提交 29ddf6c5 编写于 作者: F fengjiayi

Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into dev_add_doc

...@@ -64,7 +64,8 @@ class OpConverter { ...@@ -64,7 +64,8 @@ class OpConverter {
(*it)(op, scope, test_mode); (*it)(op, scope, test_mode);
} }
// convert fluid block to tensorrt network // Convert a fluid block to tensorrt network, NOTE it just convert operators,
// the INetwork's inputs and outputs should specified in some other modules.
void ConvertBlock(const framework::proto::BlockDesc& block, void ConvertBlock(const framework::proto::BlockDesc& block,
const std::unordered_set<std::string>& parameters, const std::unordered_set<std::string>& parameters,
const framework::Scope& scope, TensorRTEngine* engine) { const framework::Scope& scope, TensorRTEngine* engine) {
......
...@@ -51,11 +51,12 @@ class TensorRTEngine : public EngineBase { ...@@ -51,11 +51,12 @@ class TensorRTEngine : public EngineBase {
nvinfer1::Weights w_; nvinfer1::Weights w_;
}; };
TensorRTEngine(int max_batch, int max_workspace, cudaStream_t* stream, TensorRTEngine(int max_batch, int max_workspace,
cudaStream_t* stream = nullptr,
nvinfer1::ILogger& logger = NaiveLogger::Global()) nvinfer1::ILogger& logger = NaiveLogger::Global())
: max_batch_(max_batch), : max_batch_(max_batch),
max_workspace_(max_workspace), max_workspace_(max_workspace),
stream_(stream), stream_(stream ? stream : &default_stream_),
logger_(logger) {} logger_(logger) {}
virtual ~TensorRTEngine(); virtual ~TensorRTEngine();
...@@ -121,6 +122,8 @@ class TensorRTEngine : public EngineBase { ...@@ -121,6 +122,8 @@ class TensorRTEngine : public EngineBase {
// the max memory size the engine uses // the max memory size the engine uses
int max_workspace_; int max_workspace_;
cudaStream_t* stream_; cudaStream_t* stream_;
// If stream_ is not set from outside, hold its own stream.
cudaStream_t default_stream_;
nvinfer1::ILogger& logger_; nvinfer1::ILogger& logger_;
std::vector<Buffer> buffers_; std::vector<Buffer> buffers_;
...@@ -165,20 +168,31 @@ class TensorRTEngine : public EngineBase { ...@@ -165,20 +168,31 @@ class TensorRTEngine : public EngineBase {
*/ */
class TRT_EngineManager { class TRT_EngineManager {
public: public:
TensorRTEngine* Create(int max_batch, int max_workspace, bool HasEngine(const std::string& name) const {
cudaStream_t* stream) { return engines_.count(name) != 0;
engines_.emplace_back(new TensorRTEngine(max_batch, max_workspace, stream)); }
return engines_.back().get();
// Get an engine called `name`.
TensorRTEngine* Get(const std::string& name) const {
return engines_.at(name).get();
}
// Create or get an engine called `name`
TensorRTEngine* Create(int max_batch, int max_workspace, cudaStream_t* stream,
const std::string& name) {
auto* p = new TensorRTEngine(max_batch, max_workspace, stream);
engines_[name].reset(p);
return p;
} }
void DeleteALl() { void DeleteALl() {
for (auto& ptr : engines_) { for (auto& item : engines_) {
ptr.reset(nullptr); item.second.reset(nullptr);
} }
} }
private: private:
std::vector<std::unique_ptr<TensorRTEngine>> engines_; std::unordered_map<std::string, std::unique_ptr<TensorRTEngine>> engines_;
}; };
} // namespace tensorrt } // namespace tensorrt
......
...@@ -252,15 +252,14 @@ class SoftShrinkOpMaker : public framework::OpProtoAndCheckerMaker { ...@@ -252,15 +252,14 @@ class SoftShrinkOpMaker : public framework::OpProtoAndCheckerMaker {
AddOutput("Out", "Output of Softshrink operator"); AddOutput("Out", "Output of Softshrink operator");
AddAttr<float>("lambda", "non-negative offset").SetDefault(0.5f); AddAttr<float>("lambda", "non-negative offset").SetDefault(0.5f);
AddComment(R"DOC( AddComment(R"DOC(
Softshrink Activation Operator. :strong:`Softshrink Activation Operator`
$$ .. math::
out = \begin{cases} out = \begin{cases}
x - \lambda, \text{if } x > \lambda \\ x - \lambda, \text{if } x > \lambda \\
x + \lambda, \text{if } x < -\lambda \\ x + \lambda, \text{if } x < -\lambda \\
0, \text{otherwise} 0, \text{otherwise}
\end{cases} \end{cases}
$$
)DOC"); )DOC");
} }
...@@ -271,18 +270,18 @@ class HardShrinkOpMaker : public framework::OpProtoAndCheckerMaker { ...@@ -271,18 +270,18 @@ class HardShrinkOpMaker : public framework::OpProtoAndCheckerMaker {
void Make() override { void Make() override {
AddInput("X", "Input of HardShrink operator"); AddInput("X", "Input of HardShrink operator");
AddOutput("Out", "Output of HardShrink operator"); AddOutput("Out", "Output of HardShrink operator");
AddAttr<float>("threshold", "The value of threshold for HardShrink") AddAttr<float>("threshold",
"The value of threshold for HardShrink. [default: 0.5]")
.SetDefault(0.5f); .SetDefault(0.5f);
AddComment(R"DOC( AddComment(R"DOC(
HardShrink Activation Operator. :strong:`HardShrink activation operator`
$$ .. math::
out = \begin{cases} out = \begin{cases}
x, \text{if } x > \lambda \\ x, \text{if } x > \lambda \\
x, \text{if } x < -\lambda \\ x, \text{if } x < -\lambda \\
0, \text{otherwise} 0, \text{otherwise}
\end{cases} \end{cases}
$$
)DOC"); )DOC");
} }
...@@ -394,18 +393,18 @@ class ThresholdedReluOpMaker : public framework::OpProtoAndCheckerMaker { ...@@ -394,18 +393,18 @@ class ThresholdedReluOpMaker : public framework::OpProtoAndCheckerMaker {
void Make() override { void Make() override {
AddInput("X", "Input of ThresholdedRelu operator"); AddInput("X", "Input of ThresholdedRelu operator");
AddOutput("Out", "Output of ThresholdedRelu operator"); AddOutput("Out", "Output of ThresholdedRelu operator");
AddAttr<float>("threshold", "The threshold location of activation") AddAttr<float>("threshold",
"The threshold location of activation. [default 1.0].")
.SetDefault(1.0f); .SetDefault(1.0f);
AddComment(R"DOC( AddComment(R"DOC(
ThresholdedRelu Activation Operator. :strong:`ThresholdedRelu activation operator`
$$ .. math::
out = \begin{cases}
x, \text{if } x > threshold \\
0, \text{otherwise}
\end{cases}
$$
out = \begin{cases}
x, \text{if } x > threshold \\
0, \text{otherwise}
\end{cases}
)DOC"); )DOC");
} }
}; };
......
...@@ -23,30 +23,26 @@ class CompareOpProtoMaker : public framework::OpProtoAndCheckerMaker { ...@@ -23,30 +23,26 @@ class CompareOpProtoMaker : public framework::OpProtoAndCheckerMaker {
public: public:
void Make() override { void Make() override {
OpComment comment; OpComment comment;
AddInput("X", AddInput("X", string::Sprintf("the left hand operand of %s operator",
string::Sprintf("(LoDTensor) the left hand operand of %s operator", comment.type));
comment.type)); AddInput("Y", string::Sprintf("the right hand operand of %s operator",
AddInput("Y", string::Sprintf( comment.type));
"(LoDTensor) the right hand operand of %s operator",
comment.type));
AddAttr<bool>("force_cpu", AddAttr<bool>("force_cpu",
"(bool, default false) Force fill output variable to cpu " "Force fill output variable to cpu "
"memory. Otherwise, fill output variable to the running " "memory. Otherwise, fill output variable to the running "
"device") "device [default true].")
.SetDefault(false); .SetDefault(true);
AddOutput("Out", string::Sprintf( AddOutput("Out", string::Sprintf("n-dim bool tensor. Each element is %s",
"(LoDTensor) n-dim bool tensor. Each element is %s", comment.equation));
comment.equation)); AddComment(string::Sprintf(R"DOC(
AddComment(string::Sprintf(R"DOC(%s Operator
It operates element-wise on X and Y, and returns the Out. Each of them is a It operates element-wise on X and Y, and returns the Out. Each of them is a
N-dim tensor. X and Y could be any type. The each element of the Out tensor is N-dim tensor. X and Y could be any type. The each element of the Out tensor is
calculated by %s calculated by $%s$
)DOC", )DOC",
comment.type, comment.equation)); comment.equation));
AddAttr<int>("axis", AddAttr<int>(
"(int, default -1). The start dimension index " "axis",
"for broadcasting Y onto X.") "The start dimension index for broadcasting Y onto X. [default -1]")
.SetDefault(-1) .SetDefault(-1)
.EqualGreaterThan(-1); .EqualGreaterThan(-1);
} }
......
...@@ -30,19 +30,19 @@ class CumOp : public framework::OperatorWithKernel { ...@@ -30,19 +30,19 @@ class CumOp : public framework::OperatorWithKernel {
class CumsumOpMaker : public framework::OpProtoAndCheckerMaker { class CumsumOpMaker : public framework::OpProtoAndCheckerMaker {
public: public:
void Make() override { void Make() override {
AddInput("X", "Input of Cumsum operator"); AddInput("X", "Input of cumsum operator");
AddOutput("Out", "Output of Cumsum operator"); AddOutput("Out", "Output of cumsum operator");
AddAttr<int>("axis", AddAttr<int>("axis",
"(int, default -1). The dimenstion to accumulate along. " "The dimenstion to accumulate along. -1 means the last "
"-1 means the last dimenstion") "dimenstion [default -1].")
.SetDefault(-1) .SetDefault(-1)
.EqualGreaterThan(-1); .EqualGreaterThan(-1);
AddAttr<bool>("exclusive", AddAttr<bool>("exclusive",
"bool, default false). Whether to perform exclusive cumsum") "Whether to perform exclusive cumsum. [default false].")
.SetDefault(false); .SetDefault(false);
AddAttr<bool>("reverse", AddAttr<bool>("reverse",
"bool, default false). If true, the cumsum is performed in " "If true, the cumsum is performed in the reversed direction. "
"the reversed direction") "[default false].")
.SetDefault(false); .SetDefault(false);
AddComment(R"DOC( AddComment(R"DOC(
The cumulative sum of the elements along a given axis. The cumulative sum of the elements along a given axis.
......
...@@ -62,36 +62,33 @@ class LayerNormOp : public framework::OperatorWithKernel { ...@@ -62,36 +62,33 @@ class LayerNormOp : public framework::OperatorWithKernel {
class LayerNormOpMaker : public framework::OpProtoAndCheckerMaker { class LayerNormOpMaker : public framework::OpProtoAndCheckerMaker {
public: public:
void Make() override { void Make() override {
AddInput("X", "(LoDTensor) The input tensor."); AddInput("X", "The input tensor.");
AddInput("Scale", AddInput("Scale",
"(Tensor, optional) Scale is a 1-dimensional tensor of size " "(optional) Scale is a 1-dimensional tensor of size "
"H(`begin_norm_axis` splits the tensor(`X`) to a matrix [N,H])." "H(`begin_norm_axis` splits the tensor(`X`) to a matrix [N,H])."
"It is applied to the output.") "It is applied to the output.")
.AsDispensable(); .AsDispensable();
AddInput("Bias", AddInput("Bias",
"(Tensor, optional) Bias is a 1-dimensional tensor of size " "(optional) Bias is a 1-dimensional tensor of size "
"H(`begin_norm_axis` splits the tensor(`X`) to a matrix [N,H])." "H(`begin_norm_axis` splits the tensor(`X`) to a matrix [N,H])."
"It is applied to the output.") "It is applied to the output.")
.AsDispensable(); .AsDispensable();
AddOutput("Y", "(LoDTensor) Result after normalization."); AddOutput("Y", "Result after normalization.");
AddOutput("Mean", "(Tensor) Mean of the current mini batch.") AddOutput("Mean", "Mean of the current mini batch.").AsIntermediate();
.AsIntermediate(); AddOutput("Variance", "Variance of the current mini batch.")
AddOutput("Variance", "(Tensor) Variance of the current mini batch.")
.AsIntermediate(); .AsIntermediate();
AddAttr<float>("epsilon", AddAttr<float>("epsilon",
"(float, default 1e-5) Constant for " "Constant for numerical stability [default 1e-5].")
"numerical stability")
.SetDefault(1e-5) .SetDefault(1e-5)
.AddCustomChecker([](const float &epsilon) { .AddCustomChecker([](const float &epsilon) {
PADDLE_ENFORCE(epsilon >= 0.0f && epsilon <= 0.001f, PADDLE_ENFORCE(epsilon >= 0.0f && epsilon <= 0.001f,
"'epsilon' should be between 0.0 and 0.001."); "'epsilon' should be between 0.0 and 0.001.");
}); });
AddAttr<int>("begin_norm_axis", AddAttr<int>("begin_norm_axis",
"(int default:1), the " "the axis of `begin_norm_axis ... Rank(X) - 1` will be "
"axis of `begin_norm_axis ... Rank(X) - 1` will be "
"normalized. `begin_norm_axis` splits the tensor(`X`) to a " "normalized. `begin_norm_axis` splits the tensor(`X`) to a "
"matrix [N,H].") "matrix [N,H]. [default 1].")
.SetDefault(1) .SetDefault(1)
.AddCustomChecker([](const int &begin_norm_axis) { .AddCustomChecker([](const int &begin_norm_axis) {
PADDLE_ENFORCE_GT(begin_norm_axis, 0, PADDLE_ENFORCE_GT(begin_norm_axis, 0,
...@@ -99,10 +96,14 @@ class LayerNormOpMaker : public framework::OpProtoAndCheckerMaker { ...@@ -99,10 +96,14 @@ class LayerNormOpMaker : public framework::OpProtoAndCheckerMaker {
}); });
AddComment(R"DOC( AddComment(R"DOC(
Layer Normalization. Assume feature vectors exist on dimensions
Layer Norm has been implemented as discussed in the paper: :attr:`begin_norm_axis ... rank(input)` and calculate the moment statistics
https://arxiv.org/abs/1607.06450 along these dimensions for each feature vector :math:`a` with size
... :math:`H`, then normalize each feature vector using the corresponding
statistics. After that, apply learnable gain and bias on the normalized
tensor to scale and shift if :attr:`scale` and :attr:`shift` are set.
Refer to `Layer Normalization <https://arxiv.org/pdf/1607.06450v1.pdf>`_
)DOC"); )DOC");
} }
}; };
......
...@@ -348,7 +348,8 @@ class ListenAndServOpMaker : public framework::OpProtoAndCheckerMaker { ...@@ -348,7 +348,8 @@ class ListenAndServOpMaker : public framework::OpProtoAndCheckerMaker {
}; };
void SignalHandler::StopAndExit(int signal_num) { void SignalHandler::StopAndExit(int signal_num) {
VLOG(3) << "Catch interrupt signal: " << signal_num << ", program will exit"; // Do not use VLOG here for the device for printing maybe already released.
// exit will release interal allocated resoureces.
exit(0); exit(0);
} }
......
...@@ -33,12 +33,10 @@ class MeanOp : public framework::OperatorWithKernel { ...@@ -33,12 +33,10 @@ class MeanOp : public framework::OperatorWithKernel {
class MeanOpMaker : public framework::OpProtoAndCheckerMaker { class MeanOpMaker : public framework::OpProtoAndCheckerMaker {
public: public:
void Make() override { void Make() override {
AddInput("X", "The input of mean op"); AddInput("X", "(Tensor) The input of mean op");
AddOutput("Out", "The output of mean op").Reuse("X"); AddOutput("Out", "(Tensor) The output of mean op").Reuse("X");
AddComment(R"DOC( AddComment(R"DOC(
Mean Operator. Mean Operator calculates the mean of all elements in X.
Out is a scalar which is the mean of all elements in X.
)DOC"); )DOC");
} }
......
...@@ -62,26 +62,46 @@ class MultiplexOp : public framework::OperatorWithKernel { ...@@ -62,26 +62,46 @@ class MultiplexOp : public framework::OperatorWithKernel {
class MultiplexOpMaker : public framework::OpProtoAndCheckerMaker { class MultiplexOpMaker : public framework::OpProtoAndCheckerMaker {
public: public:
void Make() override { void Make() override {
AddInput("Ids", "The index tensor of multiplex operator."); AddInput("Ids",
AddInput("X", "The candidate tensors of multiplex operator.") "Tensor<int32>, index variable which is a 2-D tensor with shape "
"[M, 1] where M is the batch size.");
AddInput("X",
"A list of variables to gather from. All variables have the same "
"shape and the rank is at least 2.")
.AsDuplicable(); .AsDuplicable();
AddOutput("Out", "The output tensor of multiplex operator."); AddOutput("Out", "The output tensor of multiplex operator.");
AddComment(R"DOC( AddComment(R"DOC(
Multiplex Operator. Referring to the given index variable, this layer selects rows from the
input variables to construct a multiplex variable. Assuming that there are
Multiplex multiple tensors according to the index provided by the index tensor. :math:`m` input variables and :math:`I_i` represents the i-th input
variable and :math:`i` is in [0, :math:`m`). All input variables are
Ids: the index tensor. tensors with same shape [:math:`d_0`, :math:`d_1`, ..., :math:`d_R`].
X[0 : N - 1]: the candidate tensors for output (N >= 2). Please note that rank of the input tensor should be at least 2. Each input
For each index i from 0 to batchSize - 1, the output is the i-th row of the variable will be treated as a 2-D matrix with shape [:math:`M`, :math:`N`]
where :math:`M` for :math:`d_0` and :math:`N` for :math:`d_1` * :math:`d_2`
* ... * :math:`d_R`. Let :math:`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
[:math:`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 [:math:`d_0`,
:math:`d_1`, ..., :math:`d_R`]. If we treat the output tensor as a 2-D
matrix with shape [:math:`M`, :math:`N`] and let :math:`O[i]` be the i-th
row of the matrix, then `O[i]` is equal to :math:`I_{ID[i]}[i]`.
* Ids: the index tensor.
* X[0 : N - 1]: the candidate tensors for output (N >= 2).
* For each index i from 0 to batchSize - 1, the output is the i-th row of the
the (Ids[i])-th tensor. the (Ids[i])-th tensor.
For i-th row of the output tensor: For i-th row of the output tensor:
$$y[i] = x_{k}[i]$$ $$
y[i] = x_{k}[i]
$$
where `y` is the output tensor, `x_{k}` is the k-th input tensor, where $y$ is the output tensor, $x_{k}$ is the k-th input tensor,
and `k = Ids[i]`. and $k = Ids[i]$.
)DOC"); )DOC");
} }
......
...@@ -78,11 +78,15 @@ class CreateRecordIOReaderOp : public framework::OperatorBase { ...@@ -78,11 +78,15 @@ class CreateRecordIOReaderOp : public framework::OperatorBase {
class CreateRecordIOReaderOpMaker : public FileReaderMakerBase { class CreateRecordIOReaderOpMaker : public FileReaderMakerBase {
protected: protected:
void Apply() override { void Apply() override {
AddAttr<std::string>("filename", "The filename of record io reader"); AddAttr<std::string>(
"filename",
"The filename of record file. This file will given to reader.");
AddComment(R"DOC( AddComment(R"DOC(
CreateRecordIOReader Operator Open a recordio file and return the reader object. The returned reader object
is thread-safe.
Create a reader from a record io file NOTE: This is a very low-level API. It is used for debugging data file or
training. Please use `open_files` instead of this API for production usage.
)DOC"); )DOC");
} }
}; };
......
...@@ -54,7 +54,7 @@ std::unique_ptr<framework::ReaderBase> CreateReaderByFileName( ...@@ -54,7 +54,7 @@ std::unique_ptr<framework::ReaderBase> CreateReaderByFileName(
} }
void FileReaderMakerBase::Make() { void FileReaderMakerBase::Make() {
AddOutput("Out", "(ReaderHolder) The created random reader.").AsDuplicable(); AddOutput("Out", "(ReaderHolder): The created random reader.").AsDuplicable();
AddAttr<std::vector<int>>("shape_concat", "The concat of all data's shapes."); AddAttr<std::vector<int>>("shape_concat", "The concat of all data's shapes.");
AddAttr<std::vector<int>>( AddAttr<std::vector<int>>(
"ranks", "ranks",
......
...@@ -78,23 +78,23 @@ class RowConvOpMaker : public framework::OpProtoAndCheckerMaker { ...@@ -78,23 +78,23 @@ class RowConvOpMaker : public framework::OpProtoAndCheckerMaker {
public: public:
void Make() override { void Make() override {
AddInput("X", AddInput("X",
"(LoDTensor), the input(X) is a LodTensor, which supports " "the input(X) is a LodTensor, which supports "
"variable time-length input sequences. The underlying tensor " "variable time-length input sequences. The underlying tensor "
"in this LoDTensor is a matrix with shape (T x N), where T " "in this LoDTensor is a matrix with shape (T x N), where T "
"is the total time steps in this mini-batch and N is the input " "is the total time steps in this mini-batch and N is the input "
"data dimension."); "data dimension.");
AddInput("Filter", AddInput("Filter",
"(Tensor), the input(Filter) is a learnable parameter. It " "the input(Filter) is a learnable parameter. It "
"is a 2-D tensor with shape (future_context x N), where, " "is a 2-D tensor with shape (future_context x N), where, "
"future_context is the future context length and N is the data " "future_context is the future context length and N is the data "
"dimension."); "dimension.");
AddOutput("Out", AddOutput("Out",
"(LoDTensor), the output(Out) is a LodTensor, which supports " "the output(Out) is a LodTensor, which supports "
"variable time-length input sequences. The underlying tensor " "variable time-length input sequences. The underlying tensor "
"in this LodTensor is a matrix with shape T x N, i.e., the " "in this LodTensor is a matrix with shape T x N, i.e., the "
"same shape as X."); "same shape as X.");
AddComment(R"DOC( AddComment(R"DOC(
Row-convolution Operator. :strong:`Row-convolution operator`
The row convolution is called lookahead convolution. This operator was The row convolution is called lookahead convolution. This operator was
introduced in the following paper for DeepSpeech2: introduced in the following paper for DeepSpeech2:
...@@ -114,9 +114,23 @@ and a filter ($W$) of size $context \times d$, ...@@ -114,9 +114,23 @@ and a filter ($W$) of size $context \times d$,
the output sequence is convolved as: the output sequence is convolved as:
$$ $$
out_{i, :} = \sum_{j=i}^{i + context} in_{j,:} \dot W_{i-j, :} out_{i, :} = \\sum_{j=i}^{i + context} in_{j,:} \\cdot 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 design document
https://github.com/PaddlePaddle/Paddle/issues/2228#issuecomment-303903645 .
)DOC"); )DOC");
} }
}; };
......
...@@ -66,17 +66,25 @@ nvinfer1::Dims Vec2TRT_Dims(const std::vector<int64_t> &shape) { ...@@ -66,17 +66,25 @@ nvinfer1::Dims Vec2TRT_Dims(const std::vector<int64_t> &shape) {
} // namespace } // namespace
template <typename DeviceContext, typename T> template <typename DeviceContext, typename T>
void paddle::operators::TensorRTEngineKernel<DeviceContext, T>::Prepare( void TensorRTEngineKernel<DeviceContext, T>::Prepare(
const framework::ExecutionContext &context) const { const framework::ExecutionContext &context) const {
VLOG(4) << "Prepare engine"; VLOG(4) << "Prepare engine";
// Get the ProgramDesc and pass to convert. // Get the ProgramDesc and pass to convert.
framework::proto::BlockDesc block_desc; framework::proto::BlockDesc block_desc;
block_desc.ParseFromString(context.Attr<std::string>("subgraph")); block_desc.ParseFromString(context.Attr<std::string>("subgraph"));
max_batch_ = context.Attr<int>("max_batch"); int max_batch = context.Attr<int>("max_batch");
auto max_workspace = context.Attr<int>("max_workspace"); auto max_workspace = context.Attr<int>("max_workspace");
engine_ = Singleton<TRT_EngineManager>::Global().Create( auto params = context.Attr<std::vector<std::string>>("parameters");
max_batch_, max_workspace, &stream_); std::unordered_set<std::string> parameters;
engine_->InitNetwork(); for (const auto &param : params) {
parameters.insert(param);
}
// TODO(Superjomn) replace this with a different stream
auto *engine = Singleton<TRT_EngineManager>::Global().Create(
max_batch, max_workspace, nullptr /*engine hold its own stream*/,
context.Attr<std::string>("engine_uniq_key"));
engine->InitNetwork();
framework::BlockDesc block(nullptr /*programdesc*/, &block_desc); framework::BlockDesc block(nullptr /*programdesc*/, &block_desc);
// Add inputs // Add inputs
...@@ -87,24 +95,23 @@ void paddle::operators::TensorRTEngineKernel<DeviceContext, T>::Prepare( ...@@ -87,24 +95,23 @@ void paddle::operators::TensorRTEngineKernel<DeviceContext, T>::Prepare(
PADDLE_ENFORCE_EQ(var->GetType(), FluidDT::VarType_Type_LOD_TENSOR, PADDLE_ENFORCE_EQ(var->GetType(), FluidDT::VarType_Type_LOD_TENSOR,
"TensorRT engine only takes LoDTensor as input"); "TensorRT engine only takes LoDTensor as input");
auto shape = var->GetShape(); auto shape = var->GetShape();
engine_->DeclareInput( engine->DeclareInput(
input, FluidDataType2TRT( input, FluidDataType2TRT(
var->Proto()->type().lod_tensor().tensor().data_type()), var->Proto()->type().lod_tensor().tensor().data_type()),
Vec2TRT_Dims(var->GetShape())); Vec2TRT_Dims(var->GetShape()));
} }
// TODO(Superjomn) parameters should be passed after analysised from outside.
inference::Singleton<inference::tensorrt::OpConverter>::Global().ConvertBlock( inference::Singleton<inference::tensorrt::OpConverter>::Global().ConvertBlock(
block_desc, {}, context.scope(), engine_); block_desc, parameters, context.scope(), engine);
// Add outputs // Add outputs
VLOG(4) << "declare outputs"; VLOG(4) << "declare outputs";
for (auto &output : context.Outputs("Ys")) { for (auto &output : context.Outputs("Ys")) {
VLOG(4) << "declare output " << output; VLOG(4) << "declare output " << output;
engine_->DeclareOutput(output); engine->DeclareOutput(output);
} }
engine_->FreezeNetwork(); engine->FreezeNetwork();
} }
class TensorRTEngineOpMaker : public framework::OpProtoAndCheckerMaker { class TensorRTEngineOpMaker : public framework::OpProtoAndCheckerMaker {
...@@ -113,6 +120,7 @@ class TensorRTEngineOpMaker : public framework::OpProtoAndCheckerMaker { ...@@ -113,6 +120,7 @@ class TensorRTEngineOpMaker : public framework::OpProtoAndCheckerMaker {
AddInput("Xs", "A list of inputs.").AsDuplicable(); AddInput("Xs", "A list of inputs.").AsDuplicable();
AddOutput("Ys", "A list of outputs").AsDuplicable(); AddOutput("Ys", "A list of outputs").AsDuplicable();
AddAttr<std::string>("subgraph", "the subgraph."); AddAttr<std::string>("subgraph", "the subgraph.");
AddAttr<std::string>("engine_uniq_key", "unique key for the TRT engine.");
AddAttr<int>("max_batch", "the maximum batch size."); AddAttr<int>("max_batch", "the maximum batch size.");
AddAttr<int>("max_workspace", "the maximum batch size."); AddAttr<int>("max_workspace", "the maximum batch size.");
AddComment("TensorRT engine operator."); AddComment("TensorRT engine operator.");
......
...@@ -19,10 +19,14 @@ ...@@ -19,10 +19,14 @@
#include "paddle/fluid/framework/operator.h" #include "paddle/fluid/framework/operator.h"
#include "paddle/fluid/inference/analysis/helper.h" #include "paddle/fluid/inference/analysis/helper.h"
#include "paddle/fluid/inference/tensorrt/engine.h" #include "paddle/fluid/inference/tensorrt/engine.h"
#include "paddle/fluid/inference/tensorrt/engine.h"
namespace paddle { namespace paddle {
namespace operators { namespace operators {
using inference::Singleton;
using inference::tensorrt::TRT_EngineManager;
class TensorRTEngineOp : public framework::OperatorWithKernel { class TensorRTEngineOp : public framework::OperatorWithKernel {
public: public:
using framework::OperatorWithKernel::OperatorWithKernel; using framework::OperatorWithKernel::OperatorWithKernel;
...@@ -47,16 +51,18 @@ template <typename DeviceContext, typename T> ...@@ -47,16 +51,18 @@ template <typename DeviceContext, typename T>
class TensorRTEngineKernel : public framework::OpKernel<T> { class TensorRTEngineKernel : public framework::OpKernel<T> {
public: public:
void Compute(const framework::ExecutionContext& context) const override { void Compute(const framework::ExecutionContext& context) const override {
if (!engine_) { auto engine_name = context.Attr<std::string>("engine_uniq_key");
if (!Singleton<TRT_EngineManager>::Global().HasEngine(engine_name)) {
Prepare(context); Prepare(context);
} }
auto* engine = Singleton<TRT_EngineManager>::Global().Get(engine_name);
auto input_names = context.op().Inputs("Xs"); auto input_names = context.op().Inputs("Xs");
PADDLE_ENFORCE(!input_names.empty(), "should pass more than one inputs"); PADDLE_ENFORCE(!input_names.empty(), "should pass more than one inputs");
// Try to determine a batch_size // Try to determine a batch_size
auto& tensor0 = inference::analysis::GetFromScope<framework::LoDTensor>( auto& tensor0 = inference::analysis::GetFromScope<framework::LoDTensor>(
context.scope(), input_names.front()); context.scope(), input_names.front());
int batch_size = tensor0.dims()[0]; int batch_size = tensor0.dims()[0];
PADDLE_ENFORCE_LE(batch_size, max_batch_); PADDLE_ENFORCE_LE(batch_size, context.Attr<int>("max_batch"));
// Convert input tensor from fluid to engine. // Convert input tensor from fluid to engine.
for (const auto& x : context.Inputs("Xs")) { for (const auto& x : context.Inputs("Xs")) {
...@@ -64,20 +70,20 @@ class TensorRTEngineKernel : public framework::OpKernel<T> { ...@@ -64,20 +70,20 @@ class TensorRTEngineKernel : public framework::OpKernel<T> {
auto& t = inference::analysis::GetFromScope<framework::LoDTensor>( auto& t = inference::analysis::GetFromScope<framework::LoDTensor>(
context.scope(), x); context.scope(), x);
if (platform::is_cpu_place(t.place())) { if (platform::is_cpu_place(t.place())) {
engine_->SetInputFromCPU(x, static_cast<const void*>(t.data<void>()), engine->SetInputFromCPU(x, static_cast<const void*>(t.data<void>()),
t.memory_size()); t.memory_size());
} else { } else {
engine_->SetInputFromGPU(x, static_cast<const void*>(t.data<void>()), engine->SetInputFromGPU(x, static_cast<const void*>(t.data<void>()),
t.memory_size()); t.memory_size());
} }
} }
// Execute the engine. // Execute the engine.
PADDLE_ENFORCE_GT(batch_size, 0); PADDLE_ENFORCE_GT(batch_size, 0);
engine_->Execute(batch_size); engine->Execute(batch_size);
// Convert output tensor from engine to fluid // Convert output tensor from engine to fluid
for (const auto& y : context.Outputs("Ys")) { for (const auto& y : context.Outputs("Ys")) {
// convert output and copy to fluid. // convert output and copy to fluid.
nvinfer1::ITensor* trt_t = engine_->GetITensor(y); nvinfer1::ITensor* trt_t = engine->GetITensor(y);
auto dims = trt_t->getDimensions(); auto dims = trt_t->getDimensions();
// Use the output ITensor's dims to reshape the Fluid Tensor. // Use the output ITensor's dims to reshape the Fluid Tensor.
std::vector<int> ddim(dims.d, dims.d + dims.nbDims); std::vector<int> ddim(dims.d, dims.d + dims.nbDims);
...@@ -89,27 +95,22 @@ class TensorRTEngineKernel : public framework::OpKernel<T> { ...@@ -89,27 +95,22 @@ class TensorRTEngineKernel : public framework::OpKernel<T> {
auto size = inference::analysis::AccuDims(dims.d, dims.nbDims); auto size = inference::analysis::AccuDims(dims.d, dims.nbDims);
if (platform::is_cpu_place(fluid_t->place())) { if (platform::is_cpu_place(fluid_t->place())) {
// TODO(Superjomn) change this float to dtype size. // TODO(Superjomn) change this float to dtype size.
engine_->GetOutputInCPU( engine->GetOutputInCPU(
y, fluid_t->mutable_data<float>(platform::CPUPlace()), y, fluid_t->mutable_data<float>(platform::CPUPlace()),
size * sizeof(float)); size * sizeof(float));
} else { } else {
engine_->GetOutputInGPU( engine->GetOutputInGPU(
y, fluid_t->mutable_data<float>(platform::CUDAPlace()), y, fluid_t->mutable_data<float>(platform::CUDAPlace()),
size * sizeof(float)); size * sizeof(float));
} }
} }
cudaStreamSynchronize(stream_); cudaStreamSynchronize(*engine->stream());
} }
protected: protected:
// Build the engine. // Build the engine.
void Prepare(const framework::ExecutionContext& context) const; void Prepare(const framework::ExecutionContext& context) const;
private:
mutable cudaStream_t stream_;
mutable inference::tensorrt::TensorRTEngine* engine_{nullptr};
mutable int max_batch_{0};
}; };
} // namespace operators } // namespace operators
......
...@@ -79,6 +79,17 @@ void SetAttr<int64_t>(framework::proto::OpDesc* op, const std::string& name, ...@@ -79,6 +79,17 @@ void SetAttr<int64_t>(framework::proto::OpDesc* op, const std::string& name,
attr->set_type(paddle::framework::proto::AttrType::LONG); attr->set_type(paddle::framework::proto::AttrType::LONG);
attr->set_l(data); attr->set_l(data);
} }
template <>
void SetAttr<std::vector<std::string>>(framework::proto::OpDesc* op,
const std::string& name,
const std::vector<std::string>& data) {
auto* attr = op->add_attrs();
attr->set_name(name);
attr->set_type(paddle::framework::proto::AttrType::STRINGS);
for (const auto& s : data) {
attr->add_strings(s.c_str());
}
}
} // namespace } // namespace
...@@ -123,11 +134,15 @@ TEST(TensorRTEngineOp, manual) { ...@@ -123,11 +134,15 @@ TEST(TensorRTEngineOp, manual) {
engine_op_desc.SetOutput("Ys", std::vector<std::string>({"z0"})); engine_op_desc.SetOutput("Ys", std::vector<std::string>({"z0"}));
SetAttr<std::string>(engine_op_desc.Proto(), "subgraph", SetAttr<std::string>(engine_op_desc.Proto(), "subgraph",
block_->SerializeAsString()); block_->SerializeAsString());
SetAttr<int>(engine_op_desc.Proto(), "max_batch", 30); SetAttr<int>(engine_op_desc.Proto(), "max_batch", 100);
SetAttr<int>(engine_op_desc.Proto(), "max_workspace", 1 << 10); SetAttr<int>(engine_op_desc.Proto(), "max_workspace", 1 << 10);
SetAttr<std::string>(engine_op_desc.Proto(), "engine_uniq_key", "a_engine");
SetAttr<std::vector<std::string>>(engine_op_desc.Proto(), "parameters",
std::vector<std::string>({}));
LOG(INFO) << "create engine op"; LOG(INFO) << "create engine op";
auto engine_op = framework::OpRegistry::CreateOp(*engine_op_desc.Proto()); auto engine_op = framework::OpRegistry::CreateOp(*engine_op_desc.Proto());
LOG(INFO) << "engine_op " << engine_op.get();
framework::Scope scope; framework::Scope scope;
platform::CPUPlace place; platform::CPUPlace place;
...@@ -145,6 +160,88 @@ TEST(TensorRTEngineOp, manual) { ...@@ -145,6 +160,88 @@ TEST(TensorRTEngineOp, manual) {
engine_op->Run(scope, place); engine_op->Run(scope, place);
} }
void Execute(int batch_size, int input_dim, int output_dim, int nlayers = 1) {
framework::ProgramDesc program;
framework::Scope scope;
platform::CPUPlace place;
platform::CPUDeviceContext ctx(place);
auto* block_ = program.Proto()->add_blocks();
block_->set_idx(0);
block_->set_parent_idx(-1);
using shape_t = std::vector<int64_t>;
LOG(INFO) << "create block desc";
framework::BlockDesc block_desc(&program, block_);
auto AddFCLayer = [&](const std::string& x_name, const std::string& y_name,
const std::string& z_name, bool x_created,
const shape_t& x_shape, const shape_t& y_shape,
const shape_t& z_shape) {
LOG(INFO) << "create fc op";
auto* fc = block_desc.AppendOp();
fc->SetType("mul");
fc->SetInput("X", std::vector<std::string>({x_name}));
fc->SetInput("Y", std::vector<std::string>({y_name}));
fc->SetOutput("Out", std::vector<std::string>({z_name}));
// Set inputs' variable shape in BlockDesc
if (!x_created) {
AddTensorToBlockDesc(block_, x_name,
std::vector<int64_t>({batch_size, input_dim, 1, 1}));
}
AddTensorToBlockDesc(block_, y_name,
std::vector<int64_t>({input_dim, output_dim}));
AddTensorToBlockDesc(block_, z_name,
std::vector<int64_t>({batch_size, output_dim}));
// Prepare variables.
if (!x_created) {
CreateCPUTensor(&scope, x_name, std::vector<int64_t>(x_shape));
}
CreateCPUTensor(&scope, y_name, std::vector<int64_t>(y_shape));
CreateCPUTensor(&scope, z_name, std::vector<int64_t>(z_shape));
// It is wired, need to copy manually.
*block_->add_ops() = *fc->Proto();
};
// Test with 4 layer FC
AddFCLayer("x0", "y0", "z0", false, {batch_size, input_dim},
{input_dim, output_dim}, {batch_size, output_dim});
AddFCLayer("z0", "y1", "z1", true, {}, {output_dim, output_dim},
{batch_size, output_dim});
AddFCLayer("z1", "y2", "z2", true, {}, {output_dim, output_dim},
{batch_size, output_dim});
AddFCLayer("z2", "y3", "z3", true, {}, {output_dim, output_dim},
{batch_size, output_dim});
LOG(INFO) << "create tensorrt desc";
framework::OpDesc engine_op_desc(nullptr);
engine_op_desc.SetType("tensorrt_engine");
engine_op_desc.SetInput("Xs", std::vector<std::string>({"x0"}));
engine_op_desc.SetOutput("Ys", std::vector<std::string>({"z3"}));
SetAttr<std::string>(engine_op_desc.Proto(), "subgraph",
block_->SerializeAsString());
SetAttr<int>(engine_op_desc.Proto(), "max_batch", batch_size);
SetAttr<int>(engine_op_desc.Proto(), "max_workspace", 2 << 10);
SetAttr<std::vector<std::string>>(
engine_op_desc.Proto(), "parameters",
std::vector<std::string>({"y0", "y1", "y2", "y3"}));
SetAttr<std::string>(engine_op_desc.Proto(), "engine_uniq_key", "b_engine");
auto engine_op = framework::OpRegistry::CreateOp(*engine_op_desc.Proto());
// Execute them.
engine_op->Run(scope, place);
}
// Test with a larger FC layer.
TEST(TensorRTEngineOp, fc) { Execute(40, 256, 256); }
} // namespace operators } // namespace operators
} // namespace paddle } // namespace paddle
......
...@@ -86,32 +86,24 @@ class UniformRandomOp : public framework::OperatorWithKernel { ...@@ -86,32 +86,24 @@ class UniformRandomOp : public framework::OperatorWithKernel {
class UniformRandomOpMaker : public framework::OpProtoAndCheckerMaker { class UniformRandomOpMaker : public framework::OpProtoAndCheckerMaker {
public: public:
void Make() override { void Make() override {
AddOutput("Out", "(Tensor) The output tensor of uniform random op"); AddOutput("Out", "The output tensor of uniform random op");
AddComment(R"DOC( AddComment(R"DOC(
Uniform random operator.
This operator initializes a tensor with random values sampled from a This operator initializes a tensor with random values sampled from a
uniform distribution. uniform distribution. The random result is in set [min, max].
)DOC"); )DOC");
AddAttr<std::vector<int>>("shape", AddAttr<std::vector<int>>("shape", "The shape of the output tensor");
"(vector<int>) The shape of the output tensor"); AddAttr<float>("min", "Minimum value of uniform random. [default -1.0].")
AddAttr<float>("min",
"(float, default -1.0) "
"Minimum value of uniform random")
.SetDefault(-1.0f); .SetDefault(-1.0f);
AddAttr<float>("max", AddAttr<float>("max", "Maximun value of uniform random. [default 1.0].")
"(float, default 1.0) "
"Maximun value of uniform random")
.SetDefault(1.0f); .SetDefault(1.0f);
AddAttr<int>("seed", AddAttr<int>("seed",
"(int, default 0) "
"Random seed used for generating samples. " "Random seed used for generating samples. "
"0 means use a seed generated by the system." "0 means use a seed generated by the system."
"Note that if seed is not 0, this operator will always " "Note that if seed is not 0, this operator will always "
"generate the same random numbers every time.") "generate the same random numbers every time. [default 0].")
.SetDefault(0); .SetDefault(0);
AddAttr<int>("dtype", "(int, default 5(FP32)) Output tensor data type") AddAttr<int>("dtype", "Output tensor data type. [default 5(FP32)].")
.SetDefault(framework::proto::VarType::FP32); .SetDefault(framework::proto::VarType::FP32);
} }
}; };
......
...@@ -20,6 +20,7 @@ from ..framework import Program, Variable, Operator ...@@ -20,6 +20,7 @@ from ..framework import Program, Variable, Operator
from ..layer_helper import LayerHelper, unique_name from ..layer_helper import LayerHelper, unique_name
from ..initializer import force_init_on_cpu from ..initializer import force_init_on_cpu
from ops import logical_and, logical_not, logical_or from ops import logical_and, logical_not, logical_or
import numpy
__all__ = [ __all__ = [
'split_lod_tensor', 'split_lod_tensor',
...@@ -706,7 +707,7 @@ def lod_rank_table(x, level=0): ...@@ -706,7 +707,7 @@ def lod_rank_table(x, level=0):
.. code-block:: python .. code-block:: python
x = fluid.layers.data(name='x', shape=[10], x = fluid.layers.data(name='x', shape=[10],
dtype='float32', lod_level=1) dtype='float32', lod_level=1)
out = layers.lod_rank_table(x=x, level=0) out = layers.lod_rank_table(x=x, level=0)
""" """
helper = LayerHelper("lod_rank_table", **locals()) helper = LayerHelper("lod_rank_table", **locals())
...@@ -917,37 +918,40 @@ def create_array(dtype): ...@@ -917,37 +918,40 @@ def create_array(dtype):
dtype=dtype) dtype=dtype)
def less_than(x, y, force_cpu=True, cond=None, **ignored): @templatedoc()
def less_than(x, y, force_cpu=None, cond=None, **ignored):
""" """
**Less than** ${comment}
This layer returns the truth value of :math:`x < y` elementwise. >>> import paddle.fluid as fluid
>>> less = fluid.layers.less_than(x=label, y=limit)
Args: Args:
x(Variable): First operand of *less_than* x(${x_type}): ${x_comment}.
y(Variable): Second operand of *less_than* y(${y_type}): ${y_comment}.
force_cpu(Bool|True): The output data will be on CPU if set true. force_cpu(${force_cpu_type}): ${force_cpu_comment}.
cond(Variable|None): Optional output variable to store the result of *less_than* cond(Variable|None): Optional output variable to store the result of *less_than*
Returns: Returns:
Variable: The tensor variable storing the output of *less_than*. ${out_comment}.
Examples:
.. code-block:: python
less = fluid.layers.less_than(x=label, y=limit)
""" """
helper = LayerHelper("less_than", **locals()) helper = LayerHelper("less_than", **locals())
if cond is None: if cond is None:
cond = helper.create_tmp_variable(dtype='bool') cond = helper.create_tmp_variable(dtype='bool')
cond.stop_gradient = True cond.stop_gradient = True
attrs = dict()
if force_cpu is not None:
attrs['force_cpu'] = force_cpu
elif force_init_on_cpu():
attrs['force_cpu'] = force_init_on_cpu()
helper.append_op( helper.append_op(
type='less_than', type='less_than',
inputs={'X': [x], inputs={'X': [x],
'Y': [y]}, 'Y': [y]},
outputs={'Out': [cond]}, outputs={'Out': [cond]},
attrs={'force_cpu': force_cpu or force_init_on_cpu()}) attrs=attrs)
return cond return cond
...@@ -1012,8 +1016,28 @@ def array_read(array, i): ...@@ -1012,8 +1016,28 @@ def array_read(array, i):
def shrink_memory(x, i, table): def shrink_memory(x, i, table):
""" """
This function creates an operator to shrink_rnn_memory using the RankTable This function creates an operator to shrink rnn memory using the RankTable
as mentioned in the input parameter. as mentioned in the input parameter.
NOTE: This API is very low-level API. It is used by DynamicRNN only.
Since the Dynamic RNN uses no-padding way to implement RNN. The sequence
will be sorted by order, and the length of valid memory will be shrink after
each time step.
Args:
x(Variable): The memory object in the previous time step.
i(Variable): The step count variable. A int scalar as LoDTensor.
table(Variable): The RNNRankTable object.
Returns:
the memory variable after shrink.
Examples:
Since this API is very low level API. The example is not provided.
Please reference the implementation of class DynamicRNN for detail
usage.
""" """
helper = LayerHelper('shrink_memory', **locals()) helper = LayerHelper('shrink_memory', **locals())
out = helper.create_tmp_variable(dtype=x.dtype) out = helper.create_tmp_variable(dtype=x.dtype)
...@@ -1354,6 +1378,38 @@ class IfElse(object): ...@@ -1354,6 +1378,38 @@ class IfElse(object):
class DynamicRNN(object): class DynamicRNN(object):
"""
The dynamic RNN can process a batch of sequence data. The length of each
sample sequence can be different. This API automatically process them in
batch.
The input lod must be set. Please reference `lod_tensor`
>>> import paddle.fluid as fluid
>>> data = fluid.layers.data(name='sentence', dtype='int64', lod_level=1)
>>> embedding = fluid.layers.embedding(input=data, size=[65535, 32],
>>> is_sparse=True)
>>>
>>> drnn = fluid.layers.DynamicRNN()
>>> with drnn.block():
>>> word = drnn.step_input(embedding)
>>> prev = drnn.memory(shape=[200])
>>> hidden = fluid.layers.fc(input=[word, prev], size=200, act='relu')
>>> drnn.update_memory(prev, hidden) # set prev to hidden
>>> drnn.output(hidden)
>>>
>>> # last is the last time step of rnn. It is the encoding result.
>>> last = fluid.layers.sequence_last_step(drnn())
The dynamic RNN will unfold sequence into timesteps. Users need to define
how to process each time step during the :code:`with` block.
The `memory` is used staging data cross time step. The initial value of
memory can be zero or another variable.
The dynamic RNN can mark multiple variables as its output. Use `drnn()` to
get the output sequence.
"""
BEFORE_RNN = 0 BEFORE_RNN = 0
IN_RNN = 1 IN_RNN = 1
AFTER_RNN = 2 AFTER_RNN = 2
...@@ -1376,6 +1432,15 @@ class DynamicRNN(object): ...@@ -1376,6 +1432,15 @@ class DynamicRNN(object):
self.mem_link = [] self.mem_link = []
def step_input(self, x): def step_input(self, x):
"""
Mark a sequence as a dynamic RNN input.
Args:
x(Variable): The input sequence.
Returns:
The current timestep in the input sequence.
"""
self._assert_in_rnn_block_("step_input") self._assert_in_rnn_block_("step_input")
if not isinstance(x, Variable): if not isinstance(x, Variable):
raise TypeError( raise TypeError(
...@@ -1419,6 +1484,15 @@ class DynamicRNN(object): ...@@ -1419,6 +1484,15 @@ class DynamicRNN(object):
return array_read(array=input_array, i=self.step_idx) return array_read(array=input_array, i=self.step_idx)
def static_input(self, x): def static_input(self, x):
"""
Mark a variable as a RNN input. The input will not be scattered into
time steps.
Args:
x(Variable): The input variable.
Returns:
The input variable that can access in RNN.
"""
self._assert_in_rnn_block_("static_input") self._assert_in_rnn_block_("static_input")
if not isinstance(x, Variable): if not isinstance(x, Variable):
raise TypeError( raise TypeError(
...@@ -1440,6 +1514,10 @@ class DynamicRNN(object): ...@@ -1440,6 +1514,10 @@ class DynamicRNN(object):
@contextlib.contextmanager @contextlib.contextmanager
def block(self): def block(self):
"""
The block for user to define operators in RNN. See the class docstring
for more details.
"""
if self.status != DynamicRNN.BEFORE_RNN: if self.status != DynamicRNN.BEFORE_RNN:
raise ValueError("rnn.block() can only be invoke once") raise ValueError("rnn.block() can only be invoke once")
self.step_idx = fill_constant( self.step_idx = fill_constant(
...@@ -1466,6 +1544,9 @@ class DynamicRNN(object): ...@@ -1466,6 +1544,9 @@ class DynamicRNN(object):
x=each_array, table=self.lod_rank_table)) x=each_array, table=self.lod_rank_table))
def __call__(self, *args, **kwargs): def __call__(self, *args, **kwargs):
"""
Get the output of RNN. This API should only be invoked after RNN.block()
"""
if self.status != DynamicRNN.AFTER_RNN: if self.status != DynamicRNN.AFTER_RNN:
raise ValueError(("Output of the dynamic RNN can only be visited " raise ValueError(("Output of the dynamic RNN can only be visited "
"outside the rnn block.")) "outside the rnn block."))
...@@ -1480,6 +1561,70 @@ class DynamicRNN(object): ...@@ -1480,6 +1561,70 @@ class DynamicRNN(object):
value=0.0, value=0.0,
need_reorder=False, need_reorder=False,
dtype='float32'): dtype='float32'):
"""
Create a memory variable for dynamic rnn.
If the :code:`init` is not None, :code:`memory` will be initialized by
this variable. The :code:`need_reorder` is used to reorder the memory as
the input variable. It should be set to true when the initialized memory
depends on the input sample.
For example,
>>> import paddle.fluid as fluid
>>> sentence = fluid.layers.data(
>>> name='sentence', dtype='float32', shape=[32])
>>> boot_memory = fluid.layers.data(
>>> name='boot', dtype='float32', shape=[10])
>>>
>>> drnn = fluid.layers.DynamicRNN()
>>> with drnn.block():
>>> word = drnn.step_input(sentence)
>>> memory = drnn.memory(init=boot_memory, need_reorder=True)
>>> hidden = fluid.layers.fc(
>>> input=[word, memory], size=10, act='tanh')
>>> drnn.update_memory(ex_mem=memory, new_mem=hidden)
>>> drnn.output(hidden)
>>> rnn_output = drnn()
Otherwise, if :code:`shape`, :code:`value`, :code:`dtype` are set, the
:code:`memory` will be initialized by this :code:`value`.
For example,
>>> import paddle.fluid as fluid
>>> sentence = fluid.layers.data(
>>> name='sentence', dtype='float32', shape=[32])
>>>
>>> drnn = fluid.layers.DynamicRNN()
>>> with drnn.block():
>>> word = drnn.step_input(sentence)
>>> memory = drnn.memory(shape=[10], dtype='float32', value=0)
>>> hidden = fluid.layers.fc(
>>> input=[word, memory], size=10, act='tanh')
>>> drnn.update_memory(ex_mem=memory, new_mem=hidden)
>>> drnn.output(hidden)
>>> rnn_output = drnn()
Args:
init(Variable|None): The initialized variable.
shape(list|tuple): The memory shape. NOTE the shape does not contain
batch_size.
value(float): the initalized value.
need_reorder(bool): True if the initialized memory depends on the
input sample.
dtype(str|numpy.dtype): The data type of the initialized memory.
Returns:
the memory variable.
"""
self._assert_in_rnn_block_('memory') self._assert_in_rnn_block_('memory')
if init is not None: if init is not None:
if not isinstance(init, Variable): if not isinstance(init, Variable):
...@@ -1547,6 +1692,16 @@ class DynamicRNN(object): ...@@ -1547,6 +1692,16 @@ class DynamicRNN(object):
return self.memory(init=init) return self.memory(init=init)
def update_memory(self, ex_mem, new_mem): def update_memory(self, ex_mem, new_mem):
"""
Update the memory from ex_mem to new_mem. NOTE that the shape and data
type of :code:`ex_mem` and :code:`new_mem` must be same.
Args:
ex_mem(Variable): the memory variable.
new_mem(Variable): the plain variable generated in RNN block.
Returns:
None
"""
self._assert_in_rnn_block_('update_memory') self._assert_in_rnn_block_('update_memory')
if not isinstance(ex_mem, Variable): if not isinstance(ex_mem, Variable):
raise TypeError("The input arg `ex_mem` of update_memory() must " raise TypeError("The input arg `ex_mem` of update_memory() must "
...@@ -1564,6 +1719,15 @@ class DynamicRNN(object): ...@@ -1564,6 +1719,15 @@ class DynamicRNN(object):
self.mem_link.append((new_mem, mem_array)) self.mem_link.append((new_mem, mem_array))
def output(self, *outputs): def output(self, *outputs):
"""
mark the RNN output variables.
Args:
outputs: The output variables.
Returns:
None
"""
self._assert_in_rnn_block_('output') self._assert_in_rnn_block_('output')
parent_block = self._parent_block_() parent_block = self._parent_block_()
for each in outputs: for each in outputs:
......
...@@ -210,53 +210,68 @@ def bipartite_match(dist_matrix, ...@@ -210,53 +210,68 @@ def bipartite_match(dist_matrix,
dist_threshold=None, dist_threshold=None,
name=None): name=None):
""" """
**Bipartite matchint operator** This operator implements a greedy bipartite matching algorithm, which is
used to obtain the matching with the maximum distance based on the input
This operator is a greedy bipartite matching algorithm, which is used to
obtain the matching with the maximum distance based on the input
distance matrix. For input 2D matrix, the bipartite matching algorithm can distance matrix. For input 2D matrix, the bipartite matching algorithm can
find the matched column for each row, also can find the matched row for find the matched column for each row (matched means the largest distance),
each column. And this operator only calculate matched indices from column also can find the matched row for each column. And this operator only
to row. For each instance, the number of matched indices is the number of calculate matched indices from column to row. For each instance,
of columns of the input ditance matrix. the number of matched indices is the column number of the input distance
matrix.
There are two outputs to save matched indices and distance.
A simple description, this algothrim matched the best (maximum distance) There are two outputs, matched indices and distance.
A simple description, this algorithm matched the best (maximum distance)
row entity to the column entity and the matched indices are not duplicated row entity to the column entity and the matched indices are not duplicated
in each row of ColToRowMatchIndices. If the column entity is not matched in each row of ColToRowMatchIndices. If the column entity is not matched
any row entity, set -1 in ColToRowMatchIndices. any row entity, set -1 in ColToRowMatchIndices.
Please note that the input DistMat can be LoDTensor (with LoD) or Tensor. NOTE: the input DistMat can be LoDTensor (with LoD) or Tensor.
If LoDTensor with LoD, the height of ColToRowMatchIndices is batch size. If LoDTensor with LoD, the height of ColToRowMatchIndices is batch size.
If Tensor, the height of ColToRowMatchIndices is 1. If Tensor, the height of ColToRowMatchIndices is 1.
NOTE: This API is a very low level API. It is used by :code:`ssd_loss`
layer. Please consider to use :code:`ssd_loss` instead.
Args: Args:
dist_matrix(Variable): This input is a 2-D LoDTensor with shape dist_matrix(Variable): This input is a 2-D LoDTensor with shape
[K, M]. It is pair-wise distance matrix between the entities [K, M]. It is pair-wise distance matrix between the entities
represented by each row and each column. For example, assumed one represented by each row and each column. For example, assumed one
entity is A with shape [K], another entity is B with shape [M]. The entity is A with shape [K], another entity is B with shape [M]. The
dist_matirx[i][j] is the distance between A[i] and B[j]. The bigger dist_matrix[i][j] is the distance between A[i] and B[j]. The bigger
the distance is, the better macthing the pairs are. Please note, the distance is, the better matching the pairs are.
This tensor can contain LoD information to represent a batch of
inputs. One instance of this batch can contain different numbers of NOTE: This tensor can contain LoD information to represent a batch
entities. of inputs. One instance of this batch can contain different numbers
of entities.
match_type(string|None): The type of matching method, should be match_type(string|None): The type of matching method, should be
'bipartite' or 'per_prediction', 'bipartite' by defalut. 'bipartite' or 'per_prediction'. [default 'bipartite'].
dist_threshold(float|None): If `match_type` is 'per_prediction', dist_threshold(float|None): If `match_type` is 'per_prediction',
this threshold is to determine the extra matching bboxes based this threshold is to determine the extra matching bboxes based
on the maximum distance, 0.5 by defalut. on the maximum distance, 0.5 by default.
Returns: Returns:
match_indices(Variable): A 2-D Tensor with shape [N, M] in int type. tuple: a tuple with two elements is returned. The first is
N is the batch size. If match_indices[i][j] is -1, it matched_indices, the second is matched_distance.
means B[j] does not match any entity in i-th instance.
Otherwise, it means B[j] is matched to row The matched_indices is a 2-D Tensor with shape [N, M] in int type.
match_indices[i][j] in i-th instance. The row number of N is the batch size. If match_indices[i][j] is -1, it
i-th instance is saved in match_indices[i][j]. means B[j] does not match any entity in i-th instance.
match_distance(Variable): A 2-D Tensor with shape [N, M] in float type. Otherwise, it means B[j] is matched to row
N is batch size. If match_indices[i][j] is -1, match_indices[i][j] in i-th instance. The row number of
match_distance[i][j] is also -1.0. Otherwise, assumed i-th instance is saved in match_indices[i][j].
match_distance[i][j] = d, and the row offsets of each instance
are called LoD. Then match_distance[i][j] = dist_matrix[d+LoD[i]][j]. The matched_distance is a 2-D Tensor with shape [N, M] in float type
. N is batch size. If match_indices[i][j] is -1,
match_distance[i][j] is also -1.0. Otherwise, assumed
match_distance[i][j] = d, and the row offsets of each instance
are called LoD. Then match_distance[i][j] =
dist_matrix[d+LoD[i]][j].
Examples:
>>> x = fluid.layers.data(name='x', shape=[4], dtype='float32')
>>> y = fluid.layers.data(name='y', shape=[4], dtype='float32')
>>> iou = fluid.layers.iou_similarity(x=x, y=y)
>>> matched_indices, matched_dist = fluid.layers.bipartite_match(iou)
""" """
helper = LayerHelper('bipartite_match', **locals()) helper = LayerHelper('bipartite_match', **locals())
match_indices = helper.create_tmp_variable(dtype='int32') match_indices = helper.create_tmp_variable(dtype='int32')
...@@ -364,7 +379,7 @@ def ssd_loss(location, ...@@ -364,7 +379,7 @@ def ssd_loss(location,
normalize=True, normalize=True,
sample_size=None): sample_size=None):
""" """
**Multi-box loss layer for object dection algorithm of SSD** **Multi-box loss layer for object detection algorithm of SSD**
This layer is to compute dection loss for SSD given the location offset This layer is to compute dection loss for SSD given the location offset
predictions, confidence predictions, prior boxes and ground-truth boudding predictions, confidence predictions, prior boxes and ground-truth boudding
...@@ -372,21 +387,35 @@ def ssd_loss(location, ...@@ -372,21 +387,35 @@ def ssd_loss(location,
is a weighted sum of the localization loss (or regression loss) and is a weighted sum of the localization loss (or regression loss) and
confidence loss (or classification loss) by performing the following steps: confidence loss (or classification loss) by performing the following steps:
1. Find matched boundding box by bipartite matching algorithm. 1. Find matched bounding box by bipartite matching algorithm.
1.1 Compute IOU similarity between ground-truth boxes and prior boxes. 1.1 Compute IOU similarity between ground-truth boxes and prior boxes.
1.2 Compute matched boundding box by bipartite matching algorithm. 1.2 Compute matched boundding box by bipartite matching algorithm.
2. Compute confidence for mining hard examples 2. Compute confidence for mining hard examples
2.1. Get the target label based on matched indices. 2.1. Get the target label based on matched indices.
2.2. Compute confidence loss. 2.2. Compute confidence loss.
3. Apply hard example mining to get the negative example indices and update 3. Apply hard example mining to get the negative example indices and update
the matched indices. the matched indices.
4. Assign classification and regression targets 4. Assign classification and regression targets
4.1. Encoded bbox according to the prior boxes. 4.1. Encoded bbox according to the prior boxes.
4.2. Assign regression targets. 4.2. Assign regression targets.
4.3. Assign classification targets. 4.3. Assign classification targets.
5. Compute the overall objective loss. 5. Compute the overall objective loss.
5.1 Compute confidence loss. 5.1 Compute confidence loss.
5.1 Compute localization loss. 5.1 Compute localization loss.
5.3 Compute the overall weighted loss. 5.3 Compute the overall weighted loss.
Args: Args:
...@@ -421,39 +450,36 @@ def ssd_loss(location, ...@@ -421,39 +450,36 @@ def ssd_loss(location,
mining_type (str): The hard example mining type, should be 'hard_example' mining_type (str): The hard example mining type, should be 'hard_example'
or 'max_negative', now only support `max_negative`. or 'max_negative', now only support `max_negative`.
normalize (bool): Whether to normalize the SSD loss by the total number normalize (bool): Whether to normalize the SSD loss by the total number
of output locations, True by defalut. of output locations, True by default.
sample_size (int): The max sample size of negative box, used only when sample_size (int): The max sample size of negative box, used only when
mining_type is 'hard_example'. mining_type is 'hard_example'.
Returns: Returns:
Variable: The weighted sum of the localization loss and confidence loss, The weighted sum of the localization loss and confidence loss, with \
with shape [N * Np, 1], N and Np are the same as they are shape [N * Np, 1], N and Np are the same as they are in `location`.
in `location`.
Raises: Raises:
ValueError: If mining_type is 'hard_example', now only support ValueError: If mining_type is 'hard_example', now only support mining \
mining type of `max_negative`. type of `max_negative`.
Examples: Examples:
.. code-block:: python >>> pb = fluid.layers.data(
>>> name='prior_box',
pb = layers.data( >>> shape=[10, 4],
name='prior_box', >>> append_batch_size=False,
shape=[10, 4], >>> dtype='float32')
append_batch_size=False, >>> pbv = fluid.layers.data(
dtype='float32') >>> name='prior_box_var',
pbv = layers.data( >>> shape=[10, 4],
name='prior_box_var', >>> append_batch_size=False,
shape=[10, 4], >>> dtype='float32')
append_batch_size=False, >>> loc = fluid.layers.data(name='target_box', shape=[10, 4], dtype='float32')
dtype='float32') >>> scores = fluid.layers.data(name='scores', shape=[10, 21], dtype='float32')
loc = layers.data(name='target_box', shape=[10, 4], dtype='float32') >>> gt_box = fluid.layers.data(
scores = layers.data(name='scores', shape=[10, 21], dtype='float32') >>> name='gt_box', shape=[4], lod_level=1, dtype='float32')
gt_box = layers.data( >>> gt_label = fluid.layers.data(
name='gt_box', shape=[4], lod_level=1, dtype='float32') >>> name='gt_label', shape=[1], lod_level=1, dtype='float32')
gt_label = layers.data( >>> loss = fluid.layers.ssd_loss(loc, scores, gt_box, gt_label, pb, pbv)
name='gt_label', shape=[1], lod_level=1, dtype='float32')
loss = layers.ssd_loss(loc, scores, gt_box, gt_label, pb, pbv)
""" """
helper = LayerHelper('ssd_loss', **locals()) helper = LayerHelper('ssd_loss', **locals())
......
...@@ -22,9 +22,9 @@ from ..executor import global_scope ...@@ -22,9 +22,9 @@ from ..executor import global_scope
from layer_function_generator import generate_layer_fn, templatedoc from layer_function_generator import generate_layer_fn, templatedoc
__all__ = [ __all__ = [
'data', 'BlockGuardServ', 'ListenAndServ', 'Send', 'open_recordio_file', 'data', 'BlockGuardServ', 'ListenAndServ', 'Send', 'Recv',
'open_files', 'read_file', 'shuffle', 'batch', 'double_buffer', 'open_recordio_file', 'open_files', 'read_file', 'shuffle', 'batch',
'random_data_generator', 'Preprocessor', 'load' 'double_buffer', 'random_data_generator', 'Preprocessor', 'load'
] ]
...@@ -177,18 +177,17 @@ class ListenAndServ(object): ...@@ -177,18 +177,17 @@ class ListenAndServ(object):
}) })
def Send(endpoints, send_vars, get_vars=None): def Send(endpoints, send_vars, sync=True):
""" """
Send layer Send variables to the server side, and get vars from server
side when server have finished running server side program.
Args: Args:
endpoints: comma seperated IP:PORT pairs in the order endpoints (str): comma seperated IP:PORT pairs in the order
of send_vars to send of send_vars to send
send_vars: vars to send send_vars (list): variables to send to server
get_vars: vars to get from server after send completes. sync (bool): whether to wait the request finish
Send variables to the server side, and get vars from server
side when server have finished running server side program.
""" """
assert (type(send_vars) == list) assert (type(send_vars) == list)
...@@ -196,40 +195,33 @@ def Send(endpoints, send_vars, get_vars=None): ...@@ -196,40 +195,33 @@ def Send(endpoints, send_vars, get_vars=None):
endpoints = list(set(epmap)) endpoints = list(set(epmap))
helper = LayerHelper("Send", **locals()) helper = LayerHelper("Send", **locals())
if not get_vars:
get_vars = []
for s in send_vars:
v = helper.create_tmp_variable(dtype=s.dtype, stop_gradient=True)
get_vars.append(v)
rpc_op_role_name = core.op_proto_and_checker_maker.kOpRoleAttrName() rpc_op_role_name = core.op_proto_and_checker_maker.kOpRoleAttrName()
helper.append_op( helper.append_op(
type="send", type="send",
inputs={"X": send_vars}, inputs={"X": send_vars},
outputs={"Out": get_vars},
attrs={ attrs={
"endpoints": endpoints, "endpoints": endpoints,
"epmap": epmap, "epmap": epmap,
rpc_op_role_name: core.op_proto_and_checker_maker.OpRole.RPC rpc_op_role_name: core.op_proto_and_checker_maker.OpRole.RPC
}) })
if sync:
return get_vars helper.append_op(type="send_barrier", attrs={"endpoints": endpoints})
def Recv(endpoints, get_vars): def Recv(endpoints, get_vars, sync=True):
""" """
Recv layer Receive variables from server side
Args: Args:
endpoints: comma seperated IP:PORT pairs in the order endpoints (str): comma seperated IP:PORT pairs in the order
of send_vars to send of send_vars to send
send_vars: vars to send get_vars (list): vars to get from server after send completes.
get_vars: vars to get from server after send completes. sync (bool): whether to wait the request finish
Send variables to the server side, and get vars from server Returns:
side when server have finished running server side program. list: list of received variables
""" """
assert (type(send_vars) == list)
assert (type(get_vars) == list) assert (type(get_vars) == list)
epmap = endpoints.split(",") epmap = endpoints.split(",")
...@@ -242,6 +234,9 @@ def Recv(endpoints, get_vars): ...@@ -242,6 +234,9 @@ def Recv(endpoints, get_vars):
outputs={"Out": get_vars}, outputs={"Out": get_vars},
attrs={"endpoints": endpoints, attrs={"endpoints": endpoints,
"epmap": epmap}) "epmap": epmap})
if sync:
helper.append_op(type="fetch_barrier", attrs={"endpoints": endpoints})
return get_vars
def monkey_patch_reader_methods(reader): def monkey_patch_reader_methods(reader):
...@@ -292,6 +287,7 @@ def _copy_reader_create_op_(block, op): ...@@ -292,6 +287,7 @@ def _copy_reader_create_op_(block, op):
return new_op return new_op
@templatedoc(op_type='create_recordio_file_reader')
def open_recordio_file(filename, def open_recordio_file(filename,
shapes, shapes,
lod_levels, lod_levels,
...@@ -299,34 +295,30 @@ def open_recordio_file(filename, ...@@ -299,34 +295,30 @@ def open_recordio_file(filename,
pass_num=1, pass_num=1,
for_parallel=True): for_parallel=True):
""" """
Open a RecordIO file ${comment}
This layer takes a RecordIO file to read from and returns a Reader Variable.
Via the Reader Variable, we can get data from the given RecordIO file.
Args: Args:
filename(str): The RecordIO file's name. filename(${filename_type}): ${filename_comment}.
shapes(list): List of tuples which declaring data shapes. shapes(list): List of tuples which declaring data shapes.
lod_levels(list): List of ints which declaring data lod_level. lod_levels(${lod_levels_type}): ${lod_levels_comment}.
dtypes(list): List of strs which declaring data type. dtypes(list): List of strs which declaring data type.
pass_num(int): Number of passes to run. pass_num(int): Number of passes to run.
for_parallel(Bool): Set it as True if you are going to run for_parallel(Bool): Set it as True if you are going to run
subsequent operators in parallel. subsequent operators in parallel.
Returns: Returns:
Variable: A Reader Variable via which we can get RecordIO file data. ${out_comment}.
Examples: Examples:
.. code-block:: python
reader = fluid.layers.io.open_recordio_file( >>> import paddle.fluid as fluid
filename='./data.recordio', >>> reader = fluid.layers.io.open_recordio_file(
shapes=[(3,224,224), (1)], >>> filename='./data.recordio',
lod_levels=[0, 0], >>> shapes=[(3,224,224), (1)],
dtypes=['float32', 'int64']) >>> lod_levels=[0, 0],
>>> dtypes=['float32', 'int64'])
# Via the reader, we can use 'read_file' layer to get data: >>> # Via the reader, we can use 'read_file' layer to get data:
image, label = fluid.layers.io.read_file(reader) >>> image, label = fluid.layers.io.read_file(reader)
""" """
dtypes = [convert_np_dtype_to_dtype_(dt) for dt in dtypes] dtypes = [convert_np_dtype_to_dtype_(dt) for dt in dtypes]
shape_concat = [] shape_concat = []
...@@ -544,6 +536,9 @@ def __create_unshared_decorated_reader__(op_type, reader, attrs, name=None): ...@@ -544,6 +536,9 @@ def __create_unshared_decorated_reader__(op_type, reader, attrs, name=None):
def shuffle(reader, buffer_size): def shuffle(reader, buffer_size):
"""
Shuffle the reader.
"""
return __create_unshared_decorated_reader__( return __create_unshared_decorated_reader__(
'create_shuffle_reader', reader, {'buffer_size': int(buffer_size)}) 'create_shuffle_reader', reader, {'buffer_size': int(buffer_size)})
...@@ -589,6 +584,29 @@ def batch(reader, batch_size): ...@@ -589,6 +584,29 @@ def batch(reader, batch_size):
def double_buffer(reader, place=None, name=None): def double_buffer(reader, place=None, name=None):
"""
Wrap a double buffer reader. The data will copy to target place with a
double buffer queue. If the target place is None, the place that executor
perform on will be used.
Args:
reader(Variable): the reader variable need to be wrapped.
place(Place): the place of target data. Default is the sample place of
executor perform.
name(str): Variable name. None if the user does not care.
Returns:
wrapped reader with double buffer.
Examples:
>>> reader = fluid.layers.open_files(filenames=['somefile'],
>>> shapes=[[-1, 784], [-1, 1]],
>>> dtypes=['float32', 'int64'])
>>> reader = fluid.layers.double_buffer(reader)
>>> img, label = fluid.layers.read_file(reader)
"""
attrs = dict() attrs = dict()
if place is not None: if place is not None:
attrs['place'] = str(place).upper() attrs['place'] = str(place).upper()
......
...@@ -44,6 +44,11 @@ def _type_to_str_(tp): ...@@ -44,6 +44,11 @@ def _type_to_str_(tp):
return framework_pb2.AttrType.Name(tp) return framework_pb2.AttrType.Name(tp)
_two_dollar_pattern_ = re.compile(r"\$\$([^\$]+)\$\$")
_single_dollar_pattern_ = re.compile(r"\$([^\$]+)\$")
_two_bang_pattern_ = re.compile(r"!!([^!]+)!!")
def _generate_doc_string_(op_proto): def _generate_doc_string_(op_proto):
""" """
Generate docstring by OpProto Generate docstring by OpProto
...@@ -55,22 +60,26 @@ def _generate_doc_string_(op_proto): ...@@ -55,22 +60,26 @@ def _generate_doc_string_(op_proto):
str: the document string str: the document string
""" """
def escape_math(text):
return _two_bang_pattern_.sub(
r'$$\1$$',
_single_dollar_pattern_.sub(
r':math:`\1`', _two_dollar_pattern_.sub(r"!!\1!!", text)))
if not isinstance(op_proto, framework_pb2.OpProto): if not isinstance(op_proto, framework_pb2.OpProto):
raise TypeError("OpProto should be `framework_pb2.OpProto`") raise TypeError("OpProto should be `framework_pb2.OpProto`")
buf = cStringIO.StringIO() buf = cStringIO.StringIO()
buf.write(op_proto.comment) buf.write(escape_math(op_proto.comment))
buf.write('\nArgs:\n') buf.write('\nArgs:\n')
for each_input in op_proto.inputs: for each_input in op_proto.inputs:
line_begin = ' {0}: '.format(_convert_(each_input.name)) line_begin = ' {0}: '.format(_convert_(each_input.name))
buf.write(line_begin) buf.write(line_begin)
buf.write(each_input.comment) buf.write(escape_math(each_input.comment))
buf.write('\n') if each_input.duplicable:
buf.write(' ' * len(line_begin)) buf.write(" Duplicatable.")
buf.write('Duplicable: ') if each_input.dispensable:
buf.write(str(each_input.duplicable)) buf.write(" Optional.")
buf.write(' Optional: ')
buf.write(str(each_input.dispensable))
buf.write('\n') buf.write('\n')
skip_attrs = OpProtoHolder.generated_op_attr_names() skip_attrs = OpProtoHolder.generated_op_attr_names()
...@@ -83,7 +92,7 @@ def _generate_doc_string_(op_proto): ...@@ -83,7 +92,7 @@ def _generate_doc_string_(op_proto):
buf.write(' (') buf.write(' (')
buf.write(_type_to_str_(each_attr.type)) buf.write(_type_to_str_(each_attr.type))
buf.write('): ') buf.write('): ')
buf.write(each_attr.comment) buf.write(escape_math(each_attr.comment))
buf.write('\n') buf.write('\n')
if len(op_proto.outputs) != 0: if len(op_proto.outputs) != 0:
...@@ -92,7 +101,7 @@ def _generate_doc_string_(op_proto): ...@@ -92,7 +101,7 @@ def _generate_doc_string_(op_proto):
for each_opt in op_proto.outputs: for each_opt in op_proto.outputs:
if not each_opt.intermediate: if not each_opt.intermediate:
break break
buf.write(each_opt.comment) buf.write(escape_math(each_opt.comment))
return buf.getvalue() return buf.getvalue()
......
...@@ -225,11 +225,11 @@ def embedding(input, ...@@ -225,11 +225,11 @@ def embedding(input,
have two elements which indicate the size of the dictionary of have two elements which indicate the size of the dictionary of
embeddings and the size of each embedding vector respectively. embeddings and the size of each embedding vector respectively.
is_sparse(bool): The flag indicating whether to use sparse update. is_sparse(bool): The flag indicating whether to use sparse update.
is_distributed (bool): Whether to run lookup table from remote parameter server. is_distributed(bool): Whether to run lookup table from remote parameter server.
padding_idx(int|long|None): If :attr:`None`, it makes no effect to lookup. padding_idx(int|long|None): If :attr:`None`, it makes no effect to lookup.
Otherwise the given :attr:`padding_idx` indicates padding the output Otherwise the given :attr:`padding_idx` indicates padding the output
with zeros whenever lookup encounters it in :attr:`input`. If with zeros whenever lookup encounters it in :attr:`input`. If
:math:`padding_idx < 0`, the padding_idx to use in lookup is :math:`padding_idx < 0`, the :attr:`padding_idx` to use in lookup is
:math:`size[0] + dim`. :math:`size[0] + dim`.
param_attr(ParamAttr): Parameters for this layer param_attr(ParamAttr): Parameters for this layer
dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32, float_16, int etc dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32, float_16, int etc
...@@ -1235,14 +1235,17 @@ def conv2d(input, ...@@ -1235,14 +1235,17 @@ def conv2d(input,
act=None, act=None,
name=None): name=None):
""" """
**Convlution2D Layer**
The convolution2D layer calculates the output based on the input, filter The convolution2D layer calculates the output based on the input, filter
and strides, paddings, dilations, groups parameters. Input(Input) and and strides, paddings, dilations, groups parameters. Input and
Output(Output) are in NCHW format. Where N is batch size, C is the number of 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. 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, Filter is in MCHW format, where M is the number of output image channels,
<http://ufldl.stanford.edu/tutorial/supervised/FeatureExtractionUsingConvolution/>`_ . C is the number of input image channels, H is the height of the filter,
and W is the width of the filter. If the groups is greater than 1,
C will equal the number of input image channels divided by the groups.
Please refer to UFLDL's `convolution
<http://ufldl.stanford.edu/tutorial/supervised/FeatureExtractionUsingConvolution/>`_
for more detials.
If bias attribution and activation type are provided, bias is added to the If bias attribution and activation type are provided, bias is added to the
output of the convolution, and the corresponding activation function is output of the convolution, and the corresponding activation function is
applied to the final result. applied to the final result.
...@@ -1253,15 +1256,14 @@ def conv2d(input, ...@@ -1253,15 +1256,14 @@ def conv2d(input,
Out = \sigma (W \\ast X + b) Out = \sigma (W \\ast X + b)
In the above equation: Where:
* :math:`X`: Input value, a tensor with NCHW format. * :math:`X`: Input value, a tensor with NCHW format.
* :math:`W`: Filter value, a tensor with MCHW format. * :math:`W`: Filter value, a tensor with MCHW format.
* :math:`\\ast`: Convolution operation. * :math:`\\ast`: Convolution operation.
* :math:`b`: Bias value, a 2-D tensor with shape [M, 1]. * :math:`b`: Bias value, a 2-D tensor with shape [M, 1].
* :math:`\\sigma`: Activation function. * :math:`\\sigma`: Activation function.
* :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be different.
different.
Example: Example:
...@@ -1272,6 +1274,7 @@ def conv2d(input, ...@@ -1272,6 +1274,7 @@ def conv2d(input,
Filter shape: :math:`(C_{out}, C_{in}, H_f, W_f)` Filter shape: :math:`(C_{out}, C_{in}, H_f, W_f)`
- Output: - Output:
Output shape: :math:`(N, C_{out}, H_{out}, W_{out})` Output shape: :math:`(N, C_{out}, H_{out}, W_{out})`
Where Where
...@@ -1283,7 +1286,7 @@ def conv2d(input, ...@@ -1283,7 +1286,7 @@ def conv2d(input,
Args: Args:
input (Variable): The input image with [N, C, H, W] format. 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 num_filters(int): The number of filter. It is as same as the output
image channel. image channel.
filter_size (int|tuple|None): The filter size. If filter_size is a tuple, 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). it must contain two integers, (filter_size_H, filter_size_W).
...@@ -1306,7 +1309,8 @@ def conv2d(input, ...@@ -1306,7 +1309,8 @@ def conv2d(input,
bias_attr (ParamAttr): Bias parameter for 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 use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn
library is installed. Default: True library is installed. Default: True
use_mkldnn (bool): Use mkldnn kernels or not. use_mkldnn (bool): Use mkldnn kernels or not, it is valid only when compiled
with mkldnn library. Default: False
act (str): Activation type. Default: None act (str): Activation type. Default: None
name (str|None): A name for this layer(optional). If set None, the layer name (str|None): A name for this layer(optional). If set None, the layer
will be named automatically. will be named automatically.
...@@ -1974,6 +1978,7 @@ def batch_norm(input, ...@@ -1974,6 +1978,7 @@ def batch_norm(input,
return helper.append_activation(batch_norm_out) return helper.append_activation(batch_norm_out)
@templatedoc()
def layer_norm(input, def layer_norm(input,
scale=True, scale=True,
shift=True, shift=True,
...@@ -1984,20 +1989,11 @@ def layer_norm(input, ...@@ -1984,20 +1989,11 @@ def layer_norm(input,
act=None, act=None,
name=None): name=None):
""" """
**Layer Normalization** ${comment}
Assume feature vectors exist on dimensions
:attr:`begin_norm_axis ... rank(input)` and calculate the moment statistics
along these dimensions for each feature vector :math:`a` with size
:math:`H`, then normalize each feature vector using the corresponding
statistics. After that, apply learnable gain and bias on the normalized
tensor to scale and shift if :attr:`scale` and :attr:`shift` are set.
Refer to `Layer Normalization <https://arxiv.org/pdf/1607.06450v1.pdf>`_
The formula is as follows: The formula is as follows:
.. math:: .. math::
\\mu & = \\frac{1}{H}\\sum_{i=1}^{H} a_i \\mu & = \\frac{1}{H}\\sum_{i=1}^{H} a_i
...@@ -2005,6 +2001,15 @@ def layer_norm(input, ...@@ -2005,6 +2001,15 @@ def layer_norm(input,
h & = f(\\frac{g}{\\sigma}(a - \\mu) + b) h & = f(\\frac{g}{\\sigma}(a - \\mu) + b)
* :math:`a`: the vector representation of the summed inputs to the neurons
in that layer.
* :math:`H`: the number of hidden units in a layers
* :math:`g`: the trainable scale parameter.
* :math:`b`: the trainable bias parameter.
Args: Args:
input(Variable): The input tensor variable. input(Variable): The input tensor variable.
scale(bool): Whether to learn the adaptive gain :math:`g` after scale(bool): Whether to learn the adaptive gain :math:`g` after
...@@ -2023,14 +2028,13 @@ def layer_norm(input, ...@@ -2023,14 +2028,13 @@ def layer_norm(input,
name (str): The name of this layer. It is optional. name (str): The name of this layer. It is optional.
Returns: Returns:
Variable: A tensor variable with the same shape as the input. ${y_comment}
Examples: Examples:
.. code-block:: python
data = fluid.layers.data( >>> data = fluid.layers.data(name='data', shape=[3, 32, 32],
name='data', shape=[3, 32, 32], dtype='float32') >>> dtype='float32')
x = fluid.layers.layer_norm(input=data, begin_norm_axis=1) >>> x = fluid.layers.layer_norm(input=data, begin_norm_axis=1)
""" """
helper = LayerHelper('layer_norm', **locals()) helper = LayerHelper('layer_norm', **locals())
dtype = helper.input_dtype() dtype = helper.input_dtype()
...@@ -3739,29 +3743,13 @@ def im2sequence(input, filter_size=1, stride=1, padding=0, name=None): ...@@ -3739,29 +3743,13 @@ def im2sequence(input, filter_size=1, stride=1, padding=0, name=None):
return out return out
@templatedoc()
def row_conv(input, future_context_size, param_attr=None, act=None): def 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]. ${comment}
Parameters with shape [future_context_size + 1, D] will be created. The math
equation of row convolution is as follows:
.. math::
Out_{i} = \sum_{j = i} ^ {i + \\tau} X_{j} \odot W_{i - j}
In the above equation:
* :math:`Out_{i}`: The i-th row of output variable with shape [1, D].
* :math:`\\tau`: Future context size.
* :math:`X_{j}`: The j-th row of input variable with shape [1, D].
* :math:`W_{i-j}`: The (i-j)-th row of parameters with shape [1, D].
More details about row_conv please refer to the paper \
(http://www.cs.cmu.edu/~dyogatam/papers/wang+etal.iclrworkshop2016.pdf) and
the design document \
(https://github.com/PaddlePaddle/Paddle/issues/2228#issuecomment-303903645).
Args: Args:
input (Variable): Input variable, a 2D LoDTensor with shape [T, D]. input (${x_type}): ${x_comment}.
future_context_size (int): Future context size. Please note, the shape future_context_size (int): Future context size. Please note, the shape
of convolution kernel is [future_context_size + 1, D]. of convolution kernel is [future_context_size + 1, D].
param_attr (ParamAttr): Attributes of parameters, including param_attr (ParamAttr): Attributes of parameters, including
...@@ -3769,14 +3757,13 @@ def row_conv(input, future_context_size, param_attr=None, act=None): ...@@ -3769,14 +3757,13 @@ def row_conv(input, future_context_size, param_attr=None, act=None):
act (str): Non-linear activation to be applied to output variable. act (str): Non-linear activation to be applied to output variable.
Returns: Returns:
Variable: The output tensor with same shape as input tensor. ${out_comment}.
Examples: Examples:
.. code-block:: python >>> import paddle.fluid as fluid
>>> x = fluid.layers.data(name='x', shape=[16],
x = fluid.layers.data(name='x', shape=[16], >>> dtype='float32', lod_level=1)
dtype='float32', lod_level=1) >>> out = fluid.layers.row_conv(input=x, future_context_size=2)
out = fluid.layers.row_conv(input=x, future_context_size=2)
""" """
helper = LayerHelper('row_conv', **locals()) helper = LayerHelper('row_conv', **locals())
dtype = helper.input_dtype() dtype = helper.input_dtype()
...@@ -3792,42 +3779,23 @@ def row_conv(input, future_context_size, param_attr=None, act=None): ...@@ -3792,42 +3779,23 @@ def row_conv(input, future_context_size, param_attr=None, act=None):
return helper.append_activation(out) return helper.append_activation(out)
@templatedoc()
def multiplex(inputs, index): def multiplex(inputs, index):
""" """
**Multiplex Layer** ${comment}
Referring to the given index variable, this layer selects rows from the >>> import paddle.fluid as fluid
input variables to construct a multiplex variable. Assuming that there are >>> x1 = fluid.layers.data(name='x1', shape=[4], dtype='float32')
:math:`m` input variables and :math:`I_i` represents the i-th input >>> x2 = fluid.layers.data(name='x2', shape=[4], dtype='float32')
variable and :math:`i` is in [0, :math:`m`). All input variables are >>> index = fluid.layers.data(name='index', shape=[1], dtype='int32')
tensors with same shape [:math:`d_0`, :math:`d_1`, ..., :math:`d_R`]. >>> out = fluid.layers.multiplex(inputs=[x1, x2], index=index)
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 [:math:`M`, :math:`N`]
where :math:`M` for :math:`d_0` and :math:`N` for :math:`d_1` * :math:`d_2`
* ... * :math:`d_R`. Let :math:`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
[:math:`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 [:math:`d_0`,
:math:`d_1`, ..., :math:`d_R`]. If we treat the output tensor as a 2-D
matrix with shape [:math:`M`, :math:`N`] and let :math:`O[i]` be the i-th
row of the matrix, then `O[i]` is equal to :math:`I_{ID[i]}[i]`.
Args: Args:
inputs (list): A list of variables to gather from. All variables have the inputs (list): ${x_comment}.
same shape and the rank is at least 2. index (${ids_type}): ${ids_comment}.
index (Variable): Tensor<int32>, index variable which is a 2-D tensor
with shape [M, 1] where M is the batch size.
Returns: Returns:
Variable: Multiplex variable gathered from input variables. ${out_comment}.
Examples:
.. code-block:: python
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)
""" """
helper = LayerHelper('multiplex', **locals()) helper = LayerHelper('multiplex', **locals())
......
...@@ -40,8 +40,6 @@ __activations__ = [ ...@@ -40,8 +40,6 @@ __activations__ = [
'relu6', 'relu6',
'pow', 'pow',
'stanh', 'stanh',
'hard_shrink',
'thresholded_relu',
'hard_sigmoid', 'hard_sigmoid',
'swish', 'swish',
] ]
...@@ -64,11 +62,9 @@ __all__ = [ ...@@ -64,11 +62,9 @@ __all__ = [
'logical_or', 'logical_or',
'logical_xor', 'logical_xor',
'logical_not', 'logical_not',
'uniform_random',
'uniform_random_batch_size_like', 'uniform_random_batch_size_like',
'gaussian_random', 'gaussian_random',
'gaussian_random_batch_size_like', 'gaussian_random_batch_size_like',
'cumsum',
'scatter', 'scatter',
'sum', 'sum',
'slice', 'slice',
...@@ -79,3 +75,88 @@ __all__ = [ ...@@ -79,3 +75,88 @@ __all__ = [
for _OP in set(__all__): for _OP in set(__all__):
globals()[_OP] = generate_layer_fn(_OP) globals()[_OP] = generate_layer_fn(_OP)
__all__ += ["uniform_random"]
_uniform_random_ = generate_layer_fn('uniform_random')
def uniform_random(shape, dtype=None, min=None, max=None, seed=None):
kwargs = dict()
for name in locals():
val = locals()[name]
if val is not None:
kwargs[name] = val
return _uniform_random_(**kwargs)
uniform_random.__doc__ = _uniform_random_.__doc__ + """
Examples:
>>> result = fluid.layers.uniform_random(shape=[32, 784])
"""
__all__ += ['hard_shrink']
_hard_shrink_ = generate_layer_fn('hard_shrink')
def hard_shrink(x, threshold=None):
kwargs = dict()
for name in locals():
val = locals()[name]
if val is not None:
kwargs[name] = val
return _hard_shrink_(**kwargs)
hard_shrink.__doc__ = _hard_shrink_.__doc__ + """
Examples:
>>> data = fluid.layers.data(name="input", shape=[784])
>>> result = fluid.layers.hard_shrink(x=data, threshold=0.3)
"""
__all__ += ['cumsum']
_cum_sum_ = generate_layer_fn('cumsum')
def cumsum(x, axis=None, exclusive=None, reverse=None):
kwargs = dict()
for name in locals():
val = locals()[name]
if val is not None:
kwargs[name] = val
return _cum_sum_(**kwargs)
cumsum.__doc__ = _cum_sum_.__doc__ + """
Examples:
>>> data = fluid.layers.data(name="input", shape=[32, 784])
>>> result = fluid.layers.cumsum(data, axis=0)
"""
__all__ += ['thresholded_relu']
_thresholded_relu_ = generate_layer_fn('thresholded_relu')
def thresholded_relu(x, threshold=None):
kwargs = dict()
for name in locals():
val = locals()[name]
if val is not None:
kwargs[name] = val
_thresholded_relu_(**kwargs)
thresholded_relu.__doc__ = _thresholded_relu_.__doc__ + """
Examples:
>>> data = fluid.layers.data(name="input", shape=[1])
>>> result = fluid.layers.thresholded_relu(data, threshold=0.4)
"""
...@@ -6,7 +6,7 @@ ...@@ -6,7 +6,7 @@
# #
# http://www.apache.org/licenses/LICENSE-2.0 # http://www.apache.org/licenses/LICENSE-2.0
# #
# Unless required by applicable law or agreed to in writing, software # Unlessf required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, # distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and # See the License for the specific language governing permissions and
...@@ -51,7 +51,12 @@ def create_parameter(shape, ...@@ -51,7 +51,12 @@ def create_parameter(shape,
is_bias=False, is_bias=False,
default_initializer=None): default_initializer=None):
""" """
Create a parameter Create a parameter. The parameter is a learnable variable, which can have
gradient, and can be optimized.
NOTE: this is a very low-level API. This API is useful when you create
operator by your self. instead of using layers.
Args: Args:
shape(list[int]): shape of the parameter shape(list[int]): shape of the parameter
dtype(string): element type of the parameter dtype(string): element type of the parameter
...@@ -63,7 +68,12 @@ def create_parameter(shape, ...@@ -63,7 +68,12 @@ def create_parameter(shape,
default_initializer(Initializer): initializer for the parameter default_initializer(Initializer): initializer for the parameter
Returns: Returns:
Parameter: the created parameter the created parameter.
Examples:
>>> W = fluid.layers.create_parameter(shape=[784, 200], dtype='float32')
>>> data = fluid.layers.data(name="img", shape=[64, 784], append_batch_size=False)
>>> hidden = fluid.layers.matmul(x=data, y=W)
""" """
helper = LayerHelper("create_parameter", **locals()) helper = LayerHelper("create_parameter", **locals())
if attr is None: if attr is None:
...@@ -207,6 +217,7 @@ def assign(input, output): ...@@ -207,6 +217,7 @@ def assign(input, output):
Examples: Examples:
.. code-block:: python .. code-block:: python
out = fluid.layers.create_tensor(dtype='float32') out = fluid.layers.create_tensor(dtype='float32')
hidden = fluid.layers.fc(input=data, size=10) hidden = fluid.layers.fc(input=data, size=10)
fluid.layers.assign(hidden, out) fluid.layers.assign(hidden, out)
......
...@@ -16,6 +16,7 @@ import os ...@@ -16,6 +16,7 @@ import os
import time import time
import unittest import unittest
from multiprocessing import Process from multiprocessing import Process
import signal
import numpy import numpy
...@@ -24,9 +25,6 @@ import paddle.fluid.layers as layers ...@@ -24,9 +25,6 @@ import paddle.fluid.layers as layers
class TestSendOp(unittest.TestCase): class TestSendOp(unittest.TestCase):
@unittest.skip(
"This test is buggy. We cannot use time.sleep to sync processes, the connection may fail in unittest."
)
def test_send(self): def test_send(self):
# Run init_serv in a thread # Run init_serv in a thread
place = fluid.CPUPlace() place = fluid.CPUPlace()
...@@ -35,7 +33,9 @@ class TestSendOp(unittest.TestCase): ...@@ -35,7 +33,9 @@ class TestSendOp(unittest.TestCase):
p.daemon = True p.daemon = True
p.start() p.start()
time.sleep(10) self.ps_timeout = 5
self._wait_ps_ready(p.pid)
with open("/tmp/paddle.%d.port" % p.pid, "r") as fn: with open("/tmp/paddle.%d.port" % p.pid, "r") as fn:
selected_port = int(fn.readlines()[0]) selected_port = int(fn.readlines()[0])
self.init_client(place, selected_port) self.init_client(place, selected_port)
...@@ -44,9 +44,23 @@ class TestSendOp(unittest.TestCase): ...@@ -44,9 +44,23 @@ class TestSendOp(unittest.TestCase):
self.assertTrue(numpy.allclose(self.local_out, self.dist_out)) self.assertTrue(numpy.allclose(self.local_out, self.dist_out))
# FIXME(typhoonzero): find a way to gracefully shutdown the server. # FIXME(typhoonzero): find a way to gracefully shutdown the server.
os.system("kill -9 %d" % p.pid) os.kill(p.pid, signal.SIGKILL)
p.join() p.join()
def _wait_ps_ready(self, pid):
start_left_time = self.ps_timeout
sleep_time = 0.5
while True:
assert start_left_time >= 0, "wait ps ready failed"
time.sleep(sleep_time)
try:
# the listen_and_serv_op would touch a file which contains the listen port
# on the /tmp directory until it was ready to process all the RPC call.
os.stat("/tmp/paddle.%d.port" % pid)
return
except os.error:
start_left_time -= sleep_time
def init_serv(self, place): def init_serv(self, place):
main = fluid.Program() main = fluid.Program()
...@@ -84,7 +98,10 @@ class TestSendOp(unittest.TestCase): ...@@ -84,7 +98,10 @@ class TestSendOp(unittest.TestCase):
dtype="float32", dtype="float32",
persistable=False, persistable=False,
shape=[32, 32]) shape=[32, 32])
o = layers.Send("127.0.0.1:%d" % port, [x], [get_var]) fluid.initializer.Constant(value=2.3)(get_var, main.global_block())
layers.Send("127.0.0.1:%d" % port, [x])
o = layers.Recv("127.0.0.1:%d" % port, [get_var])
exe = fluid.Executor(place) exe = fluid.Executor(place)
self.dist_out = exe.run(main, fetch_list=o) # o is a list self.dist_out = exe.run(main, fetch_list=o) # o is a list
......
...@@ -57,17 +57,18 @@ class TestListenAndServOp(OpTest): ...@@ -57,17 +57,18 @@ class TestListenAndServOp(OpTest):
def setUp(self): def setUp(self):
self.ps_timeout = 5 self.ps_timeout = 5
self.ip = "127.0.0.1" self.ip = "127.0.0.1"
self.port = "6173" self.port = "0"
self.trainers = 1 self.trainers = 1
self.trainer_id = 1 self.trainer_id = 0
def _start_pserver(self, use_cuda, sync_mode): def _start_pserver(self, use_cuda, sync_mode):
p = Process( p = Process(
target=run_pserver, target=run_pserver,
args=(use_cuda, sync_mode, self.ip, self.port, self.trainers, args=(use_cuda, sync_mode, self.ip, self.port, self.trainers,
self.trainer_id)) self.trainer_id))
p.daemon = True
p.start() p.start()
return p.pid return p
def _wait_ps_ready(self, pid): def _wait_ps_ready(self, pid):
start_left_time = self.ps_timeout start_left_time = self.ps_timeout
...@@ -89,18 +90,20 @@ class TestListenAndServOp(OpTest): ...@@ -89,18 +90,20 @@ class TestListenAndServOp(OpTest):
def test_handle_signal_in_serv_op(self): def test_handle_signal_in_serv_op(self):
# run pserver on CPU in sync mode # run pserver on CPU in sync mode
pid = self._start_pserver(False, True) p1 = self._start_pserver(False, True)
self._wait_ps_ready(pid) self._wait_ps_ready(p1.pid)
# raise SIGTERM to pserver # raise SIGTERM to pserver
os.kill(pid, signal.SIGTERM) os.kill(p1.pid, signal.SIGKILL)
p1.join()
# run pserver on CPU in async mode # run pserver on CPU in async mode
pid = self._start_pserver(False, False) p2 = self._start_pserver(False, False)
self._wait_ps_ready(pid) self._wait_ps_ready(p2.pid)
# raise SIGTERM to pserver # raise SIGTERM to pserver
os.kill(pid, signal.SIGTERM) os.kill(p2.pid, signal.SIGKILL)
p2.join()
if __name__ == '__main__': if __name__ == '__main__':
......
...@@ -173,6 +173,7 @@ class TestCRFModel(unittest.TestCase): ...@@ -173,6 +173,7 @@ class TestCRFModel(unittest.TestCase):
pe.run(feed=feeder.feed(cur_batch), pe.run(feed=feeder.feed(cur_batch),
fetch_list=[avg_cost.name]))[0] fetch_list=[avg_cost.name]))[0]
@unittest.skip(reason="CI hangs")
def test_update_sparse_parameter_all_reduce(self): def test_update_sparse_parameter_all_reduce(self):
build_strategy = fluid.BuildStrategy() build_strategy = fluid.BuildStrategy()
build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.AllReduce build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.AllReduce
...@@ -181,6 +182,7 @@ class TestCRFModel(unittest.TestCase): ...@@ -181,6 +182,7 @@ class TestCRFModel(unittest.TestCase):
self.check_network_convergence( self.check_network_convergence(
is_sparse=True, build_strategy=build_strategy, use_cuda=False) is_sparse=True, build_strategy=build_strategy, use_cuda=False)
@unittest.skip(reason="CI hangs")
def test_update_dense_parameter_all_reduce(self): def test_update_dense_parameter_all_reduce(self):
build_strategy = fluid.BuildStrategy() build_strategy = fluid.BuildStrategy()
build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.AllReduce build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.AllReduce
...@@ -189,6 +191,7 @@ class TestCRFModel(unittest.TestCase): ...@@ -189,6 +191,7 @@ class TestCRFModel(unittest.TestCase):
self.check_network_convergence( self.check_network_convergence(
is_sparse=False, build_strategy=build_strategy, use_cuda=False) is_sparse=False, build_strategy=build_strategy, use_cuda=False)
@unittest.skip(reason="CI hangs")
def test_update_sparse_parameter_reduce(self): def test_update_sparse_parameter_reduce(self):
build_strategy = fluid.BuildStrategy() build_strategy = fluid.BuildStrategy()
build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.Reduce build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.Reduce
...@@ -197,6 +200,7 @@ class TestCRFModel(unittest.TestCase): ...@@ -197,6 +200,7 @@ class TestCRFModel(unittest.TestCase):
self.check_network_convergence( self.check_network_convergence(
is_sparse=True, build_strategy=build_strategy, use_cuda=False) is_sparse=True, build_strategy=build_strategy, use_cuda=False)
@unittest.skip(reason="CI hangs")
def test_update_dense_parameter_reduce(self): def test_update_dense_parameter_reduce(self):
build_strategy = fluid.BuildStrategy() build_strategy = fluid.BuildStrategy()
build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.Reduce build_strategy.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.Reduce
......
...@@ -157,9 +157,11 @@ class ControlFlowGraph(object): ...@@ -157,9 +157,11 @@ class ControlFlowGraph(object):
if op.type() == "fill_constant" and op.attr("force_cpu") == True: if op.type() == "fill_constant" and op.attr("force_cpu") == True:
self._skip_opt.update(op.output_arg_names()) self._skip_opt.update(op.output_arg_names())
def release_memory(self): def release_memory(self, skip_opt_set=None):
self._dataflow_analyze() self._dataflow_analyze()
self._update_skip_opt_set() self._update_skip_opt_set()
if skip_opt_set:
self._skip_opt.update(skip_opt_set)
fwd_id = 0 fwd_id = 0
bwd_id = 0 bwd_id = 0
for i in range(self.op_size): for i in range(self.op_size):
...@@ -183,7 +185,7 @@ class ControlFlowGraph(object): ...@@ -183,7 +185,7 @@ class ControlFlowGraph(object):
else: else:
bwd_id += 1 bwd_id += 1
def memory_optimize(self, level=0): def memory_optimize(self, skip_opt_set=None, level=0):
def compare_shape(x_shape, cache_shape, opt_level): def compare_shape(x_shape, cache_shape, opt_level):
if opt_level == 0: if opt_level == 0:
return x_shape == cache_shape return x_shape == cache_shape
...@@ -200,6 +202,9 @@ class ControlFlowGraph(object): ...@@ -200,6 +202,9 @@ class ControlFlowGraph(object):
self._dataflow_analyze() self._dataflow_analyze()
self._update_skip_opt_set() self._update_skip_opt_set()
# update skip set to meet users' demand
if skip_opt_set:
self._skip_opt.update(skip_opt_set)
self.pool = [] self.pool = []
for i in range(self.op_size): for i in range(self.op_size):
op = self._ops[i] op = self._ops[i]
...@@ -358,7 +363,7 @@ def _get_cfgs(input_program): ...@@ -358,7 +363,7 @@ def _get_cfgs(input_program):
return cfgs return cfgs
def memory_optimize(input_program, print_log=False, level=0): def memory_optimize(input_program, skip_opt_set=None, print_log=False, level=0):
"""Optimize memory by reusing var memory. """Optimize memory by reusing var memory.
Note: it doesn't not support subblock nested in subblock. Note: it doesn't not support subblock nested in subblock.
...@@ -374,10 +379,10 @@ def memory_optimize(input_program, print_log=False, level=0): ...@@ -374,10 +379,10 @@ def memory_optimize(input_program, print_log=False, level=0):
PRINT_LOG = print_log PRINT_LOG = print_log
cfgs = _get_cfgs(input_program) cfgs = _get_cfgs(input_program)
for cfg in cfgs: for cfg in cfgs:
cfg.memory_optimize(level) cfg.memory_optimize(skip_opt_set=skip_opt_set, level=level)
def release_memory(input_program): def release_memory(input_program, skip_opt_set=None):
cfgs = _get_cfgs(input_program) cfgs = _get_cfgs(input_program)
for cfg in cfgs: for cfg in cfgs:
cfg.release_memory() cfg.release_memory(skip_opt_set=skip_opt_set)
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册