CNTK v2 library: Add support for executing eval before grad.

jeffinstance · May 16, 2017 · 9d19a25 · 9d19a25
1 parent be99d3e
commit 9d19a25
Show file tree

Hide file tree

Showing 5 changed files with 39 additions and 11 deletions.
diff --git a/Source/CNTKv2LibraryDll/API/CNTKLibrary.h b/Source/CNTKv2LibraryDll/API/CNTKLibrary.h
@@ -3294,11 +3294,6 @@ namespace CNTK
         CNTK_API static FunctionPtr Load(std::istream& inputStream, 
                                          const DeviceDescriptor& computeDevice = DeviceDescriptor::UseDefaultDevice());
 
-        ///
-        /// Prints the entire graph underlying this Function to stderr
-        ///
-        CNTK_API void PrintGraph() const;
-
         ///
         /// Returns a string representation of this Function
         ///

diff --git a/Source/CNTKv2LibraryDll/CompositeFunction.cpp b/Source/CNTKv2LibraryDll/CompositeFunction.cpp
@@ -1317,6 +1317,11 @@ namespace CNTK
                                                                    const std::unordered_set<Variable>& inputsToExcludeGradientsFor,
                                                                    bool allocateNetworkMatrices)
     {
+        // Lets purge the current computation network and regenerate the network if the CompositeFunction
+        // was previously compiled just for evaluation and not for gradient backpropagation.
+        if ((m_computationNetwork != nullptr) && (m_currentBackpropRoots.empty() && !backpropRoots.empty()))
+            PurgeComputationNetwork();
+
         if (m_computationNetwork != nullptr)
         {
             // TODO: We should either invalidate and readapt the network if the backpropRoots change compared to what was specified when the network

diff --git a/Source/CNTKv2LibraryDll/CompositeFunction.h b/Source/CNTKv2LibraryDll/CompositeFunction.h
@@ -338,6 +338,18 @@ namespace CNTK
             m_existingNetworkStorageReferences.clear();
         }
 
+        void PurgeComputationNetwork()
+        {
+            m_currentBackpropRoots.clear();
+            m_inputsExcludedFromGradientComputation.clear();
+            m_variableToNodeMap.clear();
+            m_currentOutputsToEvaluate.clear();
+            m_lastRecordedTimeStamps.clear();
+
+            m_networkMatricesAllocated = false;
+            m_computationNetwork = nullptr;
+        }
+
     private:
 
         // Set of all primitive functions in the graph underlying 'this' Function. Also keeps the primitive Function objects alive 

diff --git a/Source/CNTKv2LibraryDll/Function.cpp b/Source/CNTKv2LibraryDll/Function.cpp
@@ -925,12 +925,6 @@ namespace CNTK
         return CompositeFunction::Deserialize(modelDictionary, device);
     }
 
-    void Function::PrintGraph() const
-    {
-        CompositeFunction::PreorderTraverseFunctions(RootFunction(), [](const FunctionPtr& function) {
-        });
-    }
-
     std::wstring Function::AsString(bool doNotInferOutputs) const
     {
         wstringstream wss;

diff --git a/bindings/python/cntk/ops/tests/evaluation_test.py b/bindings/python/cntk/ops/tests/evaluation_test.py
@@ -85,3 +85,25 @@ def test_input_without_dynamic_axes():
     assert np.allclose(eval_result, [3.006, 2.992])
     assert np.allclose(grad_result, [.01, .01])
 
+
+def test_grad_after_eval():
+    x = C.input_variable((C.FreeDimension, 2))
+    w = C.parameter(init=np.asarray([[2, 5], [1, 3]], dtype=np.float32))
+    t = C.times(x, w)
+
+    x_data = np.asarray([[0.5, 0.2]], np.float32)
+    t_val = t.eval({x : x_data})
+    assert np.array_equal(t_val, np.asarray([[[1.2, 3.1]]], dtype=np.float32))
+
+    w_grad, t_val = t.grad({x : x_data}, wrt=[w], outputs=[t])
+    assert np.array_equal(t_val, np.asarray([[[1.2, 3.1]]], dtype=np.float32))
+    assert np.array_equal(w_grad, np.asarray([[0.5, .5], [.2, .2]], dtype=np.float32))
+
+    x_data = np.asarray([[0.5, 0.2], [0.1, .6]], np.float32)
+    t_val = t.eval({x : x_data})
+    assert np.allclose(t_val, np.asarray([[[1.2, 3.1], [0.8, 2.3]]], dtype=np.float32))
+
+    w_grad, t_val = t.grad({x : x_data}, wrt=[w], outputs=[t])
+    assert np.allclose(t_val, np.asarray([[[1.2, 3.1], [0.8, 2.3]]], dtype=np.float32))
+    assert np.array_equal(w_grad, np.asarray([[0.6, .6], [.8, .8]], dtype=np.float32))
+