],pleas:[2,4,5,6,11,18,26,31,35,36,37,40,52,56,67,68,77,78,79,86,87,90,91,92,94,97,98,99,100,101,103,104,106,109,111,112,122,123,124],plot:26,plu:[4,27],plug:[71,72],pne:101,pnpairvalidationlay:118,pnpairvalidationpredict_fil:117,pod:[33,39,40,112,113],pod_nam:112,point:[18,29,31,39,41,50,60,63,71,87,98,99,101,103,106,107,122,126],pointer:[29,36,56,60,67,77,79,83,87,103],polar:10,polici:[18,112],poll:41,pollut:38,polyak:72,ponit:56,pool3:100,pool:[5,18,24,43,60,86],pool_attr:5,pool_bias_attr:5,pool_layer_attr:5,pool_pad:[5,18],pool_siz:[4,5,18,19],pool_size_i:4,pool_strid:[5,18,19],pool_typ:[4,5,18,19],pooled_height:4,pooled_width:4,pooling_lay:5,pooling_typ:4,poolingtyp:7,pop:[29,47],popul:36,popular:[30,54,56,74],port:[30,40,106,109,111,112,113,117,118],port_num:117,portabl:56,portal:104,ports_num:[109,118],ports_num_for_spars:[34,109,117,118,120],pose:31,posit:[3,4,5,18],positive_label:3,possibl:[26,29,35,41,44,60,77,81,90,107],post:[39,48],postpon:81,potenti:[50,107],power:[50,60,71,86,103,126],ppo_workspac:104,pprof:106,practic:[97,100],pre:[4,5,10,26,36,59,60,91,99,112,113,122,124],pre_activ:77,pre_bia:77,pre_stat:[68,89],preambl:77,precis:[3,46,50,72,91],precision_evalu:3,precompil:47,pred:[56,60],predecessor:60,predetermin:[4,118],predic:10,predict:[3,4,18,25,42,81,95,97,118],predict_fil:118,predict_lay:25,predict_output_dir:[117,118],prediction1:25,prediction2:25,prefer:[51,59],prefetch:[34,100],prefix:[3,5,31,33,69,86,112],pregel:41,pregrad:100,prepand:77,prepar:[27,39,73,86,97,115],prepend:[18,77],prepend_oper:77,preprocess:[10,11,86,89,113],present:[26,28,29,74,89],preserv:49,press:16,prev_batch_st:[117,118],prev_cel:18,prev_cell_data:18,prev_hidden:18,prev_hidden_data:18,prevent:[6,18,26,31,35,38,45,81,106],preview:[79,104],previou:[4,5,18,25,28,31,44,49,68,69,100,106,112,118],previous:[4,113],previous_memori:29,price:[10,79,95],prim:63,primari:[51,55],primarili:[72,81],primer:99,primit:[50,62,63,71,89],primitive_desc:63,primitivedesc:63,principl:[26,30,67],print:[2,25,26,42,51,56,77,94,95,106,114,118],print_graphviz:56,printallstatu:107,printer:3,printstatu:107,priorbox:4,prioriti:79,prite:3,privat:[29,66,74,77,83,84,85,87,89,99,103],privileg:[98,112],pro:71,prob:[3,25,95],probabilist:4,probabl:[3,4,18,25,69,86,92,97,99],problem:[4,26,27,30,38,51,54,55,72,79,81,94,98],proc:92,proce:[9,31,78,92,112],procedur:[29,64,103,123],proceed:16,process:[2,4,5,9,18,26,28,29,33,34,35,38,40,41,42,46,47,51,52,56,60,62,64,71,81,85,88,97,99,106,109,110,112,113,118,120],process_num:9,processor:[50,107],produc:[4,5,9,18,31,51,56,78],product:[4,5,18,19,39,51,100,112,113],productgraph:113,prof:106,profil:[12,49,74,86],profilerst:74,proflier:[74,107],program:[9,14,16,18,22,26,28,33,36,38,42,44,47,55,57,60,70,71,74,78,79,83,90,106,107,110,118],programdesc:[40,42,47,51,60,64,76,77,80,90],programm:[42,51,77],progress:[31,35,118],proj:4,proj_activ:18,proj_dim:18,proj_out:18,proj_siz:18,project:[4,5,18,19,39,66,86,97,100,101],promis:[4,5,69],prompt:[49,51,111],prone:26,pronunc:86,prop_kind:63,propag:[4,6,28,51,72,101,118,120],proper:[59,109],properli:[59,98,111],properti:[56,81,118],propos:[18,29,43,44,69,70,71,72,89],proprietari:62,protect:[50,85,100,101],proto:[7,41,59,64,67,75,79,85,90,101],proto_:85,protobuf:[25,29,39,40,42,47,51,52,55,56,60,64,75,77,79,80,85],protoc:[122,124],protocol:[3,108,118,126],provi:109,provid:[4,10,18,26,29,36,39,40,46,47,50,51,54,56,58,59,67,71,72,74,77,81,85,86,87,89,92,95,103,106,107,109,110,111,112,122,123,126],providermemory_threshold_on_load_data:117,provis:[112,126],prune:[4,29],ps_desir:31,pserver:[25,34,36,37,39,79,109,111,112,117,118],pserver_addr:36,pserver_cpu:39,pserver_endpoint:111,pserver_id:32,pserver_mem:39,pserver_num_thread:[34,117,118],pserver_prog:111,pserver_spec:25,pserver_startup:111,pserverstart_pserv:117,pseudo:[26,28,39,80,89],pseudocod:89,psize:100,ptr:[66,87],pub:[18,114],publish:122,pull:[30,79,82,99,122],purpos:[4,31,42,44,59,107],push:[29,47,51,74,82,99],push_back:100,put:[30,31,34,44,60,63,77,87,100,111,113,122],pvc:112,pwd:[91,92,98,104,122],pxe:126,pybind:[29,41,50],pypi:94,pyramid:4,pyramid_height:4,python2:106,python3:94,python:[18,25,26,29,37,41,46,47,51,53,54,55,56,59,65,69,73,74,79,82,87,89,91,92,94,95,97,98,99,104,108,109,111,114],pytorch:[51,74],qualcomm:50,queri:[3,4,19,112],query_id:3,question:[4,26,40,44,85,111,112],queue:[41,44],quick:[56,118],quick_start:[39,112,113,115],quick_start_data:113,quickli:[69,77,79],quickstart:113,quit:[69,107],r14b:122,r_h_val:18,r_t:4,rais:[9,18,19,45,56,109],rajathkmp:54,ran:[44,107],rand:[54,107,118,120],random:[2,4,9,16,18,33,54,67,73,77,78,101,109,118],random_crop:11,random_imag:33,randomli:[11,18,38],randomnumberse:117,rang:[4,9,16,18,33,40,42,50,54,60,74,78,85,99,111,118,120],rank0:71,rank1:71,rank:[4,18,26,89,103,112],rank_tabl:18,rankdir:56,ranktabl:18,rapid:80,raspberri:125,raspberry_pi:124,raspberrypi:124,raspbian:124,rate:[2,3,4,5,6,10,19,20,36,86,100],rather:[28,39,54,89,112],ratio:[4,118],raw:[4,18,64],rdma:118,rdma_tcp:[117,118],reach:[31,60,71,107],read:[9,11,18,25,26,28,31,33,40,41,42,44,51,52,78,79,86,89,92,97,98,104,109,112,122,126],read_from_arrai:60,read_from_realistic_imag:26,read_from_rng:26,read_lock:32,read_minibatch:51,read_mnist_imag:26,read_ranking_model_data:26,readabl:[74,79,106],reader:[0,10,25,33,42,50,54,55,75,86,106,109,111],reader_cr:33,reader_creator_bool:78,reader_creator_random_imag:[9,78],reader_creator_random_image_and_label:[9,78],readi:[31,112,113,126],readlockguard:34,readm:66,readwritebuffer_:34,readwritemani:112,real:[4,18,34,54,78,109],realist:26,realiti:86,realiz:[29,68],realli:[51,81],reaon:102,rearrang:18,reason:[5,26,27,31,41,51,99,111,113],recal:3,receiv:[31,39,41,42,44,68,111],recent:[60,72],reciev:118,recognit:[4,86],recommend:[5,18,26,91,92,93,97,99,100,104,109,110,118,122],recompil:107,record:[9,35,63,74,85,112],recordev:74,recordio:[9,10,26,33,35],recov:[31,89],recover:79,recoveri:35,rectifi:[4,16],recurr:[18,61,68,83,86],recurrent_group:[5,86,97],recurrent_lay:5,recurrent_op:89,recurrentgradientmachin:[66,69,89],recurrentgroup:3,recurrentlay:[61,118],recurs:[28,29,30,49,60,79],recv:[40,42,44,71,112],recvparametertyp:34,red:[54,106],redirect:9,reduc:[4,18,44,50,71,79,92,99,106,110,118,120],reduce_by_kei:79,reduce_mean:54,refactor:[42,44,55,69,72,73,77,81,89],refer:[2,4,5,6,11,16,18,19,22,27,29,31,35,36,37,40,50,56,63,67,68,71,75,77,79,81,83,87,89,90,91,92,97,98,100,101,103,111,113,122,123],referenc:35,reflect:35,reformat:99,refrain:101,reg:85,regard:[18,126],region:[4,83,107],regist:[41,60,67,80,87,88,100,107],register_gpu_profil:107,register_lay:100,register_op:[53,79,80,85,101],register_op_cpu_kernel:[87,101],register_op_cuda_kernel:[87,101],register_op_kernel:102,register_op_without_gradi:[79,101],register_oper:[58,80],register_tim:34,register_timer_info:107,registerop:85,registr:[101,102,108],registri:[39,58,87,113,126],regress:4,regular:[2,12,18,20,28,100,112],regularization_coeff:23,reiniti:63,reinstal:91,rel:[5,18,27,38,81,101,122],relat:[31,38,39,50,67,74,83,88,99,106,113,123,124,126],relationship:[80,87],releas:[82,86,112,122,123,124],relev:101,reli:[27,40,69,70,72,81,101,106],reliabl:[31,81],relu1:56,relu2:56,relu:[4,54,56,60,100],relwithdebinfo:106,remain:[18,89],remaind:18,rememb:[4,99],remot:[2,30,34,42,79,99,100,112,118,120],remoteparameterupdat:[34,37,118],remov:[9,18,28,42,49,51,69,99,118,122,123],ren:16,renam:[28,49,50,94],reorder:63,reorder_primit:63,reorgan:4,repeat:[18,29,55,75,76,84,85,90,106],repeatedli:[55,60],replac:[30,35,58,72,80,86],repli:99,replic:42,replicaset:39,repo:[30,99,104,124],report:[35,50,51,74,107],reportdataset:35,repositori:[4,104,122],reprenset:18,repres:[4,5,18,28,29,35,40,42,44,45,51,56,64,67,69,72,77,79,81,84,87,89,90,97,100,112],represent:[4,18,36,42,52,54,55,60,67,69,84],reproduc:98,request:[30,31,34,38,40,79,82,99,112,113,126],requir:[3,4,20,26,28,31,36,38,39,42,44,45,49,50,56,60,62,68,72,74,75,76,79,81,84,85,86,90,94,98,99,100,101,104,109,112,113,122,124,126],requisit:60,rerun:101,research:[10,18,42,51],reserv:[18,49],reserveoutput:100,reset:[4,14,18,31,46],reset_program:[14,46],resetingrad:62,resetinvalu:62,resetoutgrad:62,resetoutvalu:62,resetxxx:62,reshap:[27,78,103],reshape_s:4,resid:[18,98],resiz:[11,34,87,101,103],resize_s:11,resize_short:11,resolv:[30,99,113],resourc:[42,47,71,74,87,102,112],respect:[18,19,27,45,50,54,68,97,100,118],respons:[4,34,41,42,46,54,71,72,73,81,112,113],rest:[18,29,39,48,52,88,126],restart:[31,36,112,113,126],restartpolici:[112,113],restor:[27,72],restrict:[81,83,106,118],result:[1,3,4,17,18,22,25,27,28,35,46,54,55,56,60,64,69,70,71,73,101,103,106,107,108,112,118],result_fil:3,resum:38,retain:103,retran:112,retriev:[29,69,83,98,100,106,113],retriv:109,return_op_list:20,return_seq:5,reuqest:82,reus:[29,38,69,78,79,100,101],rev:98,revamp:42,reveal:[26,106],revers:[4,5,18,28,97],review:[10,40,113],reviews_electronics_5:113,rewrit:[30,41,101],rgb:[4,11],rho:6,rid:51,right:[4,27,28,29,30,39,46,60,79,81,99,102],ring:71,risk:28,rkt:[39,98],rmsprop:72,rmspropoptim:72,rnn:[4,5,29,51,54,69,77,79,83,86,105,117],rnn_bias_attr:97,rnn_layer_attr:97,rnn_out:97,rnn_output:89,rnn_step:4,rnn_use_batch:[61,117,118],rnnalgorithm:69,rnnlm:10,rnnstep:89,roadmap:[86,89],robust:[4,16],rocmplac:67,roi:4,role:[10,26,35,36,42,71,111,112],rollback:77,root:[6,7,28,71,112,113,123],rot:4,roughli:86,round:[50,71],routin:[50,62,71],row:[3,4,9,18,34,41,100],row_id:4,rows_:84,rpc:35,rpcserver:35,rpi:124,rpi_arm_neon:124,rpi_toolchain:124,rsize:112,rtk:126,rtype:9,rule:[3,18,28,42,45,51,55,100,112],run:[26,27,28,29,30,31,39,40,42,43,44,46,50,51,52,53,54,55,56,60,63,67,68,70,71,72,74,75,76,77,79,82,83,84,86,87,88,93,94,95,98,99,100,102,103,104,106,107,109,110,111,112,114,115,116,118,122,123,124,126],run_test:91,runinitfunct:107,runnabl:44,running_on_cloud:39,runserv:104,runtim:[22,29,40,41,42,58,68,79,90,92,102,110,122],runtime_table_:29,s_block:28,s_recurrent_group:97,safe:39,sai:[4,52,55,57,60,78,98,118,120],said:51,sake:100,same:[3,4,5,16,18,19,25,26,27,35,36,38,40,41,42,53,54,56,59,60,68,69,71,77,79,80,83,86,88,89,91,97,101,102,103,110,111,112,120],samping_id:4,sampl:[3,9,10,18,46,54,77,85,92,109,110,118,120],sample_id:3,sample_num:3,sampler:54,satifi:[3,60],satisfi:[30,63,84,94,112],save:[4,9,25,31,33,35,36,39,40,42,55,56,60,64,72,84,90,98,109,112,113,118,120],save_dir:[113,118,120],save_only_on:[117,118],save_parameter_to_tar:25,save_persist:111,saving_period:[117,118],saving_period_by_batch:[117,118,120],scalabl:79,scalar:[4,18,28,29,57,89],scale:[1,16,19,42,44,72,80,85,86,101,109],scaleop:101,scaleopmak:[79,101],scalingproject:4,scan:[18,28,35,60,79],scatter:[4,28,71],scenario:[69,117],scene:117,schdule:112,schedul:[35,39,44,112],scheduler_factor:2,scheme:[3,6,34,81,101],scienc:60,scope:[27,40,43,47,52,88,108],score:[3,4,18,69],scorer:86,scp:114,script:[10,71,91,98,101,109,110,112,114,122],sdk:123,search:[4,31,83,91,97,118],second:[4,18,26,40,49,51,54,56,68,69,75,76,78,83,85,101,110],secret:112,section:[28,44,51,77,97,99,100,106,112],see:[4,5,18,26,28,31,40,41,44,50,51,77,86,99,101,103,106,107,111,112],seed:[16,18,107,118],seem:[30,41,50,51,86,94],seen:[19,81,101],segment:[3,18,68,89,103],sel_fc:4,selcet:4,select:[4,18,69,112],selected_generation_scor:69,selected_id:[4,69],selected_indic:4,selected_row:[84,90],selected_rows_desc:[84,90],selected_scor:69,selectedrow:[58,90],selectiv:4,selector:113,self:[27,45,46,54,56,60,61,62,64,70,77,89,100,101],self_addr:40,selfnorm:4,semant:[10,26,69,82],semaphor:41,semat:26,send:[31,36,40,42,44,59,71,79,85,99,109,111,112,118],send_back_parameter_typ:34,sendbackparameterspars:34,sendbackparametertyp:34,sendparameterrequest:34,sendparameterrespons:34,sens:[72,81,99,106],sensit:4,sent:[26,36,40,42,79,85,90,113],sentenc:[4,10,51,68,69,89,97],sentence_input:89,separ:[3,18,22,36,42,53,72,80,81,109,111,118],seper:89,seq:[4,10],seq_len:89,seq_pool:4,seq_silc:4,seq_text_print:3,seq_typ:9,seqtext_evalu:3,seqtoseq:4,seqtoseq_net:4,sequenc:[1,3,4,5,7,9,10,18,28,29,40,47,51,55,61,70,75,86,89,99,100],sequence_group:4,sequence_nest_group:4,sequencesoftmaxop:18,sequencestartposit:4,sequencetextprint:3,sequencetyp:4,sequenti:[4,29,40,41,97],seri:[5,94],serial:[25,29,35,64,73,79],serializ:[79,90],serv:[42,50,79,89,92,107,109,112],server:[26,30,34,37,38,42,52,71,79,91,100,110,117,126],server_endpoint:111,serverless:31,servic:[106,109,126],sess:[54,56,70],session:[56,70,76,107],set:[2,3,4,5,9,10,11,18,19,22,25,26,28,31,39,54,58,60,63,67,68,69,74,76,77,79,80,83,86,87,89,91,97,98,99,100,101,103,104,105,106,107,109,110,111,112,113,117,118,120,123,124],set_active_typ:100,set_attr:45,set_default_parameter_nam:2,set_drop_r:100,set_float_el:27,set_input:[4,45],set_output:45,set_siz:100,set_typ:[45,100],setdatatyp:84,setdefault:101,setp:112,setq:98,settup:100,setup:[42,72,82,100,101,126],seven:86,sever:[3,4,27,34,42,44,54,68,69,71,73,74,77,84,87,89,91,109,110,112,120],sexstant:126,sgd:[20,25,26,31,39,44,72,73,84,108,109,111],sgd_optim:[108,111],sgdasync_count:117,shall:[28,30],shaoq:16,shape:[3,4,9,14,18,19,25,27,28,29,42,54,57,67,68,75,77,79,84,86,87,101,103,108,111],shard:[31,32,33,34,35,36,38,42,44,109,112],share:[4,18,30,54,66,73,77,79,81,86,87,89,98,101,107,118],shared_bia:5,shared_bias:4,shared_librari:30,shared_ptr:[63,65,66,83,87,103],shell:[92,112],shift:[4,18],shorten:4,shorter:11,should:[2,3,4,9,11,14,16,18,19,20,22,23,25,26,27,28,29,36,39,41,42,45,46,50,52,53,54,58,59,63,67,68,69,70,72,73,74,75,78,79,80,81,84,85,86,88,89,90,95,97,101,102,104,110,111,112,122],should_be_fals:26,should_be_tru:26,show:[3,6,28,29,31,49,51,57,60,64,68,71,72,75,89,94,98,103,109,112,113,118],show_check_sparse_distribution_log:[117,118],show_layer_stat:[117,118],show_parameter_stats_period:[113,117,118,120],shown:[4,26,42,46,71,74,86,97,100,103,107,112],shrink:100,shrink_rnn_memori:18,shrunk:45,shuffl:[9,42,111],sid:112,side:[4,25,42,46,60,73,103,109],sig:112,sigint:110,sigmod:85,sigmod_op:85,sigmod_output:85,sigmoid:[4,19,29,85,89,100],sigmoidactiv:5,sign:[48,64,112],signal:110,signatur:112,signific:[86,107],similar:[4,18,29,40,41,42,44,47,51,69,72,74,78,79,81,86,87,88,89,101,106,112,126],similarli:[4,9,51,60,101],simpl:[1,3,4,5,9,10,18,20,25,40,44,50,52,55,56,60,68,72,75,81,83,85,86,89,107,111,118],simple_attent:97,simple_gru:97,simple_lstm:4,simple_rnn:[4,97],simple_transform:11,simpler:73,simplest:112,simpli:[4,11,26,36,42,92,95,97,107],simplifi:[26,69,77,85,86,100,113],simul:[51,123],simultan:112,sinc:[4,5,31,35,37,38,41,42,43,44,51,58,60,63,67,72,77,78,80,81,89,103,107,111,112,123,126],sincer:99,singl:[3,5,9,18,28,31,42,44,46,50,59,71,79,83,86,91,95,100,106,109,113],singleton:[40,43],sinlg:25,sit:42,site:[30,106,112],situat:[28,76,88],size:[3,4,5,9,10,11,18,19,25,31,33,34,36,41,42,50,54,60,64,69,72,77,78,84,85,86,87,89,92,95,97,100,101,103,108,111,118,122,123,124],size_a:4,size_b:4,size_in_byt:63,size_t:[34,87,89,100],sizeof:29,skip:[28,78,99,110,112],slice:[18,40],sliceproject:4,slide:[4,6,10,31],slight:51,slightli:54,slope:[4,18],slopeinterceptlay:4,slow:107,slowli:[98,106],small:[4,10,18,27,40,52,54,62,69,99,100,118],small_messag:[117,118],smaller:[18,27,31,50,69,99],smart:83,smooth:4,snap:113,snapdragon:50,snapshot:[32,38,112],snippet:[53,70,97,100,107,112],sock:39,sock_recv_buf_s:[117,118],sock_send_buf_s:[117,118],socket:118,soft:18,soft_label:18,softmax:[4,5,18,26,29,42,44,51,56,57,69,75,97,100],softmax_param_attr:5,softmax_selfnorm_alpha:4,softmaxoutput:56,softrelu:18,softwar:[50,74,107,126],solid:54,solut:[71,126],solv:[26,28,60,79],some:[2,4,9,11,18,25,26,28,29,30,34,35,36,38,39,42,44,45,50,52,53,54,55,59,60,63,67,68,69,70,75,76,77,78,79,80,83,87,88,89,99,100,101,102,103,107,109,112,117,118,120,122,123,124,126],some_c_api_funct:66,some_inst:66,some_op:[58,68,89],some_python_class:65,somecppclass:65,somedata:25,somegotyp:65,someth:[28,34,77,98,99,106],sometim:[4,74,78,98,107],somewhat:36,somewher:83,soon:31,sophist:100,sort:[4,10,18,89,106,112,118],sort_by_length:89,sortagrad:86,sourc:[4,10,27,30,49,51,54,62,64,66,69,78,79,97,98,106,109,112,113,123],source_dict_dim:[69,97],source_dict_s:69,source_language_word:[69,97],space:[3,4,44,50,77,81,86,97,98,107],space_seperated_tokens_from_dictionary_according_to_seq:3,space_seperated_tokens_from_dictionary_according_to_sub_seq:3,span:74,spars:[2,4,6,9,18,34,41,100,103,109,112,118],sparse_binary_vector:9,sparse_binary_vector_sequ:9,sparse_float_vector:9,sparse_float_vector_sequ:9,sparse_non_value_slot:9,sparse_remot:34,sparse_upd:[2,34],sparse_value_slot:9,sparseparam:100,sparseprefetchrowcpumatrix:100,spatial:4,spatial_scal:4,speak:97,spec:[112,113],specfii:118,special:[4,18,28,36,42,50,52,58,67,69,70,101],specialvartypeinfer:58,specif:[18,19,25,28,30,31,42,45,49,52,69,79,83,87,98,101,111,120,122],specifi:[3,4,14,18,26,27,34,35,36,39,40,41,42,43,45,46,47,49,54,64,74,77,83,85,89,92,97,98,99,100,103,104,106,112,118,122,123],spectrogram:86,speech:[4,86],speed:[4,5,50,64,71,72,91,126],speedup:74,sphinx:[65,104],split:[4,9,19,38,40,43,51,57,69,79,89,109,112,120],split_count:[109,112],spread:28,sqrt:16,sqrt_x:18,squar:[4,6,7,56],square_error_cost:[108,111],srand:118,src:[30,63,109],src_backward:97,src_embed:[69,97],src_forward:97,src_primitive_desc:63,src_word_id:[69,97],src_word_vec:69,sreializ:90,srl:10,ssd:4,ssh:[112,113,114,124],ssh_server:110,sstabl:26,stabil:[4,27,60,101],stabl:[82,112],stack:[18,47,79,89,112],stackexchang:4,stage:[30,37,43,54,60,63,86,90,110,122],stale:31,stamp:107,standalon:122,standard:[2,9,16,18,41,51,79,81,86,94,98,106],stanford:[10,27,113],star:30,start:[4,5,18,25,28,30,31,34,35,36,38,39,41,42,43,69,71,73,74,91,92,94,97,98,106,107,110,114,118,121],start_mpi_train:114,start_op_idx:28,start_pass:[117,118],start_po:4,start_pserv:118,startup:[18,31,39,51,112],startup_program:[14,18,20],stat:[107,118],state:[4,5,14,18,20,29,31,46,47,68,69,74,83,86,89,97,102,113,118],state_act:[4,5],statem:60,statement:[51,55,60,100,112],static_cast:[63,103],staticinput:[4,97],statist:[4,14,16,46,74,118],statset:107,statu:[39,69,107,112,113],status:113,std:[25,30,34,56,58,59,63,65,66,74,76,79,80,83,85,87,100,101,103,118],stdbuf:109,stderr:110,stdout:[9,110],step1:18,step:[4,5,7,18,20,27,29,31,36,42,44,46,51,54,55,61,69,72,73,77,79,85,86,89,91,92,97,99,100,106,107,109,112,113,114,122,124,126],step_gradi:28,step_id:89,step_input:89,step_net:29,step_output:89,step_scop:79,stepnet:[29,68,79,83],still:[28,35,38,42,51,60,80,94,103],stirng:77,stmt1482205552000:112,stmt1482205746000:112,stochast:[6,31,35,38,72,109],stop:[4,77,98,110,113,118],stop_gradi:[18,77],storag:[48,50,109,112,113],store:[3,4,10,18,25,27,29,30,34,47,56,58,64,67,69,73,75,77,79,80,81,83,89,100,101,103,104,109,112,113,114,118,123,124],str:[11,14,18,25,28,39,89,120],straight:[75,78,84],straightforward:63,strategi:[7,31,77,118],stream:[9,42,63,74,87,102],stream_:87,streamid:22,street:4,strict:[78,109],stride:[4,5,18,63,67,86],stride_h:18,stride_i:4,stride_w:18,stride_x:4,string:[3,4,9,11,18,22,25,28,29,35,49,56,59,64,74,75,76,77,79,80,83,84,85,90,100,101,112,118],strip:106,strongli:109,struct:[35,36,48,50,58,59,66,67,74,80,85,88,102],structur:[28,29,35,51,54,64,69,75,77,79,84,110,112],sts:112,stuff:99,style:[4,79,85,91],sub:[3,4,9,18,26,28,38,40,44,54,60,68,71,73,77,97,100,122],sub_block:28,sub_nest_seq:4,sub_sequ:4,subclass:[20,77],subcommand:49,subgradi:6,subgraph:[44,54],submiss:42,submit:[63,79,104,109,112,117,118],subnet0:112,subnet:[26,112],subobjectpath:113,subscript:18,subsequ:[4,71],subsequenceinput:4,subset:[18,100],succ:60,succeed:[35,113],success:[4,36,112,113],successfulcr:113,successfulli:101,successor:118,sucess:60,sucessor:60,sudo:[98,112],suffer:27,suffici:118,suffix:[14,39,94,109],suggest:[4,30,99,107],suit:126,suitabl:[84,87,118],sum:[4,6,19,28,29,32,43,58,77,97,100],sum_op:28,sum_x:18,sume:18,summar:[54,74],summari:74,summat:18,sumopgradmak:80,sun:16,supercomput:60,suppli:[18,84],support:[2,3,4,6,7,9,11,18,27,29,31,38,39,40,41,42,44,51,53,54,60,63,64,67,69,72,73,74,76,78,79,80,81,84,86,88,91,92,94,95,97,98,100,101,103,104,107,109,112,118,122,123,124,126],support_inplac:60,suppos:[5,18,19,30,40,53,84,100],suppress:[4,49],sure:[18,91,98,100,106,111,112],surpass:[4,16],svs:85,swagger:48,swig:[37,65,66,91,122,123],switchop:29,sychron:71,symbol:[4,29,56,66,94],symbols_ready_:29,symbolt:[29,79],symlink:99,sync:[31,72,81,118],sync_with_cpp:106,syncflag:100,synchron:[31,35,41,63,71,74,109,112,118],syntax:[40,47,51,69,78],sysroot:122,system:[29,30,31,36,38,41,42,44,48,53,54,60,62,86,91,92,94,101,104,106,109,113,122],t_max:18,t_min:18,tab:94,tabl:[3,4,18,29,40,51,58,64,84,90,123],tablelookup:84,tablelookupgrad:84,tablelookupop:84,tableproject:4,tag:[3,10,82,92,97,109],tagtyp:3,tail:69,take:[3,4,5,9,18,25,26,28,29,30,31,38,40,41,42,45,47,50,52,54,55,57,58,60,63,67,72,75,76,77,78,79,80,87,88,89,91,97,98,99,100,101,102,106,107,109,111,112,113],taken:[4,45,56,60,67,89],talk:[36,52,124],tangl:106,tanh:[4,5,54,69,97,100],tanhactiv:5,tanhshrink:18,tar:[11,25,112],tarbal:112,target:[4,10,18,20,25,28,29,30,45,47,54,56,70,76,79,91,97,101,122,123,124],target_block:[28,45],target_dict_dim:97,target_dict_s:69,target_dictionary_dim:4,target_language_embed:4,target_language_word:97,target_link_librari:30,target_word:69,targetinlink:4,task13:86,task14:86,task:[3,4,42,64,69,74,85,97,111,120],task_queu:35,taskentri:35,taskqueu:35,tbd:[37,63,86],tcp:[112,118],tear:107,technic:[28,31,111],techniqu:[18,60,97,100,106],technolog:[51,98],tee:113,tell:[31,35,36,69,85,92,107,122],templat:[53,63,85,87,101,102,103,113,126],tempor:[4,18,86],temporari:[14,28,39,47,60,72,77],tempori:60,ten:98,tensor:[18,19,27,30,40,41,43,44,50,51,52,54,56,58,59,63,64,67,68,69,84,89,90,101,108],tensor_arrai:40,tensor_array_read:89,tensor_array_s:89,tensor_array_stack:89,tensor_array_unstack:89,tensor_array_writ:89,tensor_data:64,tensor_in:88,tensor_s:27,tensor_test:30,tensor_to_check:27,tensorarrai:43,tensorarraydesc:89,tensordesc:[64,84],tensorflow:[29,40,41,42,44,51,54,57,81,89,103],term:[4,5,18,31,80,81,86],termin:113,terminolog:60,tessorarrai:89,test100:10,test10:10,test1:33,test:[4,9,10,11,18,25,26,27,30,56,66,72,78,82,95,98,103,107,108,109,111,114,117],test_:101,test_all_data_in_one_period:113,test_check_grad_ingore_i:101,test_check_grad_ingore_x:101,test_check_grad_norm:101,test_check_output:101,test_data_dir:109,test_fcgrad:100,test_gpuprofil:107,test_layergrad:100,test_mkldnn:62,test_mklpack:61,test_mul_op:[91,101],test_norm:101,test_pass:[117,118,120],test_period:[117,118,120],test_recurrent_op:99,test_wait:[117,118],testa:26,testb:26,testbilinearfwdbwd:107,testcas:101,testconfig:100,testfcgrad:100,testfclay:100,testlayergrad:100,testmodel_list:117,testmulop:101,testq:26,testresult:25,testsave_dir:117,testutil:100,text1:49,text:[3,5,9,18,26,64,68,74,86,112],text_fil:9,tflop:107,tftp:126,tgz:[10,94],than:[2,3,4,5,18,28,31,39,40,45,51,52,53,54,77,79,81,89,91,97,98,100,109,111,112,122,126],the_step:51,theano:51,thehalf:18,thei:[4,14,16,18,23,26,28,30,31,36,38,40,41,44,45,49,51,54,55,59,60,69,70,74,77,79,85,89,90,97,98,99,100,101,103,107,109,111,112,117],them:[3,4,5,11,18,26,27,28,30,31,34,39,41,44,45,51,52,53,58,59,60,69,77,78,79,80,83,84,85,88,89,90,98,99,101,104,107,111,112,117,118],themselv:[28,30],theori:[51,107],therefor:[28,60,72],therein:[4,18,29],theta:54,theta_d:54,theta_g:54,thi:[2,3,4,5,6,9,10,11,14,16,17,18,20,22,23,25,26,27,28,29,30,31,34,35,36,37,38,39,40,41,42,43,44,45,46,47,50,51,52,53,54,55,56,59,60,63,67,68,69,70,71,72,73,74,75,77,78,79,80,81,84,85,86,87,88,89,92,94,95,97,98,99,100,101,102,103,104,106,107,108,109,111,112,113,118,120,122,123,124,126],thin:58,thing:[42,54,79,87,107],think:[26,30,111],third:[4,31,56,101,106,107,122,123,124],third_parti:[4,62,122,123,124],thirt:98,those:[4,29,30,31,53,55,56,57,75,122],though:[89,126],thought:[30,107],thread:[40,41,43,74,100,106,107,118,120],thread_count:43,thread_id:74,thread_id_:74,thread_local_rand_use_global_se:[117,118],thread_pool:43,threadblocks:22,threadid:120,threadloc:107,threadpool:40,three:[3,4,18,19,27,28,31,41,46,50,51,52,55,63,69,70,73,74,75,78,86,87,118,122],threshold:[2,3,4,18,31,35,45,99,118],thresholdedrelu:18,through:[4,19,28,30,31,35,37,46,60,70,72,97,100,101,104,107,108,109,123],throughout:47,throughput:[107,109],thrust:79,thu:[4,18,38,46,56,60,86,100,112],tier:113,time:[4,5,7,9,18,25,26,27,30,31,35,38,41,42,44,45,51,53,58,60,61,67,68,69,71,74,77,78,79,80,84,85,86,89,90,91,97,98,103,106,107,113,118,120,126],timelin:[4,74,79,107],timeo:112,timeout:[31,35],timeout_sec:9,timestamp:[4,32],timestep:[4,83],tip:[122,123],titan:60,titl:10,tls:48,tmp:77,to_chw:11,to_no_sequ:4,to_sequ:4,to_tar:25,todo:[3,9,10,29,31,35,38,69,85,86],toend:4,togeth:[4,5,9,25,28,89,97,111],token:[3,4,18,26,86,97],toler:[25,27,91,101],too:[10,27,40,41,45,63,88,89],took:126,tool:[74,91,94,97,98,106,111,112,122,124],toolchain:[106,122,123],toolkit:86,top:[3,18,25,68,69,86,101],top_k:[3,18,69],top_level_rnn:68,topic:63,topk_generated_scor:69,topk_id:69,topk_indic:18,topk_out:18,topk_scor:69,toplevel:98,topolog:[26,31,42,56,60,64,73],topoloi:56,topolopi:25,torch:[29,51],toronto:10,total:[18,25,31,44,46,71,74,78,106,107,109,113,126],total_pass:78,tottim:106,toward:51,trace:[29,52,54],track:[31,35,56,77],tractabl:4,tradit:[4,29,50,86],traffic:42,trail:9,train100:10,train10:10,train:[0,2,3,4,9,10,11,16,18,28,29,33,35,36,38,40,45,46,47,51,52,54,55,60,61,64,71,72,73,74,75,76,77,79,81,84,86,87,90,97,100,105,107,115,116,117,123],train_config_dir:112,train_data:109,train_data_dir:109,train_id:112,train_list:109,train_loop:51,train_read:[42,111],trainabl:[4,64,77],traindot_period:117,trainer:[26,32,33,34,35,37,42,44,52,61,62,72,73,79,100,110,111,118,120],trainer_config:[112,113],trainer_config_help:100,trainer_count:[95,109,112,113,117,118,120],trainer_cpu:39,trainer_cr:39,trainer_gpu:39,trainer_id:[109,112,118],trainer_intern:34,trainer_mem:39,trainer_packag:39,trainer_prog:42,trainerid:38,training_rol:111,trainingjob:42,trainingtest_period:117,trainonebatch:34,tran:[63,100,118],trans_var:88,transact:[31,35],transcript:86,transfer:[60,74],transform:[4,5,11,18,79,86,97,100,103],transform_param_attr:5,transformed_st:5,translat:[4,5,60],translation_id:69,translation_scor:69,transpar:[69,110],transpil:[40,111],transport:118,transpos:[4,11,100],transpose_i:18,transpose_x:18,transposedfullmatrixproject:4,travers:[28,55,60],travi:99,treat:[4,18,29,36,60],treatment:[36,50],tree:[4,29,40,47,51,77,108,118,124],trg_dic_siz:69,trg_embed:[69,97],trick:69,tricki:65,trigger:[38,73],trim:4,trivial:[69,89],true_block:[29,57,75],true_imag:78,true_label:78,true_neg:46,true_posit:46,true_read:78,truth:[3,4,18],tune:[2,86,105,106],tuninglog_barrier_abstract:117,tupl:[4,5,9,10,11,14,18,25,28,77,78],ture:4,turn:[4,18,77,78,92],tutori:[91,92,97,100,101,106,107,112,114,115,116,123],twice:[44,54,111],twine:82,two:[3,4,5,18,19,26,28,36,37,38,39,40,41,42,46,49,50,51,52,54,55,58,60,64,67,69,72,74,75,78,79,80,81,83,84,85,86,88,89,90,91,97,101,103,107,110,112,120,122,124],txt:[30,39,49,61,62,100,104,109,112,114],type:[3,4,5,7,9,10,11,14,18,19,20,25,26,28,29,31,34,35,38,39,42,48,49,50,52,58,59,63,64,65,66,68,69,75,76,77,78,79,80,81,84,85,86,87,88,90,95,97,98,100,102,103,109,112,113,118,120,123],type_nam:85,typedef:[36,50,65,66,67,87,102],typeerror:45,typeid:85,typenam:[53,85,87,101,102,103],typic:[3,42,107,123],ubuntu:[82,94,95,106],ubyt:78,uci:10,uci_h:[95,111],ufldl:[4,18],uid:113,uint16_t:50,uint32:[48,64],uint32_t:74,uint64:[64,65],uint64_t:65,unawar:36,unbalanc:118,unbound:[60,97],unchang:18,unclear:38,uncreat:28,under:[30,35,44,71,88,91,92,103,104,109,110,112],underli:[18,69],understand:[16,51,77,86,106,107,126],understand_senti:97,undeterminist:107,uni:86,unidirect:[4,86],unifi:[47,56,84,99],uniform:[2,4,9,16,33,54,77,78,118],uniform_random:77,uniniti:28,uninstal:91,uniqu:[26,29,31,38,39,63,67,77,83,101,109,111,112,118],unique_nam:77,unique_name_gener:77,unique_ptr:[80,83,87,100],unit:[4,5,18,19,30,72,74,81,87,91,97,98,103],unittest:[66,99,101],unittestcheckgrad_ep:117,unix:41,unk:[84,90],unknown:[4,18],unless:18,unlik:[4,18,69,101],unnecessari:[28,86,99],unordered_map:83,unpack:89,unrol:68,unseen:81,unseg:4,unsign:[36,50],unstack:89,unstack_from:89,unsupervis:54,unsupport:101,until:[31,36,43,44,51,60,83,111,112],unzip:122,updat:[2,4,6,18,20,28,31,35,36,42,48,50,54,68,69,70,71,72,73,83,86,89,94,100,106,109,118,120],update_equ:25,update_hook:2,update_memori:29,update_op:70,updatecallback:100,updatestack:112,upgrad:[71,91,94],upload:[31,39,41,48,82,109],upon:31,upper:4,upstream:99,uri:112,url:[9,10,99],usag:[3,4,5,11,18,25,50,57,60,73,77,101,107,109,123],use:[2,3,4,5,7,9,10,11,16,18,20,23,25,26,27,29,30,31,37,42,43,44,47,50,54,56,58,59,60,63,67,69,70,71,73,74,77,83,84,85,86,88,89,90,91,92,94,95,97,98,99,100,101,102,104,106,107,109,112,113,118,120,122,123,124],use_cpu:59,use_cudnn:[18,19,59],use_eigen_bla:122,use_eigen_for_bla:[122,123],use_etcd:25,use_global_stat:4,use_gpu:[95,109,113,117,118,120],use_mkl_pack:61,use_mkldnn:[4,59,62],use_nesterov:20,use_old_updat:[34,117,118],use_peephol:18,use_sparse_remote_updat:34,used:[3,4,5,6,7,9,10,11,16,18,22,25,26,27,29,30,31,37,38,42,45,47,50,51,54,56,60,68,69,72,73,74,77,78,79,81,83,85,87,88,89,91,94,97,98,100,101,103,106,107,112,117,118,120,122,123,124],useful:[4,5,27,50,60,77,83,88,97,100,120,122],usegpu:100,user:[2,4,5,9,10,11,14,16,18,20,22,23,25,26,27,28,29,30,33,35,38,39,40,42,43,44,45,46,47,49,53,54,55,56,58,59,63,67,69,70,71,72,74,77,78,79,80,81,83,85,87,88,89,92,99,102,104,106,109,112,117,118,122,126],user_info:10,user_nam:33,usercert:33,userinfo:10,userkei:33,usernam:33,uses:[4,31,38,40,41,42,50,60,67,68,69,73,74,87,88,91,94,97,98,99,100,103,104,109,112,118,122],using:[2,4,5,9,18,25,26,28,29,30,31,35,36,38,39,41,42,47,49,50,51,53,54,56,58,60,68,70,72,75,77,78,80,81,83,85,86,87,91,92,93,94,95,97,99,100,101,102,103,104,107,109,112,113,114,118,120,122,124],usr:[91,92,109,112,118],usual:[4,18,25,28,39,60,67,74,75,81,87,99,101,106,107,112,118,120],util:[42,61,62,71,97,100,101,102,107,126],uuid:[32,38],v7a:122,v8a:122,val:28,valid:[4,11,18,78,79,83,101,112,123],valu:[2,3,4,7,9,10,11,16,18,19,22,25,27,28,29,31,40,41,45,46,56,57,60,62,64,68,69,70,72,73,75,79,83,84,85,89,90,97,100,101,102,111,112,118,120,122,123],value1:118,value2:118,value_:84,value_evalu:3,value_rang:9,valueerror:[18,19,56],values_:89,vanilla:97,var_nam:[28,88],var_recurs:45,vardesc:[29,55,75,77,79,84],vardescbuild:29,vari:[107,112],variabl:[6,9,10,14,16,17,18,19,20,26,27,29,40,42,44,45,46,47,52,54,55,56,57,58,67,68,69,70,72,75,76,80,81,84,85,86,88,89,100,101,102,106,108,109,112,113,122,123],variablenamemap:101,varialbl:54,varianc:4,variant:[4,58,67,87,89,102],varibal:28,varibl:56,varienc:89,varient:89,variou:[29,41,50,60,81,122],varproto:85,vars_:[29,83],vartyp:[18,84,90],vartypeinfer:58,vec1:4,vec2:4,vec2seq:86,veclib:123,vecter:18,vector:[4,5,9,10,18,26,29,34,36,56,57,63,68,69,74,77,79,80,84,86,89,97,100,103],vectorenable_parallel_vector:117,veloc:20,vendor:30,verb:10,verbos:[49,99],veri:[4,7,30,35,40,44,47,51,53,54,60,63,69,73,78,81,83,86,87,89,97,106,107,110],verifi:[29,100,123],version:[4,5,28,30,39,42,45,49,52,54,56,57,64,69,82,86,91,92,95,98,100,106,107,109,111,112,113,117,118,122,123,124],versu:26,vertic:4,vgg:[5,19],via:[18,28,31,67,93,99,107,111,112,123,126],view:[4,64,67],vim:92,viriabl:109,virtual:[45,58,59,80,87,98,102],virtualenv:98,visibl:[38,83],visit:[25,28],visual:[4,69,107],vlog:[34,99],vocabulari:86,voila:95,volum:[104,113],volumemount:[112,113],volumn:112,vutbr:10,w_f:18,wai:[3,5,18,26,28,36,38,41,47,51,59,60,69,72,77,78,81,89,97,98,99,100,120],wait:[31,36,43,102,109,111,118],walk:123,wang:18,wangkuiyi:30,want:[4,26,39,40,41,46,54,59,67,72,74,76,78,81,83,87,88,89,91,92,98,99,100,104,106,109,118,120,122,124],warn:[25,49],warp:[4,107],warp_ctc:86,warpctc:4,wast:71,watch:31,wbia:112,web:[104,106],websit:104,weight:[3,4,5,6,16,18,19,23,61,64,81,97,100,118,120],weight_act:5,weightlist:100,weights_:100,weights_primitive_desc:63,weights_t:100,welcom:[30,86],well:[18,28,39,41,42,44,51,53,54,81,84,86,100,111,112,118],wer:86,were:[3,30,41,51],west:112,wget:122,wgt:63,what:[2,4,30,51,54,69,77,85,88,101,106,126],whatev:[98,109],wheel:94,when:[2,3,4,6,9,14,18,19,25,27,28,29,30,31,34,35,36,39,40,42,44,45,46,47,49,50,51,52,56,69,71,72,73,74,75,77,79,87,89,91,93,97,98,99,100,101,103,104,106,107,109,112,113,118,120,122,123,126],whenev:[18,77,86,99],where:[4,5,6,16,18,19,26,28,29,31,38,40,42,51,52,55,67,68,69,72,75,79,81,87,89,97,100,101,106,107,108,118,120],wherea:[18,29,35,53,57,87,90],whether:[3,4,11,17,18,25,27,28,29,47,74,78,84,89,91,92,100,101,118,123],which:[2,3,4,5,9,10,11,18,19,25,26,27,28,29,30,31,33,35,36,38,39,40,41,42,43,45,47,50,51,52,53,54,56,58,60,63,64,67,68,69,70,71,73,75,76,77,78,79,80,83,84,85,88,89,90,91,94,97,98,99,100,101,102,103,106,107,109,110,111,112,118,120,122,123,126],while_grad:60,while_loop:[69,89],while_op:28,whileloop:89,whileop:29,white:86,whl:91,who:[28,53,55,71,77,99],whoever:36,whole:[3,9,28,54,57,60,65,66,68,71,76,85,86,99,109,112,113,126],whose:[4,9,18,27,28,31,38,45,68,79,80,85,89,97],why:[5,27,66,98],wide:[30,45,54,94,110,114],width:[3,4,9,11,18,34,65,78,100,101],wiki:[4,30],wikipedia:[4,10],window:[4,7,10,72,86,92,98,122],wirt:56,wise:[4,11,18,44,79,86,103],wish:[91,94,104,109,111],with_avx:[91,92,109,122,123],with_bia:85,with_c_api:[91,122,123,124],with_distribut:111,with_doc:91,with_doubl:[91,100,109],with_dso:91,with_golang:[91,122],with_gpu:[91,98,109,122,123],with_mkl:[61,62,91,122],with_mkldnn:62,with_mklml:62,with_profil:107,with_python:[91,109,122,123],with_rdma:[109,122,123],with_style_check:[91,99],with_swig_pi:[91,122,123],with_test:[91,101],with_tim:[107,109],within:[4,35,42,51,86,123],without:[3,4,20,28,31,36,41,74,77,78,79,86,101,106,109,114],wloop:89,wmt14:97,wmt_shrinked_data:10,won:[107,109],word2vec:[39,109],word:[3,4,10,28,44,55,58,60,68,69,79,85,86,89,97,102,109,120],word_dict:[109,114],word_idx:10,word_vector_dim:[4,69,97],wordcount:86,words_freq_sort:10,work:[4,9,18,26,29,30,31,42,47,50,51,59,70,72,74,77,92,97,98,99,100,104,106,107,109,111,112,113,118,126],worker:[44,90,112],workercount:112,workflow:[79,112],workspac:[99,109,110,118],world:109,worth:108,would:[25,29,30,31,38,41,42,43,44,51,53,54,55,63,70,72,73,77,78,84,86,89,92,98,99,106,111,112,122,126],wouldn:[51,55],wrap:[51,53,54,71,126],wrapper:[5,30,41,53,71,72,80,89,107],write:[9,18,26,31,38,40,42,44,50,51,52,53,56,58,63,70,72,77,78,79,80,87,89,98,105,109,112],write_lock:32,write_to_arrai:60,writer:[26,77],written:[18,22,28,29,40,44,47,54,64,72,79,80,84,91,92,101,103,106,110],wrong:78,wrote:56,wsize:112,www:[10,18],x64:[122,124],x86:123,x86_64:[122,123],x_first_step:18,x_last_step:18,x_neg:27,x_num_col_dim:18,x_po:27,x_reshap:18,x_t:18,x_t_data:18,x_transpos:18,xarg:[3,92,100,114],xavier:16,xavieriniti:18,xcode:123,xcodebuild:123,xeon:102,xgbe0:118,xgbe1:118,xiangyu:16,xmap_read:9,xpu:51,xrang:[27,51,54,74,78,95,100],xx_layer:59,xxx:[26,89],xxxx:32,xxxxxxxxx:112,xxxxxxxxxx:112,xxxxxxxxxxxxx:112,xxxxxxxxxxxxxxxxxxx:112,y_dim:54,y_neg:27,y_num_col_dim:18,y_po:27,y_predict:[18,95,108,111],yaml:[30,110,112,114,126],yancey1989:39,yann:10,yapf:99,year:51,yeild:25,yep:[74,106],yet:[51,86,126],yield:[9,26,33,78],yoshua:16,you:[2,4,5,9,19,25,27,39,42,50,83,91,92,93,94,95,97,98,99,100,101,102,104,106,107,109,110,111,112,114,118,120,122,123,124,126],your:[4,9,25,26,30,34,39,49,79,91,93,94,98,99,100,104,107,109,110,111,112,120,122,123,124,126],your_access_key_id:112,your_secrete_access_kei:112,your_source_root:66,yourself:91,yuang:51,yuyang18:[9,10],yuyang:106,z_dim:54,z_size:54,zero:[2,4,5,6,9,10,14,27,28,31,54,69,73,77,84,100,112,118],zhang:16,zip:[10,77,122],zone:112,zxvf:112},titles:["API","Activation","Parameter Attribute","Evaluators","Layers","Networks","Optimizer","Pooling","Data Reader Interface and DataSets","Data Reader Interface","Dataset","Image Interface","Fluid","DataFeeder","Evaluator","Executor","Initializer","IO","Layers","Nets","Optimizer","ParamAttr","Profiler","Regularizer","Model Configuration","Training and Inference","PaddlePaddle Design Doc","Auto Gradient Checker Design","Backward Building","Design Doc: Block and Scope","Required CMake Function","Design Doc: Distributed Training","\u6a21\u578b\u53c2\u6570\u68c0\u67e5\u70b9\uff08Checkpointing\uff09","\u8bad\u7ec3\u6570\u636e\u7684\u5b58\u50a8\u548c\u5206\u53d1","Alalysis of large model distributed training in Paddle","Design Doc: Master Server","Design Doc: The Client Library of Parameter Server","Design Doc: Remote Parameter Updater for Cluster Train","Design Doc: Save Model","Submit a Distributed Training Job","Design Doc: Concurrent Programming with Fluid","Design Doc: CSP in PaddlePaddle Fluid","Design Doc: Distributed Training Architecture","Design Doc: Execute the Program with Multi CPU","Design Doc: Parameter Server","Error Clip","Evaluator Design","Executor Design Doc","FileManager\u8bbe\u8ba1\u6587\u6863","PFSClient","Design Doc: float16","Design Doc: PaddlePaddle Fluid","PaddlePaddle Fluid: Towards a Compiled Programming Language","Design Doc: Functions, Operators, and Layers","Design for GAN","Design Doc: Computations as a Graph","Survey on Graph","TheIfElseOperator","Design Doc: InferVarType","Problem","Memory Optimization","Intel\u00ae MKL Packed on PaddlePaddle: Design Doc","Intel\u00ae MKL-DNN on PaddlePaddle: Design Doc","Design Doc: Add MKLDNN Kernel in Fluid Operator","Design Doc: Model Format","Paddle\u591a\u8bed\u8a00\u63a5\u53e3\u5b9e\u73b0","C-API \u6a21\u578b\u63a8\u65ad\u5b9e\u73b0\u6587\u6863","Design Doc: The Keys of Operator Kernel Type","RNNOp design","Design: Sequence Decoder Generating LoDTensors","Optimizer Design","Design Doc: NCCL support in Paddle Fluid","Averaging Parameter in PaddlePaddle","Design Doc: The C++ ClassParameters","Introduction","Design Doc: PaddlePaddle Programs","Prune","Design Doc: Python API","Python Data Reader Design Doc","Design Doc: Refactorization Overview","Design Doc: Gradient Operators Registration","Regularization in PaddlePaddle","PaddlePaddle\u53d1\u884c\u89c4\u8303","Design of Scope in Paddle","Design Doc: Selected Rows","Interaction between C++ and Python","DeepSpeech2 on PaddlePaddle: Design Doc","Design Doc: Supporting new Device/Library","Background","Design for TensorArray","Background","Build from Sources","Run in Docker Containers","Install and Build","Install Using pip","GET STARTED","RNN Models","RNN Configuration","Build using Docker","Contribute Code","Write New Layers","How to write a new operator","Add Kernels for a New Device","How to use Eigen in Paddle","Contribute Documentation","HOW TO","Profiling the Python Code","Tune GPU Performance","PaddlePaddle Fluid Source Code Overview","Distributed Training","Cluster Training Using Fabric","Fluid Distributed Training","Distributed PaddlePaddle Training on AWS with Kubernetes","PaddlePaddle On Kubernetes","Cluster Training Using OpenMPI","<no title>","<no title>","Argument Outline","Detail Description","Set Command-line Parameters","Use Case","PaddlePaddle Documentation","Build PaddlePaddle for Android","Build PaddlePaddle for iOS","Build PaddlePaddle for Raspberry Pi","MOBILE","Cluster bootstrapping tool survey"],titleterms:{"\u4e0a\u4f20\u8bad\u7ec3\u6587\u4ef6":33,"\u4e0d\u4f7f\u7528":65,"\u4e0d\u4f7f\u7528swig\u8fd9\u79cd\u4ee3\u7801\u751f\u6210\u5668":65,"\u4e0d\u5bfc\u51fapaddle\u5185\u90e8\u7684\u7ed3\u6784\u4f53":65,"\u4e0d\u5f15\u7528\u5176\u4ed6\u52a8\u6001\u5e93":65,"\u4ec5\u4ec5\u4f7f\u7528void":65,"\u4ece\u5feb\u7167\u6062\u590d":32,"\u4f7f\u7528\u52a8\u6001\u5e93\u6765\u5206\u53d1paddl":65,"\u4f7f\u7528\u8f6c\u6362\u5e93":33,"\u5177\u4f53\u67d0\u79cd\u7c7b\u578b\u7684\u5934\u6587\u4ef6":66,"\u5177\u4f53\u67d0\u79cd\u7c7b\u578b\u7684\u5b9e\u73b0\u6587\u4ef6":66,"\u5206\u5757\u6587\u4ef6\u4f20\u8f93":48,"\u5206\u652f\u89c4\u8303":82,"\u52a0\u901f\u6267\u884c":32,"\u52a8\u6001\u5e93\u4e2d\u4e0d\u5d4c\u5165\u4efb\u4f55\u5176\u4ed6\u8bed\u8a00\u7684\u89e3\u91ca\u5668":65,"\u52a8\u6001\u6269\u5bb9":32,"\u539f\u56e0":65,"\u539f\u56e0\u5217\u8868":65,"\u53c2\u8003\u6587\u6863":48,"\u53d1\u5e03docker\u955c\u50cf":82,"\u53d1\u5e03wheel\u5305\u5230pypi":82,"\u540d\u8bcd\u89e3\u91ca":48,"\u57fa\u672c\u8981\u6c42":65,"\u5b9e\u73b0":65,"\u5b9e\u73b0\u65b9\u5f0f":66,"\u5bfc\u51fac":65,"\u5feb\u7167\u4fdd\u5b58\u7684\u8bbe\u8ba1\u5982\u4e0b":32,"\u6307\u9488\u4f5c\u4e3a\u7c7b\u578b\u7684\u53e5\u67c4":65,"\u63a8\u6d4b\u6267\u884c":32,"\u652f\u6301\u7528\u6237\u81ea\u5b9a\u4e49\u7684\u6570\u636e\u9884\u5904\u7406job":33,"\u6587\u4ef6\u4f20\u8f93\u4f18\u5316":48,"\u6587\u4ef6\u8bbf\u95ee\u65b9\u5f0f":33,"\u6587\u4ef6\u8bbf\u95ee\u7684\u6743\u9650":33,"\u6587\u4ef6\u9884\u5904\u7406":33,"\u66b4\u9732\u63a5\u53e3\u539f\u5219":66,"\u672f\u8bed":32,"\u67b6\u6784\u56fe":48,"\u6846\u67b6\u751f\u6210":48,"\u6982\u5ff5\u89e3\u91ca":33,"\u6a21\u5757":48,"\u6a21\u578b\u53c2\u6570\u68c0\u67e5\u70b9":32,"\u6a21\u578b\u63a8\u65ad\u5b9e\u73b0\u6587\u6863":66,"\u6d41\u7a0b\u4ecb\u7ecd":33,"\u751f\u6210sparse\u6587\u4ef6":48,"\u7528\u6237\u4f7f\u7528\u6d41\u7a0b":48,"\u76ee\u5f55\u7ed3\u6784":66,"\u76ee\u6807":48,"\u793a\u4f8b\u7a0b\u5e8f":33,"\u7b26\u53f7":65,"\u7c7b":65,"\u7f16\u8bd1\u9009\u9879":66,"\u7f29\u5bb9":32,"\u800c\u662f\u624b\u5199\u591a\u8bed\u8a00\u7ed1\u5b9a":65,"\u80cc\u666f":65,"\u8986\u76d6\u4e0d\u4e00\u81f4\u7684\u90e8\u5206":48,"\u8bad\u7ec3\u6570\u636e\u5b58\u50a8":33,"\u8bad\u7ec3\u6570\u636e\u7684\u5b58\u50a8\u548c\u5206\u53d1":33,"\u8f6c\u6362\u5e93":33,"\u8fd9\u4e2a\u52a8\u6001\u5e93\u4f7f\u7528c99\u6807\u51c6\u7684\u5934\u6587\u4ef6\u5bfc\u51fa\u4e00\u4e9b\u51fd\u6570":65,"\u8fdb\u884c\u8bad\u7ec3":33,"abstract":[42,43,44,71,126],"book\u4e2d\u6240\u6709\u7ae0\u8282":82,"case":[28,120],"class":[54,73,77,100],"filemanager\u8bbe\u8ba1\u6587\u6863":48,"final":59,"function":[30,53,54,77],"new":[87,100,101,102],"paddle\u52a8\u6001\u5e93\u4e2d":65,"paddle\u591a\u8bed\u8a00\u63a5\u53e3\u5b9e\u73b0":65,"paddlepaddle\u53d1\u884c\u89c4\u8303":82,"paddlepaddle\u56de\u5f52\u6d4b\u8bd5\u5217\u8868":82,"return":[77,78],"switch":[63,87],"tensor\u5230eigentensor\u7684\u8f6c\u6362":103,AWS:112,Abs:1,DNS:112,E2E:111,EFS:112,For:[30,113],KMS:112,Not:98,The:[29,36,40,47,51,54,55,57,58,67,70,73,79,80,124],Use:[29,75,104,109,113,120],Using:[30,36,94,110,114,122],With:39,about:54,abs:18,absolut:69,access:112,account:112,accuraci:18,action:[61,62],activ:[1,4,62],actor:41,adadelta:6,adagrad:6,adagradoptim:20,adam:6,adamax:6,adamaxoptim:20,adamoptim:20,add:[60,63,102,112],address:112,addto:4,advanc:87,aggreg:4,aggregatelevel:4,alalysi:34,algorithm:[27,31,42,68,76],all:[83,89],analog:40,analysi:[42,60],android:122,api:[0,42,61,62,66,70,72,77,81,85],appendix:126,approach:107,arbitrari:51,architectur:[42,74,97],argument:[49,78,109,117,120,122,124],arrai:27,array_length:18,array_read:18,array_to_lod_tensor:18,array_writ:18,asset:112,assign:18,associ:[83,112],assumpt:126,async:118,attent:97,attribut:[2,60,81],auc:3,auto:27,averag:72,avg:7,aws:112,backgraound:27,background:[44,61,87,88,89,90,101,102],backward:[28,51,55,79,101],base:[39,69],basepool:7,basic:[60,87,126],batch:78,batch_norm:[4,18],batch_siz:78,beam:[69,86],beam_search:4,beam_search_decod:18,benchmark:[61,62],benefit:[44,79],between:[26,41,77,79,85,87],bidirectional_gru:5,bidirectional_lstm:5,bilinear_interp:4,binari:29,bind:101,bla:91,block:[29,52,54,55,75,77,79],block_expand:4,blockdesc:75,book:92,bool:91,bootstrap:126,bottleneck:106,brelu:[1,18],bring:126,bucket:112,build:[28,54,79,91,93,98,104,113,122,123,124],built:107,can:83,capi:66,capi_priv:66,cast:18,ceil:18,challeng:[28,44,76],chang:69,channel:41,check:[4,27,100,110],checker:27,checkpoint:[31,32,38],choic:59,choos:[30,112],chunk:3,cifar:10,classif:3,classification_error:3,classification_error_print:3,client:36,clip:[4,45],close:27,cloudform:112,cluster:[37,109,110,111,112,114,120,126],cmake:[30,61,62,124],code:[39,52,77,99,106,108],column_sum:3,command:[109,119,120],commit:113,common:118,commun:118,compar:126,comparis:77,compat:51,compil:[29,50,52,75,79,91,101,108,122,123,124],complet:51,compos:78,comput:[29,55,63,79,81,103],con:126,concat:[4,18],concept:[77,79,112],concern:62,conclus:[38,56,126],concurr:[40,41],condit:54,config:120,configur:[24,97,105,112,123],conll05:10,connect:4,constantiniti:16,construct:55,contain:[92,113],content:[61,62,66,86,107,112],context:102,context_project:4,contribut:[99,104],control:[60,79],contruct:60,conv2d:18,conv2d_transpos:18,conv:4,conv_oper:4,conv_project:4,conv_shift:4,convert:38,convolut:86,core:[27,77,112],corner:28,cos_sim:[4,18],cost:4,cpu:[43,120],creat:[28,41,78,79,83,112,113],create_arrai:18,creation:[35,72,81],creator:78,credenti:112,crf:4,crf_decod:4,cross:[122,123,124],cross_channel_norm:4,cross_entropi:18,cross_entropy_cost:4,cross_entropy_with_selfnorm_cost:4,csp:41,ctc:[4,86],ctc_error:3,ctc_greedy_decod:18,cuda:[50,91],cudnn:91,cudnnavg:7,cudnnmax:7,current:[50,80],custom:78,data:[4,8,9,18,31,42,78,88,112,113],datafeed:[9,13],dataflow:60,dataprovid:118,dataset:[8,10,31,35,109],datatyp:[9,67],decayedadagrad:6,decayedadagradoptim:20,decod:69,decor:78,deep:[29,51],deepspeech2:86,defin:[101,112],definit:90,delet:112,demo:[54,111,112],dens:38,dep:94,depend:[54,86,91,94],deploi:39,deriv:100,describ:[51,70],descript:[49,79,118],design:[26,27,29,31,35,36,37,38,40,41,42,43,44,46,47,50,51,53,54,55,58,61,62,63,64,67,68,69,70,71,73,75,77,78,79,80,83,84,86,87,89],destroi:[83,112],detail:[34,86,118],detect:[3,4],detection_map:3,detection_output:4,develop:[79,98,105],devic:[87,102,120],devicecontext:87,dictionari:78,differ:[79,87,109,120],directori:112,discrimin:54,discuss:[44,54],dispatch:[31,35],distribut:[26,31,34,39,42,44,109,111,112,118],dnn:62,doc:[26,29,31,35,36,37,38,40,41,42,43,44,47,50,51,53,55,58,61,62,63,64,67,71,73,75,77,78,79,80,84,86,87],docker:[39,92,98,113,122],document:[104,121],doe:78,dot_prod:4,dot_product_attent:5,dotmul_oper:4,dotmul_project:4,down:112,download:[112,113],dropout:[4,18],dure:[69,78],dylib:66,dynam:[31,89],dynamic_gru:18,dynamic_lstm:18,dynamic_lstmp:18,dynet:56,each:94,ec2:112,edit_dist:18,eigen:103,elast:112,elect:38,elementwise_add:18,elementwise_div:18,elementwise_mul:18,elementwise_sub:18,els:29,elu:18,embed:[4,18],engin:54,enough:27,entri:78,environ:[39,122],eos:4,equat:100,error:45,evalu:[3,14,46],event:[25,26,74],evolut:51,examin:106,exampl:[26,30,40,41,57,66],execut:[29,43,51,75,79],executor:[15,47],exp:[1,18],expand:4,expandlevel:4,explain:27,extern:112,fabric:110,factor:4,factorization_machin:4,faq:[93,94],fault:31,file:[29,106,112,113],fill_const:18,fill_constant_batch_size_lik:18,find:112,first_seq:4,float16:50,floor:18,flow:60,fluid:[12,40,41,51,52,63,71,108,111],format:[29,31,64],forward:[55,101],frame:29,framework:[27,102,103],from:[26,38,85,91,93],full_matrix_project:4,fulli:4,functor:87,futur:[51,86],gan:54,gate:97,gated_unit:4,gener:[52,54,69,97,98,106,126],get:[95,111,113],get_output:4,give:78,global:[75,77],glu:19,gotcha:98,gpu:[92,107,118,120],grad_op:28,gradient:[27,28,36,62,80,100],gradient_print:3,graident:27,graph:[55,56,60,79,81],group:[4,112],gru:[5,118],gru_group:5,gru_step:4,gru_unit:[5,18],grumemori:4,hand:107,handler:[26,65],happen:38,hard_shrink:18,hard_sigmoid:18,hardwar:50,have:111,helper:77,hierarchi:29,high:[70,72,81,85],how:[27,34,72,78,79,87,91,101,103,104,105,107],hsigmoid:4,huber_classification_cost:4,huber_regression_cost:4,iOS:123,iam:112,ident:1,identifi:106,identity_project:4,ifels:57,ifelseop:29,im2sequ:18,imag:[4,5,11,39,92,113,122],imdb:10,img_cmrnorm:4,img_conv:4,img_conv_bn_pool:5,img_conv_group:[5,19],img_pool:4,imikolov:10,implement:[27,28,30,34,43,45,46,50,64,68,71,72,77,78,79,80,81,100,101,102,103],increment:18,infer:[25,122],infershap:[75,84],infervartyp:58,ingredi:26,ingress:48,initi:[16,36,54,112,120],insid:83,inspect:112,instal:[93,94,95,111,112,122,123,124,126],instanc:112,instead:78,integr:[87,112],intel:[61,62],interact:85,interfac:[8,9,11,27,31,36,37,47,70,78,83],intermedi:79,interpol:4,introduc:[69,89,111],introduct:[74,81,109,111],is_paramet:17,isn:78,issu:50,job:[31,39,110,112,113,114],join:4,kei:[61,67,112],kernel:[63,67,79,102],kill:110,kmax_sequence_scor:4,kube:112,kubectl:112,kubernet:[39,112,113],l1decayregular:23,l2_distanc:4,l2_normal:18,l2decayregular:23,lambda_cost:4,languag:[29,52],larg:34,last_seq:4,launch:[92,110,114],layer:[4,18,26,53,61,62,77,100,120],layout:67,leaky_relu:18,learn:[29,51],learnabl:4,less_than:18,leval:85,level:[70,72,81,85],libpaddle_capi_shar:66,libpaddle_capi_whol:66,librari:[36,50,67,79,87,122],limit:42,line:[109,119],linear:1,linear_chain_crf:18,linear_comb:4,linux:[110,122],list:[32,78],live:60,load:41,local:[42,83,112,120],lod:69,lod_rank_t:18,lod_tensor_to_arrai:18,lodtensor:[68,69,89],lodtensordesc:90,log:[1,18,99],logic:35,logsigmoid:18,look:106,low:[72,81,85],lstm:[5,118],lstm_step:4,lstm_unit:18,lstmemori:4,lstmemory_group:5,lstmemory_unit:5,machin:[4,69],macro:79,main:54,make:60,manag:[30,109],map:[78,79],master:[31,35,39,40],math:[4,87],mathemat:27,matmul:18,matrix:[62,118],max:7,max_sequence_len:18,maxframe_print:3,maxid:4,maxid_print:3,maxout:4,mean:18,member:54,memori:[4,60,68,87],merge_lod_tensor:18,messag:85,method:69,might:54,migrat:79,mileston:79,mini:78,minibatch:[9,41],misc:4,mix:[4,120],mkl:[61,62],mkldnn:63,mkldnn_helper:63,mkldnndevicecontext:63,mnist:10,mobil:125,model:[24,26,34,36,38,41,51,54,64,69,96,97,110,120],modifi:113,modul:[79,87,103],momentum:6,momentumoptim:20,more:54,motiv:[28,41,47,64,71,76],movielen:10,msrainiti:16,mul:18,multi:[43,52],multi_binary_label_cross_entropy_cost:4,multibox_loss:4,multipl:78,multiplex:[4,18],mxnet:56,name:[83,112],nativ:52,nccl:71,nce:4,necess:77,necessari:79,need:[78,98,107],nest:68,net:19,network:[5,79,97,120],neural:97,nlp:[5,118],non:111,norm:[4,81],normaliniti:16,note:27,numer:27,numpi:27,nvprof:107,nvvp:107,object:31,offset:69,ones:18,onli:[78,83],op_mak:79,openmpi:114,oper:[53,57,60,63,67,72,75,77,79,80,84,89,101],opinfomap:79,opkernel:[79,87,101,102],opproto:85,ops:81,optim:[6,20,31,36,55,60,70,77,105],option:[49,91],opwithkernel:79,order:49,org:104,origin:79,orthogon:83,other:62,out_prod:4,outlin:117,output:[4,110,112],overview:[38,45,47,61,62,79,83,86,108],pack:[61,69],packag:[30,94],pad:4,paddl:[34,71,78,83,103],paddlejob:39,paddlepaddl:[26,29,41,51,52,61,62,72,75,81,82,86,91,92,94,104,108,111,112,113,121,122,123,124],pair:112,paradigm:51,parallel_nn:120,paramattr:21,paramet:[2,4,25,26,31,36,37,39,41,44,62,72,73,77,81,109,111,112,118,119],parameteraverageoptim:72,parent:83,part:55,partit:36,pass:[91,120],path:[38,49],penalti:81,perform:[72,106,107,118],persist:35,pfsclient:[48,49],pfsserver:48,pip:94,place:[60,67,87,102],placement:42,platform:109,pnpair:3,point:[61,112],polici:60,pool2d:18,pool:[4,7],pose:[58,80],potenti:59,pow:18,power:4,precision_recal:3,prefetch:78,prelu:4,prepar:[109,110,111,112,114,123],principl:63,print:3,privat:112,pro:126,problem:[46,58,59,60,67,70,80,88],procedur:126,process:[31,36,39,70,79,98],profil:[22,106,107],program:[29,40,41,43,51,52,75,77,92,109,111],programdesc:[52,75],project:30,propos:[58,80,81],protobuf:84,protomak:101,provid:78,prune:76,pserver:38,pull:92,python:[27,39,42,61,62,68,70,72,77,78,81,85,90,100,101,106],qualiti:79,queue:[31,35],quick:95,randomnumb:118,rank:3,rank_cost:4,raspberri:124,reader:[8,9,26,78],readi:111,realiz:79,reciproc:18,recoveri:31,recurr:[4,5,97],recurrent_group:4,recv:41,reduce_max:18,reduce_mean:18,reduce_min:18,reduce_sum:18,ref:27,refactor:79,refer:[42,44,60,61,62,86,107],region:112,regist:[58,79,85,101,102],registr:[79,80],registri:79,regular:[23,36,81],rel:69,relat:[79,89],relu6:18,relu:[1,18],remark:101,remot:37,remoteexecutor:42,render:112,repeat:4,represent:[29,79],requir:[30,54],reshap:[4,18],resiz:4,result:[110,113],retri:35,reus:77,review:99,rmsprop:6,rnn:[68,89,96,97,118],rnnop:[29,68,79],roi_pool:4,rotat:4,round:18,route53:112,row:[84,86],row_conv:[4,18],row_l2_norm:4,rpc:41,run:[47,91,92,101,108,113],runtim:[39,94],sampl:4,sampling_id:4,save:38,scale:[4,18,31],scale_shift:4,scaled_dot_product_attent:19,scaling_project:4,scope:[29,68,79,83],script:[111,113],search:[69,86],secur:112,select:[36,41,84],selectedrow:84,selective_fc:4,send:41,sentiment:10,separ:79,seq_concat:4,seq_reshap:4,seq_slic:4,seqtext_print:3,sequenc:[69,97],sequence_conv:18,sequence_conv_pool:[5,19],sequence_expand:18,sequence_first_step:18,sequence_last_step:18,sequence_pool:18,sequence_reshap:18,sequence_softmax:18,sequencesoftmax:1,server:[31,35,36,39,41,44,109,111,112,118],servic:112,set:119,setup:[112,122],sextant:126,sgd:118,sgdoptim:20,shape:69,share:[26,28,60,83],should:83,shrink_memori:18,shuffl:78,sigmoid:[1,18],sigmoid_cross_entropy_with_logit:18,simpl:[69,97],simple_attent:5,simple_gru2:5,simple_gru:5,simple_img_conv_pool:[5,19],simple_lstm:5,singl:78,slice:[4,111],slice_project:4,slope_intercept:4,small_vgg:5,smooth_l1_cost:4,soft_relu:18,softmax:1,softplu:18,softrelu:1,softshrink:18,softsign:[1,18],solut:[58,59,60,61,67,76,80,88],some:98,sourc:[91,93,108],spars:[36,37,38,84,120],specifi:120,split:18,split_lod_tensor:18,spp:4,sqrt:18,squar:[1,18],square_error_cost:[4,18],squarerootn:7,stack:29,standard:99,stanh:1,start:[26,95,109,112,113],startup:113,statement:46,step:[68,93],storag:81,store:31,strategi:60,style:99,sub_nested_seq:4,sub_seq:4,subcommond:49,submit:39,suffici:78,suitabl:30,sulut:63,sum:[3,7,18],sum_cost:4,sum_to_one_norm:4,summar:[26,40],summari:64,support:[50,71,87,89],survei:[50,56,81,126],swish:18,synopsi:49,syntax:41,system:[51,112],tabl:[66,86],table_project:4,tanh:[1,18],tanh_shrink:18,task:[31,35,86],tear:112,tecton:126,templat:112,tensor:[4,79,87,103],tensorarrai:[69,89],tensordesc:90,tensorflow:56,test:[61,62,63,91,99,100,101,118,120],text_conv_pool:5,theori:27,thi:83,think:54,three:89,thresholded_relu:18,time:108,timelin:38,timer:107,tip:107,todo:[32,33,43],togeth:83,toler:31,tool:[30,104,107,109,126],toolchain:124,topic:87,topk:18,toward:52,train:[25,26,31,34,37,39,42,70,78,92,109,110,111,112,113,114,118,120],trainer:[25,31,36,38,39,41,109,112],tran:4,trans_full_matrix_project:4,transform:88,translat:69,transpil:[42,43,44,52,60,71],transpos:18,tune:[107,118],ture:51,two:27,type:[41,67,91,101],uci_h:10,uniform:89,uniforminiti:16,unit:[61,62,63,99,100,101,118],unpack:69,updat:[26,37,38,104,111,112],usag:[28,45,68,69,78,103,105],use:[34,78,103],user:31,using:98,util:3,valu:77,value_print:3,vardesc:90,variabl:[28,60,77,79,83,90],vector:118,verifi:112,version:[40,50,94],vgg_16_network:5,volum:112,vpc:112,warp_ctc:4,weightdecayregular:23,what:[34,38,98,107],when:[38,83],whl:94,why:[50,51,72,78,79,89,107],wmt14:10,work:86,worker:40,workflow:99,wrapper:100,write:[99,100,101,102,104],www:104,xavieriniti:16,yaml:113,your:[92,102],zero:18}}) \ No newline at end of file +Search.setIndex({docnames:["api/index_en","api/v2/config/activation","api/v2/config/attr","api/v2/config/evaluators","api/v2/config/layer","api/v2/config/networks","api/v2/config/optimizer","api/v2/config/pooling","api/v2/data","api/v2/data/data_reader","api/v2/data/dataset","api/v2/data/image","api/v2/fluid","api/v2/fluid/data_feeder","api/v2/fluid/evaluator","api/v2/fluid/executor","api/v2/fluid/initializer","api/v2/fluid/io","api/v2/fluid/layers","api/v2/fluid/nets","api/v2/fluid/optimizer","api/v2/fluid/param_attr","api/v2/fluid/profiler","api/v2/fluid/regularizer","api/v2/model_configs","api/v2/run_logic","design/api","design/auto_gradient_check","design/backward","design/block","design/build_system/README","design/cluster_train/README","design/cluster_train/checkpointing","design/cluster_train/data_dispatch","design/cluster_train/large_model_dist_train","design/cluster_train/master_server","design/cluster_train/pserver_client","design/cluster_train/remote_parameter_updater","design/cluster_train/save_model","design/cluster_train/submit-job","design/concurrent_programming","design/csp","design/dist_refactor/distributed_architecture","design/dist_refactor/multi_cpu","design/dist_refactor/parameter_server","design/error_clip","design/evaluator","design/executor","design/file_manager/README","design/file_manager/pfs/pfsclient","design/float16","design/fluid","design/fluid_compiler","design/functions_operators_layers","design/gan_api","design/graph","design/graph_survey","design/if_else_op","design/infer_var_type","design/kernel_hint_design","design/memory_optimization","design/mkl/mkl_packed","design/mkl/mkldnn","design/mkl/mkldnn_fluid","design/model_format","design/multi_language_interface/00.why_plain_c","design/multi_language_interface/01.inference_implementation","design/operator_kernel_type","design/ops/rnn","design/ops/sequence_decoder","design/optimizer","design/paddle_nccl","design/parameter_average","design/parameters_in_cpp","design/profiler","design/program","design/prune","design/python_api","design/reader/README","design/refactorization","design/register_grad_op","design/regularization","design/releasing_process","design/scope","design/selected_rows","design/simple_op_design","design/speech/deep_speech_2","design/support_new_device","design/switch_kernel","design/tensor_array","design/var_desc","getstarted/build_and_install/build_from_source_en","getstarted/build_and_install/docker_install_en","getstarted/build_and_install/index_en","getstarted/build_and_install/pip_install_en","getstarted/index_en","howto/deep_model/rnn/index_en","howto/deep_model/rnn/rnn_config_en","howto/dev/build_en","howto/dev/contribute_to_paddle_en","howto/dev/new_layer_en","howto/dev/new_op_en","howto/dev/new_op_kernel_en","howto/dev/use_eigen_en","howto/dev/write_docs_en","howto/index_en","howto/optimization/cpu_profiling","howto/optimization/gpu_profiling_en","howto/read_source","howto/usage/cluster/cluster_train_en","howto/usage/cluster/fabric_en","howto/usage/cluster/fluid_cluster_train_en","howto/usage/cluster/k8s_aws_en","howto/usage/cluster/k8s_en","howto/usage/cluster/openmpi_en","howto/usage/cluster/src/k8s_data/README","howto/usage/cluster/src/k8s_train/README","howto/usage/cmd_parameter/arguments_en","howto/usage/cmd_parameter/detail_introduction_en","howto/usage/cmd_parameter/index_en","howto/usage/cmd_parameter/use_case_en","index_en","mobile/cross_compiling_for_android_en","mobile/cross_compiling_for_ios_en","mobile/cross_compiling_for_raspberry_en","mobile/index_en","survey/cluster_bootstrapping_tools"],envversion:50,filenames:["api/index_en.rst","api/v2/config/activation.rst","api/v2/config/attr.rst","api/v2/config/evaluators.rst","api/v2/config/layer.rst","api/v2/config/networks.rst","api/v2/config/optimizer.rst","api/v2/config/pooling.rst","api/v2/data.rst","api/v2/data/data_reader.rst","api/v2/data/dataset.rst","api/v2/data/image.rst","api/v2/fluid.rst","api/v2/fluid/data_feeder.rst","api/v2/fluid/evaluator.rst","api/v2/fluid/executor.rst","api/v2/fluid/initializer.rst","api/v2/fluid/io.rst","api/v2/fluid/layers.rst","api/v2/fluid/nets.rst","api/v2/fluid/optimizer.rst","api/v2/fluid/param_attr.rst","api/v2/fluid/profiler.rst","api/v2/fluid/regularizer.rst","api/v2/model_configs.rst","api/v2/run_logic.rst","design/api.md","design/auto_gradient_check.md","design/backward.md","design/block.md","design/build_system/README.md","design/cluster_train/README.md","design/cluster_train/checkpointing.md","design/cluster_train/data_dispatch.md","design/cluster_train/large_model_dist_train.md","design/cluster_train/master_server.md","design/cluster_train/pserver_client.md","design/cluster_train/remote_parameter_updater.md","design/cluster_train/save_model.md","design/cluster_train/submit-job.md","design/concurrent_programming.md","design/csp.md","design/dist_refactor/distributed_architecture.md","design/dist_refactor/multi_cpu.md","design/dist_refactor/parameter_server.md","design/error_clip.md","design/evaluator.md","design/executor.md","design/file_manager/README.md","design/file_manager/pfs/pfsclient.md","design/float16.md","design/fluid.md","design/fluid_compiler.md","design/functions_operators_layers.md","design/gan_api.md","design/graph.md","design/graph_survey.md","design/if_else_op.md","design/infer_var_type.md","design/kernel_hint_design.md","design/memory_optimization.md","design/mkl/mkl_packed.md","design/mkl/mkldnn.md","design/mkl/mkldnn_fluid.md","design/model_format.md","design/multi_language_interface/00.why_plain_c.md","design/multi_language_interface/01.inference_implementation.md","design/operator_kernel_type.md","design/ops/rnn.md","design/ops/sequence_decoder.md","design/optimizer.md","design/paddle_nccl.md","design/parameter_average.md","design/parameters_in_cpp.md","design/profiler.md","design/program.md","design/prune.md","design/python_api.md","design/reader/README.md","design/refactorization.md","design/register_grad_op.md","design/regularization.md","design/releasing_process.md","design/scope.md","design/selected_rows.md","design/simple_op_design.md","design/speech/deep_speech_2.md","design/support_new_device.md","design/switch_kernel.md","design/tensor_array.md","design/var_desc.md","getstarted/build_and_install/build_from_source_en.rst","getstarted/build_and_install/docker_install_en.rst","getstarted/build_and_install/index_en.rst","getstarted/build_and_install/pip_install_en.rst","getstarted/index_en.rst","howto/deep_model/rnn/index_en.rst","howto/deep_model/rnn/rnn_config_en.rst","howto/dev/build_en.md","howto/dev/contribute_to_paddle_en.md","howto/dev/new_layer_en.rst","howto/dev/new_op_en.md","howto/dev/new_op_kernel_en.md","howto/dev/use_eigen_en.md","howto/dev/write_docs_en.rst","howto/index_en.rst","howto/optimization/cpu_profiling.md","howto/optimization/gpu_profiling_en.rst","howto/read_source.md","howto/usage/cluster/cluster_train_en.md","howto/usage/cluster/fabric_en.md","howto/usage/cluster/fluid_cluster_train_en.md","howto/usage/cluster/k8s_aws_en.md","howto/usage/cluster/k8s_en.md","howto/usage/cluster/openmpi_en.md","howto/usage/cluster/src/k8s_data/README.md","howto/usage/cluster/src/k8s_train/README.md","howto/usage/cmd_parameter/arguments_en.md","howto/usage/cmd_parameter/detail_introduction_en.md","howto/usage/cmd_parameter/index_en.rst","howto/usage/cmd_parameter/use_case_en.md","index_en.rst","mobile/cross_compiling_for_android_en.md","mobile/cross_compiling_for_ios_en.md","mobile/cross_compiling_for_raspberry_en.md","mobile/index_en.rst","survey/cluster_bootstrapping_tools.md"],objects:{"paddle.v2":{image:[11,0,0,"-"]},"paddle.v2.image":{batch_images_from_tar:[11,1,1,""],center_crop:[11,1,1,""],left_right_flip:[11,1,1,""],load_and_transform:[11,1,1,""],load_image:[11,1,1,""],load_image_bytes:[11,1,1,""],random_crop:[11,1,1,""],resize_short:[11,1,1,""],simple_transform:[11,1,1,""],to_chw:[11,1,1,""]}},objnames:{"0":["py","module","Python module"],"1":["py","function","Python function"]},objtypes:{"0":"py:module","1":"py:function"},terms:{"00m":107,"03m":107,"0424m":107,"0473v3":5,"055ee37d":112,"0630u":107,"06u":107,"0810u":107,"0957m":107,"0_cudnn5":91,"0_cudnn5_avx_mkl":[92,94],"0_cudnn7_avx_mkl":94,"0ab":4,"0rc":109,"0rc1":82,"0rc2":82,"0x10f256d50":56,"0x7ffe4de00110":56,"100gb":107,"100gi":112,"10g":39,"10m":107,"1150u":107,"11\u5b9e\u73b0\u4e86c":66,"11e6":113,"124n":107,"12gb":60,"13m":113,"1490u":107,"1550u":107,"16u":107,"173n":107,"1770u":107,"18ad":112,"18e457ce3d362ff5f3febf8e7f85ffec852f70f3b629add10aed84f930a68750":113,"197u":107,"1gb":107,"1st":18,"210u":107,"211839e770f7b538e2d8":5,"215n":107,"228u":107,"2520u":107,"2680u":107,"279n":107,"27m":107,"285m":107,"2863m":107,"28m":107,"2977m":107,"2cbf7385":112,"2nd":18,"302n":107,"30u":107,"328n":107,"32u":107,"32x32":10,"331n":107,"3320u":107,"365e":112,"36u":107,"3710m":107,"3768m":107,"387u":107,"38u":107,"3920u":107,"39u":107,"3rd":123,"4035m":107,"4090u":107,"4096mb":118,"4279m":107,"43u":107,"448a5b355b84":113,"4560u":107,"4563m":107,"45u":107,"4650u":107,"4726m":107,"473m":113,"4gb":118,"50bd":112,"50gi":112,"514u":107,"525n":107,"526u":107,"536u":107,"5460u":107,"5470u":107,"54u":107,"5690m":107,"573u":107,"578n":107,"5798m":107,"586u":107,"58s":113,"5969m":107,"5_cudnn5_avx_mkl":94,"5_cudnn5_avx_openbla":[94,95],"6080u":107,"6140u":107,"6305m":107,"639u":107,"64m":64,"655u":107,"6780u":107,"6810u":107,"682u":107,"6970u":107,"6ce9":112,"704u":107,"7090u":107,"72u":107,"73u":107,"75u":107,"760u":107,"767u":107,"783n":107,"784u":107,"78m":107,"7eamaa":10,"7kb":113,"8250u":107,"8300u":107,"830n":107,"849m":107,"861u":107,"8661m":107,"892m":107,"901n":107,"90u":107,"918u":107,"9247m":107,"924n":107,"9261m":107,"9330m":107,"94u":107,"9530m":107,"983m":107,"988u":107,"997u":107,"99u":107,"9a235":123,"9f18":113,"\u4e00\u4e2a\u5178\u578b\u7684chunk\u5982\u4e0b\u6240\u793a":48,"\u4e00\u4e2a\u795e\u7ecf\u7f51\u7edc\u7684\u6a21\u578b\u7531\u5927\u91cf\u7684\u53c2\u6570\u7ec4\u6210":32,"\u4e00\u4e2achunk\u7531\u6240\u5728\u7684\u6587\u4ef6\u504f\u79fb":48,"\u4e00\u4e2aposix\u517c\u5bb9\u7684\u6587\u4ef6\u7cfb\u7edf":48,"\u4e00\u822c\u4e0d\u5141\u8bb8\u518d\u4ece":82,"\u4e00\u822c\u7531mkl":62,"\u4e0a\u4f20\u5230cloud\u6216\u8005\u4e0b\u8f7d\u5230\u672c\u5730\u7684\u65f6\u95f4\u53ef\u80fd\u6bd4\u8f83\u957f":48,"\u4e0a\u4f20\u65b9\u6cd5":82,"\u4e0a\u6ce8\u518c\u4e00\u4e0b":48,"\u4e0a\u8ff0paddlepaddl":82,"\u4e0b\u4e00\u4e2awheel\u5305\u9700\u8981\u66f4\u65b0\u7248\u672c\u53f7\u624d\u53ef\u4ee5\u4e0a\u4f20":82,"\u4e0b\u5b58\u653e\u516c\u5171\u6570\u636e\u96c6\u5408":33,"\u4e0b\u62c9\u6846\u4e2d\u627e\u5230\u751f\u6210\u76843\u4e2a\u4e8c\u8fdb\u5236\u6587\u4ef6":82,"\u4e0b\u8f7d":48,"\u4e0b\u8f7d\u5230\u672c\u5730":48,"\u4e0b\u8f7d\u5f97\u5230":82,"\u4e0b\u9762\u5206\u522b\u4ecb\u7ecd\u67d0\u4e00\u7c7b\u6587\u4ef6\u7684\u5b9e\u73b0\u65b9\u5f0f":66,"\u4e0d\u4e00\u81f4\u7684\u7531pfsclient\u4e0b\u8f7d\u6216\u8005\u4f20\u8f93chunk\u5b8c\u6210":48,"\u4e0d\u4f7f\u7528\u9759\u6001\u5e93":65,"\u4e0d\u4f7f\u7528c":65,"\u4e0d\u4f7f\u7528swig":65,"\u4e0d\u53ef\u4ee5\u66f4\u6539":82,"\u4e0d\u540c":62,"\u4e0d\u540c\u7248\u672c\u7684\u7f16\u8bd1\u5668\u4e4b\u95f4":65,"\u4e0d\u540c\u8bed\u8a00\u7684\u63a5\u53e3\u9002\u5e94\u4e0d\u540c\u8bed\u8a00\u7684\u7279\u6027":65,"\u4e0d\u5728":66,"\u4e0d\u5bb9\u6613\u51fa\u9519":48,"\u4e0d\u5d4c\u5165\u5176\u4ed6\u8bed\u8a00\u89e3\u91ca\u5668":65,"\u4e0d\u5d4c\u5165python\u89e3\u91ca\u5668":65,"\u4e0d\u663e\u793a\u7684\u5199\u6bcf\u4e2a\u7c7b\u5177\u4f53\u5305\u542b\u4ec0\u4e48":65,"\u4e0d\u7528mount\u7684\u65b9\u5f0f\u6765\u8bbf\u95ee\u6570\u636e":33,"\u4e0e":62,"\u4e0e\u4e4b\u76f8\u5bf9\u7684\u662flocal":48,"\u4e0e\u5176\u4ed6\u7b2c\u4e09\u65b9\u5e93\u4e00\u6837":62,"\u4e0e\u529f\u80fd\u5206\u652f\u4e0d\u540c\u7684\u662f":82,"\u4e0e\u53ef\u80fd\u6709\u7684":82,"\u4e0ebatch":61,"\u4e14\u589e\u52a0\u4e00\u4e2a\u7b2c\u4e09\u65b9\u8bed\u8a00":65,"\u4e14\u8c03\u7528\u65f6\u4e0d\u80fd\u629b\u51fa\u5f02\u5e38\u6216\u51fa\u73b0\u8fd0\u884c\u65f6\u9519\u8bef":66,"\u4e14c99\u652f\u6301bool\u7c7b\u578b\u548c\u5b9a\u957f\u6574\u6570":65,"\u4e14c99\u76f8\u5bf9\u4e8ec11\u4f7f\u7528\u66f4\u52a0\u5e7f\u6cdb":65,"\u4e25\u683c\u7684\u547d\u540d\u89c4\u8303pep":82,"\u4e2a\u6027\u5316\u63a8\u8350":82,"\u4e2d":[61,62,65,66],"\u4e2d\u4f1a\u63d0\u4f9b\u4e00\u4e9b\u5fc5\u8981\u7684\u63a5\u53e3\u548c\u51fd\u6570":62,"\u4e2d\u5199\u5165json\u5185\u5bb9":32,"\u4e2d\u5b8c\u5168\u4e00\u81f4":65,"\u4e2d\u5b9e\u73b0\u4e86\u4e00\u4e2amerge\u7684\u65b9\u6cd5":62,"\u4e2d\u5b9e\u73b0\u7684\u7ed3\u6784\u4f53":66,"\u4e2d\u5bf9\u5e94\u7684layer\u5904":61,"\u4e2d\u5f15\u5165\u7684":61,"\u4e2d\u63d0\u4f9b\u4e00\u4e2a\u4e0emkl\u6709\u5173\u7684\u603b\u5f00\u5173":62,"\u4e2d\u6839\u636e":61,"\u4e2d\u6dfb\u52a0":61,"\u4e2d\u6dfb\u52a0\u4e00\u4e2a":62,"\u4e2d\u7684\u7248\u672c\u4fe1\u606f":82,"\u4e2d\u8fd0\u884c\u4efb\u52a1\u7684\u89d2\u5ea6":33,"\u4e3a":[61,62],"\u4e3a\u4e86\u5c3d\u53ef\u80fd\u5c11\u7684\u5728\u7236\u7c7blayer\u4e2d\u6dfb\u52a0\u53d8\u91cf\u6216\u8005\u51fd\u6570":62,"\u4e3a\u4e86\u5e94\u5bf9\u4ee5\u4e0a\u7684\u95ee\u9898":48,"\u4e3a\u4e86\u66b4\u9732\u7684\u63a5\u53e3\u5c3d\u91cf\u7b80\u5355":66,"\u4e3a\u4e86\u66f4\u597d\u7684\u7b26\u5408paddlepaddle\u7684\u4ee3\u7801\u98ce\u683c":62,"\u4e3a\u4e86\u6700\u5927\u7a0b\u5ea6\u51cf\u5c11\u591a\u6b21\u8c03\u7528":61,"\u4e3a\u4e86\u8fdb\u4e00\u6b65\u63d0\u5347paddlepaddle\u5728\u57fa\u672c\u6570\u5b66\u8fd0\u7b97\u7684\u8ba1\u7b97\u901f\u5ea6":62,"\u4e3b\u8981\u529f\u80fd\u5305\u62ec":48,"\u4e3b\u8981\u5305\u62ec":62,"\u4e3b\u8981\u5305\u62ec\u4e86\u6df1\u5ea6\u5b66\u4e60\u76f8\u5173\u7684\u6570\u5b66\u539f\u8bed\u4e0e\u64cd\u4f5c":62,"\u4e3b\u8981\u9488\u5bf9paddlepaddle\u5728\u91cd\u6784\u4e4b\u524d\u7684\u4ee3\u7801\u6846\u67b6\u4ee5\u53cav1\u7684api":62,"\u4e4b\u5916\u7684\u6240\u6709\u5934\u6587\u4ef6":66,"\u4e5f\u4e0d\u4f7f\u7528\u5176\u4ed6\u52a8\u6001\u5e93":65,"\u4e5f\u4e0d\u5e94\u8be5\u62a5\u9519":66,"\u4e5f\u4e0d\u751f\u6210":66,"\u4e5f\u53ef\u4ee5\u4f7f\u7528\u8fd9\u4e9b\u955c\u50cf":82,"\u4e5f\u5c31\u662f\u8bf4\u8f93\u51fa\u7684\u7ed3\u679c\u4e0d\u4f1a\u5728\u539f\u6765\u7684\u6570\u636e\u4e0a\u7d2f\u52a0":62,"\u4e66\u5199":65,"\u4eba\u8138\u8bc6\u522b":33,"\u4ec5\u4ec5\u4f7f\u7528":65,"\u4ec5\u4f1a\u5728\u652f\u6301avx2\u6307\u4ee4\u96c6\u53ca\u4ee5\u4e0a\u7684\u673a\u5668\u624d\u4f7f\u7528mkl":62,"\u4ece":82,"\u4ece\u78c1\u76d8\u6587\u4ef6\u4e2d\u52a0\u8f7duuid\u6587\u4ef6\u540d\u7684\u68c0\u67e5\u70b9\u5feb\u7167\u6587\u4ef6":32,"\u4ece\u800c\u907f\u514d\u4e86packing\u5197\u4f59":61,"\u4eceetcd\u4e2d\u8bfb\u53d6\u8282\u70b9":32,"\u4ed6\u4e3b\u8981\u5305\u542b\u4e86\u5b9e\u9645\u66b4\u9732\u7684\u7c7b\u578b\u7ed3\u6784":66,"\u4ed6\u662f\u5c06":66,"\u4ed6\u7684\u76ee\u6807\u662f\u4f7f\u7528c":65,"\u4ee3\u7801\u751f\u6210\u7684\u7b26\u53f7\u53ef\u80fd\u4e0d\u4e00\u81f4":65,"\u4ee3\u8868\u8fd9\u4e2alayer\u662f\u7528\u4e8e\u8dd1\u5728mkl":62,"\u4ee3\u8868\u8fd9\u4e2ashard\u7684\u6700\u5927index":33,"\u4ee3\u8868shard\u7684index":33,"\u4ee5\u4e0a\u4ee3\u7801\u7684reader\u8f93\u51fa\u7684data":33,"\u4ee5\u4e0a\u547d\u4ee4\u4f1a\u5728\u5f53\u524d\u76ee\u5f55\u4e0b\u751f\u6210100\u4e2a\u6587\u4ef6":33,"\u4ee5\u4e0b":33,"\u4ee5\u4e0b\u7b80\u79f0rnn":61,"\u4ee5\u4fbf\u6211\u4eec\u53ef\u4ee5\u628a\u66f4\u591a\u7684\u7cbe\u529b\u653e\u5230\u903b\u8f91\u672c\u8eab\u4e0a":48,"\u4ee5\u53ca":61,"\u4ee5\u53canumpi":33,"\u4ee5\u6b64\u8fbe\u5230\u6700\u597d\u7684\u6027\u80fd":62,"\u4ee5\u793a\u533a\u5206":[61,62],"\u4efb\u610f\u65f6\u523b\u53ea\u53ef\u80fd\u540c\u65f6\u6709\u4e00\u53f0\u670d\u52a1\u5668\u6545\u969c":32,"\u4f18\u5316\u524d":61,"\u4f18\u5316\u540e":61,"\u4f1a\u4ee5":[61,62],"\u4f1a\u4f7f\u7528\u76f8\u540c\u7684\u539f\u6570\u636e":61,"\u4f1a\u5148\u4e34\u65f6\u4fdd\u5b58\u5728":62,"\u4f1a\u5728":62,"\u4f1a\u5728\u7f16\u8bd1paddlepaddle\u7684\u65f6\u5019\u4e0b\u8f7d\u5e76\u7f16\u8bd1mkl":62,"\u4f1a\u5bfc\u81f4\u4e0d\u540c\u7248\u672cpython\u5728\u4e00\u4e2a\u8fdb\u7a0b\u91cc\u7684bug":65,"\u4f1a\u5f15\u5165":62,"\u4f1a\u628acpu\u7684buffer\u5bf9\u9f50\u4e3a4096":62,"\u4f1a\u6dfb\u52a0\u76f8\u5e94\u7684\u811a\u672c\u5728":62,"\u4f1a\u6dfb\u52a0\u76f8\u5e94\u7684\u811a\u672c\u7528\u4e8e\u6d4b\u8bd5\u548c\u5bf9\u6bd4\u5728\u4f7f\u7528mkl":61,"\u4f1a\u76f4\u63a5\u62a5\u9519\u9000\u51fa":65,"\u4f1a\u81ea\u52a8\u4f7f\u7528mklml\u5e93\u4f5c\u4e3apaddlepaddle\u7684cblas\u548clapack\u5e93":62,"\u4f1a\u81ea\u52a8\u6839\u636e\u786c\u4ef6\u914d\u7f6e":62,"\u4f1a\u88abpickle\u5e8f\u5217\u5316\u6210\u5b57\u7b26\u4e32":33,"\u4f20\u5165":33,"\u4f46":66,"\u4f46\u4e0d\u66b4\u9732":66,"\u4f46\u5e76\u6ca1\u6709\u7ecf\u8fc7\u56de\u5f52\u6d4b\u8bd5":82,"\u4f46\u6240\u6709fork\u7684\u7248\u672c\u5e93\u7684\u6240\u6709\u5206\u652f\u90fd\u76f8\u5f53\u4e8e\u7279\u6027\u5206\u652f":82,"\u4f46\u662f\u53c8\u8fc7\u4e8e\u7410\u788e":66,"\u4f46\u662f\u5728mkl":62,"\u4f46\u662f\u5728paddlepaddle\u4e2d":62,"\u4f46\u662f\u6574\u4e2a\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u4e0d\u9700\u8981\u4efb\u4f55\u8f6c\u6362":62,"\u4f46\u662f\u6ce8\u610f\u7684\u662f":62,"\u4f46\u662f\u89e3\u91ca\u6027\u8bed\u8a00":65,"\u4f5c\u4e3a\u53e6\u4e00\u4e2a\u7b2c\u4e09\u65b9\u5e93\u96c6\u6210\u8fdbpaddlepaddl":62,"\u4f5c\u4e3a\u5b58\u50a8\u7cfb\u7edf":33,"\u4f5c\u4e3a\u7c7b\u53e5\u67c4":65,"\u4f7f\u7528":[62,66,82],"\u4f7f\u7528\u4e0b\u9762\u547d\u4ee4":33,"\u4f7f\u7528\u52a8\u6001\u5e93":65,"\u4f7f\u7528\u540c\u6837\u7684\u8bad\u7ec3\u6570\u636eblock":32,"\u4f7f\u7528\u667a\u80fd\u6307\u9488\u7684\u539f\u56e0\u662f":66,"\u4f7f\u7528\u7684\u53c2\u6570\u4e0epaddlepaddle\u7533\u8bf7\u7684buffer\u5171\u7528\u4e00\u5757\u5185\u5b58":62,"\u4f7f\u7528\u76f8\u5bf9\u8def\u5f84\u7684\u5f15\u7528\u65b9\u5f0f":66,"\u4f7f\u7528\u8fd9\u4e2a\u795e\u7ecf\u7f51\u7edc\u53ef\u4ee5\u5b8c\u6210\u5bf9\u65b0\u6570\u636e\u7684\u9884\u6d4b":32,"\u4f7f\u7528\u9759\u6001\u5e93\u548c\u52a8\u6001\u5e93\u96be\u5ea6\u5dee\u4e0d\u591a":65,"\u4f7f\u7528c":66,"\u4f7f\u7528c99\u505a\u63a5\u53e3":65,"\u4f7f\u7528c99\u800c\u4e0d\u4f7f\u7528c11\u7684\u539f\u56e0\u662f":65,"\u4f7f\u7528c99\u800c\u4e0d\u4f7f\u7528c89":65,"\u4f7f\u7528regress":82,"\u4f7f\u7528swig\u53ea\u652f\u6301cpython\u89e3\u91ca\u5668":65,"\u4f7f\u7528swig\u9700\u8981\u591a\u8bed\u8a00\u7ed1\u5b9a\u7684\u5f00\u53d1\u4eba\u5458\u719f\u7ec3\u638c\u63e1swig\u914d\u7f6e":65,"\u4f7f\u7528void":65,"\u4f8b\u5982":[33,65,66,82],"\u4f8b\u5982\u5728deepspeech2":61,"\u4f8b\u5982\u5bf9\u4e8ejava\u6216\u8005python":65,"\u4f8b\u5982\u5bf9\u4e8ejava\u6765\u8bf4":65,"\u4f8b\u5982\u5bf9\u4e8epython":65,"\u4f8b\u5982c":65,"\u4f8b\u5982java\u4e0epython\u7684\u9519\u8bef\u5904\u7406\u662f\u76f4\u63a5\u6254\u51fa\u6765except":65,"\u4f8b\u5982python\u53ef\u4ee5\u4f7f\u7528":65,"\u4f8b\u5982python\u7684":65,"\u4f8b\u5982rnn":61,"\u4f9d\u6b21\u7c7b\u63a8":82,"\u4fbf\u662f\u5c06\u9759\u6001\u5e93\u52a0\u5165jvm\u4e2d":65,"\u4fee\u590d\u6240\u6709bug\u540e":82,"\u4fee\u590ddocker\u7f16\u8bd1\u955c\u50cf\u95ee\u9898":82,"\u4fee\u6539":[62,82],"\u4fee\u6539\u6210":82,"\u505a\u53ea\u8bfb\u6302\u8f7d":33,"\u505a\u5982\u4e0b\u51e0\u4e2a\u64cd\u4f5c":82,"\u505a\u63a5\u53e3":65,"\u505c\u6b62\u4fdd\u5b58\u68c0\u67e5\u70b9\u7684\u7ebf\u7a0b":32,"\u5145\u5206\u53d1\u6325\u82f1\u7279\u5c14\u5e73\u53f0\u7684\u4f18\u52bf":61,"\u5145\u5206\u5c55\u73b0\u82f1\u7279\u5c14\u5e73\u53f0\u7684\u4f18\u52bf":62,"\u5148\u5b8c\u6210\u5bf9\u6743\u91cd\u7684packing\u64cd\u4f5c":61,"\u5148\u5b9e\u73b0\u6a21\u578b\u63a8\u65ad\u7684api":66,"\u5171\u4eab\u5185\u5b58":62,"\u5171\u4eab\u540c\u4e00\u4e2a\u6743\u91cd":61,"\u5176\u4e2d":[65,82],"\u5176\u4ed6\u51fd\u6570\u5747\u8fd4\u56de":66,"\u5176\u4ed6\u7528\u6237\u7684fork\u7248\u672c\u5e93\u5e76\u4e0d\u9700\u8981\u4e25\u683c\u9075\u5b88":82,"\u5176\u8f6c\u6362\u6b21\u6570\u51cf\u5c11\u81f3":61,"\u5177\u4f53\u4f7f\u7528\u65b9\u6cd5\u4e3a":66,"\u5177\u4f53\u539f\u56e0\u53c2\u8003":66,"\u5177\u4f53\u53ef\u4ee5\u53c2\u8003mkl":62,"\u5177\u4f53\u5b9e\u73b0\u65b9\u5f0f\u6bd4\u5982":[61,62],"\u5177\u4f53\u7684\u5b8c\u6210\u72b6\u6001\u53ef\u4ee5\u53c2\u89c1":62,"\u5177\u4f53\u8bf7\u53c2\u8003":66,"\u5185\u90e8\u5b58\u50a8":62,"\u5185\u90e8\u9a71\u52a8python\u89e3\u91ca\u5668\u8fdb\u884c\u6a21\u578b\u914d\u7f6e\u89e3\u6790\u548c\u6570\u636e\u8bfb\u53d6":65,"\u518d\u5728\u6bcf\u4e00\u4e2aapi\u4e2d\u81ea\u5df1\u68c0\u67e5\u7c7b\u578b":65,"\u518d\u57fa\u4e8e":82,"\u518d\u628a\u5df2\u8f6c\u6362\u4e3apacked\u683c\u5f0f\u7684\u6570\u636e\u4f20\u9012\u7ed9\u90a3\u4e9b\u590d\u7528\u540c\u4e00\u6570\u636e\u7684gemm":61,"\u5199\u4ee3\u7801":65,"\u5199\u5165\u5feb\u7167\u6570\u636e":32,"\u51fd\u6570":[61,62],"\u51fd\u6570\u5373\u53ef\u5b8c\u6210\u8f6c\u6362":33,"\u51fd\u6570\u540d\u4e3a":66,"\u51fd\u6570\u547d\u540d":65,"\u5206\u522b\u4ee3\u8868\u8f93\u5165\u6570\u636e":62,"\u5206\u522b\u5bf9\u5e94capi":82,"\u5206\u5e03\u5f0f\u5b58\u50a8\u670d\u52a1":32,"\u5206\u652f":82,"\u5206\u652f\u4e00\u65e6\u5efa\u7acb":82,"\u5206\u652f\u4e2d":82,"\u5206\u652f\u4e3a\u5f00\u53d1":82,"\u5206\u652f\u4e3a\u6bcf\u4e00\u6b21release\u65f6\u5efa\u7acb\u7684\u4e34\u65f6\u5206\u652f":82,"\u5206\u652f\u4e3a\u7a33\u5b9a":82,"\u5206\u652f\u529f\u80fd\u7684\u5c01\u95ed":82,"\u5206\u652f\u5408\u5165":82,"\u5206\u652f\u5408\u5165master\u5206\u652f":82,"\u5206\u652f\u540c\u6b65\u4e3b\u7248\u672c\u5e93\u7684":82,"\u5206\u652f\u540d\u4e3a":82,"\u5206\u652f\u5b58\u5728\u7684\u65f6\u5019":82,"\u5206\u652f\u6d3e\u751f\u51fa\u65b0\u7684\u5206\u652f":82,"\u5206\u652f\u7684\u7248\u672c\u90fd\u662f\u7ecf\u8fc7\u5355\u5143\u6d4b\u8bd5\u548c\u56de\u5f52\u6d4b\u8bd5\u7684\u7248\u672c":82,"\u5206\u652f\u7684\u7248\u672c\u90fd\u7ecf\u8fc7\u5355\u5143\u6d4b\u8bd5":82,"\u5206\u7247":32,"\u5219\u4f7f\u7528\u542f\u52a8\u53c2\u6570\u5b9a\u4e49\u7684\u521d\u59cb\u5316\u65b9\u6cd5\u521d\u59cb\u5316\u53c2\u6570":32,"\u5219\u5ffd\u7565":32,"\u5219\u628a\u53e6\u4e00\u4e2a\u6162\u901f\u7684kill\u6389":32,"\u5219\u76f4\u63a5\u5f15\u5165\u53e6\u4e00\u79cd\u7c7b\u578b\u7684\u5934\u6587\u4ef6":66,"\u5219\u9700\u8981\u56de\u6eda\u5230\u4e0a\u4e00\u4e2a\u68c0\u67e5\u70b9":32,"\u521b\u5efa":62,"\u5220\u9664\u78c1\u76d8\u76ee\u5f55\u4e2d\u4e0d\u662f\u5f53\u524duuid\u7684\u5feb\u7167\u6587\u4ef6":32,"\u5230":32,"\u5230\u7b2c\u4e8c\u6b65":82,"\u524d\u540e\u7684\u7f51\u7edc\u6027\u80fd":61,"\u529f\u80fd":48,"\u529f\u80fd\u7684\u6b63\u786e\u6027\u5305\u62ec\u9a8c\u8bc1paddlepaddle\u76ee\u524d\u7684":82,"\u52a8\u6001\u5e93":65,"\u5305\u542b\u4e86\u67d0\u79cd\u7c7b\u578b\u7684\u7c7b\u578b\u5b9a\u4e49\u548c\u66b4\u9732\u7684\u5168\u90e8\u51fd\u6570":66,"\u5305\u62ec":[33,61,62],"\u5305\u62ec\u6743\u91cdw\u548c\u504f\u7f6eb":32,"\u5305\u62ecmkl":62,"\u534f\u540c\u5b8c\u6210releas":82,"\u5355\u4e2a\u503c":33,"\u5355\u70b9\u6545\u969c":32,"\u5373":66,"\u5373\u4f7f\u7528":66,"\u5373\u4f7f\u7528\u6237\u76f4\u63a5\u5f15\u7528\u67d0\u79cd\u7c7b\u578b\u7684\u5934\u6587\u4ef6":66,"\u5373\u4f7fc":66,"\u5373\u4f8b\u5982":66,"\u5373\u4fbfpaddl":66,"\u5373\u5b8c\u6210\u67d0\u4e00\u4e2a\u4efb\u52a1\u7684\u6700\u5c11\u51fd\u6570":66,"\u5373\u66b4\u9732":66,"\u5373\u8868\u793a\u4e0d\u9700\u8981\u8f6c\u6362":62,"\u5373\u8fd9\u4e2a\u52a8\u6001\u5e93\u662f\u4e0d\u4f9d\u8d56\u4e8e\u5176\u4ed6\u4efb\u4f55\u6587\u4ef6\u7684":65,"\u539f\u6765\u7684\u65b9\u6848":62,"\u53c2\u6570":65,"\u53c2\u8003":[48,65],"\u53c2\u8003\u4e0b\u56fe":82,"\u53c8\u53ef\u4ee5\u907f\u514d\u4e0d\u5fc5\u8981\u7684\u8f6c\u6362":62,"\u53cc\u5411\u9a8c\u8bc1":48,"\u53d1\u578b\u7248":82,"\u53d1\u5e03\u5230dockerhub":82,"\u53d1\u5e03docker\u955c\u50cf\u53ea\u9700\u8981\u5bf9\u81ea\u52a8push\u7684\u955c\u50cf\u6253\u4e0a":82,"\u53d8\u91cf\u6765\u533a\u5206layer\u7684\u5c5e\u6027":62,"\u53ea\u5bf9\u7279\u6b8a\u5728\u7ebf\u7cfb\u7edf\u8003\u8651\u4e24\u53f0\u4ee5\u4e0a\u540c\u65f6\u6545\u969c\u7684\u5bb9\u707e":32,"\u53ea\u66b4\u9732\u6982\u5ff5\u7684\u63a5\u53e3":66,"\u53ea\u80fd\u8c03\u7528paddle\u7684\u52a8\u6001\u5e93":65,"\u53ea\u9700\u8981\u6062\u590d\u8fd9\u53f0\u8282\u70b9":32,"\u53ef\u4ee5":82,"\u53ef\u4ee5\u51cf\u5c0f\u7cfb\u7edf\u590d\u6742\u6027":32,"\u53ef\u4ee5\u5728\u4efb\u4f55\u673a\u5668\u4e0a\u6267\u884c\u7684":65,"\u53ef\u4ee5\u5728\u6b64\u9875\u9762\u7684":82,"\u53ef\u4ee5\u628a\u672c\u5730\u7684\u6570\u636e\u4e0a\u4f20\u5230\u5b58\u50a8\u96c6\u7fa4\u4e2d":33,"\u53ef\u4ee5\u6709\u6548\u7684\u907f\u514dparamet":32,"\u53ef\u4ee5\u7528":48,"\u53ef\u4ee5\u7528\u4ee5\u4e0b\u6307\u4ee4":33,"\u53ef\u4ee5\u7ee7\u7eed\u5728\u81ea\u5df1\u7684\u529f\u80fd\u5206\u652f\u63d0\u4ea4\u4ee3\u7801":82,"\u53ef\u4ee5\u901a\u8fc7\u9636\u6bb5\u6027\u7684\u4fdd\u5b58\u6bcf\u4e2aparamet":32,"\u53ef\u80fd\u4f1a\u9020\u6210\u7f51\u7edc\u62e5\u585e":32,"\u53f3\u4fa7\u7684":82,"\u5404\u6b21\u524d\u5411\u4e4b\u95f4\u4e5f\u90fd\u4f7f\u7528\u4e86\u76f8\u540c\u7684\u6743\u91cd":61,"\u540c\u4e00\u6b21\u524d\u5411":61,"\u540c\u65f6":[61,62],"\u540c\u65f6\u4f1a\u5f00\u542fintel":62,"\u540c\u65f6\u518d\u5c06":82,"\u540c\u65f6\u53c8\u5c3d\u53ef\u80fd\u5c11\u7684\u727a\u7272mkl":62,"\u540c\u65f6\u63d0\u8d77":82,"\u540c\u65f6\u6570\u636e\u683c\u5f0f\u5c31\u662f":62,"\u540d\u5b57\u4fee\u9970":65,"\u540e\u5411":61,"\u540e\u5411\u65f6\u590d\u7528\u5df2\u7ecf\u8f6c\u6362\u8fc7\u7684\u6743\u91cd":61,"\u5411\u6307\u5b9a\u7684\u76ee\u5f55\u4e2d\u4e00\u4e2a\u65b0\u7684\u6587\u4ef6":32,"\u5411paddlepaddle\u7684\u4e3b\u7248\u672c\u5e93\u63d0\u4ea4":82,"\u5426\u5219\u5f97\u628apaddle\u9759\u6001\u5e93\u94fe\u63a5\u5230\u89e3\u91ca\u5668\u91cc":65,"\u542f\u52a8\u4e00\u4e2a\u65b0\u7684\u7ebf\u7a0b\u5f00\u59cb\u4fdd\u5b58\u68c0\u67e5\u70b9":32,"\u548c":[33,61,62,65,66,82],"\u548c\u672a\u6765\u53ef\u80fd\u8fd8\u4f1a\u7528\u5230":62,"\u548c\u79bb\u7ebf\u6570\u636e\u7684\u65b9\u5f0f":33,"\u54ea\u4e2atrainer\u5148\u5b8c\u6210block\u7684\u8bad\u7ec3":32,"\u56e0\u4e3a\u8fd9\u6837\u505a\u4e5f\u6ca1\u6cd5\u4fdd\u8bc1\u6d88\u9664\u968f\u673a\u6027":32,"\u56e0\u4e3aswig\u5728\u7b2c\u4e09\u65b9\u8bed\u8a00\u4e2d\u66b4\u9732\u7684\u51fd\u6570\u540d":65,"\u56e0\u6b64":61,"\u56fe\u50cf\u5206\u7c7b":82,"\u5728":[61,62,66,82],"\u5728\u4e00\u4e2a\u4e0d\u53ef\u4e2d\u65ad\u5e76\u7f3a\u5c11\u5907\u4efd\u7684\u8bad\u7ec3\u4efb\u52a1\u4e2d":32,"\u5728\u4e0a\u56fe\u4e2d\u663e\u793a\u4e86\u5728\u4e00\u4e2a\u5b9e\u9645\u751f\u4ea7\u73af\u5883\u4e2d\u7684\u5e94\u7528":33,"\u5728\u4f7f\u7528twine\u4e0a\u4f20\u4e4b\u524d":82,"\u5728\u51fa\u73b0\u5355\u70b9\u6545\u969c\u65f6":32,"\u5728\u5b9e\u73b0\u6bcf\u4e2a\u5b50\u7c7b\u7684\u65f6\u5019\u5c31\u4e0d\u9700\u8981\u5173\u5fc3\u5206\u652f\u7684\u4e8b\u60c5\u4e86":62,"\u5728\u5b9e\u73b0\u8fc7\u7a0b\u4e2d":66,"\u5728\u5bf9\u5e94\u7684":61,"\u5728\u5c42\u521d\u59cb\u5316\u7684\u65f6\u5019":61,"\u5728\u5f00\u59cb\u8bad\u7ec3\u4e4b\u524d":33,"\u5728\u5f02\u6784\u96c6\u7fa4\u4e2d":32,"\u5728\u5f15\u5165\u5176\u4ed6\u7c7b\u578b\u7684\u5934\u6587\u4ef6\u65f6":66,"\u5728\u5feb\u7167\u5199\u5165\u5b8c\u6210\u540e":32,"\u5728\u60a8\u7684\u5b9e\u9645\u73af\u5883\u4e2d":32,"\u5728\u6709\u666e\u901a\u7684cpu":62,"\u5728\u672c\u6587\u6863\u4e2d":48,"\u5728\u673a\u7fa4\u4e0a\u8fd0\u884c\u8f6c\u6362\u7a0b\u5e8f":33,"\u5728\u6837\u4f8b\u4e2d":66,"\u5728\u7528\u6237\u4f7f\u7528c":66,"\u5728\u7b2c\u4e8c\u4e2atab":82,"\u5728\u7ebf\u6a21\u578b\u9884\u6d4b\u670d\u52a1":33,"\u5728\u8bad\u7ec3\u7ed3\u675f\u7684\u65f6\u5019\u518d\u4fdd\u5b58\u4e3apaddlepaddle\u7684\u683c\u5f0f":62,"\u5728\u8bc4\u5ba1\u8fc7\u7a0b\u4e2d":82,"\u5728\u8fd9\u4e2a":82,"\u5728\u8fd9\u4e2a\u52a8\u6001\u5e93\u4e2d\u4e0d\u5d4c\u5165\u4efb\u4f55\u5176\u4ed6\u8bed\u8a00\u7684\u89e3\u91ca\u5668":65,"\u5728\u8fd9\u4e2a\u9636\u6bb5\u7684\u4ee3\u7801\u6b63\u5728\u7ecf\u5386\u56de\u5f52\u6d4b\u8bd5":82,"\u5728\u8fd9\u4e9b\u5934\u6587\u4ef6\u4e2d":66,"\u5728\u8fd9\u4e9b\u6587\u4ef6\u4e2d":66,"\u5728\u91cd\u6784\u524d\u7684paddlepaddle\u4e2d":62,"\u5728\u95ee\u9898\u672c\u8eab\u7684\u8ba1\u7b97\u91cf\u6bd4\u8f83\u5c0f\u7684\u65f6\u5019":61,"\u5728batch":61,"\u5728c":65,"\u5728c\u7684\u5934\u6587\u4ef6":65,"\u5728packing\u4e0a\u7684\u8017\u65f6":61,"\u5728paddle\u4e4b\u4e0a\u8fd0\u884c\u7684\u6df1\u5ea6\u5b66\u4e60\u8bad\u7ec3\u8f93\u51fa\u7684\u6a21\u578b\u4f1a\u63d0\u4f9b\u7ed9\u5728\u7ebf\u4eba\u8138\u8bc6\u522b\u7684\u5e94\u7528\u4f7f\u7528":33,"\u5728paramet":32,"\u5728rnn\u7684\u60c5\u51b5\u4e0b":61,"\u5747\u4f1a\u88ab\u5b89\u88c5\u5230includ":66,"\u5747\u662f\u5728":66,"\u57fa\u4e8e\u7c98\u6027\u4f1a\u8bdd\u7684\u8d1f\u8f7d\u5747\u8861\u529f\u80fd":48,"\u5916\u90e8\u5b58\u50a8":62,"\u591a\u4e2a\u503c":33,"\u591a\u4e2aparamet":32,"\u591a\u6b21\u8c03\u7528":61,"\u5927\u591a\u6570\u8bed\u8a00\u90fd\u652f\u6301\u4f7f\u7528c\u8bed\u8a00api":65,"\u5982\u56fe\u4e2dtrainer":32,"\u5982\u679c\u4e0a\u9762\u4e24\u6b65\u51fa\u73b0\u9519\u8bef":32,"\u5982\u679c\u4e0d\u9700\u8981\u5916\u90e8\u5b58\u50a8\u7528\u4e8e\u8f6c\u6362":62,"\u5982\u679c\u4f7f\u7528swig\u6211\u4eec\u9700\u8981\u5c06\u5728interface\u6587\u4ef6\u91cc":65,"\u5982\u679c\u5728\u4f7f\u7528mkl":62,"\u5982\u679c\u5931\u8d25":82,"\u5982\u679c\u5b58\u5728\u6570\u636e\u6392\u5217\u683c\u5f0f\u4e0d\u4e00\u6837\u7684\u60c5\u51b5\u65f6":62,"\u5982\u679c\u5b58\u5728\u67d0\u4e9btrainer\u6267\u884c\u901f\u5ea6\u8fc7\u6162\u4f1a\u5f71\u54cd\u6574\u4f53\u96c6\u7fa4\u7684\u901f\u5ea6":32,"\u5982\u679c\u5df2\u7ecf\u6b63\u5728\u6267\u884c\u4fdd\u5b58\u68c0\u67e5\u70b9\u7684\u7ebf\u7a0b":32,"\u5982\u679c\u662f\u5176\u5b83\u7c7b\u578b":33,"\u5982\u679c\u6709bugfix\u7684\u884c\u4e3a":82,"\u5982\u679c\u67d0\u4e00\u4e2a\u7c7b\u578b\u9700\u8981\u5f15\u7528\u53e6\u4e00\u4e2a\u7c7b\u578b":66,"\u5982\u679c\u67d0\u4e00\u4e2apaddl":66,"\u5982\u679c\u67d0\u4e00\u4e2apaddle\u6982\u5ff5\u5fc5\u987b\u8981\u66b4\u9732":66,"\u5982\u679c\u6ee1\u8db3\u6761\u4ef6":32,"\u5982\u679c\u7528\u6237\u8981\u628apaddle\u7684\u9759\u6001\u5e93":65,"\u5982\u679c\u8981\u4e0a\u4f20gpu\u7248\u672c\u7684\u5305":82,"\u5982\u679c\u8c03\u7528\u9759\u6001\u5e93\u53ea\u80fd\u5c06\u9759\u6001\u5e93\u4e0e\u89e3\u91ca\u5668\u94fe\u63a5":65,"\u5982\u679c\u9700\u8981\u624b\u52a8\u7f16\u8bd1":82,"\u5982\u679cmkl":62,"\u5982\u679cparamet":32,"\u5b50\u7c7b\u53ea\u9700\u8981\u4f7f\u7528\u5b9a\u4e49\u597d\u7684\u63a5\u53e3":62,"\u5b57\u6bb5\u8bbe\u4e3a":82,"\u5b57\u7b26\u4e32":33,"\u5b58\u50a8":33,"\u5b66\u4e60\u6210\u672c\u9ad8":65,"\u5b83\u4eec\u4e3b\u8981\u662f\u7528\u4e8e":62,"\u5b83\u4eec\u7684\u6587\u4ef6\u540d\u662f":33,"\u5b83\u53ea\u4f1a\u5305\u62ec\u751f\u6210\u597d\u7684\u52a8\u6001\u5e93\u548c\u5934\u6587\u4ef6":62,"\u5b83\u8d1f\u8d23\u51b3\u5b9a\u7f16\u8bd1\u65f6\u662f\u5426\u4f7f\u7528mklml\u548cmkl":62,"\u5b89\u88c5\u540e\u7684\u76ee\u5f55\u7ed3\u6784\u4e3a":66,"\u5b8c\u6210\u4e00\u4e2a\u4f20\u8f93\u52a8\u4f5c\u5b8c\u6210\u7684\u65f6\u95f4\u4e5f\u6bd4\u8f83\u77ed":48,"\u5b8c\u6210\u5e38\u7528layer\u7684mkl":62,"\u5b8c\u6210\u5e38\u89c1\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edcvgg":62,"\u5b8c\u6210\u6570\u636e\u7684\u9884\u5904\u7406":33,"\u5b8c\u6210\u81ea\u52a8\u5316\u4e8c\u8fdb\u5236\u7f16\u8bd1":82,"\u5b9a\u4e49":62,"\u5b9a\u4e49\u4e00\u4e9b\u9664\u4e86layer\u548cmemory\u76f8\u5173\u7684\u7c7b\u548c\u51fd\u6570":62,"\u5b9e\u73b0\u5177\u4f53\u7684\u51fd\u6570\u529f\u80fd\u5373\u53ef":62,"\u5b9e\u73b0\u7b80\u5355":65,"\u5bf9\u4e8e\u4e0d\u540c\u8bed\u8a00":65,"\u5bf9\u4e8e\u540c\u4e00\u6bb5c":65,"\u5bf9\u4e8e\u540c\u6837\u8bbe\u7f6e\u7684\u7f51\u7edc\u6a21\u578b":61,"\u5bf9\u4e8e\u591a\u8bed\u8a00\u63a5\u53e3":65,"\u5bf9\u4e8e\u5927\u591a\u6570\u8bed\u8a00":65,"\u5bf9\u4e8e\u5e8f\u5217\u957f\u5ea6":61,"\u5bf9\u4e8e\u6709\u53c2\u6570\u7684\u5c42":62,"\u5bf9\u4e8e\u6bcf\u4e00\u4e2a\u65b0\u52a0\u7684rnn":61,"\u5bf9\u4e8e\u6bcf\u79cd\u7c7b\u578b":66,"\u5bf9\u4e8e\u6bcf\u79cdc":66,"\u5bf9\u65b0\u7684\u6743\u91cd\u8fdb\u884c\u8f6c\u6362\u7528\u4e8e\u4e0b\u6b21\u8fed\u4ee3":61,"\u5bf9\u6bd4":65,"\u5bf9\u6bd4\u4f18\u5316\u540elayer\u4e0e\u76f8\u5bf9\u5e94\u7684paddlepaddle\u539f\u6709lay":61,"\u5bf9\u6bd4\u4f18\u5316\u540elayer\u81ea\u8eab":61,"\u5bf9\u8f93\u5165\u53c2\u6570\u7684\u5b89\u5168\u6027\u8fdb\u884c\u4e86\u5fc5\u8981\u7684\u5224\u65ad":66,"\u5bf9\u8fd9\u4e2a\u7248\u672c\u7684\u63d0\u4ea4":82,"\u5bfb\u627e\u6709\u6ca1\u6709\u5176\u4ed6\u53ef\u4ee5\u4f18\u5316\u7684\u53ef\u80fd":62,"\u5bfc\u51fa\u8fd9\u4e9b\u63a5\u53e3":66,"\u5c06":82,"\u5c06\u4e00\u4e2a\u795e\u7ecf\u7f51\u7edc\u53c2\u6570\u62c6\u5206\u6210\u591a\u4efd":32,"\u5c06\u5927\u91cf\u7684":65,"\u5c06\u65b0\u5206\u652f\u7684\u7248\u672c\u6253\u4e0atag":82,"\u5c06master\u5206\u652f\u7684\u5408\u5165commit\u6253\u4e0atag":82,"\u5c0f\u4e8e\u67d0\u4e2a\u6bd4\u8f83\u5c0f\u7684\u9608\u503c\u8ba4\u4e3a\u901a\u8fc7":62,"\u5c31\u9700\u8981\u5bf9\u8fd9\u4e2a\u7b2c\u4e09\u65b9\u8bed\u8a00\u589e\u52a0\u4e00\u4e9b\u5b9a\u4e49":65,"\u5de5\u5177\u4e0a\u4f20\u5373\u53ef":82,"\u5e73\u5747\u6545\u969c\u4fee\u590d\u65f6\u95f4":32,"\u5e73\u5747\u6545\u969c\u7387":32,"\u5e76\u4e14\u4f1a\u5199\u597d":62,"\u5e76\u4e14\u4f7f\u7528":66,"\u5e76\u4e14\u53ea\u9700\u8981\u5728\u5fc5\u8981\u7684\u65f6\u5019\u8f6c\u6362\u8fd9\u79cd\u683c\u5f0f":62,"\u5e76\u4e14\u5728\u5e38\u89c1\u7684\u5e73\u53f0\u4e0a":65,"\u5e76\u4e14\u5f53\u7f16\u8bd1\u65f6":61,"\u5e76\u4e14\u628a\u7cfb\u7edf\u751f\u6210\u7684ca":48,"\u5e76\u4e14\u628a\u7ed3\u679c\u8fd4\u56depfsclient\u7aef":48,"\u5e76\u4e14\u8ba9\u63a5\u53e3\u8131\u79bb\u5b9e\u73b0\u7ec6\u8282":65,"\u5e76\u4e14\u8f93\u5165\u8f93\u51fa\u90fd\u662f\u5171\u7528\u4e00\u5757\u5185\u5b58":62,"\u5e76\u5220\u9664":82,"\u5e76\u5220\u9664\u66f4\u65e9\u7684\u5feb\u7167":32,"\u5e76\u52a0\u8f7d\u5176\u4e2d\u7684\u53c2\u6570":32,"\u5e76\u53d1\u5e03\u5230pypi":82,"\u5e76\u5728\u6bcf\u6b21\u6743\u91cd\u66f4\u65b0\u540e":61,"\u5e76\u5728\u96c6\u7fa4\u4e2d\u8fd0\u884c\u591a\u4e2a\u5206\u5e03\u5f0f\u6570\u636e\u5904\u7406\u4efb\u52a1":33,"\u5e76\u5c06":82,"\u5e76\u5c06c":66,"\u5e76\u628a\u5feb\u7167\u4fdd\u5b58\u5230\u8fd9\u4e2a\u76ee\u5f55\u4e0b":32,"\u5e76\u628a\u7ed3\u679c\u653e\u5230\u5f53\u524d\u5c42\u7684":62,"\u5e76\u6ca1\u6709paddle\u7279\u522b\u9700\u8981\u7684\u7279\u6027":65,"\u5e76\u6dfb\u52a0\u5934\u6587\u4ef6":61,"\u5e76\u88ab\u5b58\u50a8\u5728\u8bf8\u5982hadoop":33,"\u5e76\u9002\u5e94github\u7684\u7279\u6027\u505a\u4e86\u4e00\u4e9b\u533a\u522b":82,"\u5e76\u91cd\u65b0\u6253\u5305wheel\u5305":82,"\u5efa\u8bae":82,"\u5f00\u53d1\u4e86\u6a21\u578b\u9884\u6d4b\u7684\u6837\u4f8b\u4ee3\u7801":66,"\u5f00\u53d1\u8005\u4fee\u6539\u81ea\u5df1\u7684\u4ee3\u7801":82,"\u5f00\u53d1\u8005fork\u7684\u7248\u672c\u5e93\u4e2d":82,"\u5f00\u53d1\u8005fork\u7684\u7248\u672c\u5e93\u4f7f\u7528":82,"\u5f00\u5934":[61,62],"\u5f00\u59cb\u63d0\u4f9b\u670d\u52a1":32,"\u5f15\u5165\u4e86\u4ee5\u4e0b\u56db\u4e2aapi":61,"\u5f15\u5165\u4e86\u7c7b\u578b\u7684\u5934\u6587\u4ef6":66,"\u5f39\u51fa\u4e0b\u9762\u7684\u9009\u62e9\u6846":82,"\u5f53\u529f\u80fd\u5206\u652f\u5f00\u53d1\u5b8c\u6bd5\u540e":82,"\u5f53\u53ea\u505a\u63a8\u65ad":61,"\u5f53\u5f00\u542f":62,"\u5f53\u6253\u5f00":62,"\u5f53\u6570\u636e\u683c\u5f0f\u4e0epaddlepaddle\u9ed8\u8ba4\u7684":62,"\u5f53\u7136\u8fd9\u4e24\u8005\u4e5f\u53ef\u4ee5\u76f8\u7b49":62,"\u5f53\u7528\u6237\u4f7f\u7528\u5b8c\u8fd9\u4e2a\u53c2\u6570\u540e":66,"\u5f53\u7f51\u7edc\u51fa\u73b0\u5206\u652f\u4e14\u5728":62,"\u5f53destination\u6587\u4ef6\u4e0d\u5b58\u5728\u6216\u8005\u5927\u5c0f\u548csource\u6587\u4ef6\u4e0d\u4e00\u81f4\u65f6":48,"\u5f88\u96be\u4fdd\u8bc1\u591a\u8bed\u8a00\u4ee3\u7801\u98ce\u683c\u7684\u4e00\u81f4\u6027":65,"\u5f97\u4f7f\u7528":65,"\u5fc5\u8981":66,"\u5fc5\u987b\u5206\u522b\u4e0e":62,"\u60c5\u611f\u5206\u6790":82,"\u6211\u4eec\u4e5f\u53ef\u4ee5\u786e\u5b9a\u6bcf\u4e00\u4e2a\u53c2\u6570\u7684\u7c7b\u578b":66,"\u6211\u4eec\u4e5f\u5c06mklml\u5373":62,"\u6211\u4eec\u4f1a\u4fdd\u8bc1":62,"\u6211\u4eec\u4f1a\u5728\u7f51\u7edc\u8bad\u7ec3\u4e4b\u524d\u628a\u683c\u5f0f\u8f6c\u6362\u4e3amkl":62,"\u6211\u4eec\u4f1a\u5bf9\u6bd4\u5982\u4e0b2\u4e2a\u65b9\u9762":61,"\u6211\u4eec\u4f1a\u628amkl":62,"\u6211\u4eec\u4f1a\u6dfb\u52a0":[61,62],"\u6211\u4eec\u4f7f\u7528\u52a8\u6001\u5e93\u6765\u5206\u53d1paddl":65,"\u6211\u4eec\u51b3\u5b9a\u4f7f\u7528\u5df2\u6709\u7684":62,"\u6211\u4eec\u53ef\u4ee5\u5148\u5b8c\u6210\u5bf9\u539f\u6570\u636e\u7684packing\u64cd\u4f5c":61,"\u6211\u4eec\u603b\u7ed3\u51fa\u4e00\u4e9b\u7279\u522b\u9700\u8981\u6ce8\u610f\u7684\u70b9":62,"\u6211\u4eec\u63d0\u4f9b\u4e24\u4e2a\u8f6c\u6362\u65b9\u5f0f":33,"\u6211\u4eec\u63d0\u51fa\u4e86chunk\u7684\u6982\u5ff5":48,"\u6211\u4eec\u6700\u7ec8\u7684\u52a8\u6001\u5e93\u4e2d\u4e0d\u5d4c\u5165python\u6216\u8005\u5176\u4ed6\u4efb\u4f55\u8bed\u8a00\u7684\u89e3\u91ca\u5668":65,"\u6211\u4eec\u8ba1\u5212\u5c06":61,"\u6211\u4eec\u8ba1\u5212\u5c06\u82f1\u7279\u5c14\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\u6570\u5b66\u5e93":62,"\u6211\u4eec\u8bbe\u8ba1\u8bf4\u660e\u4e86\u540d\u4e3afilemanager\u7cfb\u7edf":48,"\u6211\u4eec\u9009\u62e9":33,"\u6211\u4eec\u901a\u8fc7\u4f7f\u7528\u65b0\u5f15\u5165\u7684gemm":61,"\u6211\u4eec\u90fd\u63d0\u4f9bpython\u7684\u8f6c\u6362\u5e93":33,"\u6216\u8005":[62,65,66],"\u6216\u8005\u5c06\u8fd9\u53f0\u8282\u70b9\u8fc1\u79fb\u5230\u53e6\u4e00\u4e2a\u8282\u70b9\u5e76\u542f\u52a8\u5373\u53ef\u6062\u590d\u8bad\u7ec3\u4efb\u52a1":32,"\u6216\u8005\u7528tuple\u8868\u793a\u7684\u591a\u4e2a\u503c":33,"\u6216\u8005\u7531\u5b83\u4eec\u7ec4\u6210\u7684list":33,"\u6216activ":62,"\u6240\u4ee5":[62,82],"\u6240\u4ee5\u4e00\u4e2a\u7248\u672c\u53f7\u7684wheel\u5305\u53d1\u5e03\u4e4b\u540e":82,"\u6240\u4ee5\u4e0d\u5b58\u5728\u8fd9\u4e2a\u95ee\u9898":62,"\u6240\u4ee5\u5728":62,"\u6240\u4ee5\u5728\u5199\u5165\u5feb\u7167\u7684\u8fc7\u7a0b\u4e2d":32,"\u6240\u4ee5\u6211\u4eec\u5b9a\u4e49\u4e86\u4e00\u4e2a":62,"\u6240\u4ee5\u6574\u4f53\u4e0a":62,"\u6240\u4ee5\u6dfb\u52a0\u4e86\u5bf9\u5e94\u7684":62,"\u6240\u4ee5\u7528\u6237\u9700\u8981\u9996\u5148\u5728":48,"\u6240\u4ee5\u9700\u8981\u5f15\u5165\u4e00\u4e2a\u8f6c\u6362\u65b9\u6cd5":62,"\u6240\u6709\u4e0e\u7c7b\u578b\u76f8\u5173\u7684\u51fd\u6570":66,"\u6240\u6709\u5916\u90e8\u7684\u8f6c\u6362\u5de5\u4f5c\u90fd\u4f1a\u5728reset\u7cfb\u5217\u51fd\u6570\u4e2d\u90fd\u51c6\u5907\u597d":62,"\u6240\u6709\u7684":61,"\u6240\u6709\u7684\u63a5\u53e3\u5747\u4e3ac\u63a5\u53e3":66,"\u6240\u6709\u76f8\u5173\u7684":61,"\u6240\u6709\u7c7b\u578b\u540d\u4e3a":66,"\u6240\u6709mkl":62,"\u624b\u5199\u591a\u8bed\u8a00\u7ed1\u5b9a":65,"\u624d\u80fd\u66f4\u597d\u7684\u53d1\u6325mkl":62,"\u6253\u5f00\u8fd9\u4e2a\u7f16\u8bd1\u9009\u9879":66,"\u6267\u884c":82,"\u628a":33,"\u628a\u4e4b\u524d\u793a\u4f8b\u4e2d\u8f6c\u6362\u5b8c\u6bd5\u7684random":33,"\u6307\u6df1\u5ea6\u5b66\u4e60\u8bad\u7ec3\u4e4b\u540e\u5f97\u5230\u7684\u6240\u6709\u53c2\u6570":32,"\u6309\u94ae":82,"\u63a5\u53e3":[65,66],"\u63a5\u53e3\u5c42\u505a\u8fc7\u591a\u5c01\u88c5":66,"\u63a5\u53e3\u662f":33,"\u63a5\u6536\u5904\u7406pfsclient\u7aef\u7684\u6587\u4ef6\u7ba1\u7406\u8bf7\u6c42":48,"\u63a7\u5236\u662f\u5426\u4f7f\u7528mkl":62,"\u63a7\u5236\u662f\u5426\u4f7f\u7528mklml\u5e93":62,"\u63a7\u5236\u7528\u6237\u6743\u9650":33,"\u63d0\u4f9b\u4e03\u5c42\u534f\u8bae\u7684\u53cd\u5411\u4ee3\u7406":48,"\u63d0\u4f9b\u5e38\u7528\u7684\u547d\u4ee4\u884c\u7ba1\u7406\u547d\u4ee4\u7ba1\u7406\u6587\u4ef6\u548c\u76ee\u5f55":48,"\u63d0\u4f9b\u7528\u6237\u7ba1\u7406\u6587\u4ef6\u7684\u547d\u4ee4":48,"\u63d0\u4f9b\u7ed9paddle\u4f5c\u4e3a\u8bad\u7ec3\u6570\u636e":33,"\u652f\u6301\u5927\u6587\u4ef6\u7684\u65ad\u70b9\u4e0a\u4f20":48,"\u6570\u636e":48,"\u6570\u636e\u8bfb\u53d6\u5747\u4ea4\u7531\u5176\u4ed6\u8bed\u8a00\u5b8c\u6210":65,"\u6570\u636e\u957f\u5ea6\u53ca\u6821\u9a8c\u503c\u7ec4\u6210":48,"\u6570\u636e\u96c6\u9700\u8981\u9884\u5148\u88ab\u8f6c\u6362\u6210paddlepaddle\u5206\u5e03\u5f0f\u8bad\u7ec3\u4f7f\u7528\u7684\u5b58\u50a8\u683c":33,"\u6570\u636e\u9884\u5904\u7406\u4efb\u52a1":33,"\u6587\u4ef6":65,"\u6587\u4ef6\u4f20\u8f93\u7684\u7684\u5173\u952e\u5728\u4e8e\u9700\u8981pfsclient\u7aef\u5bf9\u6bd4source\u548cdestination\u7684\u6587\u4ef6chunks\u7684checksum\u662f\u5426\u4fdd\u6301\u4e00\u81f4":48,"\u6587\u4ef6\u5185\u5bb9\u4e3a":65,"\u6587\u4ef6\u540d\u4e3a\u6b64uuid":32,"\u6587\u4ef6\u5bf9\u5e94\u7684data":33,"\u6587\u4ef6\u7684\u4e0a\u4f20\u548c\u4e0b\u8f7d\u90fd\u662f\u901a\u8fc7\u5bf9chunk\u7684\u64cd\u4f5c\u6765\u5b9e\u73b0\u7684":48,"\u65b0\u624b\u5165\u95e8\u7ae0\u8282":82,"\u65b0\u7248\u672c":62,"\u65b9\u4fbf\u6d4b\u8bd5\u4eba\u5458\u6d4b\u8bd5paddlepaddle\u7684\u884c\u4e3a":82,"\u65b9\u4fbf\u7528\u6237\u4e0a\u4f20\u81ea\u5df1\u7684\u8bad\u7ec3\u6570\u636e\u4ee5\u8fdb\u884c\u5206\u5e03\u5f0f\u8bad\u7ec3":48,"\u65b9\u4fbf\u7528\u6237\u5728python\u7aef\u9009\u62e9\u662f\u5426\u542f\u7528\u8fd9\u4e2a\u529f\u80fd":61,"\u65b9\u4fbf\u7528\u6237\u9009\u62e9\u4f7f\u7528mkl":62,"\u65b9\u5f0f\u7c7b\u4f3c\u4e8e":62,"\u65e0\u6cd5\u505a\u5230\u5bf9\u4e8e\u5404\u79cd\u8bed\u8a00\u9519\u8bef\u5904\u7406\u65b9\u5f0f\u7684\u9002\u914d":65,"\u65e0\u8bba\u5728\u672c\u5730\u8fd8\u662f\u5728\u4e91\u7aef":33,"\u65e0\u8bba\u662f\u4ece":33,"\u65e0\u8bba\u662f\u5728\u672c\u5730\u6216\u662f\u4e91\u7aef\u8f6c\u6362":33,"\u65e0\u8bba\u662f\u91cd\u6784\u524d\u7684layer\u8fd8\u662f\u91cd\u6784\u540e\u7684op":62,"\u65f6":[32,61,62],"\u65f6\u4e00\u8d77\u66f4\u65b0":62,"\u662f":[48,62],"\u662f\u4e00\u4e2a\u591a\u8bed\u8a00\u63a5\u53e3\u7684\u4ee3\u7801\u751f\u6210\u5668":65,"\u662f\u4e00\u4e2a\u7c7b\u578b\u7684\u6807\u5fd7":66,"\u662f\u4e0d\u5e38\u89c1\u7684\u505a\u6cd5":65,"\u662f\u5404\u4e2a\u5b9e\u73b0\u4e2d\u5171\u4eab\u7684\u5934\u6587\u4ef6":66,"\u662f\u5426\u6253\u5f00":61,"\u662f\u56e0\u4e3ac99\u652f\u6301":65,"\u662f\u5bf9\u7528\u6237\u6587\u4ef6\u5b58\u50a8\u7a7a\u95f4\u7684\u62bd\u8c61":48,"\u662f\u6307":66,"\u662f\u7528\u6237\u4f7f\u7528c":66,"\u662fc":66,"\u663e\u5f97\u76f8\u5bf9\u6765\u8bf4\u8f83\u4e3a\u8017\u65f6":61,"\u6682\u65f6\u4e0d\u8003\u8651\u591a\u4e2aparamet":32,"\u66b4\u9732\u8fd9\u4e2a\u6982\u5ff5\u5fc5\u8981\u51fd\u6570":66,"\u6700\u540e\u5220\u9664":82,"\u6700\u5e38\u89c1\u7684\u9519\u8bef\u5904\u7406\u65b9\u5f0f\u662fexcept":65,"\u6709\u6548\u63d0\u5347paddlepaddle\u5728\u82f1\u7279\u5c14\u67b6\u6784\u4e0a\u7684\u6027\u80fd":[61,62],"\u6709\u6807\u51c6\u7684":65,"\u6709\u7684\u65f6\u5019":65,"\u672c\u5217\u8868\u8bf4\u660epaddlepaddle\u53d1\u7248\u4e4b\u524d\u9700\u8981\u6d4b\u8bd5\u7684\u529f\u80fd\u70b9":82,"\u672c\u6587\u6863\u63cf\u8ff0paddl":66,"\u673a\u5668\u7ffb\u8bd1":82,"\u6765\u4fdd\u8bc1\u8bad\u7ec3\u8fc7\u7a0b\u53ef\u4ee5\u4ece\u4e2d\u95f4\u72b6\u6001\u91cd\u65b0\u542f\u52a8":32,"\u6765\u51b3\u5b9a\u662f\u5426\u5f00\u542fmkl":61,"\u6765\u5b9e\u73b0":62,"\u6765\u786e\u4fdd\u628a":65,"\u6765\u8868\u793apaddle\u5185\u90e8\u7c7b":65,"\u6765\u8bbf\u95ee\u7528\u6237\u81ea\u5df1\u7684\u6570\u636e":33,"\u6765\u8fdb\u884c\u8ba8\u8bba":66,"\u67e5\u770blatest":82,"\u6807\u51c6\u8868\u793apaddlepaddle\u7248\u672c\u53f7":82,"\u683c\u5f0f\u4e0d\u5339\u914d\u65f6":62,"\u68c0\u67e5\u70b9\u4fdd\u5b58\u7a0b\u5e8f\u6d41\u7a0b":32,"\u6a21\u578b\u53c2\u6570\u68c0\u67e5\u70b9\u901a\u8fc7\u5b9a\u671f\u5411\u78c1\u76d8\u4e0a\u4fdd\u5b58\u4e00\u4efd\u5b58\u50a8\u5728paramet":32,"\u6a21\u578b\u6570\u636e\u68c0\u67e5\u70b9\u7684\u5b9e\u73b0":32,"\u6a21\u578b\u914d\u7f6e\u89e3\u6790":65,"\u6b21\u8fed\u4ee3\u6267\u884c\u7684\u8f6c\u6362\u6b21\u6570\u4e3a":61,"\u6b64\u65f6\u6bcf\u4e2a\u5c0f\u5206\u652f\u7684":62,"\u6b64\u65f6master\u5c06\u8d1f\u8d23\u542f\u52a8\u4e00\u4e2a\u65b0\u7684train":32,"\u6bcf\u4e00\u4e2a":82,"\u6bcf\u4e00\u4e2a\u6587\u4ef6\u662f\u6570\u636e\u96c6\u7684\u4e00\u4e2ashard":33,"\u6bcf\u4e2a":62,"\u6bcf\u4e2a\u503c\u7684\u7c7b\u578b\u53ef\u4ee5\u662f\u6574\u5f62":33,"\u6bcf\u4e2a\u6d4b\u8bd5\u4f1a\u5bf9\u6bd4paddlepaddle\u4e2dcpu\u7b97\u51fa\u7684\u7ed3\u679c\u4e0emkl":62,"\u6bcf\u4e2adata":33,"\u6bcf\u4e2amkldnnlayer\u90fd\u5305\u542b\u7528\u4e8e\u5185\u90e8\u5b58\u50a8\u548c\u5916\u90e8\u5b58\u50a8\u7684\u4e00\u7cfb\u5217mkldnnmatrix":62,"\u6bcf\u4e2aparamet":32,"\u6bcf\u4e2ashard\u5206\u522b\u5b58\u50a8\u5728\u5176\u4e2d\u4e00\u53f0paramet":32,"\u6bcf\u6b21\u8c03\u7528\u65f6\u5bf9\u539f\u6570\u636e\u7684\u91cd\u590dpacking\u4fbf\u6210\u4e3a\u4e86\u5197\u4f59":61,"\u6bcf\u6b21\u8f93\u51fa\u4e00\u4e2adata":33,"\u6bcf\u969410\u5206\u949f":32,"\u6bd4\u5982":[33,62],"\u6bd4\u5982\u53ef\u80fd\u4f1a\u7528openmp\u6539\u8fdbsgd\u7684\u66f4\u65b0\u6027\u80fd":62,"\u6bd4\u5982\u5c06":82,"\u6bd4\u5982\u6bcf\u969410\u5206\u949f\u6700\u65b0\u7684\u5feb\u7167":32,"\u6bd4\u5982\u6d41\u5f0f\u6570\u636e\u5904\u7406":33,"\u6bd4\u5982imagenet\u8fd9\u4e2a\u6570\u636e\u96c6\u53ef\u80fd\u88ab\u5206\u62101000\u4e2ashard":33,"\u6ca1\u6709\u5fc5\u8981\u5728\u6bcf\u6b21\u524d\u5411\u4e2d\u6bcf\u4e2a\u65f6\u95f4\u6b65\u7684\u8ba1\u7b97\u65f6\u5bf9\u6743\u91cd\u8fdb\u884c\u91cd\u590d\u7684packing\u64cd\u4f5c":61,"\u6ce8":[32,82],"\u6ce8\u518clayer\u7684\u65f6\u5019\u4fdd\u8bc1":[61,62],"\u6ce8\u610f":62,"\u6d4b\u8bd5\u5206\u4e3a\u6bcf\u4e2alayer":62,"\u6d4b\u8bd5\u672c\u6b21release\u7684\u6b63\u786e\u6027":82,"\u6d4b\u8bd5\u7684\u6027\u80fd\u5bf9\u6bd4\u7ed3\u679c\u4f1a\u5728":62,"\u6d6e\u70b9\u578b\u6570\u636e":33,"\u6df1\u5165paddlepaddl":62,"\u6dfb\u52a0":61,"\u6dfb\u52a0\u7684\u76f8\u5173\u6587\u4ef6\u548c\u76ee\u5f55\u7ed3\u6784\u5982\u4e0b":[61,62],"\u6fc0\u6d3b\u51fd\u6570\u662f\u72ec\u7acb\u4e8e":62,"\u70b9\u51fb":82,"\u7136\u540e\u5728\u524d\u5411":61,"\u7136\u540e\u5728etcd\u7684":32,"\u7136\u540e\u5c31\u53ef\u4ee5\u5e76\u53d1\u5199\u5165\u591a\u4e2achunk":48,"\u7136\u540e\u624d\u80fd\u4f7f\u7528pfsclient":48,"\u7136\u540e\u6309\u7167\u4e0a\u8ff0\u7684\u65b9\u6cd5":82,"\u7136\u540e\u70b9\u51fb":82,"\u7248\u672c\u5206\u652f":82,"\u7248\u672c\u53f7":82,"\u7248\u672c\u53f7\u5bf9\u5e94\u7684tag\u5373\u53ef":82,"\u7248\u672c\u53f7rc":82,"\u7248\u672cfork\u51fa\u81ea\u5df1\u7684\u529f\u80fd\u5206\u652f":82,"\u7279\u6709\u7684\u8bbe\u5907id":62,"\u73b0\u9636\u6bb5\u7684\u4f18\u5316\u4e3b\u8981\u9488\u5bf9":61,"\u73b0\u9636\u6bb5paddle\u6709\u4e00\u4e2a\u95ee\u9898\u662f":65,"\u751f\u4ea7\u73af\u5883\u4e2d\u7684\u8bad\u7ec3\u6570\u636e\u96c6\u901a\u5e38\u4f53\u79ef\u5f88\u5927":33,"\u751f\u4ea7\u73af\u5883\u7684\u65e5\u5fd7\u6570\u636e\u4f1a\u901a\u8fc7\u5b9e\u65f6\u6d41\u7684\u65b9\u5f0f":33,"\u751f\u6210\u5404\u79cd\u8bed\u8a00\u7684\u7ed1\u5b9a\u4ee3\u7801":65,"\u751f\u6210\u6587\u6863":65,"\u751f\u6210\u7684":33,"\u751f\u6210\u7ed9\u5b9a":33,"\u751f\u6210api\u6587\u6863":65,"\u751f\u6210pfsclient\u548cpfsserver\u7684\u6846\u67b6\u90e8\u5206":48,"\u7528":48,"\u7528\u4e8e\u6d4b\u8bd5\u548c\u5bf9\u6bd4\u5728\u4f7f\u7528mkl":62,"\u7528\u4e8e\u7ba1\u7406mkl":62,"\u7528\u4e8e\u9009\u62e9\u662f\u5426\u4f7f\u7528\u76f8\u5173\u529f\u80fd":61,"\u7528\u4e8e\u9009\u62e9\u662f\u5426\u4f7f\u7528mkl":62,"\u7528\u4e8emkl":[61,62],"\u7528\u6237\u4e0a\u4f20\u6570\u636e\u540e":33,"\u7528\u6237\u4e5f\u53ef\u4ee5\u4e0a\u4f20label":33,"\u7528\u6237\u53ef\u4ee5\u5b89\u5168\u7684\u91ca\u653e\u67d0\u4e2ac":66,"\u7528\u6237\u53ef\u4ee5\u628a\u81ea\u5df1\u7684\u6570\u636e\u5206\u4eab\u7ed9\u522b\u4eba":33,"\u7528\u6237\u53ef\u4ee5\u76f4\u63a5\u4f7f\u7528\u8fd9\u4e2a\u52a8\u6001\u5e93\u6765\u5f15\u5165paddl":66,"\u7528\u6237\u5728\u672c\u5730\u8f6c\u6362\u597d\u518d\u4e0a\u4f20":33,"\u7528\u6237\u6587\u4ef6\u53ef\u80fd\u662f\u6bd4\u8f83\u5927\u7684":48,"\u7528\u6237\u901a\u8fc7c":66,"\u7531\u4e8e\u5728\u73b0\u6709\u7684\u67d0\u4e9b\u60c5\u51b5\u4e0b":61,"\u7531\u4e8e\u5bf9parameters\u7684\u66f4\u65b0\u9700\u8981\u83b7\u53d6parameters\u5185\u5b58\u7684":32,"\u7531\u4e8e\u96c6\u7fa4\u4e2d\u540c\u65f6\u5b58\u5728\u4e24\u53f0\u673a\u5668\u6545\u969c\u7684\u6982\u7387\u6781\u4f4e":32,"\u7531\u4e8ec":65,"\u7531\u4e8echunk\u6bd4\u8f83\u5c0f":48,"\u7531\u4e8emkl":62,"\u7531\u4e8epypi":82,"\u7531\u5206\u652f\u5904\u7684layer\u8d1f\u8d23\u6c42\u548c":62,"\u7533\u8bf7\u7528\u6237\u7a7a\u95f4":48,"\u7684\u4e00\u4e2a\u5b50\u96c6":62,"\u7684\u4fe1\u606f":62,"\u7684\u5355\u5143\u6d4b\u8bd5\u548c\u7b80\u5355\u7f51\u7edc\u7684\u6574\u4f53\u6d4b\u8bd5":62,"\u7684\u547d\u540d\u98ce\u683c\u5e76\u4e0d\u80fd\u9002\u5e94\u5176\u4ed6\u7b2c\u4e09\u65b9\u8bed\u8a00":65,"\u7684\u57fa\u672c\u903b\u8f91":62,"\u7684\u5934\u6587\u4ef6":65,"\u7684\u5b50\u7c7b\u53ea\u9700\u8981\u4f7f\u7528\u5185\u90e8\u5b58\u50a8\u5c31\u53ef\u4ee5\u4e86":62,"\u7684\u60c5\u51b5\u4e0b":61,"\u7684\u63a5\u53e3\u6837\u5f0f":65,"\u7684\u6570\u636e\u6d41\u56fe":33,"\u7684\u65f6\u5019":62,"\u7684\u683c\u5f0f\u59cb\u7ec8\u662f":62,"\u7684\u683c\u5f0f\u5b58\u50a8":62,"\u7684\u6982\u5ff5":62,"\u7684\u6e90\u7801\u91cc\u4f7f\u7528\u4e86":65,"\u7684\u7248\u672c":82,"\u7684\u7ed3\u679c":61,"\u7684\u7f29\u5199":48,"\u7684\u7f51\u7edc\u6a21\u578b":61,"\u7684\u89c4\u8303":65,"\u7684\u89d2\u5ea6":33,"\u7684\u914d\u7f6e\u5199\u5230\u914d\u7f6e\u6587\u4ef6\u4e2d":33,"\u7684flag":[61,62],"\u7684vanilla":61,"\u76ee\u524d\u53ea\u8003\u8651":62,"\u76ee\u524d\u53ea\u8003\u8651\u52a8\u6001\u6269\u5bb9trainer\u6570\u91cf":32,"\u76ee\u524d\u5728paddlepaddle\u4e2d":62,"\u76ee\u524d\u5728paddlepaddle\u4e2d\u6570\u636e\u90fd\u662f\u4ee5":62,"\u76ee\u524d\u5d4c\u5165python\u89e3\u91ca\u5668":65,"\u76ee\u524d\u6211\u4eec\u7528cephfs\u6765\u642d\u5efa":48,"\u76ee\u524d\u7684\u4f18\u5316":62,"\u76ee\u524dpaddle\u7684\u8fdb\u7a0b\u6a21\u578b\u662fc":65,"\u76ee\u524dpaddlepaddle\u91c7\u7528\u4e86":61,"\u76ee\u5f55\u4e0b":66,"\u76ee\u5f55\u4e0b\u5bf9\u5e94\u7684\u5730\u65b9":62,"\u76f4\u63a5\u4f7f\u7528c\u8bed\u8a00\u7684":65,"\u76f4\u63a5\u5220\u9664\u8fd9\u4e2a\u53c2\u6570\u5373\u53ef":66,"\u76f4\u63a5\u5bfc\u51fa\u5230c\u7684\u63a5\u53e3\u6bd4\u8f83\u56f0\u96be":65,"\u76f8\u5173\u5c42":61,"\u77e9\u9635\u5927\u5c0f\u662f":61,"\u793e\u533a\u53c2\u4e0e\u56f0\u96be":65,"\u793e\u533a\u8d21\u732e\u4ee3\u7801\u5b66\u4e60\u6210\u672c\u9ad8":65,"\u795e\u7ecf\u7f51\u7edc\u4e2d\u7684\u53c2\u6570":32,"\u79bb\u7ebf\u6279\u5904\u7406":33,"\u7b2c\u4e00\u4e2atag\u4e3a":82,"\u7b2c\u4e09\u6b65\u5b8c\u6210\u540e":82,"\u7b2c\u4e8c\u4e2a\u4e3a":82,"\u7b49":[62,66],"\u7b49\u5168\u90e8\u9759\u6001\u5e93\u4e2d\u7684\u76ee\u6807\u6587\u4ef6\u5168\u90e8\u6253\u5305\u540e\u4ea7\u751f\u7684\u6587\u4ef6":66,"\u7b49\u5f85\u7f16\u8bd1\u5b8c\u6210\u540e":82,"\u7b49\u6587\u4ef6":66,"\u7c7b\u4f3c":66,"\u7c7b\u540d\u548cc":65,"\u7c7b\u578b":65,"\u7ed3\u8bba":65,"\u7edf\u4e00\u7528":33,"\u7f16\u8bd1\u5668\u6ca1\u6709":65,"\u7f16\u8bd1\u578b\u8bed\u8a00":65,"\u7f16\u8bd1\u65f6\u4f1a\u628a\u5bf9\u5e94\u7684\u5934\u6587\u4ef6\u548c\u5e93\u653e\u5728":62,"\u7f16\u8bd1\u8fd9\u4e2a\u7248\u672c\u7684docker\u53d1\u884c\u955c\u50cf":82,"\u7f16\u8bd1\u8fd9\u4e2a\u7248\u672c\u7684python":82,"\u7f16\u8bd1c":66,"\u800c\u4e0d\u5fc5\u5728\u610fpaddl":66,"\u800c\u4e0d\u652f\u6301pypy\u89e3\u91ca\u5668":65,"\u800c\u4e0d\u66b4\u9732\u6982\u5ff5\u7684\u5b9e\u73b0":66,"\u800c\u4e14\u5728\u4f20\u8f93\u7684\u8fc7\u7a0b\u4e2d\u4e5f\u53ef\u80fd\u51fa\u73b0\u7f51\u7edc\u4e0d\u7a33\u5b9a\u7684\u60c5\u51b5":48,"\u800c\u51fa\u73b0\u9636\u6bb5\u6027\u7684\u8fd0\u884c\u505c\u6ede":32,"\u800c\u5728cpp\u91cc\u9762\u5b9e\u73b0\u8fd9\u4e2ac\u7684\u63a5\u53e3":65,"\u800c\u591a\u8bed\u8a00\u63a5\u53e3\u9700\u8981\u76f4\u63a5\u8bfb\u53d6\u751f\u6210\u7684\u4e8c\u8fdb\u5236":65,"\u800c\u5bf9\u4e8egolang":65,"\u800c\u5bf9\u4e8egolang\u9519\u8bef\u5904\u7406\u5e94\u8be5\u4f7f\u7528\u8fd4\u56de\u503c":65,"\u800c\u662f\u76f4\u63a5\u4fee\u6539paddl":66,"\u800c\u662f\u76f4\u63a5\u7528api\u7684\u63a5\u53e3\u8fdc\u7a0b\u8bbf\u95ee":33,"\u800cswig\u53ea\u80fd\u7b80\u5355\u7684\u66b4\u9732c":65,"\u81ea\u52a8\u6302\u8f7d\u5206\u5e03\u5f0f\u5b58\u50a8\u76ee\u5f55":32,"\u81f3\u4e8e\u4e3a\u4ec0\u4e48\u9700\u8981c":66,"\u826f\u597d\u7684\u6587\u6863":65,"\u8282\u7701\u4e86\u4e0d\u5fc5\u8981\u7684\u64cd\u4f5c":62,"\u83b7\u53d6\u6700\u65b0\u7684\u68c0\u67e5\u70b9\u7684\u6587\u4ef6uuid":32,"\u867d\u7136\u4e0d\u9f13\u52b1\u8fd9\u6837":66,"\u8868\u793a\u5bf9\u8f93\u5165\u6570\u636e":62,"\u89e3\u91ca\u578b\u8bed\u8a00\u53ea\u80fd\u8c03\u7528\u52a8\u6001\u5e93":65,"\u89e3\u91ca\u6027\u8bed\u8a00\u5b9e\u9645\u8fd0\u884c\u7684\u4e8c\u8fdb\u5236\u662f\u89e3\u91ca\u5668\u672c\u8eab":65,"\u8ba1\u5212\u5728":[61,62],"\u8ba1\u7b97\u8fd9\u4e2a\u6587\u4ef6\u7684md5":32,"\u8ba9paddle\u6838\u5fc3\u4e2d":66,"\u8bad\u7ec3\u4efb\u52a1\u7684\u8fd0\u884c\u53ef\u80fd\u4f1a\u5360\u6ee1trainer\u548cparamet":32,"\u8bad\u7ec3\u548c\u7eaf\u4f7f\u7528":82,"\u8bad\u7ec3\u6a21\u578b\u6b63\u786e\u6027":82,"\u8bb0\u5f55\u4e0b\u6240\u6709\u5931\u8d25\u7684\u4f8b\u5b50":82,"\u8bbe\u7f6e":66,"\u8bc6\u522b\u6570\u5b57":82,"\u8bcd\u5411\u91cf":82,"\u8be5\u6587\u4ef6\u5bf9\u76f8\u5173gemm":61,"\u8be5\u7c7b\u7ee7\u627f\u4e8epaddlepaddle\u7684\u57fa\u7c7b":62,"\u8be6\u7ec6\u8bbe\u8ba1":48,"\u8bed\u610f\u89d2\u8272\u6807\u6ce8":82,"\u8bf4\u660e":32,"\u8bf7\u53c2\u8003":66,"\u8f6c\u6362\u5185\u5b58\u7684\u5de5\u4f5c":62,"\u8f6c\u6362\u5197\u4f59":61,"\u8f6c\u6362\u51fd\u6570":62,"\u8f6c\u6362\u751f\u6210\u7684\u6587\u4ef6\u540d\u4f1a\u662f\u4ee5\u4e0b\u683c\u5f0f":33,"\u8f6c\u6362\u8017\u65f6":61,"\u8f93\u5165\u68af\u5ea6":62,"\u8f93\u51fa\u6570\u636e\u548c\u8f93\u51fa\u68af\u5ea6":62,"\u8f93\u51fa\u6570\u636e\u548c\u8f93\u51fa\u68af\u5ea6\u7684\u8f6c\u6362":62,"\u8fbe\u5230\u5bb9\u707e\u7684\u76ee\u7684":32,"\u8fc7\u7a0b\u4e2d\u6240\u6709\u65f6\u95f4\u6b65":61,"\u8fd1\u671f\u76ee\u6807":62,"\u8fd4\u56de\u7b2c\u4e8c\u6b65":82,"\u8fd8\u662f\u4ece":33,"\u8fd9\u4e00\u5c42\u8fdb\u884c\u5c01\u88c5":66,"\u8fd9\u4e00\u6570\u636e\u683c\u5f0f\u7684\u8f6c\u6362\u64cd\u4f5c":61,"\u8fd9\u4e00\u6982\u5ff5\u4e0d\u518d\u7410\u788e":66,"\u8fd9\u4e09\u4e2a\u5206\u652f":82,"\u8fd9\u4e2a\u51fd\u6570\u672c\u8eab\u4f1a\u5728\u8ba1\u7b97\u524d\u5c06\u539f\u6570\u636e\u8f6c\u6362\u4e3a\u66f4\u9002\u5408\u82f1\u7279\u5c14\u5e73\u53f0\u7684\u5185\u90e8\u683c\u5f0f":61,"\u8fd9\u4e2a\u52a8\u6001\u5e93\u7684\u8fde\u63a5\u53c2\u6570\u4e0epaddle\u7684\u5176\u4ed6\u4e8c\u8fdb\u5236":66,"\u8fd9\u4e2a\u53c2\u6570\u4e5f\u4e0d\u4f1a\u4e00\u5e76\u5220\u9664":66,"\u8fd9\u4e2a\u5934\u6587\u4ef6\u4e0d\u5047\u8bbe\u5176\u4ed6\u6587\u4ef6\u7684\u5f15\u7528\u987a\u5e8f":66,"\u8fd9\u4e2a\u63a5\u53e3\u9700\u8981\u505a\u5230":65,"\u8fd9\u4e2a\u6587\u4ef6\u5177\u6709\u72ec\u7279\u7684\u8bed\u6cd5":65,"\u8fd9\u4e2a\u76ee\u5f55\u4e2d\u9664\u4e86":66,"\u8fd9\u4e2a\u7ed3\u6784\u4f53\u4e2d\u7684\u53e6\u4e00\u4e2a\u9879\u76ee\u662f":66,"\u8fd9\u4e2a\u7ed3\u6784\u4f53\u5305\u542b\u4e24\u4e2a\u9879\u76ee":66,"\u8fd9\u4e2a\u9009\u62e9":[61,62],"\u8fd9\u4e2a\u9759\u6001\u5e93\u5305\u542b\u4e86paddle\u7684\u5168\u90e8\u7b26\u53f7":66,"\u8fd9\u4e2ainstance\u53ef\u4ee5\u662f\u5355\u4e2a\u503c":33,"\u8fd9\u4e9b\u4f1a\u5728":[61,62],"\u8fd9\u4e9b\u51fd\u6570\u4f1a\u6839\u636e\u8f93\u5165\u53c2\u6570\u91cd\u65b0\u8bbe\u7f6e\u5185\u90e8\u548c\u5916\u90e8\u5b58\u50a8":62,"\u8fd9\u4e9b\u5206\u5e03\u5f0f\u5b58\u50a8\u670d\u52a1\u901a\u5e38\u4f1a\u628a\u6570\u636e\u5207\u5272\u6210\u591a\u4e2a\u5206\u7247\u5206\u5e03\u5f0f\u7684\u5b58\u50a8\u5728\u591a\u4e2a\u8282\u70b9\u4e4b\u4e0a":33,"\u8fd9\u4e9b\u955c\u50cf\u4e5f\u53ef\u4ee5\u4ece":82,"\u8fd9\u5bf9\u4e8e\u901a\u5e38\u7684java\u7684\u5f00\u53d1\u8005\u6765\u8bf4":65,"\u8fd9\u662f\u56e0\u4e3a":65,"\u8fd9\u6837":66,"\u8fd9\u6837\u4e0b\u4e00\u4e2acpu":62,"\u8fd9\u6837\u4fdd\u8bc1":82,"\u8fd9\u6837\u5c31\u53ef\u4ee5\u5728\u4e91\u7aef\u6267\u884c\u591a\u79cd\u6570\u636e\u7c7b\u8ba1\u7b97\u4efb\u52a1":33,"\u8fd9\u6837\u5df2\u7ecf\u4f20\u8f93\u6210\u529f\u7684\u90e8\u5206\u5c31\u4e0d\u7528\u91cd\u65b0\u4f20\u8f93\u4e86":48,"\u8fd9\u6837\u5e26\u6765\u7684\u597d\u5904\u5c31\u662f\u4e0d\u9700\u8981\u4e00\u76f4\u6e05\u7a7amemori":62,"\u8fd9\u6837\u65e2\u4f7f\u5f97\u6700\u7ec8\u4fdd\u5b58\u7684\u53c2\u6570\u683c\u5f0f\u4e0epaddlepaddle\u4e00\u81f4":62,"\u8fd9\u90fd\u9700\u8981\u8fd9\u4e2a\u63a5\u53e3\u6309\u7167\u7ea6\u5b9a\u4fd7\u6210\u7684\u89c4\u5219\u6765\u6ce8\u91ca\u5b8c\u5907":65,"\u8fd9\u91cc":62,"\u8fd9\u91cc\u7684dockerimage\u4f5c\u4e3a\u7f16\u8bd1\u73af\u5883\u4ee5\u652f\u6301\u66f4\u591a\u7684linux":82,"\u8fd9\u91cc\u9009\u62e90":82,"\u8fd9\u91cc\u9700\u8981\u7528\u6237\u989d\u5916\u6ce8\u610f":32,"\u8fdb\u4e00\u6b65\u4f18\u5316":62,"\u8fdb\u5165":82,"\u8fdb\u800c\u8fdb\u884c\u4ee3\u7801\u8bc4\u5ba1":82,"\u9009\u62e9\u662f\u5426\u7f16\u8bd1mkl":62,"\u9009\u62e9\u9700\u8981\u53d1\u5e03\u7684\u7248\u672c":82,"\u900f\u4f20\u7528\u6237\u8eab\u4efd\u7684\u529e\u6cd5":48,"\u901a\u5e38":66,"\u901a\u5e38\u5305\u542b\u4e00\u4e2acpu\u7248\u672c\u548c\u4e00\u4e2agpu\u7248\u672c":82,"\u901a\u5e38\u6307\u5c06\u4e00\u4e2a\u6574\u4f53\u62c6\u5206\u6210\u591a\u4efd\u7684\u5176\u4e2d\u7684\u4e00\u4efd":32,"\u901a\u8fc7\u4f7f\u7528\u8fd9\u4e9bapi":61,"\u901a\u8fc7\u6a21\u578b\u63a8\u65adapi\u7684\u5b9e\u73b0\u4f5c\u4e3a\u4e00\u4e2a\u6837\u4f8b":66,"\u903b\u8f91\u5212\u4e0a\u6587\u4ef6\u5206\u5757\u7684\u5355\u4f4d":48,"\u9075\u5faa\u4ee5\u4e0b\u6d41\u7a0b":82,"\u90a3\u4e48":66,"\u90a3\u4e48\u5bf9\u5e94\u7684\u5185\u90e8\u5b58\u50a8\u4e5f\u4f1a\u4e0e\u5b83\u4eec\u5171\u4eab\u5185\u5b58":62,"\u90a3\u4e48\u5c31\u4f1a\u4f7f":62,"\u90fd\u4e0d\u4f1a\u60f3\u8981\u77e5\u9053next":62,"\u90fd\u662f\u4e94\u4f4d\u7684\u6570\u5b57":33,"\u90fd\u662f\u4ee5ext\u5f00\u5934":62,"\u90fd\u662fabi\u8c03\u7528\u6807\u51c6\u7684":65,"\u90fd\u7ee7\u627f\u4e8epaddlepaddle\u7684\u57fa\u7c7b":61,"\u914d\u7f6e\u7684\u65b9\u6cd5\u53c2\u8003":48,"\u91ca\u653e\u5bf9paramters\u5185\u5b58\u7684\u9501\u5b9a":32,"\u91cc\u6240\u6709\u7684\u7b26\u53f7\u90fd\u5199\u5165\u81ea\u5df1\u7684\u7a0b\u5e8f\u7684\u4e8c\u8fdb\u5236\u6587\u4ef6\u91cc":65,"\u91cc\u9009\u62e9\u9700\u8981\u53d1\u5e03\u7684\u5206\u652f":82,"\u91cc\u9762\u6dfb\u52a0":62,"\u91cd\u5199\u7236\u7c7blayer\u7684":62,"\u91cd\u547d\u540d\u6210":65,"\u94fe\u63a5\u5230\u81ea\u5df1\u7684\u7a0b\u5e8f\u91cc":65,"\u9519\u8bef\u5904\u7406":65,"\u9519\u8bef\u5904\u7406\u65b9\u5f0f\u662f\u8fd4\u56de\u503c":65,"\u9519\u8bef\u5904\u7406\u7684\u65b9\u5f0f\u4e5f\u4e0d\u5c3d\u76f8\u540c":65,"\u9664\u6784\u9020\u67d0\u79cd\u7c7b\u578b\u7684\u51fd\u6570":66,"\u96c6\u6210\u5230":61,"\u96c6\u6210\u5230paddlepaddl":62,"\u9700\u8981":33,"\u9700\u8981\u4fee\u6539build":82,"\u9700\u8981\u53ef\u4ee5\u8de8\u5e73\u53f0\u6267\u884c":48,"\u9700\u8981\u5728cmake\u7684\u65f6\u5019":66,"\u9700\u8981\u5c06bugfix\u7684\u5206\u652f\u540c\u65f6merge\u5230":82,"\u9700\u8981\u5f15\u7528":66,"\u9700\u8981\u6709\u7a33\u5b9a\u7684\u5bfc\u51fa\u7b26\u53f7":65,"\u9700\u8981\u6ce8\u610f\u7684\u662f":[62,82],"\u9700\u8981\u7d2f\u52a0\u4e0d\u540clayer\u4f20\u8fc7\u6765\u7684\u68af\u5ea6":62,"\u9700\u8981\u88ab\u66b4\u9732\u5230\u5176\u4ed6\u8bed\u8a00":66,"\u9700\u8981\u91cd\u547d\u540dwheel\u5305\u4e2dplatform\u76f8\u5173\u7684\u540e\u7f00":82,"\u9ed8\u8ba4256k":48,"\u9ed8\u8ba4\u8bbe\u7f6e\u4e3a":61,"abstract":[40,47,51,72,81,83,100,118],"api\u4e2d\u4f7f\u7528":65,"api\u5bfc\u51fa\u7684\u52a8\u6001\u5e93":66,"api\u5bfc\u51fa\u7684\u9759\u6001\u5e93":66,"api\u63a5\u53d7\u7684\u7c7b\u578b\u5168\u662f":66,"api\u63a5\u53e3":48,"api\u63a5\u53e3\u7684\u53c2\u6570\u8f6c\u53d1\u7ed9":66,"api\u65f6":66,"api\u65f6\u6240\u552f\u4e00\u9700\u8981\u5f15\u5165\u7684\u5934\u6587\u4ef6":66,"api\u662f\u591a\u8bed\u8a00api\u7684\u57fa\u7840\u90e8\u5206":66,"api\u66b4\u9732\u7684\u7c7b\u578b":66,"api\u751f\u6210\u7684\u4e8c\u8fdb\u5236\u6587\u4ef6\u4f1a\u88ab\u5b89\u88c5\u5230":66,"api\u7684\u5b9e\u4f8b":66,"api\u7684\u5b9e\u73b0\u7ec6\u8282":66,"api\u7684\u63a5\u53e3":66,"api\u7684\u65f6\u5019\u63a8\u8350paddle\u4e0d\u5d4c\u5165python\u89e3\u91ca\u5668":66,"api\u7684\u7f16\u8bd1\u9009\u9879\u9ed8\u8ba4\u5173\u95ed":66,"api\u76ee\u5f55\u7ed3\u6784\u5982\u4e0a\u56fe\u8868\u6240\u793a":66,"api\u83b7\u5f97\u4e86\u795e\u7ecf\u7f51\u7edc\u7684\u53c2\u6570\u5b9e\u4f8b":66,"apis\u505a\u4e86\u5c01\u88c5":61,"block\u6784\u6210\u4e00\u4e2amodel":32,"book\u4e2d\u6240\u6709\u7ae0\u8282\u529f\u80fd\u7684\u6b63\u786e\u6027":82,"boolean":[18,47,49,57,65],"break":[9,30,86,89,90],"bugfix\u5206\u652f\u4e5f\u662f\u5728\u5f00\u53d1\u8005\u81ea\u5df1\u7684fork\u7248\u672c\u5e93\u7ef4\u62a4":82,"bugfix\u5206\u652f\u9700\u8981\u5206\u522b\u7ed9\u4e3b\u7248\u672c\u5e93\u7684":82,"byte":[9,11,48,64],"c99\u662f\u76ee\u524dc\u6700\u5e7f\u6cdb\u7684\u4f7f\u7528\u6807\u51c6":65,"c\u6709\u6807\u51c6\u7684abi":65,"c\u8bed\u8a00\u662f\u6709\u5bfc\u51fa\u7b26\u53f7\u7684\u6807\u51c6\u7684":65,"case":[4,18,34,40,42,47,51,60,66,72,76,78,79,97,100,101,107,111,112,119,126],"char":36,"ci\u73af\u5883\u4f7f\u7528":82,"ci\u7f16\u8bd1wheel\u5b8c\u6210\u540e\u4f1a\u81ea\u52a8\u5c06docker\u955c\u50cfpush\u5230dockerhub":82,"class":[1,2,3,4,5,6,7,9,10,13,14,15,18,21,25,26,29,40,41,42,45,46,50,51,52,53,55,56,58,60,65,69,70,74,75,79,80,81,83,84,85,87,89,101,102,103,108,117],"compute\u51fd\u6570":61,"const":[29,34,36,50,52,58,59,73,74,76,80,83,85,87,88,89,100,101,102,103],"core\u4e2d\u7684\u6a21\u578b\u8fd8\u5728\u4f7f\u7528\u8fd9\u4e2a\u53c2\u6570":66,"core\u4e2d\u8fd9\u4e00\u7c7b\u578b\u63a5\u53e3\u7684\u667a\u80fd\u6307\u9488":66,"core\u662f\u5426\u8fd8\u5728\u4f7f\u7528\u8fd9\u4e2a\u5b9e\u4f8b":66,"core\u6982\u5ff5":66,"data\u5230\u5206\u5e03\u5f0f\u5b58\u50a8\u8865\u5145\u8bad\u7ec3\u6570\u636e":33,"default":[2,3,4,5,6,7,9,10,11,17,18,19,22,25,26,29,30,40,45,54,58,64,67,68,76,77,83,84,85,90,91,92,94,98,99,101,102,106,109,110,112,113,118,120,122,123,124,126],"device\u5c31\u80fd\u62ff\u5230\u6b63\u786e\u7684\u6570\u636e":62,"dnn\u4e09\u8005\u5173\u7cfb\u5982\u4e0b\u8868":62,"dnn\u4e2d\u7684":62,"dnn\u4e2d\u7684\u6392\u5217\u65b9\u5f0f\u4e0d\u6b62\u8fd9\u4e00\u79cd":62,"dnn\u4f1a\u4f5c\u4e3a\u7b2c\u4e09\u65b9\u5e93\u96c6\u6210\u8fdbpaddlepaddl":62,"dnn\u4f1a\u7528\u5230":62,"dnn\u5171\u540c\u4f7f\u7528":62,"dnn\u524d\u540e\u7684cnn\u7f51\u7edc\u6027\u80fd":62,"dnn\u5728\u53d1\u5e03":62,"dnn\u5b9e\u73b0":62,"dnn\u5e0c\u671b\u7684\u683c\u5f0f":62,"dnn\u6570\u636e\u7684\u4e0d\u540c\u683c\u5f0f\u4ee5\u53ca\u76f8\u4e92\u4e4b\u95f4\u7684\u8f6c\u6362":62,"dnn\u7684":62,"dnn\u7684\u5e93\u76ee\u524d\u53ea\u6709\u52a8\u6001\u5e93":62,"dnn\u7684\u6027\u80fd":62,"dnn\u7684\u60c5\u51b5\u4e0b":62,"dnn\u7684\u64cd\u4f5c\u90fd\u662f\u76f4\u63a5\u8986\u76d6\u7684\u5f62\u5f0f":62,"dnn\u7684\u6d4b\u8bd5":62,"dnn\u7684\u73af\u5883\u4e0b":62,"dnn\u7684\u76f8\u5173\u529f\u80fd":62,"dnn\u7684\u7ed3\u679c":62,"dnn\u7684\u9ad8\u6027\u80fd\u683c\u5f0f\u4e0epaddlepaddle\u539f\u6709\u7684":62,"dnn\u7684layer":62,"dnn\u7684layers\u90fd\u4f1a\u7ee7\u627f\u4e8e":62,"enum":[34,36,41,67,74,75,84,85,90,102],"export":[51,56,92,104,109],"final":[4,5,18,27,28,42,56,68,69,86,89,100,101],"float":[2,3,4,6,9,18,19,45,50,58,85,87,88,100,101,102,103,107,120],"function":[4,5,9,18,19,23,26,28,29,31,35,36,37,39,40,41,42,45,46,50,52,55,58,63,68,69,72,73,74,75,76,78,79,80,81,83,85,89,97,100,101,102,103,106,107,109,111,118,126],"golang\u53ef\u4ee5\u4f7f\u7528":65,"golang\u7684":65,"gpu\u7b49":82,"h\u5e76\u4e0d\u56f0\u96be":65,"images\u6570\u636e\u96c6\u4e0a\u4f20\u5230\u4e91\u7aef\u7684":33,"import":[3,4,26,29,30,40,41,44,52,54,56,57,63,68,69,75,83,86,94,95,97,101,107,109,111,112,124],"ingress\u9700\u8981\u628apfsclient\u7684\u8eab\u4efd\u4fe1\u606f\u4f20\u7ed9pfsserv":48,"instance\u4e0e\u751f\u6210\u6570\u636e\u96c6\u65f6":33,"instance\u5305\u6db5\u4e24\u4e2a\u503c":33,"instance\u662f\u4e00\u6a21\u4e00\u6837\u7684":33,"int":[2,3,4,5,9,10,11,18,19,28,29,34,35,36,39,40,41,43,57,58,61,62,63,65,66,74,75,77,78,84,85,87,89,90,100,102,103,109,120],"interface\u6587\u4ef6\u7684\u5199\u6cd5\u975e\u5e38":65,"layer\u65f6":62,"layer\u7684\u540e\u9762\u63a5\u6709cpu":62,"list\u4f5c\u4e3a\u68c0\u67e5\u5217\u8868":82,"long":[4,5,9,18,41,107],"mkl\u5e93\u7684":61,"mklml\u4ee5\u53camkl":62,"mklml\u53ef\u4ee5\u4e0emkl":62,"mklml\u7684\u5e93\u76ee\u524d\u90fd\u662f\u52a8\u6001\u5e93":62,"mode\u4e0b\u7684\u7ed3\u679c":61,"model\u505a\u5206\u652f\u7ba1\u7406":82,"ndarray\u7c7b\u578b\u7684\u503c\u548c\u6574\u578b\u7684\u503c":33,"new":[4,9,18,27,28,29,30,31,34,35,36,37,38,41,42,45,50,51,60,61,63,67,69,72,77,78,79,81,85,86,89,95,98,99,105,112,113,126],"note\u7684\u4e66\u5199":82,"null":[56,100,118],"op\u7684\u4fe1\u606f":62,"openmp\u7528\u4e8e\u63d0\u9ad8mklml\u7684\u6027\u80fd":62,"org\u76ee\u524d\u9075\u5faa":82,"packed\u4f18\u5316\u540elayer\u7684\u6d4b\u8bd5":61,"packed\u76f8\u5173\u529f\u80fd":61,"paddle\u4e00\u4e2a\u52a8\u6001\u5e93\u53ef\u4ee5\u5728\u4efb\u4f55linux\u7cfb\u7edf\u4e0a\u8fd0\u884c":65,"paddle\u5185\u5d4c\u7684python\u89e3\u91ca\u5668\u548c\u5916\u90e8\u4f7f\u7528\u7684python\u5982\u679c\u7248\u672c\u4e0d\u540c":65,"paddle\u5185\u90e8\u7684\u7c7b\u4e3ac":65,"paddle\u7684\u591a\u8bed\u8a00\u63a5\u53e3\u5b9e\u73b0\u5305\u62ec\u4e00\u4e0b\u51e0\u4e2a\u65b9\u9762":65,"paddle\u7684\u7c7b\u578b\u5168\u90e8\u9000\u5316\u6210":66,"paddle\u7684\u94fe\u63a5\u65b9\u5f0f\u6bd4\u8f83\u590d\u6742":65,"paddle\u7684c":66,"paddle\u8bad\u7ec3\u4efb\u52a1":33,"paddle\u8def\u5f84\u4e0b":66,"paddle\u9700\u8981\u4e00\u4e2a\u591a\u8bed\u8a00\u63a5\u53e3":65,"paddle\u9700\u8981\u66b4\u9732\u7684api\u5f88\u591a":66,"paddle\u9759\u6001\u5e93\u94fe\u63a5\u590d\u6742":65,"paddle_\u7c7b\u578b\u540d":66,"paddle_\u7c7b\u578b\u540d_\u51fd\u6570\u540d":66,"paddlepaddle\u4e2d\u7684cudnn\u90e8\u5206\u4f7f\u7528\u7684\u4e5f\u662f":62,"paddlepaddle\u4f7f\u7528git":82,"paddlepaddle\u5f00\u53d1\u8fc7\u7a0b\u4f7f\u7528":82,"paddlepaddle\u63d0\u4f9b\u4e13\u7528\u7684":33,"paddlepaddle\u6bcf\u6b21\u53d1\u65b0\u7684\u7248\u672c":82,"paddlepaddle\u6bcf\u6b21\u53d1\u7248\u672c\u9996\u5148\u8981\u4fdd\u8bc1paddlepaddl":82,"paddlepaddle\u7684\u4e3b\u7248\u672c\u5e93\u9075\u5faa":82,"paddlepaddle\u7684activation\u4f1a\u76f4\u63a5\u4f7f\u7528":62,"patch\u53f7":82,"patch\u53f7\u52a0\u4e00":82,"pfsclient\u9700\u8981\u548cingress\u4e4b\u95f4\u505a\u53cc\u5411\u9a8c\u8bc1":48,"pfsclient\u9700\u8981\u5728\u4f20\u8f93\u5b8c\u6bd5\u6700\u540e\u4e00\u4e2achunk\u7684\u65f6\u5019\u68c0\u67e5destination\u6587\u4ef6\u7684md5\u503c\u662f\u5426\u548csource\u6587\u4ef6\u4e00\u81f4":48,"pfsserver\u63d0\u4f9brest":48,"public":[10,18,29,50,53,58,74,80,83,85,86,87,89,100,101,102,103,109,112,113],"py\u4e2d":82,"pypi\u4e0a\u7684package\u540d\u79f0\u4e3apaddlepaddle\u548cpaddlepaddl":82,"pypi\u4e0d\u652f\u6301\u8986\u76d6\u4e0a\u4f20":82,"reader\u7684\u4f7f\u7528\u65b9\u5f0f\u90fd\u662f\u4e00\u81f4\u7684":33,"reader\u8f93\u51fa\u7684data":33,"resnet\u7684mkl":62,"return":[2,3,4,5,7,9,10,11,17,18,19,23,25,26,27,28,29,33,34,36,39,40,46,50,52,53,54,56,58,59,60,63,68,69,70,74,75,76,80,83,85,87,89,97,100,101,102,103,111,112],"rnn\u90e8\u5206\u4e2d":61,"s3\u4e4b\u7c7b\u7684\u5206\u5e03\u5f0f\u5b58\u50a8\u4e4b\u4e0a":33,"server\u4e4b\u4e0a":32,"server\u4e4b\u95f4\u7684\u7f51\u7edc\u5e26\u5bbd":32,"server\u4f1a\u6682\u505c\u53c2\u6570\u66f4\u65b0\u5e76\u7b49\u5f85":32,"server\u4f1a\u83b7\u53d6parameters\u5185\u5b58\u7684":32,"server\u5185\u5b58\u4e2d\u7684\u6a21\u578b\u6570\u636e\u7684\u5b8c\u6574\u955c\u50cf":32,"server\u540c\u6b65\u7684\u4fdd\u5b58\u4e00\u4e2a\u7279\u5b9a\u65f6\u95f4\u70b9\u7684\u5168\u5c40\u68c0\u67e5\u70b9":32,"server\u5728\u96c6\u7fa4\u4e2d\u542f\u52a8\u540e":32,"server\u6545\u969c\u540e\u88abkubernetes\u91cd\u65b0\u542f\u52a8":32,"server\u6b64\u65f6\u8fd8\u9700\u8981\u901a\u8fc7\u7f51\u7edc\u8bbf\u95ee\u5206\u5e03\u5f0f\u5b58\u50a8\u4ee5\u4fdd\u5b58\u5feb\u7167":32,"server\u751f\u6210\u4e00\u4e2auuid":32,"server\u7684\u5355\u70b9\u6216\u591a\u70b9\u540c\u65f6\u6545\u969c":32,"server\u7684\u6570\u636e\u5feb\u7167":32,"server\u7684\u68c0\u67e5\u70b9\u5404\u81ea\u72ec\u7acb\u4fdd\u5b58":32,"server\u7b2c\u4e00\u6b21\u542f\u52a8\u6216\u4efb\u610f\u65f6\u95f4paramet":32,"short":[4,5,50,54,77,83,86,89,101],"static":[25,36,66,83,85,112,123,126],"super":[77,100],"swig\u652f\u6301\u7684\u8bed\u8a00\u6216\u8005\u89e3\u91ca\u5668\u6709\u5c40\u9650":65,"swig\u66b4\u9732\u7684\u63a5\u53e3\u4fdd\u7559\u4e86c":65,"swig\u751f\u6210\u7684\u4ee3\u7801\u4e0d\u80fd\u4fdd\u8bc1\u591a\u8bed\u8a00\u4ee3\u7801\u98ce\u683c\u7684\u4e00\u81f4\u6027":65,"swig\u76f4\u63a5\u8bfb\u53d6c":65,"swig\u9700\u8981\u5199\u4e00\u4e2ainterface\u6587\u4ef6":65,"switch":[29,66,112],"tag\u4e3a":82,"tag\u53ef\u4ee5\u662flatest\u6216latest":82,"tag\u7684\u66f4\u65b0\u65f6\u95f4\u662f\u5426\u5728\u4e0a\u8ff0\u7f16\u8bd1wheel\u5305\u5b8c\u6210\u540e\u662f\u5426\u6700\u65b0":82,"throw":112,"true":[2,3,4,5,6,7,9,11,17,18,19,21,25,26,28,29,34,51,57,61,70,75,76,77,78,82,85,89,97,100,109,112,118,120],"try":[30,31,34,35,36,51,56,60,63,78,83,86,92,94,98,107,111,123],"type\u5b57\u6bb5\u5747\u4e0d\u5c3d\u76f8\u540c":66,"var":[17,18,28,29,40,45,52,53,55,57,70,75,76,77,79,83,88,89,104],"void":[29,34,36,47,50,52,53,58,63,64,65,66,75,76,84,85,87,88,100,101,102,103],"wheel\u5305":82,"while":[2,4,9,29,38,41,51,56,59,60,69,72,73,78,81,83,87,97,101,103,118],AGE:[112,113],AWS:[9,33,109,115,116],Abs:18,Added:86,And:[3,4,6,9,10,11,18,25,27,34,38,39,47,54,56,67,71,74,78,83,87,97,112,120],But:[4,5,9,18,27,53,59,67,74,83,94,99,109,126],EOS:4,For:[3,4,5,6,9,18,25,26,28,29,35,36,37,39,40,42,45,46,51,52,53,55,58,60,64,67,68,69,72,73,74,75,76,77,78,79,80,81,84,85,86,87,88,90,91,92,95,97,99,100,101,102,103,106,107,111,117,118,120,122,124,126],IDE:[98,123],IDs:[10,18,38,41,69],IPs:109,IRs:42,Into:112,Its:[3,4,52,84,97,101,112],K8s:126,NMS:4,NOT:[18,77,101],Not:[26,31,60,86,94,126],OPs:[42,44,111],One:[3,5,18,25,27,38,64,67,83,86,88,97,100,118],Ops:[79,81,85,101],PFS:48,PRs:104,QoS:113,Such:[58,77,86,89],TLS:[26,48,112],That:[4,9,71,92,118,120],The:[1,2,3,4,5,6,9,10,11,17,18,19,22,25,26,27,28,30,31,35,37,38,39,41,42,44,45,46,49,50,52,56,59,60,63,64,66,68,69,71,72,74,75,76,77,78,81,83,84,85,86,87,88,89,90,92,94,97,98,99,100,101,102,103,104,106,107,108,109,110,111,112,113,118,120,122],Their:[4,31],Then:[4,5,18,40,42,53,58,60,71,74,76,91,92,94,97,100,106,107,109,111,112,113,114],There:[3,4,10,18,25,26,29,30,31,36,38,39,41,42,49,50,51,56,60,67,68,69,72,73,74,77,81,83,84,87,101,107,112,122,123],These:[3,11,18,28,29,45,50,55,70,81,84,85,86,91,109,120],Use:[3,9,18,26,43,49,78,79,86,91,100,106,107,112,118,119,124],Used:[5,21,79,87],Uses:[60,123],Using:[18,31,51,72,78,79,81,83,93,98,103,113],VMs:98,VPS:112,WITH:124,Will:[9,25],With:[4,5,40,45,51,71,75,86,89],YES:39,Yes:[62,92,98],___fc_layer_0__:112,__align__:50,__cuda_align__:50,__device__:50,__doc__:85,__file__:39,__forceinline__:50,__fp16:50,__global__:50,__gradient_machines__:25,__hadd:50,__half:50,__half_raw:50,__impl__:85,__init__:[45,46,54,60,70,77,89,100,106],__main__:54,__metaclass__:101,__name__:54,__param_conf__:25,__rnn_step__:97,__tmp_params__:25,__va_args__:80,__x:50,_addup_repetitive_outputs_:28,_append_backward_ops_:[28,45],_append_backward_vars_:28,_binari:30,_create_global_var:77,_def:60,_dtype:56,_filer:63,_filter:63,_fwd:63,_fwd_pd:63,_input:63,_librari:30,_link:5,_live_in:60,_live_out:60,_loss:54,_op:[56,101],_output:63,_presucessor:60,_program:60,_proj:4,_remove_no_grad_branch_:28,_reorder_input:63,_source_language_embed:97,_src_input:63,_sucessor:60,_target_language_embed:97,_test:30,_update_op:46,_use:60,_value_index:56,a75:50,a_op:101,a_prev:86,aaaaa:33,aaaaaaaaaaaaa:112,abbrevi:11,abc:4,abi:122,abil:54,abl:[4,26,28,41,42,58,70,74,77,126],about:[5,11,18,29,30,39,44,49,52,60,68,78,83,85,86,87,92,94,95,101,106,107,112,117,118,122,124],abov:[3,4,18,26,28,29,30,31,35,40,41,42,50,51,52,53,55,63,68,69,70,71,72,74,75,77,85,86,88,89,92,93,95,98,99,101,102,103,106,107,111,112,113,122,126],abs:[5,27,54],abs_numerical_grad:27,absolut:[92,122,124],acc:42,acceler:[4,32,62,71,72,92,120],accept:[2,4,9,18,26,79,122],access:[4,5,18,26,30,35,38,39,40,42,77,94,97,109],accessmod:112,accessor:77,accord:[3,4,11,18,27,28,36,42,44,55,69,79,89,101,109,117,118,120],accordingli:[3,4,100],account:[79,99,126],accoust:86,accrodingli:34,accumul:[14,31,36,46,71,72,86],accur:[27,38,74],accuraci:[3,46,86,100],achiev:[44,71,72,86,87,107],ack:118,acquir:51,across:[4,9,28,42,68,86],act1:56,act2:56,act:[4,5,18,19,29,42,56,69,77,89,95,97,108,111],act_output:85,act_typ:56,actgat:18,action:112,activ:[5,18,19,24,30,56,60,69,74,77,81,85,95,97,100,103,104,118],activi:5,actnod:18,actual:[4,18,34,45,51,54,56,63,67,72,85,87,88,103],actual_layout:63,adagrad:[72,84],adagradoptim:[20,70],adam:[26,36,42,54],adamaxoptim:20,adamoptim:20,adapt:[3,6,18,25,74,88],add:[4,5,9,18,22,23,25,27,28,29,30,34,38,41,42,44,46,50,53,57,59,70,72,76,77,79,81,83,87,91,98,99,100,101,103,107,108,120,124],add_activ:77,add_bia:77,add_depend:30,add_execut:30,add_input:[68,100],add_memori:68,add_output:68,add_scalar:[29,69,75],add_sum:77,add_test:[30,100],add_two:[29,68],add_unittest_without_exec:100,addattr:[85,101],addbia:100,addcom:[85,101],added:[3,4,18,25,29,44,45,50,67,71,72,81,99,100,101,102],adding:[81,99],addinput:[85,101],addit:[4,5,28,41,71,74,79,81,89,101,103,111],addition:68,addmemori:63,addop:[53,103],addoutput:101,addprimit:63,addprimitivedesc:63,addr:31,address:[31,36,40,42,92,107,109,110,111,114,118,126],addrow:100,addtolay:4,addtyp:85,adjust:[28,45],admin:126,administr:[38,98,126],adopt:[50,54],advanc:[27,97,107,118],advantag:[27,50,51,72,78,109],adversari:[54,78],advic:107,affect:[4,18,29],affili:69,afford:35,aforement:30,after:[4,5,10,11,18,28,29,30,35,36,38,42,43,44,45,47,49,50,60,63,71,73,74,77,86,91,92,97,99,100,101,102,103,106,109,110,112,113,118,120,122,123],aftern:74,again:[26,31,72,107],against:112,age:10,agg_level:4,aggreg:[46,71,112],ago:30,ahead:86,aid:107,alex:18,alexnet_pass1:120,alexnet_pass2:120,algo:63,algorithm:[4,6,18,28,35,45,60,63,69,72,81,86,88,97],alia:[1,2,4,16,18,20,23],alias:18,align:[4,5,9],all:[2,3,4,17,18,19,25,26,28,29,30,31,34,36,38,39,40,41,42,43,45,47,49,51,54,55,56,59,60,63,64,66,67,68,69,70,71,72,74,75,77,79,85,86,87,92,97,98,100,103,107,108,109,110,111,112,113,117,118,120,122,123,124,126],all_output_nam:28,alloc:[2,36,39,60,63,87,100,103,108,120],allow:[26,36,40,42,45,51,72,81,99,100,107,112,118],allow_only_one_model_on_one_gpu:[117,118,120],allreduc:71,almost:[18,98,110],along:[4,11,18,19],alpha:[18,30,81],alreadi:[18,22,30,31,51,63,77,83,92,107,110,112,118,124],alreali:117,also:[4,5,10,18,21,26,28,29,30,34,37,41,42,50,51,53,54,55,56,59,60,67,68,69,72,73,74,75,76,77,78,81,83,85,86,87,89,90,91,92,94,97,98,100,101,104,107,109,111,113,122,126],altern:[101,106],although:[22,28,71],altogeth:126,alwai:[4,5,18,25,30,64,84,112,118],amazon:[112,113],amazonaw:112,amazonec2fullaccess:112,amazonelasticfilesystemfullaccess:112,amazonroute53domainsfullaccess:112,amazonroute53fullaccess:112,amazons3fullaccess:112,amazonvpcfullaccess:112,ambigu:[78,86],amd64:112,amd:67,amend:99,amodei:86,among:[18,112],amort:71,amount:[18,107],analys:74,analysi:[74,106,107],analyz:60,ancestor:[75,77],andd:112,andrew:60,android:[124,125],android_abi:122,android_api:122,android_arm_neon:122,android_native_api_level:122,android_standalone_toolchain:122,android_toolchain:122,ani:[4,5,9,18,21,26,30,31,36,38,39,40,41,42,47,50,51,58,60,64,69,71,72,77,78,80,81,86,88,91,98,101,103,107,111,112,114,124],annoi:110,announc:50,anoth:[4,9,18,25,26,28,29,39,41,51,52,63,69,77,83,85,87,88,98,112,118],anroid_arm_mod:122,ans:112,answer:[40,51,99,112],anthor:18,anymor:71,anyth:[9,69,78,112],anytim:54,anywai:[106,122],apach:62,apart:18,api:[10,25,26,28,30,36,37,39,40,46,48,53,54,56,68,73,74,79,82,89,90,91,94,100,101,106,107,109,111,112,121,122,123,124,126],api_shar:30,api_test:30,api_trainer_config_helpers_lay:97,apiserv:112,apivers:[112,113],app:123,appar:28,appear:[40,51,55,87],appel:60,append:[18,25,28,45,46,69,77,78,86,97,99,100,109],append_backward:[28,70,106],append_batch_s:18,append_clip_op:45,append_gradient_machin:25,append_op:[45,59,77],append_oper:77,appl:123,appleyard:107,appli:[4,18,23,54,55,71,74,83,97,100],applic:[22,40,41,50,51,52,55,77,79,99,101,106,107,109,112,113,126],applyl1:34,appoint:101,appreci:[86,99],approach:[4,42,43,44,71,72,73,81,86,122,124,126],approxim:[18,72],apt:[92,106],arbitrari:[4,42,64,103],arch:122,archetectur:86,architectur:[50,86,91,109,122,123],archiv:[10,18,65,66],area:54,arg:[3,5,15,22,28,56,70,85,101,117],arg_nam:4,argmax:18,argu:76,argument:[4,9,11,18,22,28,29,34,35,42,70,73,76,77,91,97,99,100,118,119],arithmet:50,arm64:[122,123],arm64_standalone_toolchain:122,arm:[50,122,123,124],arm_standalone_toolchain:122,armeabi:122,armv7:[50,123],armv8:50,arn:112,around:[4,38,60,77,112,126],arrai:[2,4,9,11,18,25,36,40,41,55,69,75,77,78,79,89,101],arrang:[18,89],arrari:18,array_to_lod_tensor:60,arrow:54,articl:[18,52,55,98,109,111,113,124],artifact:[82,94,112],artifici:60,arxiv:[5,18,54,86],as_row_vector:4,as_step_input:29,asap:102,asgd:72,ask:[28,31,38,98],asr:86,assgin:60,assign:[3,4,19,28,35,40,43,45,50,52,71,86,103,109,111,112,118,126],assigne:86,assignmemt:60,associ:[73,80,103],assum:[3,4,18,29,42,63,92,97,111,120],assumpt:42,ast:40,astyp:[78,101],asyc:31,async:[31,44,117],async_count:118,async_lagged_grad_discard_ratio:118,async_lagged_ratio_default:[117,118],async_lagged_ratio_min:[117,118],asynchron:[31,41,71,74,109,118],atom:43,att_seq:5,attach:5,attend:5,attended_sequ:5,attenion:5,attent:[4,5,19],attr1:4,attr2:4,attr:[2,4,5,18,29,40,56,59,63,75,76,77,85,97,101],attr_map:85,attrdesc:75,attribu:63,attribut:[4,5,18,24,28,29,44,45,59,75,77,79,83,85,89,100,101],attributemap:101,attrproto:85,attrtyp:[75,85,101],attrvalu:85,auc:[46,117],aucvalidationlay:118,audio:86,augment:86,authent:112,author:[48,86,112],auto:[29,34,43,52,63,65,76,79,83,88,89,91,99,100,101,103,107],autom:[109,112],automat:[4,18,26,28,36,42,44,45,53,70,79,85,86,91,97,99,100,101,106,110,111,112,117,118,123],avail:[31,36,44,50,51,60,94,112,126],ave:22,averag:[3,4,7,14,18,22,25,35,118],average_test_period:[117,118],avg:[107,111],avg_cost:[42,108,111],avg_loss_valu:111,avg_x:18,avgpool:4,avoid:[18,27,29,31,42,59,63,71,72,73,91,107],avx2:91,avx:[91,92],awai:51,await:113,awar:[26,40,46,52,68,77,98,106,112],awk:114,awni:86,aws:48,aws_account_id:112,awsaccountid:112,awskeymanagementservicepowerus:112,axi:[4,18],axis:4,b363:113,b8561f5c79193550d64fa47418a9e67ebdd71546186e840f88de5026b8097465:113,ba5f:112,back:[4,18,25,28,31,42,50,54,72,92,101],background:[3,4,81,86,109],background_id:[3,4],backpropag:[27,28],backward:[1,4,5,18,23,27,29,34,36,45,54,61,62,70,72,73,76,80,81,97,100,108,118,120],backward_first:97,backward_op:27,backwardactiv:100,baidu:[18,51,86,113],bake:42,balanc:[44,71,112,118],bandwidth:[50,71],bare:[111,113,126],barrier:[109,118],barrierstatset:107,basci:56,base:[3,4,7,9,10,18,26,35,45,46,50,51,58,63,67,70,71,72,74,79,80,81,87,89,97,98,100,101,106,107,108,112,118,122,123],baseactiv:5,baseerrorclipattr:45,baseev:25,baselin:86,basematrix:100,basenam:3,basepoolingtyp:[4,5],basestr:[2,3,4,5,7,25],bash:[91,92,98,109,112,113],basic:[4,25,42,56,63,74,75,79,80,86,89,100],batch:[4,5,9,11,14,18,19,25,26,29,31,33,34,41,42,46,47,51,54,57,67,68,69,71,72,86,89,99,100,111,112,113,118],batch_id:[25,54],batch_im:54,batch_images_from_tar:11,batch_label:54,batch_norm:[54,86],batch_norm_lay:5,batch_norm_typ:4,batch_read:[33,78],batch_ref:18,batch_siz:[9,18,42,54,61,69,111],batch_szi:54,batch_z:54,batchnorm:[18,54,86],batchsiz:[4,100],bazel:30,bbbbb:33,bbox:3,bcd:4,bcebo:10,bcm2708:124,bdist_wheel:82,beacus:56,beam:[4,18,97,118],beam_gen:[4,97],beam_search:[25,69,97],beam_siz:[4,18,69,97,117,118,120],becaus:[3,4,10,26,29,30,31,36,50,69,73,77,78,81,83,84,88,89,90,97,98,100,103,106,111,112,120,122],becom:[18,43,44,83,87,107],been:[4,5,18,28,30,35,41,51,98,99],befor:[4,5,18,28,31,38,41,45,49,52,55,67,72,73,74,78,81,91,92,99,101,102,106,112,122,123,126],begin:[3,4,18,22,28,34,36,46,49,55,69,71,100,109],beginiter:[25,26],beginn:97,beginpass:[25,26],begintrain:26,behavior:[18,107],behind:[51,89,111],being:[18,28,38,45,51,76,78,103,106],belong:[3,4,42,83],below:[18,29,31,36,42,44,50,51,64,73,78,81,89,90,91,94,97,100,107,109,112,122,123],benchmark:[64,86],benefit:[5,38,39,69],besid:[4,10,21,42,60,67,71,94],best:[30,63,92,98,118],besteffort:113,beta1:6,beta2:6,beta:[18,54],better:[5,30,51,60,63,69,88,112,123,126],between:[3,4,11,18,25,28,30,31,36,42,44,50,51,63,66,71,73,80,83,92,101,103,112],bgr:11,bi_gru:5,bi_lstm:5,bia:[4,5,18,69,77,97,100],bias:[4,18,100],bias_attr:[4,5,18,77,97],bias_initi:18,bias_param_attr:5,biases_:100,biasparameter_:100,biassiz:100,bidi:113,bidirect:[4,5,86,97],big:[40,44,60,107,126],bigger:[18,31],bilinear:4,bilinear_interpol:4,bilinearfwdbwd:107,bin:[92,109,112,113],binari:[3,4,9,18,30,39,42,50,52,54,64,91,93,94,98,106,107,112],bind:[40,41,50,53,83,87],bioinf:18,bit:50,bitcod:123,bla:[92,123],black:[54,123],blank:[4,18,112,123],block0:60,block1:60,block2:60,block:[4,18,28,32,34,36,40,41,42,43,44,45,46,47,51,58,60,67,68,70,87,90,100,101,103,107,118],block_expand:86,block_i:[4,18],block_id:[40,47],block_x:[4,18],blockdesc:[23,29,55,77,79],blockdescbind:58,blockingcount:43,blueprint:69,bn_bias_attr:5,bn_layer_attr:5,bn_param_attr:5,book:[10,79,86,97,104,108],book_distribut:111,bool:[2,3,4,5,6,7,9,11,17,18,25,29,50,57,59,61,62,63,76,77,84,85,89,90,100,102,118,120],boost:[67,86,87,102],boot:[4,18,97,126],boot_bia:4,boot_bias_active_typ:4,boot_lay:97,boot_stat:89,boot_with_const_id:4,bootstrapp:126,borrow:[54,89],bos_id:[4,97],both:[1,2,4,5,11,18,22,26,29,30,31,38,42,44,50,51,54,58,60,67,69,71,74,76,84,86,87,97,100,101,103,104,107,109,111,112,123],bottl:71,bottleneck:[74,107],bottom:[25,86],bound:[4,18,60],boundari:42,boundri:3,box:[4,54,107],brace:[29,55],brain:38,branch:[4,18,26,29,30,42,51,57,75,82,94,99,101,104],breadth:118,break_if:89,brief:[30,36,50,87,103],briefli:107,bring:[51,60,103],broadcast:[18,31,71,79,126],broken:99,browser:[92,104,106,112],bsd:[41,71,98],bsp:41,bucket_nam:112,buddy_alloc:99,buf:34,buf_siz:[9,42,111],buffer:[9,34,41,63,64,72,78,83,108,118],buffer_s:9,buffered_read:78,bufsiz:9,bug:[99,112],build:[4,10,17,30,39,42,55,56,60,62,72,81,82,85,86,92,94,95,99,101,106,109,111,112,115,116,118,125],build_android:122,build_dict:10,build_model:54,builder:99,buildtool:82,built:[30,40,42,50,52,60,67,71,85,86,89,91,94,98,106,122,124,126],bulk:41,bunch:[64,107,109],button:[99,104,112],c11:65,c703c041:99,c99:66,c99e:112,cach:[50,91],cacul:[5,46,109],caff:[29,51],caffe2:[29,40,41,51],calcul:[3,4,5,18,27,28,31,36,43,46,50,60,92,97,100,107,109,118,120],calcut:60,calendar:74,call:[3,4,5,9,18,22,25,26,27,28,29,34,35,36,37,39,40,41,42,45,52,54,55,60,68,69,70,74,77,79,80,83,85,87,89,95,97,98,100,101,102,103,106,107,109,111,112,118],callabl:[2,4,9,10,17],callback:[45,100],caller:[27,106,112],can:[2,3,4,5,9,10,11,17,18,19,21,22,25,26,27,28,29,30,31,34,35,38,39,40,41,42,44,45,47,50,51,52,53,54,55,56,58,59,60,63,67,68,69,70,71,72,74,75,76,77,78,79,80,81,85,87,88,89,90,91,92,93,94,95,97,98,99,100,101,102,103,104,106,107,109,110,111,112,113,114,117,118,120,122,123,124,126],cancel:38,candid:[4,18,69,86],candidate_activ:18,cannot:[79,83,88,89,100,101],cantain:56,capabl:[50,73,79],capac:[81,112],capi:[65,91],capi_prvi:66,caption:69,captur:[4,110],card:[71,109,111],care:[5,39,60,78,86,87,92,117,118,126],carefulli:[86,118],caret:25,carpedm20:54,carri:18,cast:[50,88],cast_to_op_attr:85,cat:[9,11,92,114],categor:101,categori:[4,10,31],categorig:10,categoryfil:113,caus:[31,49,94,101],caution:[112,113],cbla:61,cc_:30,cc_binari:30,cc_test:30,cclient:37,cde:4,cdn:10,cduadevicecontext:[67,87],ceil:4,ceil_mod:4,cell:[4,5,18],cell_activ:18,cell_t_prev:18,cell_valu:18,center:11,center_crop:11,cento:[91,94,95,126],central:[81,111],ceph:[9,33,113],cephf:[33,39,48],cer:86,certain:[18,59,67,70,74,83,87,102,117],certif:[26,48,112],cffi:65,cfg:[60,113],cgo:65,ch1:41,chain:[9,28,55,100],challeng:[4,31,51,57,87],chan:41,chanc:[26,50,100],chang:[4,10,30,35,39,42,51,63,73,75,78,80,82,83,86,91,97,99,100,101,103,107,109,111,112,118,122],changes:63,channel:[4,5,11,18,40,107],channel_shar:4,chapter:[68,69,86,111],chapter_data:68,chapter_out:68,charact:86,characterist:120,check:[9,28,29,30,45,63,76,79,91,94,99,101,104,109,112,118,120],check_align:9,check_attr:85,check_eq:100,check_grad:[27,101],check_l:100,check_output:101,check_sparse_distribution_batch:[117,118],check_sparse_distribution_in_pserv:[117,118],check_sparse_distribution_ratio:[117,118],check_sparse_distribution_unbalance_degre:[117,118],check_styl:99,checker:79,checkgrad:118,checkgrad_ep:118,checkmark:126,checkout:99,checkpoint:[44,76],checksum:48,child:29,china:92,chines:104,chip:51,chmod:112,choic:[18,30,51,92],choos:[18,59,91,92,93,98,102,118,122],chosen:[18,54,67],chunk:[18,35,48],chunk_ev:14,chunk_schem:[3,14,18],chunktyp:3,chw:11,circl:55,circular:41,circumst:87,claim:112,claimnam:112,clang:[50,65,99,122],clarifi:[3,18],clariti:69,classdim:18,classic:[4,60,86],classif:[4,18,55,120],classifi:[4,54],classification_error_evalu:3,classification_evalu:3,claster:112,clean:[29,30,47,73,79,91,98,99],clear:[3,22,30,69,73,83,88,124],clearer:[73,77],clearli:83,cli:112,click:[94,99,104,106,107,112],client:[18,34,37,79],clip:[2,5,118],clip_op:45,clip_op_desc:45,clipbynorm:18,clock:4,clone:[4,91,98,99,104,106,122,124],close:[78,99],cloud:[30,31,39,48,49,79,126],cloud_read:9,cludform:112,cluster:[9,25,26,29,31,36,42,86,113,117,118],cluster_test_fil:109,cluster_train:110,cluster_train_fil:109,cluster_train_v2:[110,114],cm469:112,cmake:[66,91,98,99,100,101,104,106,107,111,122,123],cmake_build_typ:[106,122,123,124],cmake_c:123,cmake_install_prefix:122,cmake_system_nam:[122,123,124],cmakelist:[30,61,62,100],cmatrix:[65,66],cmd:113,cmu:18,cname:112,cnn:[4,18,113],coars:53,code:[4,9,26,28,30,38,41,42,44,47,50,53,54,55,59,64,67,70,72,73,74,76,78,79,80,81,85,89,91,92,94,95,97,98,100,101,102,103,104,105,107,109,111,112,113],codebas:[79,99],coeff:4,collabor:31,collect:[4,10,25,74],collectbia:100,color:11,colour:10,column:[3,4,18,55,78,100,106],column_evalu:3,com:[4,5,10,18,30,54,82,91,92,98,99,104,106,108,112,113,122,124,126],combin:[3,4,5,9,25,60,70,79,83],come:[42,46,60,75,86,89],comma:[18,22,25,36,111,118],command:[9,22,30,34,39,49,91,92,94,95,98,99,100,101,104,105,106,107,110,111,112,113,114,115,116,122,123,124],commandlin:107,comment:[18,30,56,85,86,99,101],commit:[30,99],common:[11,33,81,87,97,100,117],commonli:[49,81,97,106,107,120,124],commun:[31,36,37,41,42,44,71,99,100,109,111,112],compani:51,compar:[27,30,40,79,98,100,101],comparison:[30,51,102],compat:[19,50,53,71],compil:[4,30,42,51,56,58,60,67,71,80,84,85,90,98,100,104,109],complaint:30,complet:[4,5,10,18,25,28,29,31,35,36,45,48,55,64,67,79,93,100,101,103,106,111,112,113,126],complex:[5,18,38,41,60,69,79,97,107],complic:[4,42,53,78,88,89],compon:[41,42,56,86,89,90,100,102],compos:[9,19,26,41,53,56,68,77,79],composenotalign:9,composit:53,compress:35,compris:28,compromis:98,comput:[4,5,14,18,19,22,26,27,31,41,42,44,47,50,51,52,56,60,64,67,70,71,72,74,80,83,86,87,88,90,92,97,98,99,100,101,102,106,107,108,109,111,112,120,122,123,124],computation:[4,97],computationgraph:56,con:71,concat:[54,97],concaten:[4,5,18,54,68,89],concentr:79,concept:[3,26,40,41,51,53,54,56,63,68,69,72,73,75,83,89,90,97],conceptu:[41,47,51,54,56],concern:[26,41,46,123],concis:[54,89],conckerneltrac:22,conclud:101,concret:[79,87,101],concurr:[31,38,44,74,109],concurrentremoteparameterupdat:118,cond:[18,29,51,57,75],condit:[4,18,35,42,51,57,63,86,97,113],condtion:54,conduct:107,conf:[4,110],conf_paddle_gradient_num:112,conf_paddle_n:112,conf_paddle_port:112,conf_paddle_ports_num:112,conf_paddle_ports_num_spars:112,confid:4,confidence_threshold:4,config:[2,4,22,33,49,69,100,112,113,117,118,126],config_:[34,118],config_arg:[117,118,120],config_bas:[3,4,25],config_lay:100,config_len:36,config_pars:[61,62,100],config_proto:36,configmap:42,configur:[0,4,18,25,28,34,36,38,39,42,44,51,56,59,77,86,87,88,95,96,98,99,100,101,103,107,109,111,118,122,124,126],confirm:49,conflict:[83,99],confus:[11,54,59],congest:118,conll:10,connect:[5,18,39,40,42,44,86,100,109,111,112,113,126],connectionist:[4,18],consequ:[4,5],consid:[3,4,28,76,87,98,107,120,126],consider:[4,5,67,86],consist:[3,4,10,11,18,35,41,52,64,75,78,79,80,85,86,90,101],consol:[107,112],consolid:[29,104],constant:[4,18,56,58,59,67,100,101],constantiniti:[16,18],constraint:83,construct:[3,18,26,47,56,60,68,77,79,83,85,97,102],constructbackwardgraph:55,constructoptimizationgraph:55,constructor:[18,45,50,74,77,79,83,85,100,101],consum:[31,106],consumpt:60,contact:38,contain:[3,4,5,7,9,10,11,18,25,26,28,29,35,47,54,56,63,64,67,73,74,77,79,80,83,84,85,86,89,90,93,94,97,98,101,109,111,112,114,123],container:109,containerport:112,content:[18,36,49,64,69,104,113],content_dir:104,content_len:36,context:[4,5,10,18,19,45,63,83,84,87,97,101,103,108],context_attr:5,context_len:[4,5],context_proj_layer_nam:5,context_proj_param_attr:5,context_project:5,context_start:[4,5],contin:112,continu:[3,28,31,64,86,109,118,122],contrast:[4,18],contrib:81,contribut:[81,86,98,105],contributor:79,control:[2,29,40,41,112,113,118,126],controlflowgraph:60,conv2d:[54,102],conv:[5,18,54,63,88],conv_act:5,conv_batchnorm_drop_r:5,conv_bias_attr:5,conv_filter_s:5,conv_fwd:63,conv_layer_attr:5,conv_num_filt:5,conv_op:4,conv_pad:5,conv_param_attr:5,conv_pool_2:42,conv_strid:5,conv_with_batchnorm:5,conveni:[26,28,56,70,85,86],convent:[18,28,36,99,101],converg:110,convers:[50,51],convert:[10,18,33,42,43,44,50,51,52,63,78,80,86,111],convlay:4,convlut:86,convlution2d:18,convolut:[4,5,9,18,19,54,67,77,87],convolution2d:18,convolution_algorithm_opt:63,convoper:4,convproject:4,convtranslay:4,convtransproject:4,cool:99,cooper:86,coordin:[31,36],copi:[18,25,26,35,38,49,55,68,69,71,72,89,91,99,109,112,114],copy_from:45,copyvariablewithtensor:88,core:[2,18,28,56,59,66,72,73,89,98,108,118],coreo:[112,126],corespond:18,corner:79,corpu:[10,86],correct:[4,18,27,28,50,71,100,101,102,112],correctli:[3,9,28,50,54,100],corresond:50,correspend:18,correspoind:26,correspond:[4,18,19,26,28,29,30,45,50,56,57,63,67,68,69,77,79,80,81,85,87,100,101,102,103,106,123],correspondingli:123,corss_entropi:26,cortex:50,cos:[4,85],cosin:[4,18,85],cosineop:85,cosineopproto:85,cosineopprotomak:85,cost:[18,25,26,28,42,55,70,71,75,76,88,108,111,118],cost_id:4,cost_np:76,could:[4,9,25,26,27,35,40,41,42,43,44,50,51,52,68,70,72,73,75,77,78,80,98,102,106,107,110,112,122],count:[3,18,31,39,46,76,78,86,107,109,111,113,118,120],counter:[14,22,31,35,43,55],cours:[3,39,67,98],covari:4,cover:[51,86,103],cp27:94,cp27m:[82,94],cp27mu:[82,94],cpp:[27,34,53,61,62,65,66,73,79,90,100,102,107],cprofil:106,cprofilev:106,cpu:[2,4,22,27,39,50,59,67,72,73,74,79,81,82,87,88,91,92,98,101,102,103,106,107,108,113,118],cpu_avx_mkl:[92,94],cpu_avx_openbla:[94,95],cpu_kernel:59,cpu_noavx_openbla:94,cpu_ns_:74,cpu_per_pserv:42,cpu_per_train:42,cpudevicecontext:[67,87,101,102],cpuelapsedu:74,cpuengin:62,cpuinfo:92,cpuplac:[22,42,59,63,67,87,88,101,102,103,108,111],cpusparsematrix:66,crash:[31,107,110,118],creat:[2,9,18,23,25,26,27,29,31,36,40,43,45,46,47,48,49,50,51,53,54,55,63,67,68,70,71,72,73,77,80,81,86,92,95,98,99,100,101,104,109,111,114,118,122,126],create_backward_pass:70,create_bias_paramet:100,create_block:77,create_doc_str:85,create_input_paramet:100,create_local_scop:47,create_oper:53,create_optimization_pass:70,create_paramet:77,create_python_ops_creatation_funct:85,create_rnn:29,create_rnn_op:68,create_tmp_var:77,create_tmp_vari:77,create_var:77,create_whileloop:89,creategradientoper:80,creatememori:63,createop:85,createoper:29,createprimitivedesc:63,createstack:112,createvari:29,creation:[53,112],creationd:112,creator:[9,10,33,79,80],creator_:80,credenti:49,crf:[87,88],critic:[54,106],crlf:99,crop:[11,87],crop_grad:87,crop_siz:11,crope:11,cropgradkernel:87,cropkernel:87,cross:[4,18,77,101],cross_compil:124,cross_entropi:[4,26,42,54,60,88],cross_entropy_with_selfnorm:4,crt:48,csc:100,csr:100,csv:22,ctc:[3,18],ctc_error_evalu:86,ctc_evalu:3,ctest:[91,98,101],ctor:77,ctrl:[98,110],ctx:[63,88,101,103],cubla:[67,102],cublas_handle_:87,cublashandle_t:87,cuda7:[94,95],cuda8:[91,92,94],cuda:[22,30,52,67,74,79,87,92,94,98,101,102,107,109,111,118],cuda_context:52,cuda_dir:[117,118],cuda_fp16:50,cuda_so:92,cudaconfigurecal:107,cudadevicecontext:[52,67,87,101],cudadevicegetattribut:107,cudaelapsedu:74,cudaevent_t:74,cudaeventcr:107,cudaeventcreatewithflag:107,cudafre:107,cudagetdevic:107,cudagetdevicecount:107,cudagetdeviceproperti:107,cudagetlasterror:107,cudahostalloc:107,cudalaunch:107,cudamalloc:107,cudamemcpi:107,cudaplac:[22,67,87,88,102],cudaprofilerstart:107,cudaprofilerstop:107,cudaruntimegetvers:107,cudasetdevic:107,cudasetupargu:107,cudastream_t:87,cudastreamcr:107,cudastreamcreatewithflag:107,cudastreamsynchron:107,cudeviceget:107,cudevicegetattribut:107,cudevicegetcount:107,cudevicegetnam:107,cudevicetotalmem:107,cudnn:[4,7,18,30,59,63,67,87,88,102,118],cudnn_batch_norm:4,cudnn_conv:4,cudnn_conv_workspace_limit_in_mb:[117,118],cudnn_convt:4,cudnn_dir:[117,118],cudnn_kernel:59,cudnnavginclpadpool:4,cudnnavgpool:4,cudnnconvopkernel:102,cudnnv5:91,cudrivergetvers:107,cuinit:107,cumtim:106,cumul:4,cur_mem:69,curl:112,curli:[29,55],current:[4,18,28,29,30,31,34,36,40,44,46,51,59,67,68,69,72,73,74,77,83,88,89,92,94,97,98,100,104,110,111,112,118,123],current_block:[75,77],current_endpoint:111,current_oper:75,current_word:97,curv:26,custom:[26,39,50,54,69,72,79,86,100,112],custom_batch_read:78,custom_neg_class:18,cut:[9,89],cut_lin:9,cutoff:10,cv2:11,cxx:123,cxx_compil:[122,123,124],cxx_flag:123,cxxabi_1:94,cycl:31,cyclic:4,cython:65,d3e0:112,d_b0:54,d_b1:54,d_b2:54,d_block:54,d_f:54,d_g:54,d_h0:54,d_h0_bn:54,d_h0_relu:54,d_h1:54,d_h1_bn:54,d_h1_relu:54,d_h2:54,d_loss:54,d_loss_fak:54,d_loss_real:54,d_optim:54,d_step:54,d_t:54,d_w0:54,d_w1:54,d_w2:54,daili:99,dandroid_abi:122,dandroid_arm_mod:122,dandroid_arm_neon:122,dandroid_standalone_toolchain:122,dangl:98,dario:86,darwin:112,dash:54,dat:33,data:[0,3,10,11,17,25,26,27,29,33,34,35,41,44,46,48,50,51,54,55,56,58,59,60,63,64,67,68,69,70,71,72,73,75,77,79,81,83,84,85,86,87,89,90,95,97,100,101,103,107,108,109,111,114,115,117,118,120],data_feed:12,data_fil:11,data_i:54,data_lay:34,data_layout:18,data_layout_:88,data_read:[9,78],data_reader_creator_random_imag:78,data_shar:89,data_typ:[9,10,64,84,86,88,90,95,97,102],data_type_:[59,67,88],data_x:54,databas:10,datacent:[33,49],datacenter1:33,datacenter2:33,datacenter_1:33,datacenter_2:33,datacenter_nam:33,datadim:4,datafeed:[108,111],dataflow:56,dataflow_analysi:60,datalayout:88,dataparallel:42,dataprovider_convert:86,datasci:4,dataset:[0,18,33,39,42,72,78,86,95,97,106,111,118],dataset_nam:11,datatransform:88,datatyp:[10,18,59,63,84,86,88,90],date:109,dcgan:54,dcmake_install_prefix:[122,123,124],dcmake_system_nam:[122,123,124],dcuda_arch_nam:91,dcudnn_root:91,ddim:[67,87,103],dead:31,deal:[28,126],debug:[27,28,42,49,51,77,92,99,106],debug_str:56,decai:[6,23],decar:9,decayedadagradoptim:20,decayr:34,decent:35,decid:[26,38,54,64,72,80,81,84],declar:[18,29,54,68],decod:[4,5,18,86,97],decoder_boot:97,decoder_dim:69,decoder_group_nam:97,decoder_input:[69,97],decoder_mem:[69,97],decoder_prev:5,decoder_s:97,decoder_st:[5,97],deconv:[4,54],deconvolut:4,decor:[9,100],decrement:43,decrementcount:43,decrypt:112,deduc:79,deep:[4,18,28,38,41,47,54,55,60,62,74,79,81,86,87,92,107,111,123],deeper:[52,92],deepspeech2:61,def:[4,9,26,27,28,33,39,45,46,53,54,56,59,60,68,69,70,77,78,85,89,97,100,101],def_block:54,defalut:[18,22,118,120],default_block:54,default_devic:120,default_initi:18,default_main_program:[17,108,111],default_param_attr:77,default_st:89,default_startup_program:[108,111],default_valu:120,defaultdict:60,defaultinfervartyp:58,defect:73,defer:38,defin:[4,5,9,25,26,28,29,30,31,38,40,43,44,45,50,51,52,53,54,56,59,60,67,68,71,75,77,78,79,83,85,87,89,97,100,103,106,108,109,110,111,118],definit:[28,29,31,35,42,47,52,59,75,80,85,89,92,101,106,108],definiton:53,degener:18,degre:4,deivc:102,delai:[72,87,103,118],delet:[18,39,48,99],deletestack:112,delimit:3,deliv:126,delta:[4,27],delv:4,demand:[31,87],demo:[4,10,79,110,113,115],demolish:113,demonstr:[97,103],denot:[18,101,120],dens:[4,9,36,37,84,86,100,112],dense_arrai:9,dense_vector:[9,95],dense_vector_sequ:9,densescann:86,dep:30,depart:86,depend:[18,29,30,31,39,42,44,56,71,76,84,92,98,101,109,120,122,123,124,126],dependent_var:76,deploi:[4,110,120,126],deploy:[56,64,79,109,110,112,123,126],deprec:[4,86],depth:[29,51,86],dequeu:44,deriv:[1,26,42,45,57,70,122],desc:[29,45,63,64,77,85,89],desc_:29,descend:[18,89],descent:[4,31,72,109],descproto:64,describ:[17,26,28,29,30,35,40,42,47,52,59,63,64,68,69,73,75,77,79,84,85,88,90,100,101,102,103,112,113],describestack:112,describestackev:112,describestackresourc:112,descripotor:63,descript:[3,29,30,58,62,64,67,80,84,86,88,90,91,94,99,101,109,112,119],descriptor:[41,63,88],deseri:[25,64,73],deserializ:79,desgin:55,design:[4,9,18,28,34,59,60,65,72,74,81,101,126],desir:[9,31,42,72,112,113],destin:[18,36,49],destroi:[29,47],destruct:83,destructor:[74,100],det_output:3,detail:[2,3,4,5,6,18,27,28,35,39,42,44,49,51,54,56,60,63,64,67,68,74,77,81,83,87,88,89,90,91,95,97,98,100,101,102,103,104,106,107,110,111,112,113,119,120,124,126],detect:[18,58,91,99,122],detection_evalu:3,detection_output:3,determin:[4,9,18,29,42,60,67,79,100],dev:[79,92,98,106,122,126],dev_ctx:[29,63,74],devel:82,develop:[28,30,51,58,73,74,77,80,82,86,92,94,99,103,104,106,108,109,117,118,123,124],deverlop:118,deviat:2,devic:[2,40,42,46,50,56,62,63,67,71,73,74,79,88,92,101,103,108,118,123],device_:74,device_context:[63,101],device_count:18,device_typ:18,devicecontext:[29,67,74,101,102],deviceid:[62,120],deviceid_:62,deviceplac:87,devid:[4,118],devot:86,devtools2:91,dhcp:126,diagnos:110,diagon:18,diagram:[68,109],diamond:54,dic:11,dict:[3,10,25,28,77,109,114],dict_fil:3,dict_siz:[10,18,34,69],dictionari:[3,4,10,18,25,26,27,77,120],did:[73,92],diff_mat:27,differ:[3,4,18,22,25,28,29,30,31,36,38,42,43,44,45,46,47,50,51,54,56,57,60,63,67,69,71,72,74,76,80,83,86,88,89,90,92,97,100,101,102,103,106,110,112,113,118,123],differenti:53,difficult:[3,27,51,98],dig:[92,107,112],digit:[4,109],digraph:56,dilat:[4,18,63],dilation_h:18,dilation_i:4,dilation_w:18,dim0:101,dim1:101,dim:[4,9,18,19,21,34,63,64,68,79,84,87,90,100,101,103],dim_:[87,103],dimens:[1,4,5,7,9,18,19,21,54,79,84,86,87,89,100,101,103,120],dimension:[4,18,97,100,103],dimes:4,dios_arch:123,dios_enable_bitcod:123,dios_platform:123,dios_use_veclib_for_bla:123,dir:122,direcit:86,direct:[4,5,11,18,51,60,72,86,106],directli:[5,30,37,39,42,50,59,73,85,88,89,91,93,110,113,122],director:101,directori:[4,17,30,33,38,48,49,87,91,92,98,103,104,107,109,110,113,114,118,122,123,124],dirnam:17,disabl:[18,74],disadvantag:[72,77],discard:[9,31,35,69,118],discount:4,discov:31,discoveri:112,discrep:107,discret:4,discrim:54,discuss:[26,29,35,36,37,42,63,86],disk:[64,98,113],dispatch:[42,73,109,110,118],dispens:18,displac:18,displai:[39,49,99],dissimilar:18,dist:[82,91],dist_train:[26,39],distanc:[3,4,18],distinguish:[30,110],distribut:[4,18,29,35,36,37,38,40,41,46,52,71,79,86,90,94,105,110,113,115,116,126],distribute_test:[117,118],distributedli:[42,100],distributetranspil:111,disucss:26,div:18,divid:[4,6,18,28,46,85,90,106,117],divisor:18,diy_beam_search_prob_so:[117,118],django:104,dnn:[63,86,91],dns:112,do_forward_backward:78,doc:[9,56,68,89,101,103,104,109],doc_cn:104,dockefil:98,docker:[82,91,93,99,104,109,112,115,116,126],docker_build:26,docker_clust:[110,114],docker_push:26,dockerfil:[98,122,124],dockerhub:92,document:[4,5,18,27,42,48,55,68,69,74,79,86,91,98,99,101,102,103,105,109,111,120,123],doe:[5,18,31,35,36,38,39,40,42,44,47,50,56,60,68,73,77,79,80,81,94,98,100,101,103,107,108],doesn:[2,4,9,26,29,40,41,78,92,98,99,106,107,113,122,123],doing:[34,38,42,55,107],domain:112,don:[5,26,30,53,55,60,78,86,91,92,98,99,101,104,112],done:[3,4,5,28,30,31,35,36,42,43,58,60,64,72,80,81,86,99,106,107,112],dot:[4,5,19,101,118],dot_period:[118,120],dotmuloper:4,dotmulproject:4,doubl:[18,42,50,55,74,88,91,101,102,118],down:[86,107],download:[10,30,31,34,38,48,91,92,94,109,123,126],dozen:30,draw:69,drive:83,driver:[92,109,111],drop:[4,5,18,19,69],drop_rat:2,drope:18,dropout:[2,5,19,100],dropout_prob:18,dropout_r:[4,18,19],drpi_arm_neon:124,drpi_toolchain:124,drwxr:113,ds2:86,dst:[36,63],dst_primitive_desc:63,dtoh:107,dtype:[18,41,42,56,77,108,111],due:[35,38,54,60,69,77,106],dummi:[25,35],dump:64,duplic:[18,44],durat:[35,107],dure:[4,5,17,18,28,29,31,35,38,39,46,51,60,71,72,74,77,79,86,90,100,101,112,117,118,126],duse_eigen_for_bla:122,dwith_c_api:[66,122,123,124],dwith_distribut:111,dwith_doc:111,dwith_gpu:[91,111,124],dwith_profil:107,dwith_python:[66,111,124],dwith_swig_pi:[66,111,122,123,124],dwith_test:[91,101,123],dwith_tim:107,dynam:[18,36,66,68,77,78,91,107,118],dynamic_cast:100,dynamic_recurrent_op:89,dyogatam:18,e2e:126,each:[3,4,5,7,9,10,18,19,25,27,28,30,31,34,35,36,38,39,40,41,42,45,46,47,52,55,58,60,63,67,68,69,71,73,74,76,77,78,79,80,83,84,85,86,87,88,89,90,97,100,102,106,109,110,111,112,118,120,126],each_feature_vector:1,each_time_step_output:1,eager:51,earli:[50,52,99,101],eas:[9,58,101],easi:[27,28,69,72,78,79,81,99,100,110],easier:[26,44,50,51,78,89,98,99,100],easili:[26,54,71,74,78,80,83,87],echo:92,edg:[11,60],edit:[3,18,41,92,98,112],editdist:18,editor:[77,98],edu:[10,18,112,113],eeoi3ezpr86c:112,effect:[4,18,25,91,112,118,123],effici:[4,42,64,78,86,87,97,98,100],effort:[42,86],efg:4,efs:112,efs_dns_nam:112,efsvol:112,egd:60,eigen:[50,67,72,79,81,87,101,122,123],eigen_device_:87,eigen_test:103,eigen_use_gpu:101,eigenmatrix:103,eigenscalar:103,eigentensor:103,eigenvector:103,either:[4,5,9,18,25,26,42,54,57,58,68,72,81,93,107,123],elabor:86,elb:112,elbapis:112,electr:60,electron:113,elem_dim:4,elememt:4,element:[3,4,5,9,11,18,25,27,35,41,44,56,69,79,101,103],element_typ:[36,102],elementari:79,elementwis:[18,19],elif:[26,85],els:[26,34,39,41,42,44,45,51,54,57,58,59,60,83,85,91,92,98,100,101],elsewher:74,emac:98,email:99,emailweixu:30,emb1:34,emb2:34,emb:113,embed:[26,29,34,44,58,69,84,89,97,109],embedding_lay:34,embedding_nam:[4,97],embedding_s:[4,97],emphas:107,empir:[4,18],emplace_back:100,emploi:[28,45,85,97],empti:[3,9,18,28,31,69,101],emul:50,enabl:[2,4,18,29,30,35,40,44,45,56,74,98,99,107,109,112,118,123],enable_grad_shar:[117,118],enable_parallel_vector:118,enableonstart:22,enc_proj:[5,97],enc_seq:5,enc_vec:97,encapsul:36,encod:[5,35,69,97],encoded_proj:[5,97],encoded_sequ:[5,97],encoded_vector:97,encoder_ctx:69,encoder_ctx_expand:69,encoder_ctx_proj:69,encoder_dim:69,encoder_last:4,encoder_out_seq:69,encoder_s:97,encount:[18,34],encourag:[42,47],encrypt:112,encrypt_decrypt:112,end2end:126,end:[3,4,18,22,25,28,29,42,45,52,56,60,69,73,74,78,83,86,94,97,98,99,118],end_id:18,end_pass:26,end_po:4,endforwardbackward:25,endian:64,endif:[67,74],enditer:[25,26],endpass:[25,26],endpoint:[9,18,33,112],endtrain:26,enforc:123,engin:[39,62,63,86,107],english:[4,86,104],enjoi:92,enough:[28,29,59,60,67,98],enqueu:44,ensembl:5,ensur:[31,63,71,83,92,94,98,100,123],enter:[29,47],enterpris:79,entir:[4,5,36,38,101],entiti:[3,29,83],entranc:47,entri:[9,18,35,39,58,98,99,100,112,122],entropi:[4,18,77],entry_point:39,enumer:[1,67],env:[104,106,112],environ:[26,42,91,94,98,99,106,107,109,112,113,117,118,123],environmenterror:109,eos_id:[4,97],epoch:54,epol:41,epsilon:[4,6,18],equal:[4,5,18,19,31,89,101,102,118],equat:[3,4,5,6,18,60,101],equip:97,equival:[26,29,40,45,51,57,85,126],erlang:41,error:[2,3,4,5,18,26,27,35,49,50,51,63,83,86,100,101,110,112,118],error_clip:45,error_clip_callback:45,error_clipping_threshold:2,errorclipbyvalu:45,espeaci:18,especi:[4,5,17,62,98],essenc:[26,28],essenti:[4,26,47,50,103],establish:40,estim:[4,18,26,44,72],eta:113,etal:18,etc:[3,9,18,29,41,42,46,63,71,72,78,83,86,91,102,109,112,117,120,126],etcd:[9,25,31,35,36,38],etcd_addr:36,etcd_endpoint:9,eth0:112,etyp:41,euclidean:4,eval:[3,29,46,54,79],eval_program:46,eval_result:46,evalu:[4,12,18,24,25,38,56,76,86,107,108],evaluate_difficult:3,even:[26,50,71,77,78,98,99,107,118,123],evenli:[36,112],event:[22,113],event_:74,event_block:74,event_handl:[25,26],eventkind:74,eventlist:74,eventu:[42,89],everi:[3,4,5,9,18,26,31,35,36,38,45,46,55,56,58,60,63,67,68,71,77,83,85,88,97,99,100,101,103,108,109,114,118],everyon:99,everyth:[42,44,54,122],everywher:98,evid:73,evolv:51,exactli:[4,5,112],exampl:[3,4,5,9,10,11,18,19,25,29,39,42,44,46,49,51,52,53,54,55,56,58,60,63,67,68,69,73,74,75,77,78,79,80,81,84,87,88,89,97,98,99,100,101,102,103,106,107,108,109,111,112,113,117,118,120,124],example_read:9,exceed:4,except:[4,10,21,23,38,40,51,55,74,86,89,120],excess:60,exchang:73,exclud:4,exclude_mod:4,exclude_param:25,excluded_chunk_typ:[3,14,18],exclus:18,exconv:4,exconvt:4,exdb:10,exe:[42,108,111],exec:118,execut:[4,18,22,30,31,35,39,40,41,42,46,47,52,54,56,60,63,71,74,80,90,98,100,106,107,112],executioncontext:[63,88,101,102,103],executor:[12,17,40,42,46,50,51,52,54,70,75,77,88,90,106,108,111],exist:[26,29,31,49,51,69,77,78,80,85,87,89,94,98,100,103,112,118,123],exit:[36,49,111,113,118],expand:[18,69,92,100],expand_a:4,expand_level:4,expandconvlay:4,expans:4,expect:[4,88,107],expected_desc:63,expected_kernel_kei:88,experi:[64,86,120],experienc:99,expert:30,expir:31,explain:[3,18,31,40,51,53,55,99,102,106,109,111],explan:[4,18,27,39,40,42,83,88],explicit:[74,89,100,102],explicitli:[26,42,47,101,103,123],explod:45,explor:[4,69,81],expon:4,exponenti:[1,18],expos:[28,37,41,63,64,87,89,112],express:[26,44,46,56,60,101,112],extend:[3,72,89],extens:[38,44,69,101,122],extent:66,extern:[30,62,65,66,79,86],external_librari:30,extingrad_:62,extinval_:62,extoutgrad_:62,extoutval_:62,extra:[2,4,5,21,42,81,87,126],extra_lay:25,extraattr:[2,120],extraattribut:4,extraattributenon:4,extract:[3,4,18,51,73,86,101,112],extralayerattribut:[2,5],extralayeroutput:5,extrem:[4,40,51,107],f120da72:113,f7e3:112,fa0wx:113,fabric:109,face:[30,81],fact:[18,40,51,71,75,77],factor:[2,6,18],factor_s:4,factori:65,fail:[31,35,69,101,113,118,120],failur:[31,36,101],fake:54,fake_imag:78,faked_imag:54,fall:[50,76],falloc:48,fals:[2,3,4,5,6,9,18,27,28,29,51,57,59,61,68,75,76,78,84,90,95,97,100,101,109,113,118,120],false_block:[29,57,75],false_label:78,false_neg:46,false_posit:46,false_read:78,fan_in:18,far:[45,89],fashion:42,fast:[4,35,51,107],faster:[4,5,18,31,51,92,97,107],fastest:51,fastli:99,fat:123,father:28,fault:[25,35,79,91],favorit:98,fbd1f2bb71f4:113,fc1:[56,100,120],fc1_bia:56,fc1_weight:56,fc2:[56,120],fc3:[56,120],fc4:120,fc8a365:112,fc8a:112,fc_act:5,fc_attr:5,fc_bias_attr:5,fc_layer:[77,85,100,120],fc_layer_nam:5,fc_mat:25,fc_op:85,fc_out:[18,29],fc_output:85,fc_param_attr:5,fc_without_b:29,fclayer:100,fcop:53,feasibl:72,featur:[1,4,9,10,18,28,42,50,56,71,74,86,99,118],feed:[5,17,25,26,42,55,68,81,108,111],feed_dict:54,feed_list:[13,108,111],feed_target_nam:17,feeded_var_nam:17,feeder:[9,42,108,111],feel:99,fetch:[10,31,34,42,76,97,100,108],fetch_list:[42,77,108,111],fetch_op:76,fetch_target:17,few:[30,31,41,42,60,72,78,84,86,98],fewer:[4,18,41,77],fft:86,fg0:4,field1:25,field2:25,field:[4,21,25,29,56,58,64,76,77,80,84,85,107,112],fifth:55,figur:[26,30,42,44,54,62,68,74,77,86,97,100,107],file:[3,4,9,10,11,22,25,26,28,30,31,33,35,36,38,39,41,48,49,51,52,56,64,66,78,79,86,87,90,92,94,95,97,98,99,100,101,102,103,108,109,110,111,114,118,122,123,124,126],file_typ:9,filelist:86,filenam:[11,33,77,106],fileoffset:48,filesystem:[38,39,42,48,112],fill:[4,31,35,67,77,112],fill_zero_grad:79,fill_zeros_like_op:28,filter:[4,5,17,18,45,63],filter_s:[4,5,18,19],filter_size_h:18,filter_size_i:4,filter_size_w:18,filter_strid:18,find:[4,18,29,31,38,41,50,56,63,69,83,88,91,94,107,111,114,122,123],find_var:27,findmemori:63,findop:29,findprimit:63,findprimitivedesc:63,findvar:[29,83],fine:[2,35,53],fingerprint:112,finish:[18,31,35,38,39,47,60,71,85,91,109,110,112,113],finit:100,finnal:92,first:[4,18,22,25,26,28,29,31,35,38,39,40,42,47,49,51,54,55,56,63,68,69,75,76,77,79,84,85,86,87,89,91,97,98,99,100,101,103,107,111,112,118,120,126],first_n:18,first_seq:97,firstli:[3,4,102],firstn:9,firstseen:113,fit:[10,17,50,59,60,64,69,79,111],fit_a_lin:111,five:[75,107],fix:[2,4,18,42,60,65,77,86,99,106],flag:[4,10,18,22,61,62,74,99,101,104,118],flatten0:56,flatten:[18,56,75,77,103],flexibl:[4,5,22,26,36,42,51,55,59,68,69,72,78,87,89,97],flip:11,flist:109,fliud:40,float16:41,float16_t:50,float32:[9,18,42,50,53,54,77,78,101,108,111],float64:18,float_16:18,float_to_half_rn:50,floor:4,flow:[18,29,40,41,68,74,82],fluid:[0,13,14,15,16,17,18,19,20,21,22,23,28,42,44,47,67,74,77,87,88,102,106],fluid_cuda_create_tensor:52,fluid_cuda_mult:52,fluid_cuda_read:52,fly:28,fnt03:112,focu:[41,56,106,107],focus:101,folder:[30,33,39,49,111,112],follew:18,follow:[3,4,5,6,9,11,18,19,25,26,27,28,29,30,31,35,39,40,41,42,44,47,50,51,52,53,54,55,56,57,58,60,63,67,68,69,71,72,74,75,76,77,78,79,80,81,83,84,85,86,87,88,89,91,92,94,95,97,98,99,100,101,103,104,106,107,108,111,112,113,114,115,116,120,122,123,124,126],footprint:52,forbid:26,forc:[71,77,88],force_cpu:[18,59],force_cudnn:59,force_load:65,forest:29,forget:[6,18,26],forget_bia:18,fork:[4,99],form:[4,5,18,41,46,94,107],formal:88,format:[3,9,10,11,18,22,25,27,35,42,50,51,67,69,86,89,95,99,100,101,103,109,112,118],former:[26,30,51,60,72],formul:18,formula:[4,5,6,18,27,60],formular:4,forth:54,forward:[1,4,5,18,27,28,29,34,36,45,51,54,61,62,63,64,70,73,75,78,79,80,81,84,97,100,120],forward_infer:63,forward_list:74,forward_op:27,forward_proj:18,forward_train:63,forwardactiv:100,forwardbackward:25,found:[50,75,81,83,97,102,109,111,124],four:[3,18,46,51,55,63,67],fourth:18,foward:76,fp16:[50,79,90],fp32:[67,79,88,90],fp64:[67,90],fpga:[67,108],fpgadevicecontext:87,fpgaengin:62,fpgaplac:[67,87],frac:18,frame:[3,47,79,86,89],framework:[26,28,29,41,45,46,50,51,56,67,71,72,74,75,79,81,83,85,87,99,100,101,106,108,109,123],free:[10,52,87,99,126],freememoryop:52,frequenc:[10,86,107],frequent:[35,78,79,81,87,110,122,123],fresh:38,friend:83,friendli:54,from:[3,4,5,9,10,11,14,17,18,22,25,27,28,29,30,31,33,34,35,36,40,41,42,44,45,46,49,50,51,53,54,55,56,57,59,60,63,68,69,70,71,73,75,77,78,79,80,83,86,87,88,89,92,94,97,98,99,100,101,102,103,106,107,109,111,112,113,118,120,122,123,126],from_no_sequ:4,from_sequ:4,from_tar:25,fromfil:78,front:[56,60],fuction:22,fulfil:107,full:[4,18,31,38,68,71,72,97,100,102,126],full_matrix_project:[5,97],fulli:[18,42,44,86,100,107,111,126],fullmatrixproject:4,fullsiz:34,fully_matrix_project:5,fullyconnect:[56,77],fullyconnectedlay:100,func:[9,35,40,52,80],funciton:[5,18],functor:[53,56],fundament:[41,44,50,79],funtion:18,further:[4,85,126],furthermor:18,futur:[4,18,38,42,50,60,68,79,122],future_context_s:18,fvs:85,fwd_desc:63,fwd_op:80,fwd_primit:63,fwd_primitive_desc:63,fwd_var:45,g_b0:54,g_b1:54,g_b2:54,g_block:54,g_command_config_arg:[61,62],g_h0:54,g_h0_bn:54,g_h0_relu:54,g_h1:54,g_h1_bn:54,g_h1_relu:54,g_h2:54,g_im:54,g_loss:54,g_optim:54,g_program:77,g_state:74,g_step:54,g_w0:54,g_w1:54,g_w2:54,gain:4,gan:26,gangliao:30,gap:118,gate:[4,5,18,19],gate_act:[4,5],gate_activ:18,gate_attr:4,gate_bias_attr:4,gate_param_attr:4,gate_recurr:4,gate_v:18,gatedrecurrentlay:61,gather:[4,18,28,60,71,73,100,101],gauss:2,gaussian_normal_random:54,gcc:[50,52,65,79,91,98,106,122,124],gcc_3:94,gcreators_:85,gemm:61,gemmconvkernel:102,gen:4,gender:10,gendrated_id:69,gener:[3,4,5,9,25,26,27,28,29,30,31,33,35,36,38,40,42,51,53,58,60,63,67,71,72,75,76,77,78,79,80,84,85,86,87,89,99,101,107,112,114,118,120,122,124],generated_id:69,generated_scor:69,generated_word_embed:4,generatedinput:[4,97],gereat:3,get:[3,4,10,17,18,25,27,28,29,30,31,35,36,38,39,48,51,54,56,59,60,61,62,63,67,68,69,74,77,79,80,83,85,88,89,91,92,94,97,99,100,101,102,106,107,109,110,112,114,121],get_all_op_proto:85,get_block:77,get_config_arg:120,get_data:113,get_dict:10,get_dim:27,get_embed:10,get_float_el:27,get_grad:25,get_grad_op_desc:28,get_input_lay:100,get_lin:9,get_movie_title_dict:10,get_numeric_gradi:27,get_numerical_gradi:27,get_output:27,get_program:60,get_pserver_program:111,get_shap:25,get_startup_program:111,get_support:94,get_symbol:56,get_tensor:27,get_trainer_program:111,get_var:18,get_vari:29,get_word_dict:10,get_worker_addr:40,getactualkerneltyp:59,getattr:45,getbatchs:100,getdeviceid:102,geteigendevic:103,getengin:63,getenv:[26,39,109],getexpectedkerneltyp:[59,63,88],getinfervartyp:58,getinput:100,getinputgrad:100,getinputvalu:100,getkerneltyp:50,getkerneltypeforvar:88,getlayeroutput:25,getlibrari:63,getmat:34,getoptconfig:34,getoutputgrad:100,getoutputvalu:100,getparam:34,getparameterconfig:34,getparameterptr:100,getparameterspars:34,getparametersremot:34,getplac:[63,87,101,102,103],getsiz:100,gettask:35,gettempl:112,gettensor:88,gettranspos:100,getw:100,getweight:100,getwgrad:100,gist:5,git:[82,91,98,99,104,122,124],github:[5,18,30,54,67,82,91,98,99,104,106,108,122,124],give:[18,31,68,77,79,88,98,99,100,107,112],given:[4,9,18,19,25,28,36,38,41,44,45,51,53,54,69,78,81,89,100,118],glibc:[94,122,124],glibc_2:94,glibcxx_3:94,glide:30,global:[2,18,23,26,29,30,31,52,56,59,73,74,79,83,85,87,88,98,107,112,118],global_block:77,global_learning_r:2,global_pool:18,globalstat:107,globalstatinfo:107,glog:99,glog_v:99,glog_vmodul:99,gnueabihf:124,go_librari:30,go_test:30,goal:[41,44,50,55,71,79,86,107],gob:35,godep:30,godoc:65,goe:[5,31,51,57,83,108],going:[28,53,72,106,109,126],golang:30,good:[41,54,72,77,78,81,106,107,111,126],googl:[18,26,74,79,99,106,109,122],googleapi:112,googlenet:62,goroutin:[40,41],got:[59,83],gpg2:112,gpg:112,gprotos_:85,gpu:[2,4,7,22,27,39,41,46,50,60,67,71,72,73,74,79,81,82,87,88,91,94,95,98,102,103,105,108,109,111,126],gpu_id:[118,120],gpu_per_train:42,gpudevic:87,gpugpu_id:117,gpukernel:79,gpustarttimestamp:22,grab:31,grad:[27,28,36,45,62,77,84,118],grad_info_map:28,grad_n:45,grad_nam:45,grad_op:45,grad_op_class:79,grad_op_desc:45,grad_op_maker_:80,grad_op_typ:[79,80],grad_op_type_:80,grad_s_block:28,grad_share_block_num:[117,118],grad_to_var:[28,45],grad_var_nam:27,gradient:[2,3,4,6,18,23,25,31,35,41,43,45,55,58,70,71,72,73,77,79,84,101,106,109,111,118],gradient_clip:21,gradient_clipping_threshold:2,gradient_evalu:3,gradient_flat:27,gradient_machin:[25,66],gradientmachin:[25,66,73],gradientmachine_:34,gradopdescmak:[58,80],gradopdescmakerbas:80,gradopmak:80,gradual:107,grai:11,grain:53,gram:86,grandient:25,grant:112,graph:[4,18,25,28,29,30,31,40,41,42,43,44,46,51,54,68,71,72,75,103],great:[44,86,126],greater:[4,18,45,72,109],greaterthan:85,greedi:[18,86],green:[40,54],grep:[92,114],gridsize3d:22,groudtruth:97,ground:[3,4,18],group:[5,18,35,56,63,87,101,126],group_input1:97,group_input2:97,group_input:97,grouplen:10,grow:99,grpc:126,gru:[4,18,69,86,97],gru_bias_attr:5,gru_decod:97,gru_decoder_with_attent:97,gru_layer_attr:5,gru_memori:5,gru_out:69,gru_param_attr:5,gru_step:[69,97],gru_step_lay:5,grumemori:[5,97],gserver:[4,61,62,100],gsizex:107,gtx:60,guarante:[63,77,100],guard:34,guest:[94,98],gui:[106,107],guid:[22,48,60,79,97,99,100,107,111,112,113,123],gutmann10a:18,gzip:[35,113],h0_bn:54,h_0:18,h_f:18,h_prev:29,had:98,hadoop:26,half:[4,18,19,50,112],half_to_float:50,hand:[60,79,86,87,103,109,111],handi:30,handl:[9,26,28,39,40,42,56,60,63,67,73,78,83,87,89,102,108],handler:[25,29],hannun:86,happen:[18,35,85],hard:[42,51,69,86,89,98,112],hardshrink:18,hardsigmoid:18,hardwar:[51,52,87,98,102,107],has:[3,4,5,10,18,19,22,26,27,28,29,30,31,35,36,38,41,42,44,45,46,50,51,54,56,60,64,67,69,71,74,75,79,84,85,87,88,91,97,98,99,100,107,108,112,113,123,126],has_kei:[25,28,45],has_selected_colum:4,has_var_recurs:28,hasdependentvar:76,hash:[67,71],hasn:51,have:[4,5,9,18,19,22,26,27,28,29,30,31,35,36,38,39,41,42,44,45,47,50,51,52,53,54,55,59,60,63,64,67,68,69,71,72,73,74,75,77,78,79,80,83,84,86,87,88,90,91,92,97,98,100,101,102,107,109,112,118,120,123,124,126],haven:[51,98],hdf:[9,33],head:[19,99,101,109,114],header:[36,64,66,79,87,100,102,122,123,124],headip:114,heard:98,heavi:110,height:[4,9,11,18,29,65,78,100,101],height_:84,held:31,hello:26,help:[4,18,29,49,51,56,63,69,78,79,89,98,99,106,110],helper:[18,42,63,80,89,100],henc:[42,72,77,80,81,83],here:[2,3,4,5,9,18,19,21,26,30,31,37,41,44,45,47,49,51,55,56,63,67,68,78,81,85,91,92,94,97,99,101,102,104,109,110,111,112,113,117,120,123,124,126],heterogen:[42,44,74],heurist:[4,44,69,118],hidden:[4,5,18,70,77,97,112],hidden_dim:18,hidden_out:29,hidden_s:5,hidden_t_prev:18,hidden_v:18,hidden_valu:18,hierarch:[4,75,77,79,97],hierarchi:79,high:[2,50,71,86,87,100,109,126],higher:[53,68,89,99],highest:[9,29],highli:[10,86,89,97,120],him:26,hint:[59,106],histor:[53,102],histori:6,hl_get_sync_flag:100,hold:[26,28,31,35,37,41,50,54,56,58,60,83,85,87,88,103,111,112],holder_:[87,103],home:[33,49,92,106,112,113,114],honor:35,hook:2,hookattr:2,hookattribut:2,horizont:[4,11],host:[30,39,74,112,113,122,123,124],host_c:[122,123,124],hostfil:114,hostnam:112,hostpath:113,hostport:112,hot:18,hour:98,hourli:99,hous:[10,95],how:[2,4,18,26,29,31,35,40,41,42,47,49,51,53,56,59,63,68,69,73,74,81,85,88,97,98,102,106,109,111,112,113,118,121,124],howev:[4,5,18,27,28,38,41,42,47,51,60,67,72,73,77,78,80,81,84,85,86,87,88,97,112,117,118],howto:109,hpp:[50,65],html:10,htod:107,http:[4,5,10,18,30,39,54,82,91,92,98,99,104,106,108,112,113,122,124,126],hub:82,huber:4,huge:72,human:[4,74,86],hundr:102,hwc:11,hyp:18,hyper:[4,54,100],hyperparamet:[4,81],hyperplan:9,hypothesi:18,i1117:107,i386:123,iOS:[124,125],iamfullaccess:112,iamusersshkei:112,icc:52,iclrworkshop2016:18,icml:86,ics:10,id_input:3,id_rsa:114,idea:[30,41,51,52,72,78,81,106,111],ideal:[42,88],ident:[4,18,80,101,112],identifi:[4,57,67,100],identityoffsetproject:4,identityproject:4,ids:[3,4,18,69,100],idx:[35,54,60,100],ies:49,if_else_op:28,ifdef:[67,74],ifels:[29,75],ifelseop:75,ignor:[4,17,18,118],ignored_token:18,iii:86,illustr:[3,18,31,36,42,53,68,97,100,107],im2col:18,im_siz:54,imag:[7,8,9,10,18,26,42,51,54,55,69,70,75,78,86,91,98,99,112,115,116,120,126],image_a:78,image_b:78,image_conv_lay:86,image_fil:78,image_h:18,image_lay:78,image_nam:26,image_path:78,image_reader_cr:78,image_w:18,imagenet:[4,33],imagepullpolici:112,imageri:4,images_reader_cr:78,imagin:55,img2label:11,img:[4,5],img_conv_lay:5,img_pool_lay:5,imgsiz:107,imgsizei:107,imgsizex:107,imikolov:109,immedi:[60,63,72,81,91,112],immutable_paramet:26,imper:40,imperfect:79,implement:[4,5,9,18,19,23,29,35,36,37,38,39,40,41,42,44,51,53,56,57,58,60,63,65,66,67,69,73,76,83,85,86,87,88,89,97],implemet:34,impli:30,implicitli:40,imposs:[69,126],impractic:88,improv:[4,43,44,60,79,86,106,107,112],in_fals:18,in_plac:18,in_tru:18,inarg:34,inbound:112,includ:[3,4,5,10,11,18,26,29,30,36,39,41,50,51,54,56,60,65,66,68,69,74,75,77,79,85,91,94,97,98,100,101,106,107,109,112,113,118,122,123,124],inclus:[18,69],incom:[40,59],incorpor:4,incorrect:4,increas:[31,35,50,109,118],increment:[46,55,60,118],incupd:100,inde:[9,41],independ:[4,18,27,28,36,43,83,87,126],index:[3,4,7,9,10,18,25,27,28,29,31,35,40,75,77,89,102,112],indexslot:4,indiact:18,indic:[3,4,18,21,28,29,36,47,54,68,75,80,84,87,89,110,112,122],indice_map:89,indices_map:89,individu:[31,71,112],industri:[31,64,126],ineffici:[73,88],infer:[0,11,17,18,26,28,29,31,46,51,57,58,59,60,61,65,67,76,77,79,84,86,88,95,123,124],infer_shap:77,infer_var_type_:58,inferenc:123,inferer:86,inferfac:58,inferior:38,infernec:124,infershap:[29,77,79,101,103],infershapecontext:[101,103],infervartypefn:58,info:[3,4,10,50,68,100,110,126],infom:4,inform:[4,10,18,25,29,39,49,56,59,60,63,64,67,68,71,77,81,83,84,99,100,101,103,106,107,112,118,122],infrastructur:[51,112],ingor:118,ingrad_:62,ingredi:[41,86],inherit:[22,29,70,79,87,101],ininst:26,init:[2,18,25,29,43,54,62,68,69,95,100,109,112,120],init_attr:77,init_batch_dim_idx:18,init_from_tar:25,init_model_path:[117,118,120],init_valu:18,initi:[2,4,5,10,12,18,21,25,28,30,35,40,42,43,44,46,55,68,71,72,77,81,85,89,95,97,100,101,108,118],initial_max:2,initial_mean:[2,4],initial_min:2,initial_std:[2,4],initialize_op_attr:77,initrd:126,inlcud:5,inlin:[87,102,103,112],inner:[4,100],inner_param_attr:5,inproj_attr:4,inproj_bias_attr:4,inproj_param_attr:4,input0:103,input1:[4,5,103],input2:4,input:[1,3,4,5,7,9,11,14,18,19,25,27,28,29,34,38,40,42,43,44,45,46,50,51,52,53,54,55,56,58,59,60,62,63,67,68,69,72,73,76,77,78,79,80,83,85,86,87,88,89,95,97,99,100,101,102,103,108,111,114,120],input_conf:4,input_data:100,input_data_target:100,input_dim_idx:18,input_dtyp:18,input_featur:1,input_hassub_sequence_data:100,input_id:4,input_imag:5,input_index:100,input_label:100,input_lay:100,input_loc:4,input_nam:26,input_proj_bias_attr:5,input_proj_layer_attr:5,input_seg:89,input_seq:[4,18],input_sequence_data:100,input_sequence_label:100,input_sparse_float_value_data:100,input_sparse_non_value_data:100,input_t:100,input_to_check:27,input_valu:27,input_var:[27,77],inputbuff:34,inputdef:100,inputgradi:80,inputlayers_:100,inputs_to_check:27,inputsizechang:63,inputtyp:9,insert:[18,28,45,52,71,76,79,80,99],insid:[3,5,28,31,42,44,45,46,59,63,73,74,78,79,80,92,112],inspir:74,instal:[4,18,39,62,82,91,92,98,99,104,106,109,113],install_android:122,instanc:[4,18,27,29,31,33,37,40,42,43,45,47,52,57,63,68,69,72,77,79,80,97,100,103,107,111,118],instance_ip:112,instanti:[31,47,108],instead:[4,5,7,28,30,34,39,40,41,42,50,51,55,56,86,98,99],instrins:50,instruct:[29,55,92,98,107,122],int16:90,int32:[18,67,75,89,90,118],int64:[18,42,48,67,84,88,90],int64_t:74,int8:67,integ:[3,4,9,18,35,39,40,50,65,69,100],integer_valu:9,integer_value_sequ:[9,69,86,97],integr:[3,18,91,126],intel:[51,67,87,102],intellig:60,inteloptimizedpaddl:62,intend:91,intens:86,inter:[4,18,42],interact:[4,42,92,112],intercept:4,interchang:[55,79],interconnect:71,interest:[40,50,71,107],interfac:[0,2,4,5,22,25,29,35,39,49,56,71,73,79,80,86,87,91,101,103,112,123,126],intergr:4,intermedi:[18,42,49,52,54,60,70,86,98,122,124],intern:[4,5,25,50,86,106,109,110,112],internel:62,internet:[30,31,126],interpret:[3,18,47,51,52,90,91,107],interv:18,inth:103,intrins:[40,47,50,124],introduc:[4,11,29,31,54,61,64,81,83,85,101,106,109,113],introductori:98,intuit:[38,79],inval_:62,invalid:[78,83],invent:51,invoc:[30,53,79],invok:[4,25,28,42,45,73,77,79,80,85,88,98,99,107,112],involv:[69,101],iob:3,ioe:3,ios:123,ios_arch:123,ios_deployment_target:123,ios_development_root:123,ios_enable_bitcod:123,ios_platform:123,ios_sdk_root:123,ios_use_veclib_for_bla:123,ipad:123,iphon:123,ips:112,ipt:[4,77,85,97],ipx:126,ipython:26,is_bia:18,is_color:11,is_cpu_plac:63,is_loc:25,is_mkldnn_librari:63,is_revers:18,is_seq:[4,97],is_spars:18,is_stat:2,is_target:76,is_tensor:85,is_test:[18,63],is_traget:76,is_train:11,isinst:45,ismkldnnkernel:63,isn:107,isspars:100,issu:[18,30,54,86,92,94,98,99,107],issuecom:18,istag:82,item:[4,9,18,25,38,50,78,95,126],iter:[4,5,6,9,25,26,31,42,51,52,60,63,72,74,78,86,89],iter_multiple_input_and_param:77,its:[4,5,18,25,26,27,28,29,31,35,40,41,44,45,46,51,52,54,55,56,58,60,64,68,69,71,72,73,76,77,79,80,83,84,85,87,88,94,100,101,102,103,107,109,112,118],itself:[28,31,38,52,63,72,83],ivs:85,java:[29,65,75,79],jeremi:107,jku:18,jmlr:18,job:[10,28,38,40,42,45,79,92,109,111,117,118,120],job_desc:42,job_dispatch_packag:110,job_id:10,job_nam:[39,112],job_namespac:112,job_path:112,job_workspac:110,jobdesc:42,jobnam:42,jobpath:112,jobport0:112,jobport1:112,jobport2:112,jobport3:112,jobserv:39,join:31,jointli:5,journei:92,jpg:11,json:[56,86,112,113],jth:5,judg:4,juditski:72,jupyt:[39,92],just:[1,3,4,5,10,18,30,35,36,40,42,51,52,54,58,63,72,73,77,78,79,80,81,83,84,91,94,98,99,110,112,120,122],jx4xr:112,jypyt:26,k8s:[40,126],k8s_data:112,k8s_job:26,k8s_token:26,k8s_train:112,k8s_user:26,kafka:33,kcpu:74,kcuda:74,kcudnn:102,kdisabl:74,kebilinearinterpbw:107,kebilinearinterpfw:107,keep:[4,9,11,18,31,41,51,52,55,69,72,77,83,85,91,99,126],keep_dim:18,keep_top_k:4,kei:[10,11,19,22,25,27,28,29,31,33,35,48,50,59,63,79,80,85,86,89,98,99,101,107],kenlm:86,kept:[4,60,77],kera:81,kernel0:102,kernel1:102,kernel:[4,18,22,27,41,50,52,59,62,72,74,81,84,86,87,88,101,103,107],kernel_hint:59,kernel_type_for_var:88,kerneltyp:[59,63],key1:118,key2:118,key_pair_nam:112,keyid:112,keymetadata:112,keypair:112,keyserv:112,keystat:112,keyusag:112,keyword:[18,77],kforcecpu:59,kill:[31,112],kind:[26,27,31,37,42,45,52,55,59,63,70,71,74,87,88,90,92,102,112,113],kind_:74,kitten:18,kmark:74,kmkldnn:102,kms:112,knchw8c:67,knchw:67,knhwc:67,know:[18,26,35,40,60,64,99,100,106,107,109,112,122],knowledg:86,known:[28,29,41,51,53,68],kplain:102,kpoprang:74,kpushrang:74,kqueue:41,kriz:10,krizhevski:18,kselectedrow:84,ksimonyan:5,kstate:74,kube_cluster_tl:26,kube_ctrl_start_job:26,kube_get_workers_addr:40,kube_list_containers_in_job_and_return_current_containers_rank:26,kubeconfig:112,kubectl:[110,113,114],kuberent:[31,112],kubernet:[26,31,40,42,79,109,115,116,126],kubernetes_service_host:26,kusecudnn:59,kusemkldnn:59,kvp:22,kwarg:[5,6,9,14,18,21,46,56,77,85],kwd:[15,22],l1_rate:2,l1_regularization_op:81,l1decayregular:23,l2_rate:2,l2_regularization_op:81,l2_sim:4,l2decayregular:23,l93:34,label:[3,4,9,10,11,14,18,25,42,46,51,54,55,56,70,75,78,86,88,108,111,113],label_dim:4,label_fil:78,label_lay:78,label_path:78,lag:118,lambda:[18,40,45],lambdacost:4,lambdarank:4,lan:109,languag:[4,10,19,40,41,51,55,60,74,79,83,86,120],larg:[7,10,18,42,44,45,60,64,72,86,99],larger:[2,3,4,60],larger_than:[29,57,75],largest:18,last:[3,4,5,18,28,45,60,68,74,75,97,118],last_seq:69,last_time_step_output:4,lastseen:113,latenc:[4,50,86,110,112],latent:4,later:[30,79,81,86,87,91,94,101,103,112],latest:[4,29,30,31,38,82,91,92,94,104,113,122,123],latter:[72,89,106],launch:[63,112,118],launcher:26,layer1:[4,5],layer2:4,layer3:4,layer:[2,3,5,7,9,12,24,25,28,29,34,40,42,44,51,54,55,57,70,72,75,78,79,81,85,86,87,89,95,97,105,108,111,117,118],layer_0:100,layer_attr:[4,97,120],layer_help:59,layer_num:120,layer_typ:[4,61,62],layerbas:100,layerconfig:100,layergradutil:100,layerhelp:[18,59,77],layermap:100,layeroutout:4,layeroutput:5,layout:[11,63,88],layout_:[59,67],layouttyp:59,lazi:[72,81],lbl:3,lead:[60,67,107],leaki:54,leakyrelu:18,learing_r:70,learn:[2,3,4,5,6,10,19,26,28,36,38,41,42,44,47,54,55,60,62,69,71,72,74,78,79,81,87,92,97,98,100,101,104,107,111,123],learnabl:[18,19,25],learning_r:[2,21,36,42,108,111],leas:31,least:[3,18,31,94,122],leav:[29,112],lectur:60,lecun:10,left:[4,18,29,103,123],left_right_flip:11,legaci:92,legal:85,len:[4,18,36,40,48,51,77,95,100],length:[4,5,9,10,11,18,36,50,61,64,68,69,79,86,89,97,113,118],leran:60,less:[4,18,26,45,111,126],less_than:[26,60],lesser:18,let02:113,let:[3,4,18,26,29,38,40,52,53,55,59,63,67,68,69,70,80,87,88,101,106,111,112],level:[2,4,18,50,53,56,64,68,69,74,87,89,90,99,110,118,122],levenshtein:18,lgtest:30,lgtest_main:30,lib64:[92,118],lib:[66,91,92,106,109,122,123,124],libapi:30,libari:66,libc:94,libcuda:92,libgcc_:94,libgoogl:106,libiomp5:62,libmkldnn:62,libmklml_intel:62,libnvidia:92,libpaddl:[65,66,79,106],libpaddle_capi:66,libpaddle_gserv:66,libpaddle_math:66,libpython2:91,librari:[4,18,30,37,41,42,62,63,66,71,86,88,91,94,101,102,109,111,118,123,124],library_:67,library_typ:102,library_type_:88,librarydevicecontext:67,librarytyp:[88,102],libstdc:94,licens:[62,71],life:31,lifecycl:[74,126],lifetim:[83,94],lightweight:53,like:[3,4,9,10,18,28,29,30,31,34,39,40,41,47,51,52,53,54,55,56,58,63,67,71,72,77,78,79,80,81,83,84,86,88,89,91,94,97,98,99,106,107,108,109,112,117,120,122,123,124,126],limit:[4,9,18,51,60,64,69,79,81,107,118],linaro:124,line:[3,9,22,30,34,39,41,49,55,72,75,77,79,81,98,99,105,106,107,110,111,112,120],line_break:9,linear:[4,18,19,69,95],linearli:18,lineno:106,link1:50,link2:50,link:[4,5,18,30,48,49,83,94,101,112,126],linux:[9,41,48,92,94,98,99,109,112,124],linux_x86_64:[82,94],lipo:123,list:[3,4,5,9,11,18,22,23,25,26,28,29,30,35,39,40,47,49,51,54,67,70,73,74,77,80,83,89,95,97,98,100,101,106,109,111,112,118,120,124],listdir:109,listen:[31,40,42,109,111,118],listen_and_do:40,listen_and_serv:18,listenanddo:40,littl:[36,59,64,118],live:[101,108],live_in:60,live_out:60,load:[11,17,26,31,42,54,71,77,91,101,112,118],load_and_transform:11,load_imag:11,load_image_byt:11,load_missing_parameter_strategi:[117,118,120],load_mnist:54,load_persist:111,loadsave_parameters_in_pserv:[34,117,118],loc:3,local:[2,18,25,27,29,31,37,38,41,55,60,68,75,77,79,91,92,98,99,106,110,113,117,118],local_scop:27,localhost:[92,104],localpath:49,locat:[4,18,25,30,51,67,74,87,89,97,100,109,124],lock:[30,31,35,36],lod:[18,41,64,68,84,89,90],lod_desc:[84,90],lod_expand:69,lod_level:[18,77,84,90],lod_rank_table_obj:18,lod_tensor:[18,68,84,90],lod_tensor_aarri:18,lod_tensor_arrai:18,lodrankt:18,lodtensor:[18,19,41,58,64,79,90],lodtensorarrai:18,lodtensordesc:[64,84],log:[35,42,49,54,94,100,109,110,112,113,114,118,124],log_barrier_abstract:118,log_barrier_lowest_nod:[117,118],log_barrier_show_log:[117,118],log_clip:[117,118],log_error_clip:[117,118],log_period:[113,118,120],log_period_serv:[117,118],logarithm:[1,18],logic:[38,42,44,45,54,58,70,71,73,83,89,101],login:[94,114],logist:18,logit:[18,54,88],longer:[31,42,60],look:[3,18,29,39,40,51,52,55,72,77,80,81,86,108,109,111,112,113,117],lookahead:[4,18,86],lookup:[18,58,69,108],lookup_t:60,loop:[27,29,51,60,74,78,83],loop_var:89,loss:[4,18,28,42,54,56,70,72,81,86,100],lot:[42,67,69,72,77,81,87,109,117,126],low:[4,70,71,86,87,89],low_rnn:68,lower:[4,18,50,68,69,99,110],lower_level_rnn:68,lowest:118,lpaddle_capi_shar:66,lpaddle_capi_whol:66,lrelu:54,lstm:[4,18,97,113],lstm_bias_attr:5,lstm_cell_attr:5,lstm_group:5,lstm_layer_attr:5,lstm_step:5,lstm_unit_op:18,lstmemori:[5,97],lstmemory_group:4,lstmlayer:61,lstmp:18,ltr:4,luckili:60,mac:[66,98,99,122],machin:[5,10,25,42,44,51,54,60,62,71,72,81,91,94,98,100,112,113,114,117,118,120,124,126],machine_transl:97,maco:[94,95,98,123],macro:[53,67,80,101],made:[31,36,51,97],mai:[4,5,18,22,27,29,42,46,50,52,59,60,63,71,74,78,79,83,86,88,91,92,103,104,107,109,112,124],main:[18,40,41,45,51,52,56,71,75,79,94,106,109,112],main_program:[17,18,28,46],mainli:[37,60,67,87,91,101,118],mainlin:94,maintain:[4,29,35,72,77,79,112],majel:30,major:[42,50,88,122],make:[3,4,18,26,28,29,30,31,35,36,38,41,42,43,50,51,55,68,69,72,73,77,78,79,81,86,89,91,98,99,100,101,102,106,107,109,111,112,122,123,124,126],make_chan:41,make_ddim:103,make_function_oper:53,make_vari:85,maker:[79,80],malloc:[87,100],man:48,manag:[25,31,36,37,40,41,42,49,74,83,87,94,104],mandarin:[4,86],mandatori:123,mani:[5,11,28,30,35,40,41,51,54,59,60,69,73,74,77,79,80,83,84,85,88,89,98,118],manili:56,manipul:[51,77,80,110,123],manner:[4,72,81,86,87],mantain:60,manual:[42,70,72,80,110,122,123,126],manufactur:51,manylinux1:94,manylinux1_x86_64:[82,94],manylinux:82,map:[3,4,9,19,25,26,29,35,45,63,67,77,80,83,85,87,89,98,118,126],map_fn:89,map_read:9,mapper:9,mapreduc:[26,109],margin:18,mark:[28,44,54,55,68,69,74,83,97,106,126],marker:74,market:50,mask:[2,4,18],master:[26,38,79,82,118,124],mastermind:30,mat:[65,66],mat_cache_row:34,mat_norm:34,mat_normal_shar:34,mat_param_attr:5,mat_sparse_row:34,mat_sparse_row_auto_grow:34,mat_sparse_row_id:34,mat_sparse_row_prefetch:34,mat_sparse_row_prefetch_full_s:34,mat_value_shar:34,match:[18,30,50,94,107],matchbox:126,math:[5,18,62,65,79,99,100,101,107],mathemat:81,matirx:4,matmul:[29,56,68,89,101],matric:[18,97,100],matrix:[3,4,5,9,18,19,25,34,65,66,100,101,117,120,123],matrixptr:100,matrixtyp:66,mattyp:34,matur:109,max:[2,4,9,10,18,19,22,27,43,45,60,77,107,118,120],max_diff:27,max_id:[4,25],max_job_id:10,max_length:[4,69,97],max_movie_id:10,max_norm:18,max_relative_error:[27,101],max_seq_len:18,max_sort_s:4,max_user_id:10,max_x:18,maxframe_evalu:3,maxid:3,maxid_evalu:3,maxim:[4,45],maximum:[3,4,10,18,22,29,36,97,101,107,118],maxinum:7,maxoutfunctor:87,maxpool:4,mayb:[29,63,101],mchw:18,md5:[10,32],mean:[2,3,4,5,6,7,9,11,22,25,28,30,42,43,45,56,69,76,78,83,86,88,92,97,98,101,106,107,108,111,112,118,120,126],mean_op:18,mean_var_nam:4,meant:89,measur:[18,46,107],mechan:[4,5,19,28,37,46,63,77,80,97,102,112],mem:[4,18,29,39,69],mem_per_pserv:42,mem_per_train:42,member:[4,10,26,45,55,56,67,73,77,83,101],memcpi:[73,107],memor:4,memori:[5,18,28,29,34,35,39,50,52,62,63,64,67,69,72,74,79,88,97,98,99,100,103,107,108,113,118,120],memory_boot:5,memory_nam:4,memory_optim:60,memory_test:98,memory_threshold_on_load_data:118,memoryalloc:87,memorydesc:63,mention:[18,28,30,35,42,44,51,68,71,72,74,98],mere:5,merg:[4,18,36,38,43,46,62,68,71,73,99,104,118,123],messag:[18,29,40,41,47,51,52,55,64,74,75,76,77,79,80,84,90,99,113,118],metaclass:101,metadata:[48,112,113],metal:[111,126],metaphor:55,metaplotlib:26,method:[4,6,25,27,29,38,40,42,43,45,50,54,55,56,59,70,71,77,78,79,83,84,88,89,91,92,94,100,101,103,104,106,107,118,120],methodolog:72,metric:[46,74],microarchitectur:50,might:[4,29,30,40,41,51,60,75,86,98,99,100,106,112,122],mileag:107,million:[10,120],min:[2,4,18,22,43,45,77,107,112,120],min_block:29,min_count:44,min_desc:29,min_word_freq:10,mind:106,mini:[4,9,14,18,25,29,31,41,46,47,51,57,68],mini_batch:78,minibatch:[4,18,29,46,55,57,75],minim:[29,42,44,45,51,54,70,79,108,111,118,122,123,124],minimum:[4,18,22,86,111,123],minimun:118,minsizerel:[122,123,124],minu:80,minus_grad:80,minusgradop:80,minusop:80,minusopgradmak:80,minusopprotoandcheckermak:80,minut:[31,38,92,98,112],mirror:[30,92],mislead:36,mismatch:18,miss:[54,118],mistak:51,misus:102,mit:112,mix:[5,74,89,97],mixed_lay:5,mixed_layer_attr:5,mixedlayertyp:4,mixtur:106,mkdir:[49,91,104,112,114],mkl:[63,79,87,88,91,92,102],mkl_packed_:61,mkldnn:[4,62,67,88],mkldnn_:62,mkldnn_batch_norm:4,mkldnnactiv:62,mkldnnbase:62,mkldnnlayer:62,mkldnnmatrix:62,mkldnnstream:62,mkldnntester:62,mklml:62,mklpack:61,mklpackedgatedrecurrentlay:61,mklpackedgemm:61,mklpackedlstmlay:61,mklpackedrecurrentlay:61,mlp:56,mnist:[33,42,54,55,75,78,79,106],mnist_random_image_batch_read:78,mnist_train:78,mnist_train_batch_read:78,mobil:[50,51,60,79,104,121],mod:109,mode:[4,22,25,50,61,71,99,118],model:[0,4,5,10,17,18,19,25,28,29,31,32,40,42,44,45,46,55,60,61,70,71,72,79,81,86,88,89,95,99,100,104,105,109,111,112,118],model_list:[118,120],model_path:120,modelparallel:42,modern:60,modif:[86,93],modifi:[4,23,42,50,56,81,97,100,101,109,110,112],modul:[5,10,25,42,53,54,69,86,89,101,106],modular:69,modulo:4,moment:106,momentum:[2,18,83,102],momentumop:106,momentumoptim:20,mon:113,monitor:[41,74],mono:4,month:30,more:[3,4,5,9,18,22,26,27,28,30,31,35,38,39,41,42,44,49,50,51,52,53,55,59,60,63,67,68,69,70,74,77,78,79,81,86,87,89,91,92,95,97,98,100,101,102,103,104,106,107,108,109,111,113,120,124,126],most:[4,9,25,26,28,30,38,41,42,52,55,56,67,69,72,74,78,81,86,87,94,97,100,106,107,108,117,126],mostli:[50,126],motiv:79,mount:[39,92,109,112,113],mountpath:[112,113],move:[4,31,35,49,51,72,92,107,112,126],movement:107,movi:10,movidiu:51,movie_categori:10,movie_info:10,movie_review:10,movieinfo:10,moving_average_fract:4,mpi:[41,71,114],mpirun:114,mse:[51,55,70,75],much:[4,18,31,51,63,70,78,81,89,107],mul:[53,60,77,100,101],mul_grad:101,mul_op:[18,101],mul_ratio:4,mul_result:77,mulgradkernel:101,mulkernel:101,mulop:[53,101],mulopgrad:101,mulopmak:101,mult:[40,52],multi:[4,19,46,71,73,88,100,106,110,111,117,118,126],multi_binary_label_cross_entropi:4,multidimension:18,multigradientmachin:73,multinomi:4,multip:19,multipl:[3,4,5,9,14,18,19,25,26,27,35,36,38,40,41,42,44,46,51,52,53,59,71,74,79,86,88,90,97,100,101,106,109,112,118,120],multiple_input:77,multiple_param_attr:77,multipli:[3,4,18,40,100],multiprocess:9,must:[1,3,4,5,9,11,18,28,36,45,60,63,64,67,74,76,77,78,79,85,90,97,100,101,103,109,111,112,118,120,122,124],mutabl:[87,103],mutable_data:[63,87,101,103],mutex:41,mutuable_data:[87,103],mutual:18,mxnet:[29,40,41,51],my_cluster_nam:112,my_external_dns_nam:112,my_lib:109,myerrorclip:45,myfil:9,mypaddl:113,naiv:40,name:[2,3,4,5,7,11,17,18,21,22,25,26,27,28,29,31,33,34,36,39,40,42,46,50,53,56,59,62,63,64,66,67,69,74,75,77,79,82,84,85,89,90,92,94,95,97,98,100,101,102,107,108,109,111,113,115,116,118,120,123,126],name_:74,name_prefix:33,namespac:[29,57,65,77,100,101,113],nativ:[4,18,50,99],natur:[18,35,38,44,69,89,120],navig:104,ncall:106,nccl1:71,nccl2:71,ncclinit:71,nchw8:88,nchw8c:88,nchw:[4,18,62,67],ndarrai:[11,18,25,33],ndcg:4,ndcg_num:4,ndk:122,nearest:50,nearli:27,necess:89,necessari:[4,28,29,36,38,45,46,60,64,69,73,77,85,89,100,114],necessarili:[40,100],neck:71,need:[3,4,5,9,17,18,19,23,26,27,28,30,34,35,36,38,39,41,42,44,45,46,49,51,52,53,54,59,60,63,67,69,70,71,72,73,74,76,77,79,80,81,83,84,85,86,87,89,91,92,93,94,95,97,100,101,102,103,104,109,111,112,113,117,118,120,122,123,124,126],neg:[3,4,18],neg_distribut:4,neg_overlap:4,neg_pos_ratio:4,neglect:4,neighberhood:71,neither:4,neon:[50,122,124],ner:3,nervana:51,nessesari:86,nest:[4,9,28,29,74,75,90],net:[4,5,12,18,29,54,68,83],netop:[29,79],network:[2,3,4,9,18,19,24,25,26,27,28,29,31,34,42,44,46,54,56,60,61,62,68,70,72,74,77,78,81,83,85,86,87,88,90,95,100,101,103,107,109,111,118,126],network_config:120,networkadministr:112,neural:[4,5,9,18,25,26,28,29,31,42,56,60,61,62,68,72,81,83,87,88,90,95,103,107,109,111,118],neuralnetwork:73,neuron:[18,100],never:[9,60,78,83,112,113],new_block_idx:77,new_dim:18,new_op_and_kernel:102,new_op_desc:45,new_scop:88,new_stat:68,newblock:77,newbuff:63,newer:122,newest:36,newli:[50,123,126],newop:29,newopdesc:77,newprogram:77,newscop:88,newvardesc:77,next:[4,10,28,31,37,41,45,69,71,89,97,100,101,106,107,112,113,118],nextlay:62,nfs4:112,nfs:112,nfsver:112,ngram:10,nic:[117,118],nil:35,nine:10,nlp:4,nltk:10,nms_threshold:4,nms_top_k:4,nnz:100,no_grad_dict:28,no_grad_set:[27,28,101],no_gradi:28,no_sequ:4,node1ip:114,node2ip:114,node3ip:114,node:[4,30,38,40,42,44,56,60,69,71,79,98,100,109,110,111,112,113,114,118,126],node_0:112,node_1:112,node_2:112,node_id:109,nodeattr:56,nodeentri:56,nodefil:110,nodesep:56,nohup:109,nois:[4,18,31,54,109],noisi:[4,54],non:[4,18,31,50,51,84,100,101,112,118],none:[2,3,4,5,6,7,11,13,14,17,18,19,21,22,23,25,26,27,28,29,45,46,54,56,57,68,69,70,75,77,85,89,97,108,111],noneedtran:63,nonlinear:100,nontranspos:18,nor:[40,98],norm:[5,18,54,67],norm_by_tim:[4,18],normal:[4,5,10,18,21,72,86,97,100,109,113,118],normaliniti:16,normliz:18,notat:[4,60],note:[2,4,5,7,11,18,25,26,28,29,34,35,39,60,64,67,71,78,79,87,88,91,92,101,103,104,107,109,111,112,118,120,123],notebook:[39,92],notest_dist_fit_a_lin:111,noteworthi:51,noth:[1,25,59,77,83,98,118],notic:[4,45,51,71,80,97,99,100],notif:99,notimplementederror:45,notin:88,notingradi:101,notion:89,notori:27,now:[9,28,30,31,44,54,64,67,72,79,80,81,83,111,112,118,123],np_arrai:9,nproc:98,nullptr:[63,74,80,83,100],num:[4,5,109,118],num_channel:[4,5],num_chunk_typ:[3,14,18],num_class:[4,5,18,56],num_col_dim:18,num_filt:[4,5,18,19],num_flatten_dim:18,num_gradient_serv:[109,117,118],num_head:19,num_hidden:56,num_input:99,num_neg_sampl:[4,18],num_or_sect:18,num_parameter_serv:26,num_pass:[25,113,117,118,120],num_per_batch:11,num_pserv:42,num_repeat:4,num_result:3,num_results_per_sampl:4,num_row:84,num_shard:33,num_step:89,num_total_class:18,num_train:42,num_true_class:18,number:[3,4,5,9,10,11,14,18,19,22,29,31,33,44,46,60,72,74,78,79,85,89,98,100,106,109,111,112,118],numchunktyp:3,numdevices_:120,numer:[4,101],numeric_grad:27,numerical_grad:27,numlogicaldevices_:120,numofallsampl:3,numofwrongpredict:3,numpi:[2,9,11,18,25,33,50,54,77,78,91,101],numreal:34,numsampl:107,numtagtyp:3,numtimeout:35,nv_:30,nv_gpu:98,nv_librari:30,nv_test:30,nvcc:[30,50,52],nvidia:[50,67,71,87,92,98,107,118],nvlink:71,nvprof:74,obei:3,object:[2,4,5,9,18,25,26,34,42,45,46,54,56,60,65,70,74,77,79,81,83,103,107],observ:[4,100,107],obtain:[4,18,38,72,87],obvious:[30,67,106],occup:60,occupi:[50,74],occur:[10,25,60],occurr:29,oct:113,odd:4,odoti:18,off:[66,91,92,98,109,111,122,123,124,126],offer:[29,79,85],offici:[4,30,99,104,112,122],offlin:[31,33,126],offset:[4,18,34],often:[4,34,56,60,67,99,106,109],ograd:100,old:[27,36,38,69,79,118],older:[18,51,122],omega:81,omit:18,omp_num_thread:106,ompi_comm_world_rank:109,onc:[4,31,35,40,42,44,46,51,55,72,99,100,104,112],one:[1,3,4,5,7,9,18,22,25,26,27,28,29,31,34,35,36,38,39,40,42,45,46,47,50,51,52,53,54,56,58,59,63,64,67,68,69,70,71,72,73,75,76,77,78,79,80,83,84,86,87,88,89,92,98,99,100,101,102,108,110,111,112,113,114,118,120,126],onehotcrossentropyopkernel:101,ones:[53,54,79,99],onli:[3,4,5,7,11,18,25,26,27,28,30,34,35,36,37,38,39,40,42,44,45,46,47,49,50,51,54,55,60,63,68,69,70,71,73,74,77,79,84,85,86,87,88,89,91,93,94,97,98,100,101,102,103,104,107,111,112,113,117,118,120,123,126],onlin:[4,6,31,33,60,78],only_cpu:27,onnx:51,onto:[18,42,44,112,114],op1:[60,88],op1_2_op2:88,op1_to_op2:88,op2:[60,88],op3:60,op_:101,op_check:101,op_class:[79,85],op_desc:[45,58,76],op_info:108,op_kei:63,op_maker_class:[79,85],op_proto:85,op_registri:108,op_siz:45,op_test:101,op_typ:[79,101,102],op_unique_kei:63,opattrcheck:101,opcreat:85,opdesc:[29,45,55,75,76,77,79,80,85,90],opdescbind:[58,80],opdescbuild:29,open:[4,11,18,26,33,51,54,62,78,99,106,109,112],openbla:[91,92,122],opencv:11,openmp:106,openmpi:109,opensourc:71,oper:[4,5,9,11,18,22,23,27,29,40,41,42,43,44,46,47,50,51,52,54,55,56,58,59,68,69,70,71,74,76,81,83,86,87,88,90,97,99,100,102,103,107,108,112,118,122],operand:[18,50],operator_grad:27,operator_list:74,operatorbas:[29,53,79,80,85,101],operatorwithkernel:[88,101],opinfo:[58,79,80],opinfomak:58,opinfomap:80,opkernel:103,opkernelbas:102,opkernelkei:79,opkerneltyp:[67,88,102],opmak:85,opposit:18,opproto:101,opprotoandcheckermak:[80,101],opprotomak:[85,101],opregist:85,opregistri:85,ops:[27,28,29,30,40,52,55,56,72,75,76,77,79,87,101,126],ops_:29,ops_test:30,opt:[26,70,76,85,91],opt_op_list:70,optest:101,optestmeta:101,optim:[2,12,23,24,25,27,28,42,43,44,52,54,71,72,73,75,79,81,84,86,100,106,107,108,109,111,122,124],optimis:70,optimize_op:111,optimize_op_attr:77,optimizer_mod:18,optimizer_op:111,option:[3,4,18,22,26,30,42,54,59,64,75,76,77,79,84,85,86,90,98,100,106,109,110,111,120,122,123,126],optmization_op_list:70,opts_np:76,optyp:[58,85],opwithkernel:84,order:[4,5,9,11,18,22,25,28,55,64,74,78,81,89,91,100,106,109,112,113,118,126],ordereddict:25,orderli:18,oregon:112,org:[3,4,5,10,18,33,48,54],organ:[3,4],orient:85,origin:[4,5,9,10,17,18,27,50,54,83,89,99,103],other:[3,4,5,9,18,22,29,31,36,40,49,50,51,52,58,60,63,67,68,72,76,81,83,85,86,87,88,91,97,98,99,106,108,109,111,112,113,120,122,123,124,126],otherchunktyp:3,otherwis:[4,9,10,11,18,22,25,26,28,31,36,38,54,58,63,78,86,97,99,110,120],our:[18,26,28,30,41,42,44,54,58,60,67,71,72,83,89,91,94,97,98,99,100,106,111,112,113,122],out:[4,18,25,26,29,30,35,38,42,45,51,56,60,63,68,69,77,88,95,97,101,103,106,107,112,113,114,118],out_dir:112,out_fals:18,out_left:4,out_mem:97,out_memori:5,out_right:4,out_size_i:4,out_size_x:4,out_tru:18,outer:4,outgrad_:62,outlier:4,outlin:119,outout_lay:25,outout_layer1:25,outout_layer2:25,output:[1,2,3,5,7,9,14,18,19,22,25,26,27,28,29,33,38,40,44,45,49,52,53,54,55,56,57,58,60,63,64,68,69,72,75,76,77,78,79,80,83,84,85,87,88,89,91,97,98,99,100,101,102,103,106,107,109,113,118,120,122],output_:[4,62,100],output_all_step:68,output_arg_nam:45,output_dim_idx:18,output_dtyp:18,output_fil:22,output_height:18,output_id:4,output_lay:[25,95],output_max_index:7,output_mem:[4,97],output_mod:22,output_nam:27,output_num:68,output_path:33,output_s:18,output_seg:89,output_width:18,outputbuff:34,outputgradi:80,outputh:4,outputw:4,outsid:[4,5,42,83],outupt:89,outv:100,outval_:62,over:[4,5,18,25,26,51,60,71,72,89,99,100,107],overal:[54,72,74,99,126],overfit:[18,81],overflow:18,overhead:107,overlap:[3,4,100],overlap_threshold:[3,4],overload:[50,59],overrid:[29,31,49,63,87,100,101,103],overview:[35,36,37,87],overwhelm:99,overwrit:[49,109],own:[4,18,28,36,38,45,47,56,58,70,71,72,81,85,101,109,110,112,122,123],owner:[98,99],pack:[89,122],packag:[9,10,35,39,40,53,62,82,91,92,99,101,106,112],pad:[5,18,63,86],pad_c:4,pad_h:4,pad_w:4,padding_attr:4,padding_down:18,padding_h:18,padding_i:4,padding_idx:18,padding_left:18,padding_right:18,padding_up:18,padding_w:18,padding_x:4,paddl:[1,2,3,4,5,6,7,9,10,11,13,14,15,16,17,18,19,20,21,22,23,25,26,29,30,31,33,39,42,49,52,53,54,57,61,62,63,64,65,66,68,69,73,75,79,81,82,85,86,87,89,91,92,94,95,97,98,99,100,101,102,104,106,107,108,109,110,111,112,113,114,118,120,122,126],paddle_begin_init_param:36,paddle_dir:101,paddle_element_typ:36,paddle_element_type_float32:36,paddle_element_type_float64:36,paddle_element_type_int32:36,paddle_element_type_int64:36,paddle_element_type_uint32:36,paddle_element_type_uint64:36,paddle_enforc:[29,63],paddle_enforce_eq:[101,103],paddle_error:[65,66],paddle_exampl:39,paddle_finish_init_param:36,paddle_get_param:36,paddle_gradi:36,paddle_init_num_gradient_serv:109,paddle_init_param:36,paddle_init_port:109,paddle_init_ports_num:109,paddle_init_ports_num_for_spars:109,paddle_init_pserv:109,paddle_init_trainer_count:109,paddle_init_trainer_id:109,paddle_init_use_gpu:109,paddle_job:39,paddle_manylinux_devel:91,paddle_matrix:[65,66],paddle_matrix_cr:66,paddle_matrix_get_shap:65,paddle_matrix_shap:65,paddle_new_etcd_pserver_cli:36,paddle_new_pserver_cli:36,paddle_on_cloud:39,paddle_output:113,paddle_paramet:36,paddle_pserver2:110,paddle_pserver_cli:36,paddle_pserver_client_releas:36,paddle_save_model:36,paddle_send_grad:36,paddle_train:[66,82,110],paddle_with_cuda:74,paddle_with_mkldnn:67,paddlepaddl:[4,5,9,10,11,18,25,30,31,33,36,37,38,39,40,42,48,49,53,54,55,57,59,64,68,69,70,73,74,77,78,79,83,89,90,93,95,97,98,99,100,101,102,106,107,109,110,114,115,116,125,126],paddlepaddle_gpu:94,paddlepaddlebook:92,paddlepaddlehub:92,page:[99,112],pair:[3,18,19,22,23,28,29,42,55,70,74,79],pairwis:4,pakcag:30,paper:[4,18,54,86],para:34,paradigm:[40,47,79],paragraph:68,paragraph_data:68,paragraph_out:68,parallel:[18,40,41,42,44,71,74,79,88,98,107,109,112,113,118,120],parallel_for:40,parallel_nn:[2,117,118],paralleldo:43,parallelfor:40,paralleliz:86,param:[2,4,5,9,17,18,27,29,36,73,77,87,103],param_attr:[4,5,12,18,19,34,77,97],param_config_proto:36,param_initi:18,paramattr:[2,4,18,97],paramet:[3,5,6,7,9,10,11,17,18,19,22,23,24,27,28,29,30,32,34,38,40,42,43,45,47,49,51,52,54,55,56,58,64,68,70,71,75,78,83,85,86,89,91,95,99,100,101,102,103,105,108,110,120,123],parameter_block_s:[117,118],parameter_block_size_for_spars:[117,118],parameter_learning_r:2,parameter_list:[28,70],parameter_nam:[25,26],parameter_serv:26,parameter_valu:34,parameterattribut:[2,4,5,34],parameterclient_:34,parametermap:100,parametermutex_:34,parameters_:100,parameters_and_grad:[23,70],parameterserver2:34,parameterset:26,parameterupdat:73,parameterupdater_:34,parametr:4,params_grad:[70,111],paramt:112,paraspars:100,parent:[29,40,75,77,79,100],parent_:[29,83],parent_idx:77,parenthes:79,pars:[9,10,30,42,56,98,112,120],parser:9,part:[3,4,18,19,28,29,38,42,51,63,64,75,77,86,87,97,100,106,107,109,111,126],parti:[98,107,122,123,124],partial:[4,25],partial_sum:4,particip:101,particular:[55,64,79,86,88,107],partit:[31,33,42,44,79,109,112],pass:[4,18,25,28,29,31,41,45,46,51,54,60,64,70,72,73,76,77,78,79,81,83,86,89,99,100,107,109,110,111,112,113,118],pass_gener:4,pass_id:[25,42,111],pass_idx:78,pass_num:111,passtyp:100,password:114,past:[26,92,95,112],patch:[18,48],path:[3,9,10,11,17,25,31,35,36,39,60,69,78,86,91,92,109,112,113,118,120,122,123,124],path_to_paddlepaddle_working_directori:104,pattern:[10,31,65,72,81,112],paus:[31,38],pcie:71,pdf:[5,18],peephol:18,peer:71,pem:[26,33,112],pend:[31,35],peopl:98,pep425tag:94,pep8:99,per:[3,4,10,11,18,31,36,71,72,78,81,101,118],percal:106,perf_test:106,perfectli:86,perfom:[118,120],perform:[4,5,18,27,36,41,42,46,50,51,54,60,71,73,74,78,79,81,86,87,88,97,98,100,101,105,109,111,117,122,123,124],perftool:[74,106],period:[31,38,118],perm:18,permiss:112,permut:18,peroid:[4,11],persist:[17,18,47,84,86,90,112],persistentvolum:112,persistentvolumeclaim:112,person:[3,26,59],perspect:[79,107],perturb:[27,100],pex:126,pfs:[33,49],pfsclient:33,pfspath:49,pgp:112,phase:[18,63,69,71,72,78,80,86,126],philosophi:[72,81],photo:54,phrase:18,physic:[123,126],pick:[18,112,123],pickl:[109,114],pictur:71,piec:[5,40,74,103,111],pil:[11,109],pillow:39,ping:99,pip:[82,91,93,95,99,104,106],pipe:9,pipelin:[46,86],piperead:9,pivot:63,pixel:[4,9,10,42],place:[13,15,18,22,28,29,31,38,42,44,47,59,63,71,79,88,100,103,107,108,111],place_:[59,67,87,88],place_typ:102,placehold:[54,87,103],placement:44,plain:[3,4,9,39,64,66,67],plan:[31,40,63,79,86,100,122],platform:[29,52,63,67,74,87,88,94,99,101,102,103,108,112,122,123,124],pleas:[2,4,5,6,11,18,26,31,35,36,37,40,52,56,67,68,77,78,79,86,87,90,91,92,94,97,98,99,100,101,103,104,106,109,111,112,122,123,124],plot:26,plu:[4,27],plug:[71,72],pne:101,pnpairvalidationlay:118,pnpairvalidationpredict_fil:117,pod:[33,39,40,112,113],pod_nam:112,point:[18,29,31,39,41,50,60,63,71,87,98,99,101,103,106,107,122,126],pointer:[29,36,56,60,67,77,79,83,87,103],polar:10,polici:[18,112],polit:18,poll:41,pollut:38,polyak:72,ponit:56,pool3:100,pool:[5,18,24,43,60,86],pool_attr:5,pool_bias_attr:5,pool_layer_attr:5,pool_pad:[5,18],pool_siz:[4,5,18,19],pool_size_i:4,pool_strid:[5,18,19],pool_typ:[4,5,18,19],pooled_height:4,pooled_width:4,pooling_lay:5,pooling_typ:4,poolingtyp:7,pop:[29,47],popul:36,popular:[30,54,56,74],port:[18,30,40,106,109,111,112,113,117,118],port_num:117,portabl:56,portal:104,ports_num:[109,118],ports_num_for_spars:[34,109,117,118,120],pose:31,posit:[3,4,5,18],positive_label:3,possibl:[26,29,35,41,44,60,77,81,90,107],post:[39,48],postpon:81,potenti:[50,107],power:[50,60,71,86,103,126],ppo_workspac:104,pprof:106,practic:[97,100],pre:[4,5,10,26,36,59,60,91,99,112,113,122,124],pre_activ:77,pre_bia:77,pre_id:18,pre_mem:18,pre_stat:[68,89],preambl:77,precis:[3,14,18,46,50,72,91],precision_evalu:3,precompil:47,pred:[56,60],predecessor:60,predetermin:[4,118],predic:[10,17],predict:[3,4,18,25,42,81,95,97,118],predict_fil:118,predict_lay:25,predict_output_dir:[117,118],prediction1:25,prediction2:25,prefer:[51,59],prefetch:[34,100],prefix:[3,5,18,31,33,69,86,112],pregel:41,pregrad:100,prepand:77,prepar:[27,39,73,86,97,115],prepend:[18,77],prepend_oper:77,preprocess:[10,11,86,89,113],present:[26,28,29,74,89],preserv:49,prev_batch_st:[117,118],prev_cel:18,prev_cell_data:18,prev_hidden:18,prev_hidden_data:18,prevent:[6,18,26,31,35,38,45,81,106],preview:[79,104],previou:[4,5,18,22,25,28,31,44,49,68,69,100,106,112,118],previous:[4,113],previous_memori:29,price:[10,79,95],prim:63,primari:[51,55],primarili:[72,81],primer:99,primit:[50,62,63,71,89],primitive_desc:63,primitivedesc:63,principl:[18,26,30,67],print:[2,22,25,26,42,51,56,77,94,95,106,114,118],print_graphviz:56,print_phas:18,print_tensor_lod:18,print_tensor_nam:18,print_tensor_shap:18,print_tensor_typ:18,printallstatu:107,printer:3,printstatu:107,priorbox:4,prioriti:79,prite:3,privat:[29,66,74,77,83,84,85,87,89,99,103],privileg:[98,112],pro:71,prob:[3,25,95],probabalist:18,probabilist:4,probabl:[3,4,18,25,69,86,92,97,99],problem:[4,26,27,30,38,51,54,55,72,79,81,94,98],proc:92,proce:[9,31,78,92,112],procedur:[29,64,103,123],proceed:18,process:[2,4,5,9,18,26,28,29,33,34,35,38,40,41,42,46,47,51,52,56,60,62,64,71,81,85,88,97,99,106,109,110,112,113,118,120],process_num:9,processor:[50,107],produc:[4,5,9,18,31,51,56,78],product:[4,5,18,19,39,51,100,112,113],productgraph:113,prof:106,profil:[12,49,74,86],profilerst:74,proflier:[74,107],program:[9,13,17,18,22,26,28,33,36,38,42,44,47,55,57,60,70,71,74,78,79,83,90,106,107,110,118],programdesc:[40,42,47,51,60,64,76,77,80,90],programm:[42,51,77],progress:[31,35,118],proj:4,proj_activ:18,proj_dim:18,proj_out:18,proj_siz:18,project:[4,5,18,19,39,66,86,97,100,101],promis:[4,5,69],prompt:[49,51,111],prone:26,pronunc:86,prop_kind:63,propag:[4,6,28,51,72,101,118,120],proper:[59,109],properli:[59,98,111],properti:[56,81,118],propos:[18,29,43,44,69,70,71,72,89],proprietari:62,protect:[50,85,100,101],proto:[7,41,59,64,67,75,79,85,90,101],proto_:85,protobuf:[25,29,39,40,42,47,51,52,55,56,60,64,75,77,79,80,85],protoc:[122,124],protocol:[3,108,118,126],provi:109,provid:[4,10,18,26,29,36,39,40,46,47,50,51,54,56,58,59,67,71,72,74,77,81,85,86,87,89,92,95,103,106,107,109,110,111,112,122,123,126],providermemory_threshold_on_load_data:117,provis:[112,126],prune:[4,17,29],ps_desir:31,pserver:[25,34,36,37,39,79,109,111,112,117,118],pserver_addr:36,pserver_cpu:39,pserver_endpoint:111,pserver_id:32,pserver_mem:39,pserver_num_thread:[34,117,118],pserver_prog:111,pserver_spec:25,pserver_startup:111,pserverstart_pserv:117,pseudo:[26,28,39,80,89],pseudocod:89,psize:100,ptr:[66,87],pub:[18,114],publish:122,pull:[30,79,82,99,122],purpos:[4,31,42,44,59,107],push:[29,47,51,74,82,99],push_back:100,put:[30,31,34,44,60,63,77,87,100,111,113,122],pvc:112,pwd:[91,92,98,104,122],pxe:126,pybind:[29,41,50],pypi:94,pyramid:4,pyramid_height:4,python2:106,python3:94,python:[18,25,26,29,37,41,46,47,51,53,54,55,56,59,65,69,73,74,79,82,87,89,91,92,94,95,97,98,99,104,108,109,111,114],pytorch:[51,74],qualcomm:50,queri:[3,4,19,112],query_id:3,question:[4,26,40,44,85,111,112],queue:[41,44],quick:[56,118],quick_start:[39,112,113,115],quick_start_data:113,quickli:[69,77,79],quickstart:113,quit:[69,107],r14b:122,r_h_val:18,r_t:4,rais:[9,18,19,23,45,56,109],rajathkmp:54,ran:[44,107],rand:[54,107,118,120],random:[2,4,9,18,33,54,67,73,77,78,101,109,118],random_crop:11,random_imag:33,randomli:[11,18,38],randomnumberse:117,rang:[4,9,18,33,40,42,50,54,60,74,78,85,99,111,118,120],rank0:71,rank1:71,rank:[4,18,26,89,103,112],rank_tabl:18,rankdir:56,ranktabl:18,rapid:80,raspberri:125,raspberry_pi:124,raspberrypi:124,raspbian:124,rate:[2,3,4,5,6,10,19,36,86,100],rather:[28,39,54,89,112],ratio:[4,118],raw:[4,18,64],rdma:118,rdma_tcp:[117,118],reach:[31,60,71,107],read:[9,11,18,25,26,28,31,33,40,41,42,44,51,52,78,79,86,89,92,97,98,104,109,112,122,126],read_from_arrai:60,read_from_realistic_imag:26,read_from_rng:26,read_lock:32,read_minibatch:51,read_mnist_imag:26,read_ranking_model_data:26,readabl:[74,79,106],reader:[0,10,25,33,42,50,54,55,75,86,106,109,111],reader_cr:33,reader_creator_bool:78,reader_creator_random_imag:[9,78],reader_creator_random_image_and_label:[9,78],readi:[31,112,113,126],readlockguard:34,readm:66,readwritebuffer_:34,readwritemani:112,real:[4,18,34,54,78,109],realist:26,realiti:86,realiz:[29,68],realli:[51,81],reaon:102,rearrang:18,reason:[5,26,27,31,41,51,99,111,113],recal:[3,14,18],receiv:[18,31,39,41,42,44,68,111],recent:[60,72],reciev:118,recognit:[4,86],recommend:[5,18,26,91,92,93,97,99,100,104,109,110,118,122],recompil:107,record:[9,22,35,63,74,85,112],recordev:74,recordio:[9,10,26,33,35],recov:[31,89],recover:79,recoveri:35,rectifi:4,recurr:[18,61,68,83,86],recurrent_group:[5,86,97],recurrent_lay:5,recurrent_op:89,recurrentgradientmachin:[66,69,89],recurrentgroup:3,recurrentlay:[61,118],recurs:[28,29,30,49,60,79],recv:[40,42,44,71,112],recvparametertyp:34,red:[54,106],redirect:9,reduc:[4,18,44,50,71,79,92,99,106,110,118,120],reduce_by_kei:79,reduce_mean:54,ref:18,ref_batch_dim_idx:18,refactor:[42,44,55,69,72,73,77,81,89],refer:[2,4,5,6,11,18,19,22,27,29,31,35,36,37,40,50,56,63,67,68,71,75,77,79,81,83,87,89,90,91,92,97,98,100,101,103,111,113,122,123],referenc:35,reflect:35,reformat:99,reformul:18,refrain:101,reg:85,regard:[18,126],region:[4,83,107],regist:[41,60,67,80,87,88,100,107],register_gpu_profil:107,register_lay:100,register_op:[53,79,80,85,101],register_op_cpu_kernel:[87,101],register_op_cuda_kernel:[87,101],register_op_kernel:102,register_op_without_gradi:[79,101],register_oper:[58,80],register_tim:34,register_timer_info:107,registerop:85,registr:[101,102,108],registri:[39,58,87,113,126],regress:4,regular:[2,12,18,21,28,100,112],reiniti:63,reinstal:91,rel:[5,18,27,38,81,101,122],relat:[31,38,39,50,67,74,83,88,99,106,113,123,124,126],relationship:[80,87],releas:[82,86,112,122,123,124],relev:101,reli:[27,40,69,70,72,81,101,106],reliabl:[31,81],relu1:56,relu2:56,relu:[4,54,56,60,100],relwithdebinfo:106,remain:[18,89],remaind:18,rememb:[4,99],remot:[2,30,34,42,79,99,100,112,118,120],remoteparameterupdat:[34,37,118],remov:[9,18,28,42,49,51,69,99,118,122,123],renam:[28,49,50,94],reorder:[18,63],reorder_primit:63,reorderlodtensorbyrankt:18,reorgan:4,repeat:[18,29,55,75,76,84,85,90,106],repeatedli:[55,60],replac:[18,30,35,58,72,80,86],repli:99,replic:42,replicaset:39,repo:[30,99,104,124],report:[35,50,51,74,107],reportdataset:35,repositori:[4,104,122],reprenset:18,repres:[4,5,18,28,29,35,40,42,44,45,51,56,64,67,69,72,77,79,81,84,87,89,90,97,100,112],represent:[4,18,36,42,52,54,55,60,67,69,84],reproduc:98,request:[30,31,34,38,40,79,82,99,112,113,126],requir:[3,4,26,28,31,36,38,39,42,44,45,49,50,56,60,62,68,72,74,75,76,79,81,84,85,86,90,94,98,99,100,101,104,109,112,113,122,124,126],requisit:60,rerun:101,research:[10,18,42,51],reserv:[18,49],reserveoutput:100,reset:[4,18,31,46],reset_program:46,resetingrad:62,resetinvalu:62,resetoutgrad:62,resetoutvalu:62,resetxxx:62,reshap:[27,78,103],reshape_s:4,resid:[18,98],resiz:[11,34,87,101,103],resize_s:11,resize_short:11,resolv:[30,99,113],resourc:[42,47,71,74,87,102,112],respect:[18,19,27,45,50,54,68,97,100,118],respons:[4,34,41,42,46,54,71,72,73,81,112,113],rest:[18,29,39,48,52,88,126],restart:[31,36,112,113,126],restartpolici:[112,113],restor:[27,72],restrict:[81,83,106,118],result:[1,3,4,17,18,22,25,27,28,35,46,54,55,56,60,64,69,70,71,73,101,103,106,107,108,112,118],result_fil:3,resum:38,retain:103,retran:112,retriev:[29,69,83,98,100,106,113],retriv:109,return_seq:5,reuqest:82,reus:[29,38,69,78,79,100,101],rev:98,revamp:42,reveal:[26,106],revers:[4,5,18,28,97],review:[10,40,113],reviews_electronics_5:113,rewrit:[30,41,101],rgb:[4,11],rho:6,rid:51,right:[4,27,28,29,30,39,46,60,79,81,99,102],ring:71,risk:28,rkt:[39,98],rmsprop:72,rmspropoptim:72,rnn:[4,5,18,29,51,54,69,77,79,83,86,105,117],rnn_bias_attr:97,rnn_layer_attr:97,rnn_out:97,rnn_output:89,rnn_step:4,rnn_use_batch:[61,117,118],rnnalgorithm:69,rnnlm:10,rnnstep:89,roadmap:[86,89],robust:4,rocmplac:67,roi:4,role:[10,26,35,36,42,71,111,112],rollback:77,root:[6,7,28,71,112,113,123],rot:4,roughli:86,round:[50,71],routin:[50,62,71],row:[3,4,9,18,34,41,100],row_id:4,rows_:84,rpc:35,rpcserver:35,rpi:124,rpi_arm_neon:124,rpi_toolchain:124,rsize:112,rtk:126,rtype:9,rule:[3,18,28,42,45,51,55,100,112],run:[18,26,27,28,29,30,31,39,40,42,43,44,46,50,51,52,53,54,55,56,60,63,67,68,70,71,72,74,75,76,77,79,82,83,84,86,87,88,93,94,95,98,99,100,102,103,104,106,107,109,110,111,112,114,115,116,118,122,123,124,126],run_test:91,runinitfunct:107,runnabl:44,running_on_cloud:39,runserv:104,runtim:[22,29,40,41,42,58,68,79,90,92,102,110,122],runtime_table_:29,s_block:28,s_recurrent_group:97,safe:39,sai:[4,52,55,57,60,78,98,118,120],said:51,sake:100,same:[3,4,5,18,19,23,25,26,27,35,36,38,40,41,42,53,54,56,59,60,68,69,71,77,79,80,83,86,88,89,91,97,101,102,103,110,111,112,120],samping_id:4,sampl:[3,9,10,18,46,54,77,85,92,109,110,118,120],sample_id:3,sample_num:3,sample_weight:18,sampler:54,satifi:[3,60],satisfi:[30,63,84,94,112],save:[4,9,17,25,31,33,35,36,39,40,42,55,56,60,64,72,84,90,98,109,112,113,118,120],save_dir:[113,118,120],save_only_on:[117,118],save_parameter_to_tar:25,save_persist:111,saving_period:[117,118],saving_period_by_batch:[117,118,120],scalabl:79,scalar:[4,18,28,29,57,89],scale:[1,19,42,44,72,80,85,86,101,109],scale_a:18,scale_b:18,scaleop:101,scaleopmak:[79,101],scalingproject:4,scan:[18,28,35,60,79],scatter:[4,28,71],scenario:[69,117],scene:117,schdule:112,schedul:[35,39,44,112],scheduler_factor:2,scheme:[3,6,34,81,101],scienc:60,scope:[15,27,40,43,47,52,88,108],score:[3,4,14,18,69],scorer:86,scp:114,script:[10,71,91,98,101,109,110,112,114,122],sdk:123,search:[4,18,31,83,91,97,118],second:[4,18,26,40,49,51,54,56,68,69,75,76,78,83,85,101,110],secret:112,section:[28,44,51,77,97,99,100,106,112],see:[4,5,18,26,28,31,40,41,44,50,51,77,86,99,101,103,106,107,111,112],seed:[18,107,118],seem:[30,41,50,51,86,94],seen:[19,81,101],segment:[3,18,68,89,103],sel_fc:4,selcet:4,select:[4,18,69,112],selected_generation_scor:69,selected_id:[4,69],selected_indic:4,selected_row:[84,90],selected_rows_desc:[84,90],selected_scor:69,selectedrow:[58,90],selectiv:4,selector:113,self:[27,45,46,54,56,60,61,62,64,70,77,89,100,101],self_addr:40,selfnorm:4,semant:[10,26,69,82],semaphor:41,semat:26,send:[31,36,40,42,44,59,71,79,85,99,109,111,112,118],send_back_parameter_typ:34,send_var:18,sendbackparameterspars:34,sendbackparametertyp:34,sendparameterrequest:34,sendparameterrespons:34,sens:[72,81,99,106],sensit:4,sent:[26,36,40,42,79,85,90,113],sentenc:[4,10,51,68,69,89,97],sentence_input:89,separ:[3,18,22,36,42,53,72,80,81,109,111,118],seper:[18,89],seq0:18,seq1:18,seq2:18,seq3:18,seq:[4,10],seq_len:89,seq_pool:4,seq_silc:4,seq_text_print:3,seq_typ:9,seqtext_evalu:3,seqtoseq:4,seqtoseq_net:4,sequenc:[1,3,4,5,7,9,10,18,28,29,40,47,51,55,61,70,75,86,89,99,100],sequence_group:4,sequence_nest_group:4,sequencesoftmaxop:18,sequencestartposit:4,sequencetextprint:3,sequencetyp:4,sequenti:[4,29,40,41,97],seri:[5,94],serial:[25,29,35,64,73,79],serializ:[79,90],serv:[42,50,79,89,92,107,109,112],server:[18,26,30,34,37,38,42,52,71,79,91,100,110,117,126],server_endpoint:111,serverless:31,servic:[106,109,126],sess:[54,56,70],session:[56,70,76,107],set:[2,3,4,5,9,10,11,18,19,21,22,23,25,26,28,31,39,54,58,60,63,67,68,69,74,76,77,79,80,83,86,87,89,91,97,98,99,100,101,103,104,105,106,107,109,110,111,112,113,117,118,120,123,124],set_active_typ:100,set_attr:45,set_default_parameter_nam:2,set_drop_r:100,set_float_el:27,set_input:[4,45],set_output:45,set_siz:100,set_typ:[45,100],setdatatyp:84,setdefault:101,setp:112,setq:98,settup:100,setup:[42,72,82,100,101,126],seven:86,sever:[3,4,27,34,42,44,54,68,69,71,73,74,77,84,87,89,91,109,110,112,120],sexstant:126,sgd:[25,26,31,39,44,72,73,84,108,109,111],sgd_optim:[108,111],sgdasync_count:117,sgdoptim:20,shall:[28,30],shape:[3,4,9,18,19,25,27,28,29,42,54,57,67,68,75,77,79,84,86,87,101,103,108,111],shard:[31,32,33,34,35,36,38,42,44,109,112],share:[4,18,30,54,66,73,77,79,81,86,87,89,98,101,107,118],shared_bia:5,shared_bias:4,shared_librari:30,shared_ptr:[63,65,66,83,87,103],shell:[92,112],shift:[4,18],shorten:4,shorter:11,should:[2,3,4,9,11,18,19,22,25,26,27,28,29,36,39,41,42,45,46,50,52,53,54,58,59,63,67,68,69,70,72,73,74,75,78,79,80,81,84,85,86,88,89,90,95,97,101,102,104,110,111,112,122],should_be_fals:26,should_be_tru:26,show:[3,6,28,29,31,49,51,57,60,64,68,71,72,75,89,94,98,103,109,112,113,118],show_check_sparse_distribution_log:[117,118],show_layer_stat:[117,118],show_parameter_stats_period:[113,117,118,120],shown:[4,18,26,42,46,71,74,86,97,100,103,107,112],shrink:100,shrink_rnn_memori:18,shrunk:45,shuffl:[9,42,111],sid:112,side:[4,18,25,42,46,60,73,103,109],sig:112,sigint:110,sigma:18,sigmod:85,sigmod_op:85,sigmod_output:85,sigmoid:[4,19,29,85,89,100],sigmoidactiv:5,sigmoidcrossentropywithlogit:18,sign:[48,64,112],signal:110,signatur:112,signific:[86,107],similar:[4,18,29,40,41,42,44,47,51,69,72,74,78,79,81,86,87,88,89,101,106,112,126],similarli:[4,9,51,60,101],simpl:[1,3,4,5,9,10,18,25,40,44,50,52,55,56,60,68,72,75,81,83,85,86,89,107,111,118],simple_attent:97,simple_gru:97,simple_lstm:4,simple_rnn:[4,97],simple_transform:11,simpler:73,simplest:112,simpli:[4,11,26,36,42,92,95,97,107],simplifi:[26,69,77,85,86,100,113],simul:[51,123],simultan:112,sinc:[4,5,18,31,35,37,38,41,42,43,44,51,58,60,63,67,72,77,78,80,81,89,103,107,111,112,123,126],sincer:99,singl:[3,5,9,18,28,31,42,44,46,50,59,71,79,83,86,91,95,100,106,109,113],singleton:[40,43],sinlg:25,sit:[18,42],site:[30,106,112],sitten:18,sittin:18,situat:[28,76,88],size:[3,4,5,9,10,11,18,19,25,31,33,34,36,41,42,50,54,60,64,69,72,77,78,84,85,86,87,89,92,95,97,100,101,103,108,111,118,122,123,124],size_a:4,size_b:4,size_in_byt:63,size_t:[34,87,89,100],sizeof:29,skip:[28,78,99,110,112],slice0:18,slice1:18,slice2:18,slice3:18,slice:[18,40],sliceproject:4,slide:[4,6,10,31],slight:51,slightli:54,slope:[4,18],slopeinterceptlay:4,slow:107,slowli:[98,106],small:[4,10,18,27,40,52,54,62,69,99,100,118],small_messag:[117,118],smaller:[18,27,31,50,69,99],smart:83,smooth:4,snap:113,snapdragon:50,snapshot:[32,38,112],snippet:[53,70,97,100,107,112],sock:39,sock_recv_buf_s:[117,118],sock_send_buf_s:[117,118],socket:118,soft:18,soft_label:18,softmax:[4,5,18,26,29,42,44,51,56,57,69,75,97,100],softmax_param_attr:5,softmax_selfnorm_alpha:4,softmaxoutput:56,softrelu:18,softwar:[50,74,107,126],solid:54,solut:[71,126],solv:[26,28,60,79],some:[2,4,9,11,18,25,26,28,29,30,34,35,36,38,39,42,44,45,50,52,53,54,55,59,60,63,67,68,69,70,75,76,77,78,79,80,83,87,88,89,99,100,101,102,103,107,109,112,117,118,120,122,123,124,126],some_c_api_funct:66,some_inst:66,some_lay:18,some_op:[58,68,89],some_python_class:65,somecppclass:65,somedata:25,somegotyp:65,someth:[28,34,77,98,99,106],sometim:[4,74,78,98,107],somewhat:36,somewher:83,soon:31,sophist:100,sort:[4,10,18,22,89,106,112,118],sort_by_length:89,sortagrad:86,sorted_kei:22,sourc:[4,10,18,27,30,49,51,54,62,64,66,69,78,79,97,98,106,109,112,113,123],source_dict_dim:[69,97],source_dict_s:69,source_language_word:[69,97],space:[3,4,44,50,77,81,86,97,98,107],space_seperated_tokens_from_dictionary_according_to_seq:3,space_seperated_tokens_from_dictionary_according_to_sub_seq:3,span:74,spars:[2,4,6,9,18,34,41,100,103,109,112,118],sparse_binary_vector:9,sparse_binary_vector_sequ:9,sparse_float_vector:9,sparse_float_vector_sequ:9,sparse_non_value_slot:9,sparse_remot:34,sparse_upd:[2,34],sparse_value_slot:9,sparseparam:100,sparseprefetchrowcpumatrix:100,spatial:4,spatial_scal:4,speak:97,spec:[112,113],specfii:118,special:[4,18,28,36,42,50,52,58,67,69,70,101],specialvartypeinfer:58,specif:[18,19,25,28,30,31,42,45,49,52,69,79,83,87,98,101,111,120,122],specifi:[3,4,17,18,22,26,27,34,35,36,39,40,41,42,43,45,46,47,49,54,64,74,77,83,85,89,92,97,98,99,100,103,104,106,112,118,122,123],spectrogram:86,speech:[4,86],speed:[4,5,50,64,71,72,91,126],speedup:74,sphinx:[65,104],split:[4,9,19,38,40,43,51,57,69,79,89,109,112,120],split_count:[109,112],sport:18,spread:28,sqrt_x:18,squar:[4,6,7,56],square_error_cost:[108,111],srand:118,src:[30,63,109],src_backward:97,src_embed:[69,97],src_forward:97,src_primitive_desc:63,src_word_id:[69,97],src_word_vec:69,sreializ:90,srl:10,ssd:4,ssh:[112,113,114,124],ssh_server:110,sstabl:26,stabil:[4,18,27,60,101],stabl:[82,112],stack:[18,47,79,89,112],stackexchang:4,stage:[30,37,43,54,60,63,86,90,110,122],stale:31,stamp:107,stand:18,standalon:122,standard:[2,9,18,41,51,79,81,86,94,98,106],stanford:[10,27,113],star:30,start:[4,5,18,25,28,30,31,34,35,36,38,39,41,42,43,69,71,73,74,91,92,94,97,98,106,107,110,114,118,121],start_mpi_train:114,start_op_idx:28,start_pass:[117,118],start_po:4,start_pserv:118,startup:[18,31,39,51,112],startup_program:18,stat:[107,118],state:[4,5,18,22,29,31,46,47,68,69,74,83,86,89,97,102,113,118],state_act:[4,5],statem:60,statement:[51,55,60,100,112],static_cast:[63,103],staticinput:[4,97],statist:[4,18,46,74,118],statset:107,statu:[18,39,69,107,112,113],status:113,std:[25,30,34,56,58,59,63,65,66,74,76,79,80,83,85,87,100,101,103,118],stdbuf:109,stderr:110,stdout:[9,110],step1:18,step:[4,5,7,18,27,29,31,36,42,44,46,51,54,55,61,69,72,73,77,79,85,86,89,91,92,97,99,100,106,107,109,112,113,114,122,124,126],step_gradi:28,step_id:89,step_input:89,step_net:29,step_output:89,step_scop:79,stepnet:[29,68,79,83],still:[28,35,38,42,51,60,80,94,103],stirng:77,stmt1482205552000:112,stmt1482205746000:112,stochast:[6,31,35,38,72,109],stop:[4,77,98,110,113,118],stop_gradi:[18,77],storag:[48,50,109,112,113],store:[3,4,10,18,25,27,29,30,34,47,56,58,64,67,69,73,75,77,79,80,81,83,89,100,101,103,104,109,112,113,114,118,123,124],str:[11,18,25,28,39,89,120],straight:[75,78,84],straightforward:63,strategi:[7,31,77,118],stream:[9,42,63,74,87,102],stream_:87,streamid:22,street:4,strict:[78,109],stride:[4,5,18,63,67,86],stride_h:18,stride_i:4,stride_w:18,stride_x:4,string:[3,4,9,11,18,22,25,28,29,35,49,56,59,64,74,75,76,77,79,80,83,84,85,90,100,101,112,118],strip:106,strongli:109,struct:[35,36,48,50,58,59,66,67,74,80,85,88,102],structur:[28,29,35,51,54,64,69,75,77,79,84,110,112],sts:112,stuff:99,style:[4,79,85,91],sub:[3,4,9,18,26,28,38,40,44,54,60,68,71,73,77,97,100,122],sub_block:28,sub_nest_seq:4,sub_sequ:4,subclass:77,subcommand:49,subgradi:6,subgraph:[44,54],submiss:42,submit:[63,79,104,109,112,117,118],subnet0:112,subnet:[26,112],subobjectpath:113,subscript:18,subsequ:[4,71],subsequenceinput:4,subset:[18,100],substitut:18,succ:60,succeed:[35,113],success:[4,36,112,113],successfulcr:113,successfulli:101,successor:118,sucess:60,sucessor:60,sudo:[98,112],suffer:27,suffici:118,suffix:[39,94,109],suggest:[4,30,99,107],suit:126,suitabl:[84,87,118],sum:[4,6,19,28,29,32,43,58,77,97,100],sum_op:28,sum_x:18,sume:18,summar:[18,54,74],summari:74,summat:18,sumopgradmak:80,supercomput:60,suppli:[18,84],support:[2,3,4,6,7,9,11,18,27,29,31,38,39,40,41,42,44,51,53,54,60,63,64,67,69,72,73,74,76,78,79,80,81,84,86,88,91,92,94,95,97,98,100,101,103,104,107,109,112,118,122,123,124,126],support_inplac:60,suppos:[5,18,19,30,40,53,84,100],suppress:[4,49],sure:[18,91,98,100,106,111,112],surpass:4,svs:85,swagger:48,swig:[37,65,66,91,122,123],switchop:29,sychron:71,symbol:[4,29,56,66,94],symbols_ready_:29,symbolt:[29,79],symlink:99,sync:[31,72,81,118],sync_with_cpp:106,syncflag:100,synchron:[31,35,41,63,71,74,109,112,118],syntax:[40,47,51,69,78],sysroot:122,system:[29,30,31,36,38,41,42,44,48,53,54,60,62,86,91,92,94,101,104,106,109,113,122],t_max:18,t_min:18,tab:94,tabl:[3,4,18,29,40,51,58,64,84,90,123],tablelookup:84,tablelookupgrad:84,tablelookupop:84,tableproject:4,tag:[3,10,82,92,97,109],tagtyp:3,tail:69,take:[3,4,5,9,18,25,26,28,29,30,31,38,40,41,42,45,47,50,52,54,55,57,58,60,63,67,72,75,76,77,78,79,80,87,88,89,91,97,98,99,100,101,102,106,107,109,111,112,113],taken:[4,45,56,60,67,89],talk:[36,52,124],tangl:106,tanh:[4,5,54,69,97,100],tanhactiv:5,tanhshrink:18,tar:[11,25,112],tarbal:112,target:[4,10,18,25,28,29,30,45,47,54,56,70,76,79,91,97,101,122,123,124],target_block:[28,45],target_dict_dim:97,target_dict_s:69,target_dictionary_dim:4,target_language_embed:4,target_language_word:97,target_link_librari:30,target_var:17,target_word:69,targetinlink:4,task13:86,task14:86,task:[3,4,18,42,64,69,74,85,97,111,120],task_queu:35,taskentri:35,taskqueu:35,tbd:[37,63,86],tcp:[112,118],tear:107,technic:[28,31,111],techniqu:[18,60,97,100,106],technolog:[18,51,98],tee:113,tell:[22,31,35,36,69,85,92,107,122],templat:[53,63,85,87,101,102,103,113,126],tempor:[4,18,86],temporari:[28,39,47,60,72,77],tempori:60,ten:98,tensor:[19,27,30,40,41,43,44,50,51,52,54,56,58,59,63,64,67,68,69,84,89,90,101,108],tensor_arrai:40,tensor_array_read:89,tensor_array_s:89,tensor_array_stack:89,tensor_array_unstack:89,tensor_array_writ:89,tensor_data:64,tensor_in:88,tensor_s:27,tensor_test:30,tensor_to_check:27,tensorarrai:43,tensorarraydesc:89,tensordesc:[64,84],tensorflow:[29,40,41,42,44,51,54,57,81,89,103],term:[4,5,18,31,80,81,86],termin:113,terminolog:60,tessorarrai:89,test100:10,test10:10,test1:33,test:[4,9,10,11,18,25,26,27,30,56,66,72,78,82,95,98,103,107,108,109,111,114,117],test_:101,test_all_data_in_one_period:113,test_check_grad_ingore_i:101,test_check_grad_ingore_x:101,test_check_grad_norm:101,test_check_output:101,test_data_dir:109,test_fcgrad:100,test_gpuprofil:107,test_layergrad:100,test_mkldnn:62,test_mklpack:61,test_mul_op:[91,101],test_norm:101,test_pass:[117,118,120],test_period:[117,118,120],test_recurrent_op:99,test_wait:[117,118],testa:26,testb:26,testbilinearfwdbwd:107,testcas:101,testconfig:100,testfcgrad:100,testfclay:100,testlayergrad:100,testmodel_list:117,testmulop:101,testq:26,testresult:25,testsave_dir:117,testutil:100,text1:49,text:[3,5,9,18,26,64,68,74,86,112],text_fil:9,tflop:107,tftp:126,tgz:[10,94],than:[2,3,4,5,18,28,31,39,40,45,51,52,53,54,77,79,81,89,91,97,98,100,109,111,112,122,126],the_step:51,theano:51,thehalf:18,thei:[4,18,26,28,30,31,36,38,40,41,44,45,49,51,54,55,59,60,69,70,74,77,79,85,89,90,97,98,99,100,101,103,107,109,111,112,117],them:[3,4,5,11,18,26,27,28,30,31,34,39,41,44,45,51,52,53,58,59,60,69,77,78,79,80,83,84,85,88,89,90,98,99,101,104,107,111,112,117,118],themselv:[28,30],theori:[51,107],therefor:[28,60,72],therein:[4,18,29],theta:54,theta_d:54,theta_g:54,thi:[2,3,4,5,6,9,10,11,17,18,22,23,25,26,27,28,29,30,31,34,35,36,37,38,39,40,41,42,43,44,45,46,47,50,51,52,53,54,55,56,59,60,63,67,68,69,70,71,72,73,74,75,77,78,79,80,81,84,85,86,87,88,89,92,94,95,97,98,99,100,101,102,103,104,106,107,108,109,111,112,113,118,120,122,123,124,126],thin:58,thing:[42,54,79,87,107],think:[26,30,111],third:[4,18,31,56,101,106,107,122,123,124],third_parti:[4,62,122,123,124],thirt:98,those:[4,29,30,31,53,55,56,57,75,122],though:[89,126],thought:[18,30,107],thread:[40,41,43,74,100,106,107,118,120],thread_count:43,thread_id:74,thread_id_:74,thread_local_rand_use_global_se:[117,118],thread_pool:43,threadblocks:22,threadid:120,threadloc:107,threadpool:40,three:[3,4,18,19,27,28,31,41,46,50,51,52,55,63,69,70,73,74,75,78,86,87,118,122],threshold:[2,3,4,18,31,35,45,99,118],thresholdedrelu:18,through:[4,19,28,30,31,35,37,46,60,70,72,97,100,101,104,107,108,109,123],throughout:47,throughput:[107,109],thrust:79,thu:[4,18,38,46,56,60,86,100,112],tier:113,time:[4,5,7,9,18,22,25,26,27,30,31,35,38,41,42,44,45,51,53,58,60,61,67,68,69,71,74,77,78,79,80,84,85,86,89,90,91,97,98,103,106,107,113,118,120,126],timelin:[4,74,79,107],timeo:112,timeout:[31,35],timeout_sec:9,timer:22,timestamp:[4,32],timestep:[4,83],tip:[122,123],titan:60,titl:10,tls:48,tmp:77,to_chw:11,to_no_sequ:4,to_sequ:4,to_tar:25,todo:[3,9,10,29,31,35,38,69,85,86],toend:4,togeth:[4,5,9,25,28,89,97,111],token:[3,4,18,26,86,97],toler:[25,27,91,101],too:[10,27,40,41,45,63,88,89],took:126,tool:[74,91,94,97,98,106,111,112,122,124],toolchain:[106,122,123],toolkit:86,top:[3,18,25,68,69,86,101],top_k:[3,18,69],top_level_rnn:68,topic:63,topk_generated_scor:69,topk_id:69,topk_indic:18,topk_out:18,topk_scor:69,toplevel:98,topolog:[26,31,42,56,60,64,73],topoloi:56,topolopi:25,torch:[29,51],toronto:10,total:[18,22,25,31,44,46,71,74,78,106,107,109,113,126],total_pass:78,tottim:106,toward:51,trace:[29,52,54],track:[31,35,56,77],tractabl:4,tradit:[4,29,50,86],traffic:42,trail:9,train100:10,train10:10,train:[0,2,3,4,9,10,11,18,28,29,33,35,36,38,40,45,46,47,51,52,54,55,60,61,64,71,72,73,74,75,76,77,79,81,84,86,87,90,97,100,105,107,115,116,117,123],train_config_dir:112,train_data:109,train_data_dir:109,train_id:112,train_list:109,train_loop:51,train_read:[42,111],trainabl:[4,21,64,77],traindot_period:117,trainer:[26,32,33,34,35,37,42,44,52,61,62,72,73,79,100,110,111,118,120],trainer_config:[112,113],trainer_config_help:100,trainer_count:[95,109,112,113,117,118,120],trainer_cpu:39,trainer_cr:39,trainer_gpu:39,trainer_id:[109,112,118],trainer_intern:34,trainer_mem:39,trainer_packag:39,trainer_prog:42,trainerid:38,training_rol:111,trainingjob:42,trainingtest_period:117,trainonebatch:34,tran:[63,100,118],trans_var:88,transact:[31,35],transcript:86,transfer:[60,74],transform:[4,5,11,18,79,86,97,100,103],transform_param_attr:5,transformed_st:5,translat:[4,5,60],translation_id:69,translation_scor:69,transpar:[69,110],transpil:[40,111],transport:118,transpos:[4,11,100],transpose_i:18,transpose_x:18,transposedfullmatrixproject:4,travers:[28,55,60],travi:99,treat:[4,18,29,36,60],treatment:[36,50],tree:[4,29,40,47,51,77,108,118,124],trg_dic_siz:69,trg_embed:[69,97],trick:69,tricki:65,trigger:[38,73],trim:4,trivial:[69,89],true_block:[29,57,75],true_imag:78,true_label:78,true_neg:46,true_posit:46,true_read:78,truth:[3,4,18],tune:[2,86,105,106],tuninglog_barrier_abstract:117,tupl:[4,5,9,10,11,18,25,28,77,78],ture:4,turn:[4,18,77,78,92],tutori:[22,91,92,97,100,101,106,107,112,114,115,116,123],twice:[44,54,111],twine:82,two:[3,4,5,18,19,26,28,36,37,38,39,40,41,42,46,49,50,51,52,54,55,58,60,64,67,69,72,74,75,78,79,80,81,83,84,85,86,88,89,90,91,97,101,103,107,110,112,120,122,124],txt:[30,39,49,61,62,100,104,109,112,114],type:[3,4,5,7,9,10,11,18,19,23,25,26,28,29,31,34,35,38,39,42,48,49,50,52,58,59,63,64,65,66,68,69,75,76,77,78,79,80,81,84,85,86,87,88,90,95,97,98,100,102,103,109,112,113,118,120,123],type_nam:85,typedef:[36,50,65,66,67,87,102],typeerror:45,typeid:85,typenam:[53,85,87,101,102,103],typic:[3,42,107,123],ubuntu:[82,94,95,106],ubyt:78,uci:10,uci_h:[95,111],ufldl:[4,18],uid:113,uint16_t:50,uint32:[48,64],uint32_t:74,uint64:[64,65],uint64_t:65,unawar:36,unbalanc:118,unbound:[60,97],unchang:18,unclear:38,uncreat:28,under:[18,30,35,44,71,88,91,92,103,104,109,110,112],underli:[18,69],understand:[51,77,86,106,107,126],understand_senti:97,undeterminist:107,uni:86,unidirect:[4,86],unifi:[47,56,84,99],uniform:[2,4,9,18,33,54,77,78,118],uniform_random:77,uniforminiti:16,uniniti:28,uninstal:91,uniqu:[26,29,31,38,39,63,67,77,83,101,109,111,112,118],unique_nam:77,unique_name_gener:77,unique_ptr:[80,83,87,100],unit:[4,5,18,19,30,72,74,81,87,91,97,98,103],unitest:18,unittest:[66,99,101],unittestcheckgrad_ep:117,unix:41,unk:[84,90],unknown:[4,18,23],unless:18,unlik:[4,18,69,101],unnecessari:[28,86,99],unnorm:18,unordered_map:83,unpack:89,unrol:68,unscal:18,unseen:81,unseg:4,unsign:[36,50],unstack:89,unstack_from:89,unsupervis:54,unsupport:101,until:[31,36,43,44,51,60,83,111,112],unus:18,unzip:122,updat:[2,4,6,18,28,31,35,36,42,48,50,54,68,69,70,71,72,73,83,86,89,94,100,106,109,118,120],update_equ:25,update_hook:2,update_memori:29,update_op:70,updatecallback:100,updatestack:112,upgrad:[71,91,94],upload:[31,39,41,48,82,109],upon:31,upper:4,upstream:99,uri:112,url:[9,10,99],usag:[3,4,5,11,18,25,50,57,60,73,77,101,107,109,123],use:[2,3,4,5,7,9,10,11,18,22,25,26,27,29,30,31,37,42,43,44,47,50,54,56,58,59,60,63,67,69,70,71,73,74,77,83,84,85,86,88,89,90,91,92,94,95,97,98,99,100,101,102,104,106,107,109,112,113,118,120,122,123,124],use_cpu:59,use_cudnn:[18,19,59],use_eigen_bla:122,use_eigen_for_bla:[122,123],use_etcd:25,use_global_stat:4,use_gpu:[95,109,113,117,118,120],use_mkl_pack:61,use_mkldnn:[4,59,62],use_old_updat:[34,117,118],use_peephol:18,use_sparse_remote_updat:34,used:[3,4,5,6,7,9,10,11,18,22,25,26,27,29,30,31,37,38,42,45,47,50,51,54,56,60,68,69,72,73,74,77,78,79,81,83,85,87,88,89,91,94,97,98,100,101,103,106,107,112,117,118,120,122,123,124],useful:[4,5,27,50,60,77,83,88,97,100,120,122],usegpu:100,user:[2,4,5,9,10,11,18,22,25,26,27,28,29,30,33,35,38,39,40,42,43,44,45,46,47,49,53,54,55,56,58,59,63,67,69,70,71,72,74,77,78,79,80,81,83,85,87,88,89,92,99,102,104,106,109,112,117,118,122,126],user_info:10,user_nam:33,usercert:33,userinfo:10,userkei:33,usernam:33,uses:[4,18,31,38,40,41,42,50,60,67,68,69,73,74,87,88,91,94,97,98,99,100,103,104,109,112,118,122],using:[2,4,5,9,14,18,25,26,28,29,30,31,35,36,38,39,41,42,47,49,50,51,53,54,56,58,60,68,70,72,75,77,78,80,81,83,85,86,87,91,92,93,94,95,97,99,100,101,102,103,104,107,109,112,113,114,118,120,122,124],usr:[91,92,109,112,118],usual:[4,18,25,28,39,60,67,74,75,81,87,99,101,106,107,112,118,120],util:[42,61,62,71,97,100,101,102,107,126],uuid:[32,38],v7a:122,v8a:122,val:28,valid:[4,11,18,78,79,83,101,112,123],valu:[2,3,4,7,9,10,11,18,19,22,25,27,28,29,31,40,41,45,46,56,57,60,62,64,68,69,70,72,73,75,79,83,84,85,89,90,97,100,101,102,111,112,118,120,122,123],value1:118,value2:118,value_:84,value_evalu:3,value_rang:9,valueerror:[18,19,56],values_:89,vanilla:97,var_nam:[28,88],var_recurs:45,vardesc:[29,55,75,77,79,84],vardescbuild:29,vari:[107,112],variabl:[6,9,10,17,18,19,26,27,29,40,42,44,45,46,47,52,54,55,56,57,58,67,68,69,70,72,75,76,80,81,84,85,86,88,89,100,101,102,106,108,109,112,113,122,123],variablenamemap:101,varialbl:54,varianc:4,variant:[4,58,67,87,89,102],varibal:28,varibl:56,varienc:89,varient:89,variou:[29,41,50,60,81,122],varproto:85,vars_:[29,83],vartyp:[18,84,90],vartypeinfer:58,vec1:4,vec2:4,vec2seq:86,veclib:123,vecter:18,vector:[4,5,9,10,18,26,29,34,36,56,57,63,68,69,74,77,79,80,84,86,89,97,100,103],vectorenable_parallel_vector:117,vendor:30,verb:10,verbos:[49,99],veri:[4,7,30,35,40,44,47,51,53,54,60,63,69,73,78,81,83,86,87,89,97,106,107,110],verifi:[29,100,123],version:[4,5,28,30,39,42,45,49,52,54,56,57,64,69,82,86,91,92,95,98,100,106,107,109,111,112,113,117,118,122,123,124],versu:26,vertic:4,vgg:5,via:[18,28,31,67,93,99,107,111,112,123,126],view:[4,64,67],vim:92,viriabl:109,virtual:[45,58,59,80,87,98,102],virtualenv:98,visibl:[38,83],visit:[25,28],visual:[4,69,107],vlog:[34,99],vocabulari:86,voila:95,volum:[104,113],volumemount:[112,113],volumn:112,vutbr:10,w_f:18,wai:[3,5,18,26,28,36,38,41,47,51,59,60,69,72,77,78,81,89,97,98,99,100,120],wait:[31,36,43,102,109,111,118],walk:123,wang:18,wangkuiyi:30,want:[4,22,26,39,40,41,46,54,59,67,72,74,76,78,81,83,87,88,89,91,92,98,99,100,104,106,109,118,120,122,124],warn:[25,49],warp:[4,18,107],warp_ctc:86,warpctc:4,wast:71,watch:31,wbia:112,web:[104,106],websit:104,weight:[3,4,5,6,18,19,21,23,61,64,81,97,100,118,120],weight_act:5,weightlist:100,weights_:100,weights_primitive_desc:63,weights_t:100,welcom:[30,86],well:[18,28,39,41,42,44,51,53,54,81,84,86,100,111,112,118],wer:86,were:[3,30,41,51],west:112,wget:122,wgt:63,what:[2,4,30,51,54,69,77,85,88,101,106,126],whatev:[98,109],wheel:94,when:[2,3,4,6,9,18,19,25,27,28,29,30,31,34,35,36,39,40,42,44,45,46,47,49,50,51,52,56,69,71,72,73,74,75,77,79,87,89,91,93,97,98,99,100,101,103,104,106,107,109,112,113,118,120,122,123,126],whenev:[18,77,86,99],where:[4,5,6,18,19,26,28,29,31,38,40,42,51,52,55,67,68,69,72,75,79,81,87,89,97,100,101,106,107,108,118,120],wherea:[18,29,35,53,57,87,90],whether:[3,4,11,18,25,27,28,29,47,74,78,84,89,91,92,100,101,118,123],which:[2,3,4,5,9,10,11,17,18,19,21,22,25,26,27,28,29,30,31,33,35,36,38,39,40,41,42,43,45,47,50,51,52,53,54,56,58,60,63,64,67,68,69,70,71,73,75,76,77,78,79,80,83,84,85,88,89,90,91,94,97,98,99,100,101,102,103,106,107,109,110,111,112,118,120,122,123,126],while_grad:60,while_loop:[69,89],while_op:[18,28],whileloop:89,whileop:29,white:86,whl:91,who:[28,53,55,71,77,99],whoever:36,whole:[3,9,28,54,57,60,65,66,68,71,76,85,86,99,109,112,113,126],whose:[4,9,18,27,28,31,38,45,68,79,80,85,89,97],why:[5,27,66,98],wide:[30,45,54,94,110,114],width:[3,4,9,11,18,34,65,78,100,101],wiil:18,wiki:[4,30],wikipedia:[4,10],window:[4,7,10,72,86,92,98,122],wirt:56,wise:[4,11,18,44,79,86,103],wish:[91,94,104,109,111],with_avx:[91,92,109,122,123],with_bia:85,with_c_api:[91,122,123,124],with_distribut:111,with_doc:91,with_doubl:[91,100,109],with_dso:91,with_golang:[91,122],with_gpu:[91,98,109,122,123],with_mkl:[61,62,91,122],with_mkldnn:62,with_mklml:62,with_profil:107,with_python:[91,109,122,123],with_rdma:[109,122,123],with_style_check:[91,99],with_swig_pi:[91,122,123],with_test:[91,101],with_tim:[107,109],within:[4,18,35,42,51,86,123],without:[3,4,28,31,36,41,74,77,78,79,86,101,106,109,114],wloop:89,wmt14:97,wmt_shrinked_data:10,won:[107,109],word2vec:[39,109],word:[3,4,10,28,44,55,58,60,68,69,79,85,86,89,97,102,109,120],word_dict:[109,114],word_idx:10,word_vector_dim:[4,69,97],wordcount:86,words_freq_sort:10,work:[4,9,18,26,29,30,31,42,47,50,51,59,70,72,74,77,92,97,98,99,100,104,106,107,109,111,112,113,118,126],worker:[44,90,112],workercount:112,workflow:[79,112],workspac:[99,109,110,118],world:109,worth:108,would:[22,25,29,30,31,38,41,42,43,44,51,53,54,55,63,70,72,73,77,78,84,86,89,92,98,99,106,111,112,122,126],wouldn:[51,55],wrap:[18,51,53,54,71,126],wrapper:[5,30,41,53,71,72,80,89,107],write:[9,18,26,31,38,40,42,44,50,51,52,53,56,58,63,70,72,77,78,79,80,87,89,98,105,109,112],write_lock:32,write_to_arrai:60,writer:[26,77],written:[18,22,28,29,40,44,47,54,64,72,79,80,84,91,92,101,103,106,110],wrong:78,wrote:56,wsize:112,www:[10,18],x64:[122,124],x86:123,x86_64:[122,123],x_first_step:18,x_last_step:18,x_neg:27,x_num_col_dim:18,x_po:27,x_reshap:18,x_t:18,x_t_data:18,x_transpos:18,xarg:[3,92,100,114],xavier:18,xavieriniti:[16,18],xcode:123,xcodebuild:123,xeon:102,xgbe0:118,xgbe1:118,xmap_read:9,xpu:51,xrang:[27,51,54,74,78,95,100],xx_layer:59,xxx:[26,89],xxxx:32,xxxxxxxxx:112,xxxxxxxxxx:112,xxxxxxxxxxxxx:112,xxxxxxxxxxxxxxxxxxx:112,y_dim:54,y_neg:27,y_num_col_dim:18,y_po:27,y_predict:[18,95,108,111],yaml:[30,110,112,114,126],yancey1989:39,yann:10,yapf:99,year:51,yeild:25,yep:[74,106],yet:[51,86,126],yield:[9,26,33,78],you:[2,4,5,9,18,19,22,25,27,39,42,50,83,91,92,93,94,95,97,98,99,100,101,102,104,106,107,109,110,111,112,114,118,120,122,123,124,126],your:[4,9,25,26,30,34,39,49,79,91,93,94,98,99,100,104,107,109,110,111,112,120,122,123,124,126],your_access_key_id:112,your_secrete_access_kei:112,your_source_root:66,yourself:91,yuang:51,yuyang18:[9,10],yuyang:106,z_dim:54,z_size:54,zero:[2,4,5,6,9,10,27,28,31,54,69,73,77,84,100,112,118],zip:[10,77,122],zone:112,zxvf:112},titles:["API","Activation","Parameter Attribute","Evaluators","Layers","Networks","Optimizer","Pooling","Data Reader Interface and DataSets","Data Reader Interface","Dataset","Image Interface","Fluid","data_feeder","evaluator","executor","initializer","io","layers","nets","optimizer","param_attr","profiler","regularizer","Model Configuration","Training and Inference","PaddlePaddle Design Doc","Auto Gradient Checker Design","Backward Building","Design Doc: Block and Scope","Required CMake Function","Design Doc: Distributed Training","\u6a21\u578b\u53c2\u6570\u68c0\u67e5\u70b9\uff08Checkpointing\uff09","\u8bad\u7ec3\u6570\u636e\u7684\u5b58\u50a8\u548c\u5206\u53d1","Alalysis of large model distributed training in Paddle","Design Doc: Master Server","Design Doc: The Client Library of Parameter Server","Design Doc: Remote Parameter Updater for Cluster Train","Design Doc: Save Model","Submit a Distributed Training Job","Design Doc: Concurrent Programming with Fluid","Design Doc: CSP in PaddlePaddle Fluid","Design Doc: Distributed Training Architecture","Design Doc: Execute the Program with Multi CPU","Design Doc: Parameter Server","Error Clip","Evaluator Design","Executor Design Doc","FileManager\u8bbe\u8ba1\u6587\u6863","PFSClient","Design Doc: float16","Design Doc: PaddlePaddle Fluid","PaddlePaddle Fluid: Towards a Compiled Programming Language","Design Doc: Functions, Operators, and Layers","Design for GAN","Design Doc: Computations as a Graph","Survey on Graph","TheIfElseOperator","Design Doc: InferVarType","Problem","Memory Optimization","Intel\u00ae MKL Packed on PaddlePaddle: Design Doc","Intel\u00ae MKL-DNN on PaddlePaddle: Design Doc","Design Doc: Add MKLDNN Kernel in Fluid Operator","Design Doc: Model Format","Paddle\u591a\u8bed\u8a00\u63a5\u53e3\u5b9e\u73b0","C-API \u6a21\u578b\u63a8\u65ad\u5b9e\u73b0\u6587\u6863","Design Doc: The Keys of Operator Kernel Type","RNNOp design","Design: Sequence Decoder Generating LoDTensors","Optimizer Design","Design Doc: NCCL support in Paddle Fluid","Averaging Parameter in PaddlePaddle","Design Doc: The C++ ClassParameters","Introduction","Design Doc: PaddlePaddle Programs","Prune","Design Doc: Python API","Python Data Reader Design Doc","Design Doc: Refactorization Overview","Design Doc: Gradient Operators Registration","Regularization in PaddlePaddle","PaddlePaddle\u53d1\u884c\u89c4\u8303","Design of Scope in Paddle","Design Doc: Selected Rows","Interaction between C++ and Python","DeepSpeech2 on PaddlePaddle: Design Doc","Design Doc: Supporting new Device/Library","Background","Design for TensorArray","Background","Build from Sources","Run in Docker Containers","Install and Build","Install Using pip","GET STARTED","RNN Models","RNN Configuration","Build using Docker","Contribute Code","Write New Layers","How to write a new operator","Add Kernels for a New Device","How to use Eigen in Paddle","Contribute Documentation","HOW TO","Profiling the Python Code","Tune GPU Performance","PaddlePaddle Fluid Source Code Overview","Distributed Training","Cluster Training Using Fabric","Fluid Distributed Training","Distributed PaddlePaddle Training on AWS with Kubernetes","PaddlePaddle On Kubernetes","Cluster Training Using OpenMPI","<no title>","<no title>","Argument Outline","Detail Description","Set Command-line Parameters","Use Case","PaddlePaddle Documentation","Build PaddlePaddle for Android","Build PaddlePaddle for iOS","Build PaddlePaddle for Raspberry Pi","MOBILE","Cluster bootstrapping tool survey"],titleterms:{"\u4e0a\u4f20\u8bad\u7ec3\u6587\u4ef6":33,"\u4e0d\u4f7f\u7528":65,"\u4e0d\u4f7f\u7528swig\u8fd9\u79cd\u4ee3\u7801\u751f\u6210\u5668":65,"\u4e0d\u5bfc\u51fapaddle\u5185\u90e8\u7684\u7ed3\u6784\u4f53":65,"\u4e0d\u5f15\u7528\u5176\u4ed6\u52a8\u6001\u5e93":65,"\u4ec5\u4ec5\u4f7f\u7528void":65,"\u4ece\u5feb\u7167\u6062\u590d":32,"\u4f7f\u7528\u52a8\u6001\u5e93\u6765\u5206\u53d1paddl":65,"\u4f7f\u7528\u8f6c\u6362\u5e93":33,"\u5177\u4f53\u67d0\u79cd\u7c7b\u578b\u7684\u5934\u6587\u4ef6":66,"\u5177\u4f53\u67d0\u79cd\u7c7b\u578b\u7684\u5b9e\u73b0\u6587\u4ef6":66,"\u5206\u5757\u6587\u4ef6\u4f20\u8f93":48,"\u5206\u652f\u89c4\u8303":82,"\u52a0\u901f\u6267\u884c":32,"\u52a8\u6001\u5e93\u4e2d\u4e0d\u5d4c\u5165\u4efb\u4f55\u5176\u4ed6\u8bed\u8a00\u7684\u89e3\u91ca\u5668":65,"\u52a8\u6001\u6269\u5bb9":32,"\u539f\u56e0":65,"\u539f\u56e0\u5217\u8868":65,"\u53c2\u8003\u6587\u6863":48,"\u53d1\u5e03docker\u955c\u50cf":82,"\u53d1\u5e03wheel\u5305\u5230pypi":82,"\u540d\u8bcd\u89e3\u91ca":48,"\u57fa\u672c\u8981\u6c42":65,"\u5b9e\u73b0":65,"\u5b9e\u73b0\u65b9\u5f0f":66,"\u5bfc\u51fac":65,"\u5feb\u7167\u4fdd\u5b58\u7684\u8bbe\u8ba1\u5982\u4e0b":32,"\u6307\u9488\u4f5c\u4e3a\u7c7b\u578b\u7684\u53e5\u67c4":65,"\u63a8\u6d4b\u6267\u884c":32,"\u652f\u6301\u7528\u6237\u81ea\u5b9a\u4e49\u7684\u6570\u636e\u9884\u5904\u7406job":33,"\u6587\u4ef6\u4f20\u8f93\u4f18\u5316":48,"\u6587\u4ef6\u8bbf\u95ee\u65b9\u5f0f":33,"\u6587\u4ef6\u8bbf\u95ee\u7684\u6743\u9650":33,"\u6587\u4ef6\u9884\u5904\u7406":33,"\u66b4\u9732\u63a5\u53e3\u539f\u5219":66,"\u672f\u8bed":32,"\u67b6\u6784\u56fe":48,"\u6846\u67b6\u751f\u6210":48,"\u6982\u5ff5\u89e3\u91ca":33,"\u6a21\u5757":48,"\u6a21\u578b\u53c2\u6570\u68c0\u67e5\u70b9":32,"\u6a21\u578b\u63a8\u65ad\u5b9e\u73b0\u6587\u6863":66,"\u6d41\u7a0b\u4ecb\u7ecd":33,"\u751f\u6210sparse\u6587\u4ef6":48,"\u7528\u6237\u4f7f\u7528\u6d41\u7a0b":48,"\u76ee\u5f55\u7ed3\u6784":66,"\u76ee\u6807":48,"\u793a\u4f8b\u7a0b\u5e8f":33,"\u7b26\u53f7":65,"\u7c7b":65,"\u7f16\u8bd1\u9009\u9879":66,"\u7f29\u5bb9":32,"\u800c\u662f\u624b\u5199\u591a\u8bed\u8a00\u7ed1\u5b9a":65,"\u80cc\u666f":65,"\u8986\u76d6\u4e0d\u4e00\u81f4\u7684\u90e8\u5206":48,"\u8bad\u7ec3\u6570\u636e\u5b58\u50a8":33,"\u8bad\u7ec3\u6570\u636e\u7684\u5b58\u50a8\u548c\u5206\u53d1":33,"\u8f6c\u6362\u5e93":33,"\u8fd9\u4e2a\u52a8\u6001\u5e93\u4f7f\u7528c99\u6807\u51c6\u7684\u5934\u6587\u4ef6\u5bfc\u51fa\u4e00\u4e9b\u51fd\u6570":65,"\u8fdb\u884c\u8bad\u7ec3":33,"abstract":[42,43,44,71,126],"book\u4e2d\u6240\u6709\u7ae0\u8282":82,"case":[28,120],"class":[54,73,77,100],"filemanager\u8bbe\u8ba1\u6587\u6863":48,"final":59,"function":[30,53,54,77],"new":[87,100,101,102],"paddle\u52a8\u6001\u5e93\u4e2d":65,"paddle\u591a\u8bed\u8a00\u63a5\u53e3\u5b9e\u73b0":65,"paddlepaddle\u53d1\u884c\u89c4\u8303":82,"paddlepaddle\u56de\u5f52\u6d4b\u8bd5\u5217\u8868":82,"return":[77,78],"switch":[63,87],"tensor\u5230eigentensor\u7684\u8f6c\u6362":103,"while":18,AWS:112,Abs:1,DNS:112,E2E:111,EFS:112,For:[30,113],KMS:112,Not:98,The:[29,36,40,47,51,54,55,57,58,67,70,73,79,80,124],Use:[29,75,104,109,113,120],Using:[30,36,94,110,114,122],With:39,about:54,abs:18,absolut:69,access:112,account:112,accuraci:[14,18],action:[61,62],activ:[1,4,62],actor:41,adadelta:6,adagrad:[6,20],adam:[6,20],adamax:[6,20],add:[60,63,102,112],address:112,addto:4,advanc:87,aggreg:4,aggregatelevel:4,alalysi:34,algorithm:[27,31,42,68,76],all:[83,89],analog:40,analysi:[42,60],android:122,api:[0,42,61,62,66,70,72,77,81,85],append_regularization_op:23,appendix:126,approach:107,arbitrari:51,architectur:[42,74,97],argument:[49,78,109,117,120,122,124],arrai:27,array_length:18,array_read:18,array_to_lod_tensor:18,array_writ:18,asset:112,assign:18,associ:[83,112],assumpt:126,async:118,attent:97,attribut:[2,60,81],auc:3,auto:27,averag:72,avg:7,aws:112,backgraound:27,background:[44,61,87,88,89,90,101,102],backward:[28,51,55,79,101],base:[39,69],basepool:7,basic:[60,87,126],batch:78,batch_norm:[4,18],batch_siz:78,beam:[69,86],beam_search:[4,18],beam_search_decod:18,benchmark:[61,62],benefit:[44,79],between:[26,41,77,79,85,87],bidirectional_gru:5,bidirectional_lstm:5,bilinear_interp:4,binari:29,bind:101,bla:91,block:[29,52,54,55,75,77,79],block_expand:4,blockdesc:75,blockguard:18,blockguardserv:18,blockguardwithcomplet:18,book:92,bool:91,bootstrap:126,bottleneck:106,brelu:[1,18],bring:126,bucket:112,build:[28,54,79,91,93,98,104,113,122,123,124],built:107,can:83,capi:66,capi_priv:66,cast:18,ceil:18,challeng:[28,44,76],chang:69,channel:41,check:[4,27,100,110],checker:27,checkpoint:[31,32,38],choic:59,choos:[30,112],chunk:3,chunk_ev:18,chunkevalu:14,cifar:10,classif:3,classification_error:3,classification_error_print:3,client:36,clip:[4,18,45],clip_by_norm:18,close:27,cloudform:112,cluster:[37,109,110,111,112,114,120,126],cmake:[30,61,62,124],code:[39,52,77,99,106,108],column_sum:3,command:[109,119,120],commit:113,common:118,commun:118,compar:126,comparis:77,compat:51,compil:[29,50,52,75,79,91,101,108,122,123,124],complet:51,compos:78,comput:[29,55,63,79,81,103],con:126,concat:[4,18],concept:[77,79,112],concern:62,conclus:[38,56,126],concurr:[40,41],condit:54,conditionalblock:18,config:120,configur:[24,97,105,112,123],conll05:10,connect:4,constant:16,construct:55,contain:[92,113],content:[61,62,66,86,107,112],context:102,context_project:4,contribut:[99,104],control:[60,79],control_flow:18,contruct:60,conv2d:18,conv2d_transpos:18,conv:4,conv_oper:4,conv_project:4,conv_shift:4,convert:38,convolut:86,core:[27,77,112],corner:28,cos_sim:[4,18],cost:4,cpu:[43,120],creat:[28,41,78,79,83,112,113],create_arrai:18,create_global_var:18,create_paramet:18,create_tensor:18,creation:[35,72,81],creator:78,credenti:112,crf:4,crf_decod:[4,18],cross:[122,123,124],cross_channel_norm:4,cross_entropi:18,cross_entropy_cost:4,cross_entropy_with_selfnorm_cost:4,csp:41,ctc:[4,86],ctc_error:3,ctc_greedy_decod:18,cuda:[50,91],cuda_profil:22,cudnn:91,cudnnavg:7,cudnnmax:7,current:[50,80],custom:78,data:[4,8,9,18,31,42,78,88,112,113],data_feed:13,datafeed:[9,13],dataflow:60,dataprovid:118,dataset:[8,10,31,35,109],datatyp:[9,67],decayedadagrad:[6,20],decod:69,decor:78,deep:[29,51],deepspeech2:86,defin:[101,112],definit:90,delet:112,demo:[54,111,112],dens:38,dep:94,depend:[54,86,91,94],deploi:39,deriv:100,describ:[51,70],descript:[49,79,118],design:[26,27,29,31,35,36,37,38,40,41,42,43,44,46,47,50,51,53,54,55,58,61,62,63,64,67,68,69,70,71,73,75,77,78,79,80,83,84,86,87,89],destroi:[83,112],detail:[34,86,118],detect:[3,4],detection_map:3,detection_output:4,develop:[79,98,105],devic:[18,87,102,120],devicecontext:87,dictionari:78,differ:[79,87,109,120],directori:112,discrimin:54,discuss:[44,54],dispatch:[31,35],distribut:[26,31,34,39,42,44,109,111,112,118],dnn:62,doc:[26,29,31,35,36,37,38,40,41,42,43,44,47,50,51,53,55,58,61,62,63,64,67,71,73,75,77,78,79,80,84,86,87],docker:[39,92,98,113,122],document:[104,121],doe:78,dot_prod:4,dot_product_attent:5,dotmul_oper:4,dotmul_project:4,down:112,download:[112,113],dropout:[4,18],dure:[69,78],dylib:66,dynam:[31,89],dynamic_gru:18,dynamic_lstm:18,dynamic_lstmp:18,dynamicrnn:18,dynet:56,each:94,ec2:112,edit_dist:18,eigen:103,elast:112,elect:38,elementwise_add:18,elementwise_div:18,elementwise_max:18,elementwise_min:18,elementwise_mul:18,elementwise_pow:18,elementwise_sub:18,els:29,elu:18,embed:[4,18],engin:54,enough:27,entri:78,environ:[39,122],eos:4,equat:100,error:45,evalu:[3,14,46],event:[25,26,74],evolut:51,examin:106,exampl:[26,30,40,41,57,66],execut:[29,43,51,75,79],executor:[15,47],exp:[1,18],expand:4,expandlevel:4,explain:27,extern:112,fabric:110,factor:4,factorization_machin:4,faq:[93,94],fault:31,file:[29,106,112,113],fill_const:18,fill_constant_batch_size_lik:18,find:112,first_seq:4,float16:50,floor:18,flow:60,fluid:[12,40,41,51,52,63,71,108,111],format:[29,31,64],forward:[55,101],frame:29,framework:[27,102,103],from:[26,38,85,91,93],full_matrix_project:4,fulli:4,functor:87,futur:[51,86],gan:54,gate:97,gated_unit:4,gener:[52,54,69,97,98,106,126],get:[95,111,113],get_inference_program:17,get_output:4,get_plac:18,give:78,global:[75,77],global_scop:15,glu:19,gotcha:98,gpu:[92,107,118,120],grad_op:28,gradient:[27,28,36,62,80,100],gradient_print:3,graident:27,graph:[55,56,60,79,81],group:[4,112],gru:[5,118],gru_group:5,gru_step:4,gru_unit:[5,18],grumemori:4,hand:107,handler:[26,65],happen:38,hard_shrink:18,hard_sigmoid:18,hardwar:50,have:111,helper:77,hierarchi:29,high:[70,72,81,85],how:[27,34,72,78,79,87,91,101,103,104,105,107],hsigmoid:4,huber_classification_cost:4,huber_regression_cost:4,iOS:123,iam:112,ident:1,identifi:106,identity_project:4,ifels:[18,57],ifelseop:29,im2sequ:18,imag:[4,5,11,39,92,113,122],imdb:10,img_cmrnorm:4,img_conv:4,img_conv_bn_pool:5,img_conv_group:5,img_pool:4,imikolov:10,implement:[27,28,30,34,43,45,46,50,64,68,71,72,77,78,79,80,81,100,101,102,103],increment:18,infer:[25,122],infershap:[75,84],infervartyp:58,ingredi:26,ingress:48,initi:[16,36,54,112,120],insid:83,inspect:112,instal:[93,94,95,111,112,122,123,124,126],instanc:112,instead:78,integr:[87,112],intel:[61,62],interact:85,interfac:[8,9,11,27,31,36,37,47,70,78,83],intermedi:79,interpol:4,introduc:[69,89,111],introduct:[74,81,109,111],isn:78,issu:50,job:[31,39,110,112,113,114],join:4,kei:[61,67,112],kernel:[63,67,79,102],kill:110,kmax_sequence_scor:4,kube:112,kubectl:112,kubernet:[39,112,113],l1decai:23,l2_distanc:4,l2_normal:18,l2decai:23,lambda_cost:4,languag:[29,52],larg:34,last_seq:4,launch:[92,110,114],layer:[4,18,26,53,61,62,77,100,120],layout:67,leaky_relu:18,learn:[29,51],learnabl:4,less_than:18,leval:85,level:[70,72,81,85],libpaddle_capi_shar:66,libpaddle_capi_whol:66,librari:[36,50,67,79,87,122],limit:42,line:[109,119],linear:1,linear_chain_crf:18,linear_comb:4,linux:[110,122],list:[32,78],listenandserv:18,live:60,load:41,load_inference_model:17,load_param:17,load_persist:17,load_var:17,local:[42,83,112,120],lod:69,lod_rank_t:18,lod_tensor_to_arrai:18,lodtensor:[68,69,89],lodtensordesc:90,log:[1,18,99],logic:35,logsigmoid:18,look:106,low:[72,81,85],lstm:[5,118],lstm_step:4,lstm_unit:18,lstmemori:4,lstmemory_group:5,lstmemory_unit:5,machin:[4,69],macro:79,main:54,make:60,manag:[30,109],map:[78,79],master:[31,35,39,40],math:[4,87],mathemat:27,matmul:18,matrix:[62,118],max:7,max_sequence_len:18,maxframe_print:3,maxid:4,maxid_print:3,maxout:4,mean:18,member:54,memori:[4,60,68,87],merge_lod_tensor:18,messag:85,method:69,might:54,migrat:79,mileston:79,mini:78,minibatch:[9,41],misc:4,mix:[4,120],mkl:[61,62],mkldnn:63,mkldnn_helper:63,mkldnndevicecontext:63,mnist:10,mobil:125,model:[24,26,34,36,38,41,51,54,64,69,96,97,110,120],modifi:113,modul:[79,87,103],momentum:[6,20],more:54,motiv:[28,41,47,64,71,76],movielen:10,mul:18,multi:[43,52],multi_binary_label_cross_entropy_cost:4,multibox_loss:4,multipl:78,multiplex:[4,18],mxnet:56,name:[83,112],nativ:52,nccl:71,nce:[4,18],necess:77,necessari:79,need:[78,98,107],nest:68,net:19,network:[5,79,97,120],neural:97,nlp:[5,118],non:111,norm:[4,81],normal:16,note:27,numer:27,numpi:27,nvprof:107,nvvp:107,object:31,offset:69,ones:18,onli:[78,83],op_mak:79,openmpi:114,oper:[53,57,60,63,67,72,75,77,79,80,84,89,101],opinfomap:79,opkernel:[79,87,101,102],opproto:85,ops:[18,81],optim:[6,20,31,36,55,60,70,77,105],option:[49,91],opwithkernel:79,order:49,org:104,origin:79,orthogon:83,other:62,out_prod:4,outlin:117,output:[4,110,112],overview:[38,45,47,61,62,79,83,86,108],pack:[61,69],packag:[30,94],pad:4,paddl:[34,71,78,83,103],paddlejob:39,paddlepaddl:[26,29,41,51,52,61,62,72,75,81,82,86,91,92,94,104,108,111,112,113,121,122,123,124],pair:112,paradigm:51,parallel_nn:120,paralleldo:18,param_attr:21,paramattr:21,paramet:[2,4,25,26,31,36,37,39,41,44,62,72,73,77,81,109,111,112,118,119],parameteraverageoptim:72,parent:83,part:55,partit:36,pass:[91,120],path:[38,49],penalti:81,perform:[72,106,107,118],persist:35,pfsclient:[48,49],pfsserver:48,pip:94,place:[60,67,87,102],placement:42,platform:109,pnpair:3,point:[61,112],polici:60,pool2d:18,pool:[4,7],pose:[58,80],potenti:59,pow:18,power:4,precision_recal:3,prefetch:78,prelu:4,prepar:[109,110,111,112,114,123],principl:63,print:[3,18],privat:112,pro:126,problem:[46,58,59,60,67,70,80,88],procedur:126,process:[31,36,39,70,79,98],profil:[22,106,107],program:[29,40,41,43,51,52,75,77,92,109,111],programdesc:[52,75],project:30,propos:[58,80,81],protobuf:84,protomak:101,provid:78,prune:76,pserver:38,pull:92,python:[27,39,42,61,62,68,70,72,77,78,81,85,90,100,101,106],qualiti:79,queue:[31,35],quick:95,randomnumb:118,rank:3,rank_cost:4,raspberri:124,reader:[8,9,26,78],readi:111,realiz:79,reciproc:18,recoveri:31,recurr:[4,5,97],recurrent_group:4,recv:41,reduce_max:18,reduce_mean:18,reduce_min:18,reduce_sum:18,ref:27,refactor:79,refer:[42,44,60,61,62,86,107],region:112,regist:[58,79,85,101,102],registr:[79,80],registri:79,regular:[23,36,81],rel:69,relat:[79,89],relu6:18,relu:[1,18],remark:101,remot:37,remoteexecutor:42,render:112,reorder_lod_tensor_by_rank:18,repeat:4,represent:[29,79],requir:[30,54],reset_profil:22,reshap:[4,18],resiz:4,result:[110,113],retri:35,reus:77,review:99,rmsprop:6,rnn:[68,89,96,97,118],rnnop:[29,68,79],roi_pool:4,rotat:4,round:18,route53:112,row:[84,86],row_conv:[4,18],row_l2_norm:4,rpc:41,run:[47,91,92,101,108,113],runtim:[39,94],sampl:4,sampling_id:4,save:38,save_inference_model:17,save_param:17,save_persist:17,save_var:17,scale:[4,18,31],scale_shift:4,scaled_dot_product_attent:19,scaling_project:4,scope:[29,68,79,83],scope_guard:15,script:[111,113],search:[69,86],secur:112,select:[36,41,84],selectedrow:84,selective_fc:4,send:[18,41],sentiment:10,separ:79,seq_concat:4,seq_reshap:4,seq_slic:4,seqtext_print:3,sequenc:[69,97],sequence_conv:18,sequence_conv_pool:[5,19],sequence_expand:18,sequence_first_step:18,sequence_last_step:18,sequence_pool:18,sequence_reshap:18,sequence_softmax:18,sequencesoftmax:1,server:[31,35,36,39,41,44,109,111,112,118],servic:112,set:119,setup:[112,122],sextant:126,sgd:[20,118],shape:69,share:[26,28,60,83],should:83,shrink_memori:18,shuffl:78,sigmoid:[1,18],sigmoid_cross_entropy_with_logit:18,simpl:[69,97],simple_attent:5,simple_gru2:5,simple_gru:5,simple_img_conv_pool:[5,19],simple_lstm:5,singl:78,slice:[4,111],slice_project:4,slope_intercept:4,small_vgg:5,smooth_l1_cost:4,soft_relu:18,softmax:1,softplu:18,softrelu:1,softshrink:18,softsign:[1,18],solut:[58,59,60,61,67,76,80,88],some:98,sourc:[91,93,108],spars:[36,37,38,84,120],specifi:120,split:18,split_lod_tensor:18,spp:4,sqrt:18,squar:[1,18],square_error_cost:[4,18],squarerootn:7,stack:29,standard:99,stanh:[1,18],start:[26,95,109,112,113],startup:113,statement:46,staticrnn:18,staticrnnmemorylink:18,step:[68,93],storag:81,store:31,strategi:60,style:99,sub_nested_seq:4,sub_seq:4,subcommond:49,submit:39,suffici:78,suitabl:30,sulut:63,sum:[3,7,18],sum_cost:4,sum_to_one_norm:4,summar:[26,40],summari:64,support:[50,71,87,89],survei:[50,56,81,126],swish:18,switch_scop:15,synopsi:49,syntax:41,system:[51,112],tabl:[66,86],table_project:4,tanh:[1,18],tanh_shrink:18,task:[31,35,86],tear:112,tecton:126,templat:112,tensor:[4,18,79,87,103],tensorarrai:[69,89],tensordesc:90,tensorflow:56,test:[61,62,63,91,99,100,101,118,120],text_conv_pool:5,theori:27,thi:83,think:54,three:89,thresholded_relu:18,time:108,timelin:38,timer:107,tip:107,todo:[32,33,43],togeth:83,toler:31,tool:[30,104,107,109,126],toolchain:124,topic:87,topk:18,toward:52,train:[25,26,31,34,37,39,42,70,78,92,109,110,111,112,113,114,118,120],trainer:[25,31,36,38,39,41,109,112],tran:4,trans_full_matrix_project:4,transform:88,translat:69,transpil:[42,43,44,52,60,71],transpos:18,tune:[107,118],ture:51,two:27,type:[41,67,91,101],uci_h:10,uniform:[16,89],unit:[61,62,63,99,100,101,118],unpack:69,updat:[26,37,38,104,111,112],usag:[28,45,68,69,78,103,105],use:[34,78,103],user:31,using:98,util:3,valu:77,value_print:3,vardesc:90,variabl:[28,60,77,79,83,90],vector:118,verifi:112,version:[40,50,94],vgg_16_network:5,volum:112,vpc:112,warp_ctc:4,warpctc:18,weightnormparamattr:21,what:[34,38,98,107],when:[38,83],whileguard:18,whl:94,why:[50,51,72,78,79,89,107],wmt14:10,work:86,worker:40,workflow:99,wrapper:100,write:[99,100,101,102,104],www:104,xavier:16,yaml:113,your:[92,102],zero:18}}) \ No newline at end of file diff --git a/develop/doc/survey/cluster_bootstrapping_tools.html b/develop/doc/survey/cluster_bootstrapping_tools.html index 8a54bebca7c0847eb2299cdbb5d0326a1c5cc722..5286ba6b99e78acef034774a770d6bb3c71a0284 100644 --- a/develop/doc/survey/cluster_bootstrapping_tools.html +++ b/develop/doc/survey/cluster_bootstrapping_tools.html @@ -159,17 +159,17 @@- Training and Inference
- Fluid
diff --git a/develop/doc_cn/_sources/api/v2/fluid/data_feeder.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/data_feeder.rst.txt index 0fa78f7dfb04c13be7eb83b7fd35cb03f2f4a7fa..a591c7334fd31c98a94b50a4344f251560a0f2f9 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/data_feeder.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/data_feeder.rst.txt @@ -1,9 +1,14 @@ +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! + =========== -DataFeeder +data_feeder =========== DataFeeder ------------ -.. automodule:: paddle.v2.fluid.data_feeder - :members: DataFeeder +---------- + +.. autoclass:: paddle.v2.fluid.data_feeder.DataFeeder + :members: :noindex: + diff --git a/develop/doc_cn/_sources/api/v2/fluid/evaluator.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/evaluator.rst.txt index a23f3301d0331e0ea3733f06444515eb4680cd31..00dcecfd628a35d83d1c596bf0aea819a1705862 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/evaluator.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/evaluator.rst.txt @@ -1,9 +1,21 @@ -=========== -Evaluator -=========== - -Evaluator ------------ -.. automodule:: paddle.v2.fluid.evaluator - :members: Evaluator +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! + +========= +evaluator +========= + +Accuracy +-------- + +.. autoclass:: paddle.v2.fluid.evaluator.Accuracy + :members: :noindex: + +ChunkEvaluator +-------------- + +.. autoclass:: paddle.v2.fluid.evaluator.ChunkEvaluator + :members: + :noindex: + diff --git a/develop/doc_cn/_sources/api/v2/fluid/executor.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/executor.rst.txt index 3a283538c120cfa1ef646c390bb71c6251c23675..a028f6283f2ca333bdf6c9857a98661c0222b41e 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/executor.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/executor.rst.txt @@ -1,9 +1,32 @@ -=========== -Executor -=========== +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! + +======== +executor +======== Executor +-------- + +.. autoclass:: paddle.v2.fluid.executor.Executor + :members: + :noindex: + +global_scope +------------ + +.. autofunction:: paddle.v2.fluid.executor.global_scope + :noindex: + +scope_guard ----------- -.. automodule:: paddle.v2.fluid.executor - :members: Executor + +.. autofunction:: paddle.v2.fluid.executor.scope_guard + :noindex: + +switch_scope +------------ + +.. autofunction:: paddle.v2.fluid.executor.switch_scope :noindex: + diff --git a/develop/doc_cn/_sources/api/v2/fluid/initializer.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/initializer.rst.txt index 8f587837e9873370722062404f511654a9460587..c38be033fff2997930525f51c93995db09daa2b6 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/initializer.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/initializer.rst.txt @@ -1,50 +1,35 @@ +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! + =========== -Initializer +initializer =========== +Constant +-------- - -Initializer ------------ -.. automodule:: paddle.v2.fluid.initializer - :members: Initializer - :noindex: - - - -ConstantInitializer -------------------- -.. automodule:: paddle.v2.fluid.initializer - :members: ConstantInitializer +.. autoclass:: paddle.v2.fluid.initializer.Constant + :members: :noindex: +Uniform +------- - -UniformInitializer ------------------- -.. automodule:: paddle.v2.fluid.initializer - :members: UniformInitializer - :noindex: - - - -NormalInitializer ------------------ -.. automodule:: paddle.v2.fluid.initializer - :members: NormalInitializer +.. autoclass:: paddle.v2.fluid.initializer.Uniform + :members: :noindex: +Normal +------ -XavierInitializer ------------------ -.. automodule:: paddle.v2.fluid.initializer - :members: XavierInitializer +.. autoclass:: paddle.v2.fluid.initializer.Normal + :members: :noindex: +Xavier +------ -MSRAInitializer ---------------- -.. automodule:: paddle.v2.fluid.initializer - :members: MSRAInitializer +.. autoclass:: paddle.v2.fluid.initializer.Xavier + :members: :noindex: diff --git a/develop/doc_cn/_sources/api/v2/fluid/io.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/io.rst.txt index 67f68c4e9e16b379207b8de114cdf769e056f78e..37c9c273e369532e8ff596e9649cb695a98a2505 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/io.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/io.rst.txt @@ -1,10 +1,61 @@ -=========== -IO -=========== +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! +== +io +== +save_vars +--------- -is_parameter +.. autofunction:: paddle.v2.fluid.io.save_vars + :noindex: + +save_params ----------- -.. autofunction:: paddle.v2.fluid.io.is_parameter + +.. autofunction:: paddle.v2.fluid.io.save_params + :noindex: + +save_persistables +----------------- + +.. autofunction:: paddle.v2.fluid.io.save_persistables + :noindex: + +load_vars +--------- + +.. autofunction:: paddle.v2.fluid.io.load_vars + :noindex: + +load_params +----------- + +.. autofunction:: paddle.v2.fluid.io.load_params :noindex: + +load_persistables +----------------- + +.. autofunction:: paddle.v2.fluid.io.load_persistables + :noindex: + +save_inference_model +-------------------- + +.. autofunction:: paddle.v2.fluid.io.save_inference_model + :noindex: + +load_inference_model +-------------------- + +.. autofunction:: paddle.v2.fluid.io.load_inference_model + :noindex: + +get_inference_program +--------------------- + +.. autofunction:: paddle.v2.fluid.io.get_inference_program + :noindex: + diff --git a/develop/doc_cn/_sources/api/v2/fluid/layers.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/layers.rst.txt index 231ec2d4ba102a5d31c47cbc7a5d484ef17a7f3a..e24613b94b422b7cdf9c6383c359fa92a4faf6ff 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/layers.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/layers.rst.txt @@ -1,546 +1,799 @@ -========== -Layers -========== +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! +====== +layers +====== -fc ---- -.. autofunction:: paddle.v2.fluid.layers.fc +control_flow +============ + +split_lod_tensor +---------------- + +.. autofunction:: paddle.v2.fluid.layers.split_lod_tensor :noindex: -embedding ---------- -.. autofunction:: paddle.v2.fluid.layers.embedding +merge_lod_tensor +---------------- + +.. autofunction:: paddle.v2.fluid.layers.merge_lod_tensor :noindex: -dynamic_lstm ------------- -.. autofunction:: paddle.v2.fluid.layers.dynamic_lstm +BlockGuard +---------- + +.. autoclass:: paddle.v2.fluid.layers.BlockGuard + :members: :noindex: -dynamic_lstmp -------------- -.. autofunction:: paddle.v2.fluid.layers.dynamic_lstmp +BlockGuardWithCompletion +------------------------ + +.. autoclass:: paddle.v2.fluid.layers.BlockGuardWithCompletion + :members: :noindex: -dynamic_gru ------------ -.. autofunction:: paddle.v2.fluid.layers.dynamic_gru +StaticRNNMemoryLink +------------------- + +.. autoclass:: paddle.v2.fluid.layers.StaticRNNMemoryLink + :members: :noindex: -data ----- -.. autofunction:: paddle.v2.fluid.layers.data +WhileGuard +---------- + +.. autoclass:: paddle.v2.fluid.layers.WhileGuard + :members: :noindex: -mean ----- -.. autofunction:: paddle.v2.fluid.layers.mean +While +----- + +.. autoclass:: paddle.v2.fluid.layers.While + :members: :noindex: -mul ---- -.. autofunction:: paddle.v2.fluid.layers.mul +lod_rank_table +-------------- + +.. autofunction:: paddle.v2.fluid.layers.lod_rank_table :noindex: -elementwise_add ---------------- -.. autofunction:: paddle.v2.fluid.layers.elementwise_add +max_sequence_len +---------------- + +.. autofunction:: paddle.v2.fluid.layers.max_sequence_len :noindex: -elementwise_sub ---------------- -.. autofunction:: paddle.v2.fluid.layers.elementwise_sub +topk +---- + +.. autofunction:: paddle.v2.fluid.layers.topk :noindex: -elementwise_mul ---------------- -.. autofunction:: paddle.v2.fluid.layers.elementwise_mul +lod_tensor_to_array +------------------- + +.. autofunction:: paddle.v2.fluid.layers.lod_tensor_to_array :noindex: -elementwise_div ---------------- -.. autofunction:: paddle.v2.fluid.layers.elementwise_div +array_to_lod_tensor +------------------- + +.. autofunction:: paddle.v2.fluid.layers.array_to_lod_tensor :noindex: +increment +--------- -dropout -------- -.. autofunction:: paddle.v2.fluid.layers.dropout +.. autofunction:: paddle.v2.fluid.layers.increment :noindex: +array_write +----------- -reshape --------- -.. autofunction:: paddle.v2.fluid.layers.reshape +.. autofunction:: paddle.v2.fluid.layers.array_write :noindex: +create_array +------------ -sigmoid +.. autofunction:: paddle.v2.fluid.layers.create_array + :noindex: + +less_than --------- -.. autofunction:: paddle.v2.fluid.layers.sigmoid + +.. autofunction:: paddle.v2.fluid.layers.less_than :noindex: +array_read +---------- -scale ---------- -.. autofunction:: paddle.v2.fluid.layers.scale +.. autofunction:: paddle.v2.fluid.layers.array_read + :noindex: + +shrink_memory +------------- + +.. autofunction:: paddle.v2.fluid.layers.shrink_memory :noindex: +array_length +------------ -transpose +.. autofunction:: paddle.v2.fluid.layers.array_length + :noindex: + +IfElse +------ + +.. autoclass:: paddle.v2.fluid.layers.IfElse + :members: + :noindex: + +DynamicRNN +---------- + +.. autoclass:: paddle.v2.fluid.layers.DynamicRNN + :members: + :noindex: + +ConditionalBlock +---------------- + +.. autoclass:: paddle.v2.fluid.layers.ConditionalBlock + :members: + :noindex: + +StaticRNN --------- -.. autofunction:: paddle.v2.fluid.layers.transpose + +.. autoclass:: paddle.v2.fluid.layers.StaticRNN + :members: :noindex: +reorder_lod_tensor_by_rank +-------------------------- -sigmoid_cross_entropy_with_logits ---------------------------------- -.. autofunction:: paddle.v2.fluid.layers.esigmoid_cross_entropy_with_logits +.. autofunction:: paddle.v2.fluid.layers.reorder_lod_tensor_by_rank :noindex: +ParallelDo +---------- -cast +.. autoclass:: paddle.v2.fluid.layers.ParallelDo + :members: + :noindex: + +Print +----- + +.. autofunction:: paddle.v2.fluid.layers.Print + :noindex: + +device +====== + +get_places +---------- + +.. autofunction:: paddle.v2.fluid.layers.get_places + :noindex: + +io +== + +data ---- -.. autofunction:: paddle.v2.fluid.layers.cast + +.. autofunction:: paddle.v2.fluid.layers.data :noindex: +BlockGuardServ +-------------- -concat -------- -.. autofunction:: paddle.v2.fluid.layers.concat +.. autoclass:: paddle.v2.fluid.layers.BlockGuardServ + :members: :noindex: +ListenAndServ +------------- -sums +.. autoclass:: paddle.v2.fluid.layers.ListenAndServ + :members: + :noindex: + +Send ---- -.. autofunction:: paddle.v2.fluid.layers.sums + +.. autofunction:: paddle.v2.fluid.layers.Send :noindex: +nn +== -linear_chain_crf ----------------- -.. autofunction:: paddle.v2.fluid.layers.linear_chain_crf +fc +-- + +.. autofunction:: paddle.v2.fluid.layers.fc :noindex: +embedding +--------- -assign -------- .. autofunction:: paddle.v2.fluid.layers.embedding :noindex: +dynamic_lstm +------------ -split_lod_tensor ----------------- -.. autofunction:: paddle.v2.fluid.layers.split_lod_tensor +.. autofunction:: paddle.v2.fluid.layers.dynamic_lstm :noindex: +dynamic_lstmp +------------- -merge_lod_tensor +.. autofunction:: paddle.v2.fluid.layers.dynamic_lstmp + :noindex: + +dynamic_gru +----------- + +.. autofunction:: paddle.v2.fluid.layers.dynamic_gru + :noindex: + +gru_unit +-------- + +.. autofunction:: paddle.v2.fluid.layers.gru_unit + :noindex: + +linear_chain_crf ---------------- -.. autofunction:: paddle.v2.fluid.layers.merge_lod_tensor + +.. autofunction:: paddle.v2.fluid.layers.linear_chain_crf + :noindex: + +crf_decoding +------------ + +.. autofunction:: paddle.v2.fluid.layers.crf_decoding :noindex: cos_sim --------- +------- + .. autofunction:: paddle.v2.fluid.layers.cos_sim :noindex: - cross_entropy ------------- + .. autofunction:: paddle.v2.fluid.layers.cross_entropy :noindex: - - square_error_cost ----------------- + .. autofunction:: paddle.v2.fluid.layers.square_error_cost :noindex: - accuracy ---------- +-------- + .. autofunction:: paddle.v2.fluid.layers.accuracy :noindex: +chunk_eval +---------- + +.. autofunction:: paddle.v2.fluid.layers.chunk_eval + :noindex: sequence_conv ------------- + .. autofunction:: paddle.v2.fluid.layers.sequence_conv :noindex: - conv2d ------ + .. autofunction:: paddle.v2.fluid.layers.conv2d :noindex: - sequence_pool ------------- + .. autofunction:: paddle.v2.fluid.layers.sequence_pool :noindex: +pool2d +------ -sequence_first_step -------------------- -.. autofunction:: paddle.v2.fluid.layers.sequence_first_step +.. autofunction:: paddle.v2.fluid.layers.pool2d :noindex: +batch_norm +---------- + +.. autofunction:: paddle.v2.fluid.layers.batch_norm + :noindex: -sequence_last_step +beam_search_decode ------------------ -.. autofunction:: paddle.v2.fluid.layers.sequence_last_step + +.. autofunction:: paddle.v2.fluid.layers.beam_search_decode :noindex: +conv2d_transpose +---------------- -pool2d ------- -.. autofunction:: paddle.v2.fluid.layers.pool2d +.. autofunction:: paddle.v2.fluid.layers.conv2d_transpose :noindex: +sequence_expand +--------------- -batch_norm +.. autofunction:: paddle.v2.fluid.layers.sequence_expand + :noindex: + +lstm_unit +--------- + +.. autofunction:: paddle.v2.fluid.layers.lstm_unit + :noindex: + +reduce_sum ---------- -.. autofunction:: paddle.v2.fluid.layers.batch_norm + +.. autofunction:: paddle.v2.fluid.layers.reduce_sum + :noindex: + +reduce_mean +----------- + +.. autofunction:: paddle.v2.fluid.layers.reduce_mean :noindex: +reduce_max +---------- + +.. autofunction:: paddle.v2.fluid.layers.reduce_max + :noindex: -beam_search_decode +reduce_min +---------- + +.. autofunction:: paddle.v2.fluid.layers.reduce_min + :noindex: + +sequence_first_step +------------------- + +.. autofunction:: paddle.v2.fluid.layers.sequence_first_step + :noindex: + +sequence_last_step ------------------ -.. autofunction:: paddle.v2.fluid.layers.beam_search_decode + +.. autofunction:: paddle.v2.fluid.layers.sequence_last_step + :noindex: + +dropout +------- + +.. autofunction:: paddle.v2.fluid.layers.dropout :noindex: +split +----- -lod_rank_table --------------- -.. autofunction:: paddle.v2.fluid.layers.lod_rank_table +.. autofunction:: paddle.v2.fluid.layers.split :noindex: +ctc_greedy_decoder +------------------ -max_sequence_len ----------------- -.. autofunction:: paddle.v2.fluid.layers.max_sequence_len +.. autofunction:: paddle.v2.fluid.layers.ctc_greedy_decoder :noindex: +edit_distance +------------- -topk ------ -.. autofunction:: paddle.v2.fluid.layers.topk +.. autofunction:: paddle.v2.fluid.layers.edit_distance :noindex: +l2_normalize +------------ -lod_tensor_to_array -------------------- -.. autofunction:: paddle.v2.fluid.layers.lod_tensor_to_array +.. autofunction:: paddle.v2.fluid.layers.l2_normalize :noindex: +matmul +------ - -array_to_lod_tensor -------------------- -.. autofunction:: paddle.v2.fluid.layers.array_to_lod_tensor +.. autofunction:: paddle.v2.fluid.layers.matmul :noindex: +warpctc +------- +.. autofunction:: paddle.v2.fluid.layers.warpctc + :noindex: +sequence_reshape +---------------- -fill_constant -------------- -.. autofunction:: paddle.v2.fluid.layers.fill_constant +.. autofunction:: paddle.v2.fluid.layers.sequence_reshape :noindex: +transpose +--------- +.. autofunction:: paddle.v2.fluid.layers.transpose + :noindex: -fill_constant_batch_size_like ------------------------------ -.. autofunction:: paddle.v2.fluid.layers.fill_constant_batch_size_like +im2sequence +----------- + +.. autofunction:: paddle.v2.fluid.layers.im2sequence :noindex: +nce +--- -ones ----- -.. autofunction:: paddle.v2.fluid.layers.ones +.. autofunction:: paddle.v2.fluid.layers.nce :noindex: +beam_search +----------- -zeros ------ -.. autofunction:: paddle.v2.fluid.layers.zeros +.. autofunction:: paddle.v2.fluid.layers.beam_search :noindex: +row_conv +-------- -increment ---------- -.. autofunction:: paddle.v2.fluid.layers.increment +.. autofunction:: paddle.v2.fluid.layers.row_conv :noindex: +multiplex +--------- -array_write ------------ -.. autofunction:: paddle.v2.fluid.layers.array_write +.. autofunction:: paddle.v2.fluid.layers.multiplex :noindex: +ops +=== +mean +---- -create_array ------------- -.. autofunction:: paddle.v2.fluid.layers.create_array +.. autofunction:: paddle.v2.fluid.layers.mean :noindex: +mul +--- -less_than ---------- -.. autofunction:: paddle.v2.fluid.layers.less_than +.. autofunction:: paddle.v2.fluid.layers.mul :noindex: +reshape +------- -array_read ----------- -.. autofunction:: paddle.v2.fluid.layers.array_read +.. autofunction:: paddle.v2.fluid.layers.reshape :noindex: +scale +----- -shrink_memory --------------- -.. autofunction:: paddle.v2.fluid.layers.shrink_memory +.. autofunction:: paddle.v2.fluid.layers.scale :noindex: +sigmoid_cross_entropy_with_logits +--------------------------------- -array_length -------------- -.. autofunction:: paddle.v2.fluid.layers.array_length +.. autofunction:: paddle.v2.fluid.layers.sigmoid_cross_entropy_with_logits :noindex: +elementwise_add +--------------- -conv2d_transpose ----------------- -.. autofunction:: paddle.v2.fluid.layers.conv2d_transpose +.. autofunction:: paddle.v2.fluid.layers.elementwise_add :noindex: - -sequence_expand +elementwise_div --------------- -.. autofunction:: paddle.v2.fluid.layers.sequence_expand + +.. autofunction:: paddle.v2.fluid.layers.elementwise_div :noindex: +elementwise_sub +--------------- -gru_unit --------- -.. autofunction:: paddle.v2.fluid.layers.gru_unit +.. autofunction:: paddle.v2.fluid.layers.elementwise_sub :noindex: +elementwise_mul +--------------- -lstm_unit ---------- -.. autofunction:: paddle.v2.fluid.layers.lstm_unit +.. autofunction:: paddle.v2.fluid.layers.elementwise_mul :noindex: +elementwise_max +--------------- -sequence_softmax ----------------- -.. autofunction:: paddle.v2.fluid.layers.sequence_softmax +.. autofunction:: paddle.v2.fluid.layers.elementwise_max :noindex: +elementwise_min +--------------- -reduce_sum ----------- -.. autofunction:: paddle.v2.fluid.layers.reduce_sum +.. autofunction:: paddle.v2.fluid.layers.elementwise_min :noindex: +elementwise_pow +--------------- -reduce_mean ------------ -.. autofunction:: paddle.v2.fluid.layers.reduce_mean +.. autofunction:: paddle.v2.fluid.layers.elementwise_pow :noindex: +clip +---- -reduce_max ----------- -.. autofunction:: paddle.v2.fluid.layers.reduce_max +.. autofunction:: paddle.v2.fluid.layers.clip :noindex: +clip_by_norm +------------ -reduce_min ----------- -.. autofunction:: paddle.v2.fluid.layers.reduce_min +.. autofunction:: paddle.v2.fluid.layers.clip_by_norm :noindex: +sequence_softmax +---------------- -split ------ -.. autofunction:: paddle.v2.fluid.layers.split +.. autofunction:: paddle.v2.fluid.layers.sequence_softmax :noindex: +sigmoid +------- -matmul ------- -.. autofunction:: paddle.v2.fluid.layers.matmul +.. autofunction:: paddle.v2.fluid.layers.sigmoid :noindex: logsigmoid ---------- + .. autofunction:: paddle.v2.fluid.layers.logsigmoid :noindex: exp --- + .. autofunction:: paddle.v2.fluid.layers.exp :noindex: relu ---- + .. autofunction:: paddle.v2.fluid.layers.relu :noindex: tanh ---- + .. autofunction:: paddle.v2.fluid.layers.tanh :noindex: tanh_shrink ----------- + .. autofunction:: paddle.v2.fluid.layers.tanh_shrink :noindex: softshrink ---------- + .. autofunction:: paddle.v2.fluid.layers.softshrink :noindex: sqrt ---- + .. autofunction:: paddle.v2.fluid.layers.sqrt :noindex: abs ----- +--- + .. autofunction:: paddle.v2.fluid.layers.abs :noindex: ceil ---- + .. autofunction:: paddle.v2.fluid.layers.ceil :noindex: floor ----- + .. autofunction:: paddle.v2.fluid.layers.floor :noindex: round ----- + .. autofunction:: paddle.v2.fluid.layers.round :noindex: reciprocal ---------- + .. autofunction:: paddle.v2.fluid.layers.reciprocal :noindex: log --- + .. autofunction:: paddle.v2.fluid.layers.log :noindex: square ------ + .. autofunction:: paddle.v2.fluid.layers.square :noindex: softplus -------- + .. autofunction:: paddle.v2.fluid.layers.softplus :noindex: softsign ---------- +-------- + .. autofunction:: paddle.v2.fluid.layers.softsign :noindex: brelu ----- + .. autofunction:: paddle.v2.fluid.layers.brelu :noindex: leaky_relu ---------- + .. autofunction:: paddle.v2.fluid.layers.leaky_relu :noindex: soft_relu --------- + .. autofunction:: paddle.v2.fluid.layers.soft_relu :noindex: elu ----- +--- + .. autofunction:: paddle.v2.fluid.layers.elu :noindex: relu6 ----- + .. autofunction:: paddle.v2.fluid.layers.relu6 :noindex: pow ----- +--- + .. autofunction:: paddle.v2.fluid.layers.pow :noindex: +stanh +----- + +.. autofunction:: paddle.v2.fluid.layers.stanh + :noindex: + hard_shrink ----------- + .. autofunction:: paddle.v2.fluid.layers.hard_shrink :noindex: thresholded_relu ---------------- + .. autofunction:: paddle.v2.fluid.layers.thresholded_relu :noindex: hard_sigmoid -------------- +------------ + .. autofunction:: paddle.v2.fluid.layers.hard_sigmoid :noindex: swish ------- +----- + .. autofunction:: paddle.v2.fluid.layers.swish :noindex: -im2sequence +tensor +====== + +create_tensor +------------- + +.. autofunction:: paddle.v2.fluid.layers.create_tensor + :noindex: + +create_parameter +---------------- + +.. autofunction:: paddle.v2.fluid.layers.create_parameter + :noindex: + +create_global_var +----------------- + +.. autofunction:: paddle.v2.fluid.layers.create_global_var + :noindex: + +cast +---- + +.. autofunction:: paddle.v2.fluid.layers.cast + :noindex: + +concat ------ -.. autofunction:: paddle.v2.fluid.layers.im2sequence + +.. autofunction:: paddle.v2.fluid.layers.concat :noindex: -edit_distance ---------------- -.. autofunction:: paddle.v2.fluid.layers.edit_distance_error +sums +---- + +.. autofunction:: paddle.v2.fluid.layers.sums :noindex: -ctc_greedy_decoder ---------------- -.. autofunction:: paddle.v2.fluid.layers.ctc_greedy_decoder +assign +------ + +.. autofunction:: paddle.v2.fluid.layers.assign :noindex: -l2_normalize ------------- -.. autofunction:: paddle.v2.fluid.layers.l2_normalize +fill_constant_batch_size_like +----------------------------- + +.. autofunction:: paddle.v2.fluid.layers.fill_constant_batch_size_like :noindex: -sequence_reshape ----------------- -.. autofunction:: paddle.v2.fluid.layers.sequence_reshape +fill_constant +------------- + +.. autofunction:: paddle.v2.fluid.layers.fill_constant :noindex: -row_conv --------- -.. autofunction:: paddle.v2.fluid.layers.row_conv +ones +---- + +.. autofunction:: paddle.v2.fluid.layers.ones :noindex: -multiplex ---------- -.. autofunction:: paddle.v2.fluid.layers.multiplex +zeros +----- + +.. autofunction:: paddle.v2.fluid.layers.zeros :noindex: + diff --git a/develop/doc_cn/_sources/api/v2/fluid/nets.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/nets.rst.txt index 500019bc507f859c4c91de5d322a82eb1e78e2de..015581b7660848bdb0845fafe2d3fc05405e6ae6 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/nets.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/nets.rst.txt @@ -1,33 +1,31 @@ -=========== -Nets -=========== +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! + +==== +nets +==== simple_img_conv_pool -------------------- -.. autofunction:: paddle.v2.fluid.nets.simple_img_conv_pool - :noindex: - -img_conv_group ---------------- -.. autofunction:: paddle.v2.fluid.nets.img_conv_group +.. autofunction:: paddle.v2.fluid.nets.simple_img_conv_pool :noindex: - sequence_conv_pool ------------------ + .. autofunction:: paddle.v2.fluid.nets.sequence_conv_pool :noindex: - glu --- + .. autofunction:: paddle.v2.fluid.nets.glu :noindex: - scaled_dot_product_attention ---------------------------- + .. autofunction:: paddle.v2.fluid.nets.scaled_dot_product_attention :noindex: diff --git a/develop/doc_cn/_sources/api/v2/fluid/optimizer.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/optimizer.rst.txt index 19b4940f08de3e2f7dc177f2961e538946d10a78..1691ebb9a7cb16da96e04147d0adea322374f529 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/optimizer.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/optimizer.rst.txt @@ -1,54 +1,49 @@ -=========== -Optimizer -=========== - -Optimizer ------------ -.. automodule:: paddle.v2.fluid.optimizer - :members: Optimizer - :noindex: +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! +========= +optimizer +========= -SGDOptimizer ------------ -.. automodule:: paddle.v2.fluid.optimizer - :members: SGDOptimizer - :noindex: +SGD +--- +.. autoclass:: paddle.v2.fluid.optimizer.SGD + :members: + :noindex: +Momentum +-------- -MomentumOptimizer ------------------ -.. automodule:: paddle.v2.fluid.optimizer - :members: MomentumOptimizer +.. autoclass:: paddle.v2.fluid.optimizer.Momentum + :members: :noindex: +Adagrad +------- - -AdagradOptimizer ----------------- -.. automodule:: paddle.v2.fluid.optimizer - :members: AdagradOptimizer +.. autoclass:: paddle.v2.fluid.optimizer.Adagrad + :members: :noindex: +Adam +---- -AdamOptimizer -------------- -.. automodule:: paddle.v2.fluid.optimizer - :members: AdamOptimizer +.. autoclass:: paddle.v2.fluid.optimizer.Adam + :members: :noindex: +Adamax +------ -AdamaxOptimizer ------------ -.. automodule:: paddle.v2.fluid.optimizer - :members: AdamaxOptimizer +.. autoclass:: paddle.v2.fluid.optimizer.Adamax + :members: :noindex: +DecayedAdagrad +-------------- -DecayedAdagradOptimizer ------------------------ -.. automodule:: paddle.v2.fluid.optimizer - :members: DecayedAdagradOptimizer +.. autoclass:: paddle.v2.fluid.optimizer.DecayedAdagrad + :members: :noindex: diff --git a/develop/doc_cn/_sources/api/v2/fluid/param_attr.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/param_attr.rst.txt index ca0c8af9e8c4f2271de7a131ad0d27c0e8635f50..8083d0d858dafcd275eaddb9b475875ee42ef724 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/param_attr.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/param_attr.rst.txt @@ -1,11 +1,21 @@ -=========== +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! + +========== +param_attr +========== + ParamAttr -=========== +--------- +.. autoclass:: paddle.v2.fluid.param_attr.ParamAttr + :members: + :noindex: +WeightNormParamAttr +------------------- -ParamAttr ------------ -.. automodule:: paddle.v2.fluid.param_attr - :members: ParamAttr +.. autoclass:: paddle.v2.fluid.param_attr.WeightNormParamAttr + :members: :noindex: + diff --git a/develop/doc_cn/_sources/api/v2/fluid/profiler.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/profiler.rst.txt index 7d4042d1f41c12c4a551ba6576559d612116872a..4a1ff7cb6976e0054f77428b699ea679aa91394f 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/profiler.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/profiler.rst.txt @@ -1,10 +1,25 @@ -=========== -Profiler -=========== +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! +======== +profiler +======== +cuda_profiler +------------- -Profiler ------------ .. autofunction:: paddle.v2.fluid.profiler.cuda_profiler :noindex: + +reset_profiler +-------------- + +.. autofunction:: paddle.v2.fluid.profiler.reset_profiler + :noindex: + +profiler +-------- + +.. autofunction:: paddle.v2.fluid.profiler.profiler + :noindex: + diff --git a/develop/doc_cn/_sources/api/v2/fluid/regularizer.rst.txt b/develop/doc_cn/_sources/api/v2/fluid/regularizer.rst.txt index 868e225ed3d59e79aeb217fb88081ea25f80fa2c..2c17d15599baa1d02eb87c7b6c40034769ebb3a4 100644 --- a/develop/doc_cn/_sources/api/v2/fluid/regularizer.rst.txt +++ b/develop/doc_cn/_sources/api/v2/fluid/regularizer.rst.txt @@ -1,25 +1,27 @@ +.. THIS FILE IS GENERATED BY `gen_doc.{py|sh}` + !DO NOT EDIT THIS FILE MANUALLY! + =========== -Regularizer +regularizer =========== -WeightDecayRegularizer ----------------------- -.. automodule:: paddle.v2.fluid.regularizer - :members: WeightDecayRegularizer - :noindex: - +append_regularization_ops +------------------------- -L2DecayRegularizer ------------------- -.. automodule:: paddle.v2.fluid.regularizer - :members: L2DecayRegularizer +.. autofunction:: paddle.v2.fluid.regularizer.append_regularization_ops :noindex: +L1Decay +------- +.. autoclass:: paddle.v2.fluid.regularizer.L1Decay + :members: + :noindex: -L1DecayRegularizer -------------------- -.. automodule:: paddle.v2.fluid.regularizer - :members: L1DecayRegularizer +L2Decay +------- +.. autoclass:: paddle.v2.fluid.regularizer.L2Decay + :members: + :noindex: diff --git a/develop/doc_cn/api/index_cn.html b/develop/doc_cn/api/index_cn.html index e630b53aeb788cbc7ce69f70434256e8d7dda29e..92ad7605dd0cfdda9015e1d58283c635aee734ae 100644 --- a/develop/doc_cn/api/index_cn.html +++ b/develop/doc_cn/api/index_cn.html @@ -172,17 +172,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/config/activation.html b/develop/doc_cn/api/v2/config/activation.html index b4e9d3d8e420a9b3851fe860527a621672d025a2..55bc0e3171099cc47272b62ad78534b64e999d89 100644 --- a/develop/doc_cn/api/v2/config/activation.html +++ b/develop/doc_cn/api/v2/config/activation.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/config/attr.html b/develop/doc_cn/api/v2/config/attr.html index f4118fe1c293cd42c61e1e8a4995540a6ffc6547..84bcd0e6f2ed3a9c217388df2c76b5adb4f88b64 100644 --- a/develop/doc_cn/api/v2/config/attr.html +++ b/develop/doc_cn/api/v2/config/attr.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/config/evaluators.html b/develop/doc_cn/api/v2/config/evaluators.html index 6a3dd7b7ff5096dc61a342a1d6357647fc16132a..110f99429d39e6da436c49e5d7d6ac3298caec4d 100644 --- a/develop/doc_cn/api/v2/config/evaluators.html +++ b/develop/doc_cn/api/v2/config/evaluators.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/config/layer.html b/develop/doc_cn/api/v2/config/layer.html index d2fda21af9c88d8111aba5b32692116b98ea512d..87ee264489aa141422eba9dd7d1bac8d1b8cf481 100644 --- a/develop/doc_cn/api/v2/config/layer.html +++ b/develop/doc_cn/api/v2/config/layer.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/config/networks.html b/develop/doc_cn/api/v2/config/networks.html index 6435474c517087d4e5d12d82295e5dbe4a33cbc0..d97762c6ee1057559503bbb6c1ae9341e7c94f0b 100644 --- a/develop/doc_cn/api/v2/config/networks.html +++ b/develop/doc_cn/api/v2/config/networks.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/config/optimizer.html b/develop/doc_cn/api/v2/config/optimizer.html index 2010c91abc9d780cad10cd28274f13dd9a7c776e..8391ad321f5354711b9ad83d0eb41f3a3250ef94 100644 --- a/develop/doc_cn/api/v2/config/optimizer.html +++ b/develop/doc_cn/api/v2/config/optimizer.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/config/pooling.html b/develop/doc_cn/api/v2/config/pooling.html index a8aa37b0f5aa7b08b682bfc3b243b72e38484cf4..3dcaa73592b771cca3234485a63b4ffc1e1f87d0 100644 --- a/develop/doc_cn/api/v2/config/pooling.html +++ b/develop/doc_cn/api/v2/config/pooling.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/data.html b/develop/doc_cn/api/v2/data.html index 400ae40c63506e479dba1d4b555eac720b73d8c1..7a7a9d88c440c696f4ed8f6ffa0084efbcc695cb 100644 --- a/develop/doc_cn/api/v2/data.html +++ b/develop/doc_cn/api/v2/data.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/data/data_reader.html b/develop/doc_cn/api/v2/data/data_reader.html index 4ff1e824ed242afcef6452c989e950fd109e1033..683de1700651f5e133c1fae3b624c32ac7967a78 100644 --- a/develop/doc_cn/api/v2/data/data_reader.html +++ b/develop/doc_cn/api/v2/data/data_reader.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/data/dataset.html b/develop/doc_cn/api/v2/data/dataset.html index 634ab6b609f3266d76626c39fab6aa787e9a92e9..0243e7ae1c52effdf67006c323263b2a2c1f8011 100644 --- a/develop/doc_cn/api/v2/data/dataset.html +++ b/develop/doc_cn/api/v2/data/dataset.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/data/image.html b/develop/doc_cn/api/v2/data/image.html index 271b20e7f434d67a16a3215bda3057f67f984bdc..20298b47d85257806295616d7fb760d193829833 100644 --- a/develop/doc_cn/api/v2/data/image.html +++ b/develop/doc_cn/api/v2/data/image.html @@ -173,17 +173,17 @@- 训练与应用
- Fluid
diff --git a/develop/doc_cn/api/v2/fluid.html b/develop/doc_cn/api/v2/fluid.html index c90e71492dc80d2fa934dc122f43b23066feb7e7..acc7de0a9eeee46e8ea0b697cb1a8f9b1f71da42 100644 --- a/develop/doc_cn/api/v2/fluid.html +++ b/develop/doc_cn/api/v2/fluid.html @@ -34,7 +34,7 @@ - + @@ -173,17 +173,17 @@- 训练与应用
- Fluid
@@ -237,17 +237,17 @@Fluid¶
@@ -259,7 +259,7 @@ @@ -235,10 +235,15 @@--DataFeeder¶
--@@ -236,76 +236,24 @@DataFeeder¶
++@@ -249,10 +254,10 @@ diff --git a/develop/doc_cn/api/v2/fluid/evaluator.html b/develop/doc_cn/api/v2/fluid/evaluator.html index 0054570a769db743e30ff12af79d261afdd16d12..3c964240fbaef81ba58c37b1fe58e0dcb16a5f6a 100644 --- a/develop/doc_cn/api/v2/fluid/evaluator.html +++ b/develop/doc_cn/api/v2/fluid/evaluator.html @@ -8,7 +8,7 @@ -data_feeder¶
++DataFeeder¶
+-
+
-
+class
paddle.v2.fluid.data_feeder.DataFeeder(feed_list, place, program=None)
+
Evaluator — PaddlePaddle 文档 +evaluator — PaddlePaddle 文档 @@ -34,8 +34,8 @@ - - + + @@ -173,17 +173,17 @@- 训练与应用
- Fluid
@@ -226,7 +226,7 @@- Fluid >
-- Evaluator
+- evaluator
-Evaluator¶
--@@ -236,9 +236,38 @@Evaluator¶
+evaluator¶
+++Accuracy¶
-
-class
paddle.v2.fluid.evaluator.Evaluator(name, **kwargs)
- Base Class for all evaluators
--
-- - - 参数: -
-
- name (str) – The name of evaluator. such as, “accuracy”. Used for generate -temporary variable name. -
- main_program (Program, optional) – The evaluator should be added to this -main_program. Default default_main_program() -
- startup_program (Program, optional) – The parameter should be added to this -startup_program. Default default_startup_program() -
-
-
-
-
states¶
- list – The list of state variables. states will be reset to zero -when reset is invoked.
-
-
-
-
-
metrics¶
- list – The list of metrics variables. They will be calculate -every mini-batch
-
-
-
-
-
reset(executor, reset_program=None)
- reset metric states at the begin of each pass/user specified batch
-
-
-
-
-
eval(executor, eval_program=None)
- Evaluate the statistics merged by multiple mini-batches.
+classpaddle.v2.fluid.evaluator.Accuracy(input, label, k=1, **kwargs) +Average Accuracy for multiple mini-batches.
-
+
+@@ -318,10 +266,10 @@ every mini-batch diff --git a/develop/doc_cn/api/v2/fluid/executor.html b/develop/doc_cn/api/v2/fluid/executor.html index 569e950f28a5cff48852331c930db83e51b969ab..d1c95576c3e381c6fd549e0982876b7e75f80fd6 100644 --- a/develop/doc_cn/api/v2/fluid/executor.html +++ b/develop/doc_cn/api/v2/fluid/executor.html @@ -8,7 +8,7 @@ -ChunkEvaluator¶
+-
-
create_state(suffix, dtype, shape)
- Create state variable.
-NOTE: It is not a public API.
--
-- - - 参数: -
-
- suffix (str) – the state suffix. -
- dtype (str|core.DataType) – the state data type -
- shape (tuple|list) – the shape of state -
Returns: State variable
-
paddle.v2.fluid.evaluator.ChunkEvaluator(input, label, chunk_scheme, num_chunk_types, excluded_chunk_types=None) +Accumulate counter numbers output by chunk_eval from mini-batches and +compute the precision recall and F1-score using the accumulated counter +numbers.
Executor — PaddlePaddle 文档 +executor — PaddlePaddle 文档 @@ -34,8 +34,8 @@ - - + + @@ -173,17 +173,17 @@- 训练与应用
- Fluid
@@ -226,7 +226,7 @@- Fluid >
-- Executor
+- executor
@@ -236,91 +236,40 @@-@@ -249,10 +278,10 @@ diff --git a/develop/doc_cn/api/v2/fluid/initializer.html b/develop/doc_cn/api/v2/fluid/initializer.html index 8427e9a108a3f3c5b01eb45eddf71ad4aac428d3..d302e002ec41844883a2e8144c8403eb06738393 100644 --- a/develop/doc_cn/api/v2/fluid/initializer.html +++ b/develop/doc_cn/api/v2/fluid/initializer.html @@ -8,7 +8,7 @@ -Executor¶
+executor¶
+Executor¶
+-
+
-
+class
paddle.v2.fluid.executor.Executor(places)
+
++global_scope¶
+-
+
-
+
paddle.v2.fluid.executor.global_scope()
+
++scope_guard¶
+-
+
-
+
paddle.v2.fluid.executor.scope_guard(*args, **kwds)
+
+switch_scope¶
+-
+
-
+
paddle.v2.fluid.executor.switch_scope(scope)
+
Initializer — PaddlePaddle 文档 +initializer — PaddlePaddle 文档 @@ -34,8 +34,8 @@ - - + + @@ -173,17 +173,17 @@- 训练与应用
- Fluid
@@ -226,7 +226,7 @@- Fluid >
-- Initializer
+- initializer
-Initializer¶
--Initializer¶
--
+
-
-class
paddle.v2.fluid.initializer.Initializer
- Base class for variable initializers
-Defines the common interface of variable initializers. -They add operations to the init program that are used -to initialize variables. Users should not use this class -directly, but need to use one of its implementations.
+paddle.v2.fluid.initializer.Constant+ConstantInitializer的别名-
-class
paddle.v2.fluid.initializer.ConstantInitializer(value=0.0)
- Implements the constant initializer
+paddle.v2.fluid.initializer.Uniform+UniformInitializer的别名-
-class
paddle.v2.fluid.initializer.UniformInitializer(low=-1.0, high=1.0, seed=0)
- Implements the random uniform distribution initializer
+paddle.v2.fluid.initializer.Normal+NormalInitializer的别名-
-class
paddle.v2.fluid.initializer.NormalInitializer(loc=0.0, scale=1.0, seed=0)
- Implements the random Normal(Gaussian) distribution initializer
--
-class
paddle.v2.fluid.initializer.XavierInitializer(uniform=True, fan_in=None, fan_out=None, seed=0)
- Implements the Xavier initializer
-This class implements the Xavier weight initializer from the paper -Understanding the difficulty of training deep feedforward neural -networks[1] by Xavier Glorot and Yoshua Bengio.
-This initializer is designed to keep the scale of the gradients -approximately same in all the layers. In case of Uniform distribution, -the range is [-x, x], where x = sqrt(6 / (fan_in + fan_out)). -In case of Normal distribution, the mean is 0 and the standard deviation -is sqrt(2/ (fan_in + fan_out)).
-References
--
-
- [1] Understanding the difficulty of training deep feedforward neural -
- networks. International conference on artificial intelligence and -statistics. -(http://proceedings.mlr.press/v9/glorot10a.html) -
-
-class
paddle.v2.fluid.initializer.MSRAInitializer(uniform=True, fan_in=None, seed=0)
- Implements the MSRA initializer a.k.a. Kaiming Initializer
-This class implements the weight initialization from the paper -Delving Deep into Rectifiers: Surpassing Human-Level Performance on -ImageNet Classification[1] by Kaiming He, Xiangyu Zhang, Shaoqing Ren -and Jian Sun. This is a robust initialization method that particularly -considers the rectifier nonlinearities. In case of Uniform distribution, -the range is [-x, x], where x = sqrt(6 / fan_in). In case of Normal -distribution, the mean is 0 and the standard deviation -is sqrt(2/ fan_in).
-References
--
-
- [1] Delving Deep into Rectifiers: Surpassing Human-Level Performance -
- on ImageNet Classification -(https://arxiv.org/abs/1502.01852) -
paddle.v2.fluid.initializer.Xavier+XavierInitializer的别名- 训练与应用
- Fluid @@ -226,7 +226,7 @@
- Fluid > -
- IO +
- io
-
-
paddle.v2.fluid.io.is_parameter(var)
- Check whether the variable is a Parameter.
-This function checks whether the input variable is a Parameter.
+paddle.v2.fluid.io.save_vars(executor, dirname, main_program=None, vars=None, predicate=None) +Save variables to directory by executor.
+- 参数: var – The input variable. +参数: -
+
- executor – executor that save variable +
- dirname – directory path +
- main_program – program. If vars is None, then filter all variables in this +
返回: boolean result whether the variable is a Parameter. +program which fit predicate. Default default_main_program. +:param predicate: The Predicate describes a callable that returns a variable +as a bool. If it returns true, the variables will be saved. +:param vars: variables need to be saved. If specify vars, program & predicate +will be ignored +:return: None
+-
+
paddle.v2.fluid.io.save_params(executor, dirname, main_program=None)
+ Save all parameters to directory with executor.
+-
+
paddle.v2.fluid.io.save_persistables(executor, dirname, main_program=None)
+ Save all persistables to directory with executor.
+-
+
paddle.v2.fluid.io.load_vars(executor, dirname, main_program=None, vars=None, predicate=None)
+ Load variables from directory by executor.
++
++ + + 参数: -
+
- executor – executor that save variable +
- dirname – directory path +
- main_program – program. If vars is None, then filter all variables in this +
program which fit predicate. Default default_main_program(). +:param predicate: The Predicate describes a callable that returns a variable +as a bool. If it returns true, the variables will be loaded. +:param vars: variables need to be loaded. If specify vars, program & +predicate will be ignored +:return: None
+-
+
paddle.v2.fluid.io.load_params(executor, dirname, main_program=None)
+ load all parameters from directory by executor.
+-
+
paddle.v2.fluid.io.load_persistables(executor, dirname, main_program=None)
+ load all persistables from directory by executor.
-
+
paddle.v2.fluid.io.save_inference_model(dirname, feeded_var_names, target_vars, executor, main_program=None)
+ Build a model especially for inference, +and save it to directory by the executor.
++
++ + + 参数: -
+
- dirname – directory path +
- feeded_var_names – Names of variables that need to be feeded data during inference +
- target_vars – Variables from which we can get inference results. +
- executor – executor that save inference model +
- main_program – original program, which will be pruned to build the inference model. +Default default_main_program(). +
返回: None
+-
+
paddle.v2.fluid.io.load_inference_model(dirname, executor)
+ Load inference model from a directory
++
++ + + 参数: -
+
- dirname – directory path +
- executor – executor that load inference model +
返回: [program, feed_target_names, fetch_targets] +program: program especially for inference. +feed_target_names: Names of variables that need to feed data +fetch_targets: Variables from which we can get inference results.
+-
+
paddle.v2.fluid.io.get_inference_program(target_vars, main_program=None)
+ - 训练与应用
- Fluid @@ -226,7 +226,7 @@
- Fluid > -
- Layers +
- layers
-
-
paddle.v2.fluid.layers.fc(input, size, num_flatten_dims=1, param_attr=None, bias_attr=None, act=None, name=None)
- Fully Connected Layer
-The fully connected layer can take multiple tensors as its inputs. It -creates a variable (one for each input tensor) called weights for each -input tensor, which represents a fully connected weight matrix from -each input unit to each output unit. The fully connected layer -multiplies each input tensor with its coresponding weight to produce -an output Tensor. If multiple input tensors are given, the results of -multiple multiplications will be sumed up. If bias_attr is not None, -a biases variable will be created and added to the output. Finally, -if activation is not None, it will be applied to the output as well.
-This process can be formulated as follows:
--\[Out = Act({\sum_{i=0}^{N-1}W_iX_i + b})\]-In the above equation:
--
-
- \(N\): Number of the input. -
- \(X_i\): The input tensor. -
- \(W\): The weights created by this layer. -
- \(b\): The bias parameter created by this layer (if needed). -
- \(Act\): The activation funtion. -
- \(Out\): The output tensor. -
paddle.v2.fluid.layers.split_lod_tensor(input, mask, level=0) +split_lod_tensor
+This function takes in an input that contains the complete lod information, +and takes in a mask which is used to mask certain parts of the input. +The output is the true branch and the false branch with the mask applied to +the input at a certain level in the tensor.
参数: -
-
- input (Variable|list) – The input tensor(s) to the fully connected layer. -
- size (int) – The number of output units in the fully connected layer. -
- num_flatten_dims (int) – The fc layer can accept an input tensor with more -than two dimensions. If this happens, the -multidimensional tensor will first be flattened -into a 2-dimensional matrix. The parameter -num_flatten_dims determines how the input tensor -is flattened: the first num_flatten_dims -(inclusive, index starts from 1) dimensions will -be flatten to form the first dimension of the -final matrix (height of the matrix), and the rest -rank(X) - num_flatten_dims dimensions are -flattened to form the second dimension of the -final matrix (width of the matrix). For example, -suppose X is a 6-dimensional tensor with a shape -[2, 3, 4, 5, 6], and num_flatten_dims = 3. Then, -the flattened matrix will have a shape -[2 x 3 x 4, 5 x 6] = [24, 30]. By default, -num_flatten_dims is set to 1. -
- param_attr (ParamAttr|list) – The parameter attribute for learnable -parameters/weights of the fully connected -layer. -
- param_initializer (ParamAttr|list) – The initializer used for the -weight/parameter. If set None, -XavierInitializer() will be used. -
- bias_attr (ParamAttr|list) – The parameter attribute for the bias parameter -for this layer. If set None, no bias will be -added to the output units. -
- bias_initializer (ParamAttr|list) – The initializer used for the bias. -If set None, then ConstantInitializer() -will be used. -
- act (str) – Activation to be applied to the output of the fully connected -layer. -
- name (str) – Name/alias of the fully connected layer. +
- input (tuple|list|None) – The input tensor that contains complete +lod information needed to construct the output. +
- mask (list) – A bool column vector which masks the input. +
- level (int) – The specific lod level to rank.
返回: The output tensor variable.
-返回类型: Variable
+返回: The true branch of tensor as per the mask applied to input. +Variable: The false branch of tensor as per the mask applied to input.
Raises:
+ValueError– If rank of the input tensor is less than 2.返回类型: Variable
Examples
--data = fluid.layers.data(name="data", shape=[32, 32], dtype="float32") -fc = fluid.layers.fc(input=data, size=1000, act="tanh") +
x = layers.data(name='x', shape=[1]) +x.persistable = True + +y = layers.data(name='y', shape=[1]) +y.persistable = True + +out_true, out_false = layers.split_lod_tensor( + input=x, mask=y, level=level)
-embedding¶
++-merge_lod_tensor¶
-
-
paddle.v2.fluid.layers.embedding(input, size, is_sparse=False, padding_idx=None, param_attr=None, dtype='float32')
- Embedding Layer
-This layer is used to lookup embeddings of IDs, provided by
-input, in -a lookup table. The result of this lookup is the embedding of each ID in the -input.All the input variables are passed in as local variables to the LayerHelper -constructor.
+paddle.v2.fluid.layers.merge_lod_tensor(in_true, in_false, x, mask, level=0) +merge_lod_tensor
+This function takes in an input \(x\), the True branch, the False +branch and a binary \(mask\). Using this information, this function +merges the True and False branches of the tensor into a single Output +at a certain lod level indiacted by \(level\).
+参数: -
-
- input (Variable) – The tensor variable containing the IDs. -
- size (tuple|list) – The shape of the look up table parameter. It should -have two elements which indicate the size of the dictionary of -embeddings and the size of each embedding vector respectively. -
- is_sparse (bool) – The flag indicating whether to use sparse update. -
- padding_idx (int|long|None) – If
None, it makes no effect to lookup. -Otherwise the givenpadding_idxindicates padding the output -with zeros whenever lookup encounters it ininput. If -\(padding_idx < 0\), the padding_idx to use in lookup is -\(size[0] + dim\).
- - param_attr (ParamAttr) – Parameters for this layer -
- dtype (np.dtype|core.DataType|str) – The type of data : float32, float_16, int etc +
- in_true (tuple|list|None) – The True branch to be merged. +
- in_false (tuple|list|None) – The False branch to be merged. +
- x (tuple|list|None) – The input tensor that contains complete +lod information needed to construct the output. +
- mask (list) – A bool column vector which masks the input. +
- level (int) – The specific lod level to rank.
返回: The tensor variable storing the embeddings of the supplied inputs.
+返回: The merged output tensor.
返回类型: Variable
@@ -365,344 +317,209 @@ with zeros whenever lookup encounters it inExamples --dict_size = len(dataset.ids) -data = fluid.layers.data(name='ids', shape=[32, 32], dtype='float32') -fc = fluid.layers.embedding(input=data, size=[dict_size, 16]) +
x = layers.data( + name='x', shape=[1], dtype='float32', stop_gradient=False) +y = layers.data( + name='y', shape=[1], dtype='bool', stop_gradient=False) + +level = 0 + +out_true, out_false = layers.split_lod_tensor( + input=x, mask=y, level=level) +out = layers.merge_lod_tensor( + in_true=out_true, in_false=out_false, mask=y, x=x, level=level)
-dynamic_lstm¶
--
+
-
-
paddle.v2.fluid.layers.dynamic_lstm(input, size, param_attr=None, bias_attr=None, use_peepholes=True, is_reverse=False, gate_activation='sigmoid', cell_activation='tanh', candidate_activation='tanh', dtype='float32', name=None)
- Dynamic LSTM Layer
-The defalut implementation is diagonal/peephole connection -(https://arxiv.org/pdf/1402.1128.pdf), the formula is as follows:
--\[ \begin{align}\begin{aligned}i_t & = \sigma(W_{ix}x_{t} + W_{ih}h_{t-1} + W_{ic}c_{t-1} + b_i)\\f_t & = \sigma(W_{fx}x_{t} + W_{fh}h_{t-1} + W_{fc}c_{t-1} + b_f)\\\tilde{c_t} & = act_g(W_{cx}x_t + W_{ch}h_{t-1} + b_c)\\o_t & = \sigma(W_{ox}x_{t} + W_{oh}h_{t-1} + W_{oc}c_t + b_o)\\c_t & = f_t \odot c_{t-1} + i_t \odot \tilde{c_t}\\h_t & = o_t \odot act_h(c_t)\end{aligned}\end{align} \]-where the \(W\) terms denote weight matrices (e.g. \(W_{xi}\) is -the matrix of weights from the input gate to the input), \(W_{ic}, W_{fc}, W_{oc}\) are diagonal weight matrices for peephole connections. In -our implementation, we use vectors to reprenset these diagonal weight -matrices. The \(b\) terms denote bias vectors (\(b_i\) is the input -gate bias vector), \(\sigma\) is the non-linear activations, such as -logistic sigmoid function, and \(i, f, o\) and \(c\) are the input -gate, forget gate, output gate, and cell activation vectors, respectively, -all of which have the same size as the cell output activation vector \(h\).
-The \(\odot\) is the element-wise product of the vectors. \(act_g\) -and \(act_h\) are the cell input and cell output activation functions -and tanh is usually used for them. \(\tilde{c_t}\) is also called -candidate hidden state, which is computed based on the current input and -the previous hidden state.
-Set use_peepholes to False to disable peephole connection. The formula -is omitted here, please refer to the paper -http://www.bioinf.jku.at/publications/older/2604.pdf for details.
-Note that these \(W_{xi}x_{t}, W_{xf}x_{t}, W_{xc}x_{t}, W_{xo}x_{t}\) -operations on the input \(x_{t}\) are NOT included in this operator. -Users can choose to use fully-connect layer before LSTM layer.
+classpaddle.v2.fluid.layers.BlockGuard(main_program) +BlockGuard class.
+BlockGuard class is used to create a sub-block in a program by +using the Python with keyword.
+-
+class
paddle.v2.fluid.layers.BlockGuardWithCompletion(rnn)
+ BlockGuardWithCompletion class.
+BlockGuardWithCompletion class is used to create an op with a block in a program.
+-
+class
paddle.v2.fluid.layers.StaticRNNMemoryLink(init, pre_mem, mem=None)
+ StaticRNNMemoryLink class.
-- 参数: -
-
- input (Variable) – The input of dynamic_lstm layer, which supports -variable-time length input sequence. The underlying -tensor in this Variable is a matrix with shape -(T X 4D), where T is the total time steps in this -mini-batch, D is the hidden size. -
- size (int) – 4 * hidden size. -
- param_attr (ParamAttr|None) –
The parameter attribute for the learnable -hidden-hidden weights.
--
-
- Weights = {\(W_{ch}, W_{ih}, W_{fh}, W_{oh}\)} -
- The shape is (D x 4D), where D is the hidden -size. -
- - bias_attr (ParamAttr|None) –
The bias attribute for the learnable bias -weights, which contains two parts, input-hidden -bias weights and peephole connections weights if -setting use_peepholes to True.
--
-
- use_peepholes = False -
-
--
-
- Biases = {\(b_c, b_i, b_f, b_o\)}. -
- The shape is (1 x 4D). -
-
-
- use_peepholes = True -
-
--
-
- Biases = { \(b_c, b_i, b_f, b_o, W_{ic}, W_{fc}, W_{oc}\)}. -
- The shape is (1 x 7D). -
- - use_peepholes (bool) – Whether to enable diagonal/peephole connections, -default True. -
- is_reverse (bool) – Whether to compute reversed LSTM, default False. -
- gate_activation (str) – The activation for input gate, forget gate and -output gate. Choices = [“sigmoid”, “tanh”, “relu”, -“identity”], default “sigmoid”. -
- cell_activation (str) – The activation for cell output. Choices = [“sigmoid”, -“tanh”, “relu”, “identity”], default “tanh”. -
- candidate_activation (str) – The activation for candidate hidden state. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], -default “tanh”. -
- dtype (str) – Data type. Choices = [“float32”, “float64”], default “float32”. -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- init – the initial variable for Memory +
- init – Variable +
- pre_mem – the memory variable in previous time step +
- pre_mem – Variable +
- mem – the memory variable in current time step +
- mem – Variable
参数: -
+
返回: The hidden state, and cell state of LSTM. The shape of both is (T x D), and lod is the same with the input.
-返回类型: tuple
-Examples
-+-hidden_dim = 512 -forward_proj = fluid.layers.fc(input=input_seq, size=hidden_dim * 4, - act=None, bias_attr=None) -forward, _ = fluid.layers.dynamic_lstm( - input=forward_proj, size=hidden_dim * 4, use_peepholes=False) -
StaticRNNMemoryLink class is used to create a link between two +memory cells of a StaticRNN.
-
+class
paddle.v2.fluid.layers.WhileGuard(while_op)
+ -
+class
paddle.v2.fluid.layers.While(cond, name=None)
+ -
-
paddle.v2.fluid.layers.dynamic_lstmp(input, size, proj_size, param_attr=None, bias_attr=None, use_peepholes=True, is_reverse=False, gate_activation='sigmoid', cell_activation='tanh', candidate_activation='tanh', proj_activation='tanh', dtype='float32', name=None)
- Dynamic LSTMP Layer
-LSTMP (LSTM with recurrent projection) layer has a separate projection -layer after the LSTM layer, projecting the original hidden state to a -lower-dimensional one, which is proposed to reduce the number of total -parameters and furthermore computational complexity for the LSTM, -espeacially for the case that the size of output units is relative -large (https://research.google.com/pubs/archive/43905.pdf).
-The formula is as follows:
--\[ \begin{align}\begin{aligned}i_t & = \sigma(W_{ix}x_{t} + W_{ir}r_{t-1} + W_{ic}c_{t-1} + b_i)\\f_t & = \sigma(W_{fx}x_{t} + W_{fr}r_{t-1} + W_{fc}c_{t-1} + b_f)\\\tilde{c_t} & = act_g(W_{cx}x_t + W_{cr}r_{t-1} + b_c)\\o_t & = \sigma(W_{ox}x_{t} + W_{or}r_{t-1} + W_{oc}c_t + b_o)\\c_t & = f_t \odot c_{t-1} + i_t \odot \tilde{c_t}\\h_t & = o_t \odot act_h(c_t)\\r_t & = \overline{act_h}(W_{rh}h_t)\end{aligned}\end{align} \]-In the above formula:
--
-
- \(W\): Denotes weight matrices (e.g. \(W_{xi}\) is the matrix of weights from the input gate to the input). -
- \(W_{ic}\), \(W_{fc}\), \(W_{oc}\): Diagonal weight matrices for peephole connections. In our implementation, we use vectors to reprenset these diagonal weight matrices. -
- \(b\): Denotes bias vectors (e.g. \(b_i\) is the input gate bias vector). -
- \(\sigma\): The activation, such as logistic sigmoid function. -
- \(i, f, o\) and \(c\): The input gate, forget gate, output gate, and cell activation vectors, respectively, all of which have the same size as the cell output activation vector \(h\). -
- \(h\): The hidden state. -
- \(r\): The recurrent projection of the hidden state. -
- \(\tilde{c_t}\): The candidate hidden state, whose computation is based on the current input and previous hidden state. -
- \(\odot\): The element-wise product of the vectors. -
- \(act_g\) and \(act_h\): The cell input and cell output activation functions and tanh is usually used for them. -
- \(\overline{act_h}\): The activation function for the projection output, usually using identity or same as \(act_h\). -
Set use_peepholes to False to disable peephole connection. The formula -is omitted here, please refer to the paper -http://www.bioinf.jku.at/publications/older/2604.pdf for details.
-Note that these \(W_{xi}x_{t}, W_{xf}x_{t}, W_{xc}x_{t}, W_{xo}x_{t}\) -operations on the input \(x_{t}\) are NOT included in this operator. -Users can choose to use fully-connected layer before LSTMP layer.
+paddle.v2.fluid.layers.lod_rank_table(x, level=0) +LoD Rank Table Operator. Given an input variable x and a level number +of LoD, this layer creates a LodRankTable object. A LoDRankTable object +contains a list of bi-element tuples. Each tuple consists of an index and +a length, both of which are int type. Refering to specified level of LoD, +the index is the sequence index number and the length representes the +sequence length. Please note that the list is ranked in descending order by +the length. The following is an example:
++
++x is a LoDTensor: + x.lod = [[0, 2, 3], + [0, 5, 6, 7]] + x.data = [a, b, c, d, e, f, g] + +1. set level to 0: + Create lod rank table: + lod_rank_table_obj = lod_rank_table(x, level=0) + + Get: + lod_rank_table_obj.items() = [(0, 2), (1, 1)] + +2. set level to 1: + Create lod rank table: + lod_rank_table_obj = lod_rank_table(x, level=1) + + Get: + lod_rank_table_obj.items() = [(0, 5), (1, 1), (2, 1)] +
参数: -
-
- input (Variable) – The input of dynamic_lstmp layer, which supports -variable-time length input sequence. The underlying -tensor in this Variable is a matrix with shape -(T X 4D), where T is the total time steps in this -mini-batch, D is the hidden size. -
- size (int) – 4 * hidden size. -
- proj_size (int) – The size of projection output. -
- param_attr (ParamAttr|None) –
The parameter attribute for the learnable -hidden-hidden weight and projection weight.
--
-
- Hidden-hidden weight = {\(W_{ch}, W_{ih}, W_{fh}, W_{oh}\)}. -
- The shape of hidden-hidden weight is (P x 4D), -where P is the projection size and D the hidden -size. -
- Projection weight = {\(W_{rh}\)}. -
- The shape of projection weight is (D x P). -
- - bias_attr (ParamAttr|None) –
The bias attribute for the learnable bias -weights, which contains two parts, input-hidden -bias weights and peephole connections weights if -setting use_peepholes to True.
--
-
- use_peepholes = False -
-
--
-
- Biases = {\(b_c, b_i, b_f, b_o\)}. -
- The shape is (1 x 4D). -
-
-
- use_peepholes = True -
-
--
-
- Biases = { \(b_c, b_i, b_f, b_o, W_{ic}, W_{fc}, W_{oc}\)}. -
- The shape is (1 x 7D). -
- - use_peepholes (bool) – Whether to enable diagonal/peephole connections, -default True. -
- is_reverse (bool) – Whether to compute reversed LSTM, default False. -
- gate_activation (str) – The activation for input gate, forget gate and -output gate. Choices = [“sigmoid”, “tanh”, “relu”, -“identity”], default “sigmoid”. -
- cell_activation (str) – The activation for cell output. Choices = [“sigmoid”, -“tanh”, “relu”, “identity”], default “tanh”. -
- candidate_activation (str) – The activation for candidate hidden state. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], -default “tanh”. -
- proj_activation (str) – The activation for projection output. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], -default “tanh”. -
- dtype (str) – Data type. Choices = [“float32”, “float64”], default “float32”. -
- name (str|None) – A name for this layer(optional). If set None, the layer -will be named automatically. +
- x (Variable) – Input variable, a LoDTensor based which to create the lod +rank table. +
- level (int) – Specify the LoD level, on which to create the lod rank +table.
返回: The projection of hidden state, and cell state of LSTMP. The shape of projection is (T x P), for the cell state which is (T x D), and both LoD is the same with the input.
+返回: The created LoDRankTable object.
返回类型: tuple
+返回类型: Variable
Examples
--hidden_dim, proj_dim = 512, 256 -fc_out = fluid.layers.fc(input=input_seq, size=hidden_dim * 4, - act=None, bias_attr=None) -proj_out, _ = fluid.layers.dynamic_lstmp(input=fc_out, - size=hidden_dim * 4, - proj_size=proj_dim, - use_peepholes=False, - is_reverse=True, - cell_activation="tanh", - proj_activation="tanh") +
x = fluid.layers.data(name='x', shape=[10], + dtype='float32', lod_level=1) +out = layers.lod_rank_table(x=x, level=0)
-dynamic_gru¶
++max_sequence_len¶
-
-
paddle.v2.fluid.layers.dynamic_gru(input, size, param_attr=None, bias_attr=None, is_reverse=False, gate_activation='sigmoid', candidate_activation='tanh', h_0=None)
- Dynamic GRU Layer
-Refer to Empirical Evaluation of Gated Recurrent Neural Networks on -Sequence Modeling
-The formula is as follows:
--\[ \begin{align}\begin{aligned}u_t & = act_g(W_{ux}x_{t} + W_{uh}h_{t-1} + b_u)\\r_t & = act_g(W_{rx}x_{t} + W_{rh}h_{t-1} + b_r)\\\tilde{h_t} & = act_c(W_{cx}x_{t} + W_{ch}(r_t \odot h_{t-1}) + b_c)\\h_t & = (1-u_t) \odot h_{t-1} + u_t \odot \tilde{h_t}\end{aligned}\end{align} \]-The \(\odot\) is the element-wise product of the vectors. \(act_g\) -is the update gate and reset gate activation function and \(sigmoid\) -is usually used for it. \(act_c\) is the activation function for -candidate hidden state and \(tanh\) is usually used for it.
-Note that these \(W_{ux}x_{t}, W_{rx}x_{t}, W_{cx}x_{t}\) operations on -the input \(x_{t}\) are NOT included in this operator. Users can choose -to use fully-connect layer before GRU layer.
+paddle.v2.fluid.layers.max_sequence_len(rank_table) +Max Sequence Len Operator. Given a LoDRankTable object, this layer +returns the max length of a batch of sequences. In fact, a LoDRankTable +object contains a list of tuples(<sequence index, sequence length>) and +the list is already sorted by sequence length in descending order, so the +operator just returns the sequence length of the first tuple element.
- 参数: -
-
- input (Variable) – The input of dynamic_gru layer, which supports -variable-time length input sequence. The underlying tensor in this -Variable is a matrix with shape \((T \times 3D)\), where -\(T\) is the total time steps in this mini-batch, \(D\) -is the hidden size. -
- size (int) – The dimension of the gru cell. -
- param_attr (ParamAttr|None) –
The parameter attribute for the learnable -hidden-hidden weight matrix. Note:
--
-
- The shape of the weight matrix is \((T \times 3D)\), where -\(D\) is the hidden size. -
- All elements in the weight matrix can be divided into two parts. -The first part are weights of the update gate and reset gate with -shape \((D \times 2D)\), and the second part are weights for -candidate hidden state with shape \((D \times D)\). -
- - bias_attr (ParamAttr) – The parameter attribute for learnable the -hidden-hidden bias. -
- is_reverse (bool) – Whether to compute reversed GRU, default
-
False.
- - gate_activation (str) – The activation for update gate and reset gate. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], default “sigmoid”. -
- activation (str) – The activation for candidate hidden state. -Choices = [“sigmoid”, “tanh”, “relu”, “identity”], default “tanh”. -
参数: rank_table (Variable) – Input variable which is a LoDRankTable object. 返回: The hidden state of GRU. The shape is (T times D), and lod is the same with the input.
-返回: The max length of sequence. 返回类型: Variable
-返回类型: Variable Examples
--hidden_dim = 512 -x = fluid.layers.fc(input=data, size=hidden_dim * 3) -hidden = fluid.layers.dynamic_gru(input=x, dim=hidden_dim) +
x = fluid.layers.data(name='x', shape=[10], + dtype='float32', lod_level=1) +rank_table = layers.lod_rank_table(x=x, level=0) +max_seq_len = layers.max_sequence_len(rank_table)
-data¶
++topk¶
-
-
paddle.v2.fluid.layers.data(name, shape, append_batch_size=True, dtype='float32', lod_level=0, type=VarType.LOD_TENSOR, stop_gradient=True)
- Data Layer
-This function takes in the input and based on whether data has -to be returned back as a minibatch, it creates the global variable by using -the helper functions. The global variables can be accessed by all the -following operators in the graph.
-All the input variables of this function are passed in as local variables -to the LayerHelper constructor.
+paddle.v2.fluid.layers.topk(input, k) +topk
+This function performs the operation that selects the k entries in the input +vector and outputs their values and indices as vectors. Thus topk_out[j] is +the j-th largest entry in input, and its index is topk_indices[j]
参数: -
-
- name (str) – The name/alias of the function -
- shape (list) – Tuple declaring the shape. -
- append_batch_size (bool) – Whether or not to append the data as a batch. -
- dtype (int|float) – The type of data : float32, float_16, int etc -
- type (VarType) – The output type. By default it is LOD_TENSOR. -
- lod_level (int) – The LoD Level. 0 means the input data is not a sequence. -
- main_program (Program) – Name of the main program that calls this -
- startup_program (Program) – Name of the startup program -
- stop_gradient (bool) – A boolean that mentions whether gradient should flow. +
- input (Variable|list) – The input tensor that has all the data. +
- k (int) – The number of top elements that the function will pick.
返回: The global variable that gives access to the data.
+返回: -
+
- The variable of type array that contains the k largest entries +
from input.
+
+- Variable: The variable of type array that contains the indices of k +
largest entries from input.
+
+
返回类型: Variable
@@ -711,399 +528,388 @@ to the LayerHelper constructor.Examples
--data = fluid.layers.data(name='x', shape=[784], dtype='float32') +
x = fluid.layers.data(name='x', shape=[10]) +k = 5 +array = fluid.layers.topk(x, k)
-mean¶
++lod_tensor_to_array¶
-
-
paddle.v2.fluid.layers.mean(**kwargs)
- Mean Operator.
-Out is a scalar which is the mean of all elements in X.
+paddle.v2.fluid.layers.lod_tensor_to_array(x, table) +Convert a LOD_TENSOR to an LOD_TENSOR_ARRAY.
-- 参数: x – The input of mean op -Duplicable: False Optional: False -返回: The output of mean op +参数: -
+
- x (Variable|list) – The LOD tensor to be converted to a LOD tensor array. +
- table (ParamAttr|list) – The variable that stores the level of lod +which is ordered by sequence length in +descending order. +
返回: -
+
- The variable of type array that has been converted from a +
tensor.
+
+
返回类型: Variable
+
++BlockGuard¶
+++BlockGuardWithCompletion¶
+-
+
+-StaticRNNMemoryLink¶
+-
+
-dynamic_lstmp¶
+++WhileGuard¶
+-
+
++While¶
+-
+
+lod_rank_table¶
Examples
+++x = fluid.layers.data(name='x', shape=[10]) +table = fluid.layers.lod_rank_table(x, level=0) +array = fluid.layers.lod_tensor_to_array(x, table) +
-mul¶
++-array_to_lod_tensor¶
-
-
paddle.v2.fluid.layers.mul(**kwargs)
- Mul Operator.
-This operator is used to perform matrix multiplication for input $X$ and $Y$.
-The equation is:
-$$Out = X * Y$$
-Both the input $X$ and $Y$ can carry the LoD (Level of Details) information, -or not. But the output only shares the LoD information with input $X$.
+paddle.v2.fluid.layers.array_to_lod_tensor(x, table) +Convert a LoD_Tensor_Aarry to an LoDTensor.
+参数: -
-
- x – (Tensor), The first input tensor of mul op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of mul op. -Duplicable: False Optional: False -
- x_num_col_dims (INT) – (int, default 1), The mul_op can take tensors with more than two -dimensions as its inputs. If the input $X$ is a tensor with more -than two dimensions, $X$ will be flattened into a two-dimensional -matrix first. The flattening rule is: the first num_col_dims -will be flattened to form the first dimension of the final matrix -(the height of the matrix), and the rest rank(X) - num_col_dims -dimensions are flattened to form the second dimension of the final -matrix (the width of the matrix). As a result, height of the -flattened matrix is equal to the product of $X$’s first -x_num_col_dims dimensions’ sizes, and width of the flattened -matrix is equal to the product of $X$’s last rank(x) - num_col_dims -dimensions’ size. For example, suppose $X$ is a 6-dimensional -tensor with the shape [2, 3, 4, 5, 6], and x_num_col_dims = 3. -Thus, the flattened matrix will have a shape [2 x 3 x 4, 5 x 6] = -[24, 30]. -
- y_num_col_dims (INT) – (int, default 1), The mul_op can take tensors with more than two, -dimensions as its inputs. If the input $Y$ is a tensor with more -than two dimensions, $Y$ will be flattened into a two-dimensional -matrix first. The attribute y_num_col_dims determines how $Y$ is -flattened. See comments of x_num_col_dims for more details. +
- x (Variable|list) – The lod tensor array to be converted to a tensor. +
- table (ParamAttr|list) – The variable that stores the level of lod +which is ordered by sequence length in +descending order.
返回: (Tensor), The output tensor of mul op.
+返回: -
+
- The variable of type tensor that has been converted +
from an array.
+
+
返回类型: Variable
Examples
++x = fluid.layers.data(name='x', shape=[10]) +table = fluid.layers.lod_rank_table(x, level=0) +array = fluid.layers.lod_tensor_to_array(x, table) +lod_tensor = fluid.layers.array_to_lod_tensor(array, table) +
-elementwise_add¶
++-increment¶
-
-
paddle.v2.fluid.layers.elementwise_add(**kwargs)
- Limited Elementwise Add Operator.
-The equation is:
-$$Out = X + Y$$
-$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be -smaller than or equal to the dimensions of $X$.
-There are two cases for this operator: -1. The shape of $Y$ is same with $X$; -2. The shape of $Y$ is a subset of $X$.
-For case 2: -$Y$ will be broadcasted to match the shape of $X$ and axis should be -set to index of the start dimension to broadcast $Y$ onto $X$.
--
-
- For example -
- --
shape(X) = (2, 3, 4, 5), shape(Y) = (,) -shape(X) = (2, 3, 4, 5), shape(Y) = (5,) -shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) -shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 -shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 -
-
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) -information. However, the output only shares the LoD information with input $X$.
+paddle.v2.fluid.layers.increment(x, value=1.0, in_place=True) +This function performs an operation that increments each value in the +input \(x\) by an amount: \(value\) as mentioned in the input +parameter. This operation is performed in-place by default.
+参数: -
-
- x – (Tensor), The first input tensor of elementwise op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of elementwise op. -Duplicable: False Optional: False -
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. +
- 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.
返回: The output of elementwise op.
+返回: -
+
- The tensor variable storing the transformation of +
element-wise increment of each value in the input.
+
+
返回类型: Variable
Examples
++data = fluid.layers.data(name='data', shape=[32, 32], dtype='float32') +data = fluid.layers.increment(x=data, value=3.0, in_place=True) +
-elementwise_sub¶
++-array_write¶
-
-
paddle.v2.fluid.layers.elementwise_sub(**kwargs)
- Limited Elementwise Sub Operator.
-The equation is:
-$$Out = X - Y$$
-$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be -smaller than or equal to the dimensions of $X$.
-There are two cases for this operator: -1. The shape of $Y$ is same with $X$; -2. The shape of $Y$ is a subset of $X$.
-For case 2: -$Y$ will be broadcasted to match the shape of $X$ and axis should be -set to index of the start dimension to broadcast $Y$ onto $X$.
--
-
- For example -
- --
shape(X) = (2, 3, 4, 5), shape(Y) = (,) -shape(X) = (2, 3, 4, 5), shape(Y) = (5,) -shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) -shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 -shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 -
-
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) -information. However, the output only shares the LoD information with input $X$.
+paddle.v2.fluid.layers.array_write(x, i, array=None) +This function writes the given input variable to the specified position +indicating by the arrary index to an output LOD_TENSOR_ARRAY. If the +output LOD_TENSOR_ARRAY is not given(None), a new one will be created and +returned.
+参数: -
-
- x – (Tensor), The first input tensor of elementwise op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of elementwise op. -Duplicable: False Optional: False -
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. +
- x (Variable|list) – The input tensor from which the data will be read. +
- i (Variable|list) – The index of the output LOD_TENSOR_ARRAY, pointing to +the position to which the input tensor will be +written. +
- array (Variable|list) – The output LOD_TENSOR_ARRAY to which the input +tensor will be written. If this parameter is +NONE, a new LOD_TENSOR_ARRAY will be created and +returned.
返回: The output of elementwise op.
+返回: The output LOD_TENSOR_ARRAY where the input tensor is written.
+返回类型: Variable
Examples
-elementwise_mul¶
++create_array¶
-
-
paddle.v2.fluid.layers.elementwise_mul(**kwargs)
- Limited Elementwise Mul Operator.
-The equation is:
-$$Out = X odotY$$
-$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be -smaller than or equal to the dimensions of $X$.
-There are two cases for this operator: -1. The shape of $Y$ is same with $X$; -2. The shape of $Y$ is a subset of $X$.
-For case 2: -$Y$ will be broadcasted to match the shape of $X$ and axis should be -set to index of the start dimension to broadcast $Y$ onto $X$.
--
-
- For example -
- +
shape(X) = (2, 3, 4, 5), shape(Y) = (,) -shape(X) = (2, 3, 4, 5), shape(Y) = (5,) -shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) -shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 -shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 +
paddle.v2.fluid.layers.create_array(dtype) +This function creates an array of type \(LOD_TENSOR_ARRAY\) using the +LayerHelper.
++
++ + + 参数: dtype (int|float) – The data type of the elements in the array. +返回: The tensor variable storing the elements of data type. +返回类型: Variable +Examples
+-data = fluid.layers.create_array(dtype='float32')
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) -information. However, the output only shares the LoD information with input $X$.
+ + ++-less_than¶
+-
+
-
+
paddle.v2.fluid.layers.less_than(x, y, cond=None, **ignored)
+ Less than
+This layer returns the truth value of \(x < y\) elementwise.
+参数: -
-
- x – (Tensor), The first input tensor of elementwise op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of elementwise op. -Duplicable: False Optional: False -
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. +
- x (Variable) – First operand of less_than +
- y (Variable) – Second operand of less_than +
- cond (Variable|None) – Optional output variable to store the result of less_than
返回: The output of elementwise op.
+返回: The tensor variable storing the output of less_than.
+返回类型: Variable
Examples
++less = fluid.layers.less_than(x=label, y=limit) +
-elementwise_div¶
++-array_read¶
-
-
paddle.v2.fluid.layers.elementwise_div(**kwargs)
- Limited Elementwise Div Operator.
-The equation is:
-$$Out = X / Y$$
-$X$ is a tensor of any dimension and the dimensions of tensor $Y$ must be -smaller than or equal to the dimensions of $X$.
-There are two cases for this operator: -1. The shape of $Y$ is same with $X$; -2. The shape of $Y$ is a subset of $X$.
-For case 2: -$Y$ will be broadcasted to match the shape of $X$ and axis should be -set to index of the start dimension to broadcast $Y$ onto $X$.
--
-
- For example -
- --
shape(X) = (2, 3, 4, 5), shape(Y) = (,) -shape(X) = (2, 3, 4, 5), shape(Y) = (5,) -shape(X) = (2, 3, 4, 5), shape(Y) = (4, 5) -shape(X) = (2, 3, 4, 5), shape(Y) = (3, 4), with axis=1 -shape(X) = (2, 3, 4, 5), shape(Y) = (2), with axis=0 -
-
Either of the inputs $X$ and $Y$ or none can carry the LoD (Level of Details) -information. However, the output only shares the LoD information with input $X$.
+paddle.v2.fluid.layers.array_read(array, i) +This function performs the operation to read the data in as an +LOD_TENSOR_ARRAY. +:param array: The input tensor that will be written to an array. +:type array: Variable|list +:param i: The subscript index in tensor array, that points the
++
place where data will be written to.
+- 参数: -
-
- x – (Tensor), The first input tensor of elementwise op. -Duplicable: False Optional: False -
- y – (Tensor), The second input tensor of elementwise op. -Duplicable: False Optional: False -
- axis (INT) – (int, default -1). The start dimension index for broadcasting Y onto X. -
返回: The tensor type variable that has the data written to it. 返回: The output of elementwise op.
-返回类型: Variable Examples
-dropout¶
+++shrink_memory¶
-
-
paddle.v2.fluid.layers.dropout(x, dropout_prob, is_test=False, seed=None, **kwargs)
- Computes dropout.
-Drop or keep each element of x independently. Dropout is a regularization -technique for reducing overfitting by preventing neuron co-adaption during -training. The dropout operator randomly set (according to the given dropout -probability) the outputs of some units to zero, while others are remain -unchanged.
+paddle.v2.fluid.layers.shrink_memory(x, i, table) +This function creates an operator to shrink_rnn_memory using the RankTable +as mentioned in the input parameter.
+
+-array_length¶
+-
+
-
+
paddle.v2.fluid.layers.array_length(array)
+ This function performs the operation to find the length of the input +LOD_TENSOR_ARRAY.
- 参数: -
-
- x (variable) – The input tensor. -
- dropout_prob (float) – Probability of setting units to zero. -
- is_test (bool) – A flag indicating whether it is in test phrase or not. -
- seed (int) – A Python integer used to create random seeds. If this -parameter is set to None, a random seed is used. -NOTE: If an integer seed is given, always the same output -units will be dropped. DO NOT use a fixed seed in training. -
参数: array (LOD_TENSOR_ARRAY) – The input array that will be used +to compute the length. 返回: A tensor variable.
-返回: The length of the input LoDTensorArray. 返回类型: Variable
-返回类型: Variable Examples
--x = fluid.layers.data(name="data", shape=[32, 32], dtype="float32") -droped = fluid.layers.dropout(input=x, dropout_rate=0.5) -
-reshape¶
--
+
-
-
paddle.v2.fluid.layers.reshape(**kwargs)
- Reshape Operator.
-Reshape Input(X) into the shape specified by Attr(shape).
-An example: -Given a 2-D tensor X with 2 rows and 2 columns : [[1, 2], [3, 4]]
-and target shape = [1, 4], the reshape operator will transform -the tensor X into a 2-D tensor: [[1, 2, 3, 4]]
-One dimension in the target shape can be set -1, representing that its -size is unknown. In this case, the real dimension will be infered from -the original shape of Input(X) and other dimensions in the target shape.
-+class
paddle.v2.fluid.layers.IfElse(cond, name=None) + + + +++DynamicRNN¶
+-
+
-
+class
paddle.v2.fluid.layers.DynamicRNN(name=None)
+
++ConditionalBlock¶
+-
+
-
+class
paddle.v2.fluid.layers.ConditionalBlock(inputs, name=None)
+
+-StaticRNN¶
+-
+
-
+class
paddle.v2.fluid.layers.StaticRNN(name=None)
+ StaticRNN class.
+StaticRNN class is used to create a StaticRNN. The RNN will have its +own parameters like inputs, outputs, memories, status and length.
+-
+
-
+
memory(init=None, shape=None, batch_ref=None, init_value=0.0, init_batch_dim_idx=0, ref_batch_dim_idx=1)
+ - 参数: -
-
- x – The input tensor of reshape operator. -Duplicable: False Optional: False -
- shape (INTS) – (vector<int>) Target shape of reshape operator. +
- init – boot memory, if not set, a shape, batch_ref must be provided +
- shape – shape of the boot memory +
- batch_ref – batch size reference variable +
- init_value – the init value of boot memory +
- init_batch_dim_idx – the index of batch size in init’s dimension +
- ref_batch_dim_idx – the index of batch size in batch_ref’s dimension
参数: -
+
返回: The output tensor of reshape operator.
-
-
+
--sigmoid¶
--
-
-
-
paddle.v2.fluid.layers.sigmoid(**kwargs)
- Sigmoid Activation Operator
-$$out = frac{1}{1 + e^{-x}}$$
--
- - - 参数: x – Input of Sigmoid operator -Duplicable: False Optional: False -返回: Output of Sigmoid operator -
-scale¶
++reorder_lod_tensor_by_rank¶
-
-
paddle.v2.fluid.layers.scale(**kwargs)
- Scale operator
-$$Out = scale*X$$
+paddle.v2.fluid.layers.reorder_lod_tensor_by_rank(x, rank_table) +ReorderLoDTensorByRankTable operator.
+Input(X) is a batch of sequences. Input(RankTable) stores new orders of the +input sequence batch. The reorder_lod_tensor_by_rank operator reorders the +Input(X) according to the information provided by Input(RankTable).
+For example:
+If the indices stored in the Input(RankTable) are [3, 0, 2, 1], the +Input(X) will be reordered that the fourth sequence in Input(X) will become the +first one, and then followed by the original first, third, and the second one.
+This is: +X = [Seq0, Seq1, Seq2, Seq3]. The indices in RankTable are [3, 0, 2, 1]. +Out = [Seq3, Seq0, Seq2, Seq1] with a new LoD information.
+If the LoD information of Input(X) is empty, this means Input(X) is not sequence +data. This is also identical to a batch of sequences where each sequence has a +fixed length 1. In this case, the reorder_lod_tensor_by_rank operator reorders +each slice of Input(X) along the first axis according to Input(RankTable).
+This is: +X = [Slice0, Slice1, Slice2, Slice3] and its LoD information is empty. The +indices in RankTable are [3, 0, 2, 1]. +Out = [Slice3, Slice0, Slice2, Slice1] with no LoD information is appended.
+NOTE: This operator sorts Input(X) according to a given LoDRankTable which does +not need to be calculated according to Input(X). It can be calculated according +to another different sequence, and then this operator sorts Input(X) according +to the given LoDRankTable.
参数: -
-
- x – (Tensor) Input tensor of scale operator. +
- x – (LoDTensor), the input lod tensor to be reordered according to Input(RankTable). +Duplicable: False Optional: False +
- rank_table – (LoDRankTable), the rank table according to which Input(X) is reordered. Duplicable: False Optional: False -
- scale (FLOAT) – (float, default 1.0)The scaling factor of the scale operator.
返回: (Tensor) Output tensor of scale operator.
+返回: (LoDTensor), the reordered lod tensor.
-transpose¶
+++ParallelDo¶
+-
+
-
+class
paddle.v2.fluid.layers.ParallelDo(places, name=None)
+ ParallelDo class.
+ParallelDo class is used to create a ParallelDo.
+
+Print¶
-
-
paddle.v2.fluid.layers.transpose(x, perm, name=None)
- transpose Layer
-Permute the dimensions of input according to perm.
-The i-th dimension of the returned tensor will correspond to the -perm[i]-th dimension of input.
+paddle.v2.fluid.layers.Print(input, first_n=-1, message=None, summarize=-1, print_tensor_name=True, print_tensor_type=True, print_tensor_shape=True, print_tensor_lod=True, print_phase='both') +Print operator
+This creates a print op that will print when a tensor is accessed.
+Wraps the tensor passed in so that whenever that a tensor is accessed, +the message message is printed, along with the current value of the +tensor t.
参数: -
-
- input (Variable) – (Tensor), A Tensor. -
- perm (list) – A permutation of the dimensions of input. +
- input (Variable) – A Tensor to print. +
- summarize (int) – Print this number of elements in the tensor, will print +all if left is negative. +
- message (str) – A string message to print as a prefix. +
- first_n (int) – Only log first_n number of times. +
- print_tensor_name (bool) – Print the tensor name. +
- print_tensor_type (bool) – Print the tensor type. +
- print_tensor_shape (bool) – Print the tensor shape. +
- print_tensor_lod (bool) – Print the tensor lod. +
- print_phase (bool) – Which phase to displace, including ‘forward’, +‘backward’ and ‘both’. If set to ‘backward’ or ‘both’, will +print the gradients of input tensor.
返回: A transposed Tensor.
+返回: Output tensor, same data with input tensor.
返回类型: Variable
@@ -1139,210 +966,260 @@ perm[i]-th dimension of input.Examples
--x = fluid.layers.data(name='x', shape=[5, 10, 15], dtype='float32') -x_transposed = layers.transpose(x, perm=[1, 0, 2]) +
+value = some_layer(...) +Print(value, summarize=10,
++
message=”The content of some_layer: ”)--sigmoid_cross_entropy_with_logits¶
---cast¶
--
-
-
-
paddle.v2.fluid.layers.cast(x, dtype)
- This function takes in the input with input_dtype -and casts it to the output_dtype as the output.
-
-concat¶
++device¶
++-get_places¶
-
-
paddle.v2.fluid.layers.concat(input, axis=0)
- Concat
-This function concatenates the input along the axis mentioned -and returns that as the output.
+paddle.v2.fluid.layers.get_places(device_count=None, device_type=None) +Returns a list of places based on flags. The list will be used for parallel +execution.
-参数: -
-
- input (list) – List of tensors to be concatenated -
- axis (int) – Integer axis along which the tensors will be concatenated +
- device_count (INT) – device count +
- device_type (STRING) – device type
返回: Output variable of the concatenation
-返回类型: Variable
+返回: vector of Place
Examples
-+sums¶
++io¶
++-data¶
-
-
paddle.v2.fluid.layers.sums(input, out=None)
- This function performs the sum operation on the input and returns the -result as the output.
+paddle.v2.fluid.layers.data(name, shape, append_batch_size=True, dtype='float32', lod_level=0, type=VarType.LOD_TENSOR, stop_gradient=True) +Data Layer
+This function takes in the input and based on whether data has +to be returned back as a minibatch, it creates the global variable by using +the helper functions. The global variables can be accessed by all the +following operators in the graph.
+All the input variables of this function are passed in as local variables +to the LayerHelper constructor.
- 参数: input (Variable|list) – The input tensor that has the elements -that need to be summed up. +参数: -
+
- name (str) – The name/alias of the function +
- shape (list) – Tuple declaring the shape. +
- append_batch_size (bool) – Whether or not to append the data as a batch. +
- dtype (int|float) – The type of data : float32, float_16, int etc +
- type (VarType) – The output type. By default it is LOD_TENSOR. +
- lod_level (int) – The LoD Level. 0 means the input data is not a sequence. +
- main_program (Program) – Name of the main program that calls this +
- startup_program (Program) – Name of the startup program +
- stop_gradient (bool) – A boolean that mentions whether gradient should flow. +
返回: -
-
- The tensor type variable that has the sum of input -
- written to it. -
返回: The global variable that gives access to the data.
返回类型: Variable +返回类型: Variable
+Examples
++data = fluid.layers.data(name='x', shape=[784], dtype='float32') +
-linear_chain_crf¶
--
+
-
-
paddle.v2.fluid.layers.linear_chain_crf(input, label, param_attr=None)
- BlockGuardServ class.
+BlockGuardServ class is used to create an op with a block in a program.
+-
+class
paddle.v2.fluid.layers.ListenAndServ(endpoint, fan_in=1, optimizer_mode=True)
+ ListenAndServ class.
+ListenAndServ class is used to wrap listen_and_serv op to create a server +which can receive variables from clients and run a block.
+-
-
paddle.v2.fluid.layers.embedding(input, size, is_sparse=False, padding_idx=None, param_attr=None, dtype='float32')
- Embedding Layer
-This layer is used to lookup embeddings of IDs, provided by
-input, in -a lookup table. The result of this lookup is the embedding of each ID in the -input.All the input variables are passed in as local variables to the LayerHelper -constructor.
+paddle.v2.fluid.layers.Send(endpoints, send_vars, get_vars) +Send layer
- 参数:
+-BlockGuardServ¶
+paddle.v2.fluid.layers.BlockGuardServ(server) +-assign¶
+++ListenAndServ¶
+-
+
+Send¶
-
+class
+IfElse¶
+
initializer¶
++-Constant¶
+-ConstantInitializer¶
--
+
+-Uniform¶
+-UniformInitializer¶
--
+
+-Normal¶
+-@@ -236,26 +236,162 @@NormalInitializer¶
--
+
+-Xavier¶
+--XavierInitializer¶
--
-
-@@ -333,10 +282,10 @@ is sqrt(2/ fan_in). diff --git a/develop/doc_cn/api/v2/fluid/io.html b/develop/doc_cn/api/v2/fluid/io.html index a68b6e94253b7a618f3a4ed41f10640af3e7167f..0a0c51c9f540cbd755c36f33c3ac9caf023b0807 100644 --- a/develop/doc_cn/api/v2/fluid/io.html +++ b/develop/doc_cn/api/v2/fluid/io.html @@ -8,7 +8,7 @@ -MSRAInitializer¶
--
-
IO — PaddlePaddle 文档 +io — PaddlePaddle 文档 @@ -35,7 +35,7 @@ - + @@ -173,17 +173,17 @@@@ -236,127 +236,79 @@-diff --git a/develop/doc_cn/api/v2/fluid/layers.html b/develop/doc_cn/api/v2/fluid/layers.html index 7c937ccee1c20fe8bc9810621bcf517e05638a93..46fde35a24aa975f107c842cd269dcfb13b104f7 100644 --- a/develop/doc_cn/api/v2/fluid/layers.html +++ b/develop/doc_cn/api/v2/fluid/layers.html @@ -8,7 +8,7 @@ -IO¶
--@@ -269,7 +405,7 @@ Next - Previous + Previousis_parameter¶
+io¶
+++save_vars¶
++save_params¶
+-
+
++save_persistables¶
+-
+
++load_vars¶
+-
+
++load_params¶
+-
+
++load_persistables¶
+-
+
++save_inference_model¶
+-
+
++load_inference_model¶
+-
+
+get_inference_program¶
+-
+
Layers — PaddlePaddle 文档 +layers — PaddlePaddle 文档 @@ -34,7 +34,7 @@ - + @@ -173,17 +173,17 @@-Layers¶
--fc¶
+layers¶
++control_flow¶
++split_lod_tensor¶
-
+
+-Momentum¶
+-AdagradOptimizer¶
--
+
+-Adagrad¶
+-AdamOptimizer¶
--
+
+-Adam¶
+-AdamaxOptimizer¶
--
+
+-Adamax¶
+-@@ -216,10 +216,26 @@DecayedAdagradOptimizer¶
--
+
+@@ -336,10 +281,10 @@ their internal state. diff --git a/develop/doc/api/v2/fluid/param_attr.html b/develop/doc/api/v2/fluid/param_attr.html index 6ba47f50fbf23d61db970e83db64e512278d1f81..d3e178399dd7958e7842b3ff5f9ebef2e5d8850e 100644 --- a/develop/doc/api/v2/fluid/param_attr.html +++ b/develop/doc/api/v2/fluid/param_attr.html @@ -8,7 +8,7 @@ -DecayedAdagrad¶
+ParamAttr — PaddlePaddle documentation +param_attr — PaddlePaddle documentation @@ -34,8 +34,8 @@ - - + + @@ -162,17 +162,17 @@--ParamAttr¶
--@@ -217,9 +217,9 @@ParamAttr¶
++@@ -230,10 +246,10 @@ diff --git a/develop/doc/api/v2/fluid/profiler.html b/develop/doc/api/v2/fluid/profiler.html index cda12d3fe0c3ae20da61a5a910f61afd2d20a381..e2e4c0d1101370c9647d63ac982c14d2c8be93c5 100644 --- a/develop/doc/api/v2/fluid/profiler.html +++ b/develop/doc/api/v2/fluid/profiler.html @@ -8,7 +8,7 @@ -param_attr¶
+++ParamAttr¶
+-
+
+WeightNormParamAttr¶
+-
+
Profiler — PaddlePaddle documentation +profiler — PaddlePaddle documentation @@ -34,8 +34,8 @@ - - + + @@ -162,17 +162,17 @@-@@ -217,37 +217,55 @@Profiler¶
--@@ -259,10 +306,10 @@ to “Compute Command Line Profiler User Guide”. diff --git a/develop/doc/api/v2/fluid/regularizer.html b/develop/doc/api/v2/fluid/regularizer.html index b1c507121be0dcd44f5d2b6720316331db96b1ea..6ab82c163f46aeb685b95fd2dd168f02871021fc 100644 --- a/develop/doc/api/v2/fluid/regularizer.html +++ b/develop/doc/api/v2/fluid/regularizer.html @@ -8,7 +8,7 @@ -Profiler¶
+profiler¶
+++cuda_profiler¶
++reset_profiler¶
+-
+
+profiler¶
+-
+
Regularizer — PaddlePaddle documentation +regularizer — PaddlePaddle documentation @@ -34,8 +34,8 @@ - - + + @@ -162,17 +162,17 @@-Regularizer¶
--WeightDecayRegularizer¶
--
+
regularizer¶
++-append_regularization_ops¶
+-L2DecayRegularizer¶
--
+
+-L1Decay¶
+-L1DecayRegularizer¶
--
-
-
-
-
+
initializer¶
+
+
-Constant¶
+
-
ConstantInitializer¶
--
+
+
-Uniform¶
+
-
UniformInitializer¶
--
+
+
-Normal¶
+
-
@@ -217,26 +217,162 @@
NormalInitializer¶
--
+
+
-Xavier¶
+
-
-XavierInitializer¶
--
-
-
@@ -314,10 +263,10 @@ is sqrt(2/ fan_in).
diff --git a/develop/doc/api/v2/fluid/io.html b/develop/doc/api/v2/fluid/io.html
index 35c0fe2639bd14afebf1cac317cf3910ee8c4d71..a133eeccf8a1b37b98d23577beea20063bace8fd 100644
--- a/develop/doc/api/v2/fluid/io.html
+++ b/develop/doc/api/v2/fluid/io.html
@@ -8,7 +8,7 @@
- MSRAInitializer¶
--
-
-
diff --git a/develop/doc/api/v2/fluid/layers.html b/develop/doc/api/v2/fluid/layers.html
index eb2cdeea51cdf393f6a4917529702b89c00f3c51..6e8d381e5b7ed0f3671258e7a6eacf208a3719a7 100644
--- a/develop/doc/api/v2/fluid/layers.html
+++ b/develop/doc/api/v2/fluid/layers.html
@@ -8,7 +8,7 @@
- IO¶
-
-
@@ -250,7 +386,7 @@
Next
- Previous
+ Previous
is_parameter¶
+io¶
+
+
+save_vars¶
+
+save_params¶
+-
+
+
+save_persistables¶
+-
+
+
+load_vars¶
+-
+
+
+load_params¶
+-
+
+
+load_persistables¶
+-
+
+
+save_inference_model¶
+-
+
+
+load_inference_model¶
+-
+
+
get_inference_program¶
+-
+
-
Layers¶
-
-
fc¶
+layers¶
+
+
control_flow¶
+
+