CNTK v2 library: Cleaned up python examples and added some inline doc…

…umentation
coderye · Sep 5, 2016 · 12a0220 · 12a0220
1 parent fb41fbd
commit 12a0220
Show file tree

Hide file tree

Showing 11 changed files with 305 additions and 1,880 deletions.
diff --git a/Source/CNTKv2LibraryDll/API/CNTKLibrary.h b/Source/CNTKv2LibraryDll/API/CNTKLibrary.h
@@ -892,7 +892,7 @@ namespace CNTK
 
     // Forward declarations
     inline Variable PlaceholderVariable(const NDShape& shape, const std::vector<Axis>& dynamicAxes = Axis::DefaultInputVariableDynamicAxes);
-    inline Variable InputVariable(const NDShape& shape, bool isSparse, CNTK::DataType dataType, bool needsGradient, const std::wstring& name = L"", const std::vector<Axis>& dynamicAxes = Axis::DefaultInputVariableDynamicAxes);
+    inline Variable InputVariable(const NDShape& shape, bool isSparse, CNTK::DataType dataType, bool needsGradient, const std::wstring& name, const std::vector<Axis>& dynamicAxes = Axis::DefaultInputVariableDynamicAxes);
     inline Variable OutputVariable(const NDShape& shape, CNTK::DataType dataType, Function* ownerFunction, const std::vector<Axis>& dynamicAxes, const std::wstring& name = L"");
 
     ///
@@ -916,10 +916,12 @@ namespace CNTK
                                     std::unordered_map<Variable, Variable>& placeholderReplacements,
                                     std::unordered_set<FunctionPtr>& allPrimitiveFunctions);
 
+#ifndef SWIG
     private:
         friend inline Variable PlaceholderVariable(const NDShape& shape, const std::vector<Axis>& dynamicAxes /*= Axis::DefaultInputVariableDynamicAxes*/);
-        friend inline Variable InputVariable(const NDShape& shape, bool isSparse, CNTK::DataType dataType, bool needsGradient, const std::wstring& name /*= L""*/, const std::vector<Axis>& dynamicAxes /*= Axis::DefaultInputVariableDynamicAxes*/);
+        friend inline Variable InputVariable(const NDShape& shape, bool isSparse, CNTK::DataType dataType, bool needsGradient, const std::wstring& name, const std::vector<Axis>& dynamicAxes /*= Axis::DefaultInputVariableDynamicAxes*/);
         friend inline Variable OutputVariable(const NDShape& shape, CNTK::DataType dataType, Function* ownerFunction, const std::vector<Axis>& dynamicAxes, const std::wstring& name /*= L""*/);
+#endif
 
     public:
 
@@ -1012,17 +1014,24 @@ namespace CNTK
         bool NeedsGradient() const { return m_dataFields->m_needsGradient; }
 
     protected:
+#ifdef SWIG
+    public:
+#endif
         Variable(const NDShape& shape, VariableKind varType, CNTK::DataType dataType, const NDArrayViewPtr& value, bool needsGradient, const std::vector<Axis>& dynamicAxes, const std::wstring& name, const std::wstring& uid)
             : Variable(shape, varType, dataType, nullptr, value, needsGradient, dynamicAxes, /*isSparse =*/ false, name, uid)
         {}
 
+    protected:
         NDArrayViewPtr Value() const
         {
             assert(m_dataFields->m_value != nullptr);
             return m_dataFields->m_value;
         }
 
     private:
+#ifdef SWIG
+    public:
+#endif
         Variable(const NDShape& shape, bool isSparse, CNTK::DataType dataType, bool needsGradient, const std::wstring& name, const std::vector<Axis>& dynamicAxes, const std::wstring& uid)
             : Variable(shape, VariableKind::Input, dataType, nullptr, nullptr, needsGradient, dynamicAxes, isSparse, name, uid)
         {}
@@ -1113,7 +1122,7 @@ namespace CNTK
     ///
     /// Create an 'Input' Variable.
     ///
-    inline Variable InputVariable(const NDShape& shape, DataType dataType, const std::wstring& name = L"", const std::vector<Axis>& dynamicAxes = Axis::DefaultInputVariableDynamicAxes)
+    inline Variable InputVariable(const NDShape& shape, DataType dataType, const std::wstring& name, const std::vector<Axis>& dynamicAxes = Axis::DefaultInputVariableDynamicAxes)
     {
         return InputVariable(shape, dataType, /*needsGradient =*/ false, name, dynamicAxes);
     }
@@ -1137,7 +1146,7 @@ namespace CNTK
     ///
     /// Create an 'Input' Variable denoting sparse data.
     ///
-    inline Variable InputVariable(const NDShape& shape, bool isSparse, CNTK::DataType dataType, const std::wstring& name = L"", const std::vector<Axis>& dynamicAxes = Axis::DefaultInputVariableDynamicAxes)
+    inline Variable InputVariable(const NDShape& shape, bool isSparse, CNTK::DataType dataType, const std::wstring& name, const std::vector<Axis>& dynamicAxes = Axis::DefaultInputVariableDynamicAxes)
     {
         return InputVariable(shape, isSparse, dataType, /*needsGradient =*/ false, name, dynamicAxes);
     }

diff --git a/Tests/UnitTests/V2LibraryTests/TrainerTests.cpp b/Tests/UnitTests/V2LibraryTests/TrainerTests.cpp
@@ -18,17 +18,18 @@ void TrainSimpleFeedForwardClassifer(const DeviceDescriptor& device)
     const size_t numSweepsToTrainWith = 2;
     const size_t numMinibatchesToTrain = (numSamplesPerSweep * numSweepsToTrainWith) / minibatchSize;
 
-    auto minibatchSource = TextFormatMinibatchSource(L"SimpleDataTrain_cntk_text.txt", { { L"features", inputDim }, { L"labels", numOutputClasses} }, 0);
-    auto streamInfos = minibatchSource->StreamInfos();
-    auto featureStreamInfo = std::find_if(streamInfos.begin(), streamInfos.end(), [](const StreamInformation& streamInfo) { return (streamInfo.m_name == L"features"); });
-    auto labelStreamInfo = std::find_if(streamInfos.begin(), streamInfos.end(), [](const StreamInformation& streamInfo) { return (streamInfo.m_name == L"labels"); });
+    auto featureStreamName = L"features";
+    auto labelsStreamName = L"labels";
+    auto minibatchSource = TextFormatMinibatchSource(L"SimpleDataTrain_cntk_text.txt", { { featureStreamName, inputDim }, { labelsStreamName, numOutputClasses } }, 0);
+    auto featureStreamInfo = minibatchSource->StreamInfo(featureStreamName);
+    auto labelStreamInfo = minibatchSource->StreamInfo(labelsStreamName);
 
-    std::unordered_map<StreamInformation, std::pair<NDArrayViewPtr, NDArrayViewPtr>> inputMeansAndInvStdDevs = { { *featureStreamInfo, { nullptr, nullptr } } };
+    std::unordered_map<StreamInformation, std::pair<NDArrayViewPtr, NDArrayViewPtr>> inputMeansAndInvStdDevs = { { featureStreamInfo, { nullptr, nullptr } } };
     ComputeInputPerDimMeansAndInvStdDevs(minibatchSource, inputMeansAndInvStdDevs);
 
     auto nonLinearity = std::bind(Sigmoid, _1, L"");
     auto input = InputVariable({ inputDim }, DataType::Float, L"features");
-    auto normalizedinput = PerDimMeanVarianceNormalize(input, inputMeansAndInvStdDevs[*featureStreamInfo].first, inputMeansAndInvStdDevs[*featureStreamInfo].second);
+    auto normalizedinput = PerDimMeanVarianceNormalize(input, inputMeansAndInvStdDevs[featureStreamInfo].first, inputMeansAndInvStdDevs[featureStreamInfo].second);
     auto classifierOutput = FullyConnectedDNNLayer(normalizedinput, hiddenLayerDim, device, nonLinearity);
     for (size_t i = 1; i < numHiddenLayers; ++i)
         classifierOutput = FullyConnectedDNNLayer(classifierOutput, hiddenLayerDim, device, nonLinearity);
@@ -61,7 +62,7 @@ void TrainSimpleFeedForwardClassifer(const DeviceDescriptor& device)
     for (size_t i = 0; i < numMinibatchesToTrain; ++i)
     {
         auto minibatchData = minibatchSource->GetNextMinibatch(minibatchSize, device);
-        trainer.TrainMinibatch({ { input, minibatchData[*featureStreamInfo].m_data }, { labels, minibatchData[*labelStreamInfo].m_data } }, device);
+        trainer.TrainMinibatch({ { input, minibatchData[featureStreamInfo].m_data }, { labels, minibatchData[labelStreamInfo].m_data } }, device);
         PrintTrainingProgress(trainer, i, outputFrequencyInMinibatches);
     }
 }