Enforce drop_empty_grad=false When the input of an op is duplicable.

For input argument with a list of variables, drop_empty_grad is not allowed because it makes the correspondence bewteen a variable and its gradient ambiguous. Use REGISTER_OP_EX to register the op or call InputGrad(?,false) in GradOpDescMaker.

paddle/framework/grad_op_desc_maker.h

@@ -22,6 +22,14 @@
 namespace paddle {
 namespace framework {
 
+/*
+  This functor class is responsible for creating the gradient ops for the given
+  operator fwd_op. After it is called (through operator()), the pairs of
+  (gradient variable, corresponding input variable of fwd_op) will be added to
+  grad_to_var. If an input variable of fwd_op is contained in no_grad_set, its
+  gradient varialbe will be ignored or kEmptyVarName depending on the template
+  argument DropEmptyIG in the derived classes.
+ */
 class GradOpDescMakerBase {
  public:
   explicit GradOpDescMakerBase(
@@ -56,6 +64,16 @@ class GradOpDescMakerBase {
     if (!drop_empty_grad) {
       return ret_val;
     }
+    PADDLE_ENFORCE_LE(var_names.size(), 1UL,
+                      "BUG from operator developer:"
+                      " for input argument with a list of variables, "
+                      " drop_empty_grad is not allowed because it makes"
+                      " the correspondence bewteen a variable and its gradient"
+                      " ambiguous. Use REGISTER_OP_EX to register the op"
+                      " or call InputGrad(?,false) in GradOpDescMaker."
+                      " Op type %s",
+                      fwd_op_.Type());
+
     std::vector<std::string> dropped_ret_val;
     dropped_ret_val.reserve(ret_val.size());
     std::copy_if(ret_val.begin(), ret_val.end(),

paddle/framework/op_desc.h

@@ -127,7 +127,9 @@ class OpDesc {
   }
 
   proto::OpDesc desc_;
+  // input arg name => output variable names
   VariableNameMap inputs_;
+  // output arg name => output variable names
   VariableNameMap outputs_;
   AttributeMap attrs_;
 

paddle/framework/op_registry.h

@@ -126,6 +126,14 @@ class OpKernelRegistrar : public Registrar {
                              __test_global_namespace_##uniq_name##__>::value, \
                 msg)
 
+/*
+  The variadic arguments should be class types derived from one of the
+  following classes:
+    OpProtoAndCheckerMaker
+    GradOpDescMakerBase
+    VarTypeInference
+    InferShapeBase
+*/
 #define REGISTER_OPERATOR(op_type, op_class, ...)                      \
   STATIC_ASSERT_GLOBAL_NAMESPACE(                                      \
       __reg_op__##op_type,                                             \
@@ -144,20 +152,29 @@ class OpKernelRegistrar : public Registrar {
   }
 
 /**
- * Macro to register Operator.
+ * Macro to register Operator. When the input is duplicable, you should
+ * use REGISTER_OP_EX with deop_empty_grad=false instead.
  */
-#define REGISTER_OP(op_type, op_class, op_maker_class, grad_op_type,       \
-                    grad_op_class)                                         \
-  REGISTER_OPERATOR(grad_op_type, grad_op_class);                          \
-  class _GradOpDescMaker_##grad_op_type##_                                 \
-      : public ::paddle::framework::DefaultGradOpDescMaker<true> {         \
-    using ::paddle::framework::DefaultGradOpDescMaker<                     \
-        true>::DefaultGradOpDescMaker;                                     \
-                                                                           \
-   protected:                                                              \
-    virtual std::string GradOpType() const { return #grad_op_type; }       \
-  };                                                                       \
-  REGISTER_OPERATOR(op_type, op_class, _GradOpDescMaker_##grad_op_type##_, \
+#define REGISTER_OP(op_type, op_class, op_maker_class, grad_op_type, \
+                    grad_op_class)                                   \
+  REGISTER_OP_EX(op_type, op_class, op_maker_class, grad_op_type,    \
+                 grad_op_class, true)
+
+// When an argument is duplicable, we need to use this version.
+// Perhaps we can omit DropEmptyIG template parameter and
+// only have one version of REGISTER_OP.
+#define REGISTER_OP_EX(op_type, op_class, op_maker_class, grad_op_type,       \
+                       grad_op_class, drop_empty_grad)                        \
+  REGISTER_OPERATOR(grad_op_type, grad_op_class);                             \
+  class _GradOpDescMaker_##grad_op_type##_                                    \
+      : public ::paddle::framework::DefaultGradOpDescMaker<drop_empty_grad> { \
+    using ::paddle::framework::DefaultGradOpDescMaker<                        \
+        drop_empty_grad>::DefaultGradOpDescMaker;                             \
+                                                                              \
+   protected:                                                                 \
+    virtual std::string GradOpType() const { return #grad_op_type; }          \
+  };                                                                          \
+  REGISTER_OPERATOR(op_type, op_class, _GradOpDescMaker_##grad_op_type##_,    \
                     op_maker_class);
 
 #define REGISTER_OP_WITH_KERNEL(op_type, ...)                         \

paddle/operators/concat_op.cc

@@ -98,8 +98,8 @@ class ConcatOpGrad : public framework::OperatorWithKernel {
 }  // namespace paddle
 
 namespace ops = paddle::operators;
-REGISTER_OP(concat, ops::ConcatOp, ops::ConcatOpMaker, concat_grad,
-            ops::ConcatOpGrad)
+REGISTER_OP_EX(concat, ops::ConcatOp, ops::ConcatOpMaker, concat_grad,
+               ops::ConcatOpGrad, false)
 REGISTER_OP_CPU_KERNEL(concat,
                        ops::ConcatKernel<paddle::platform::CPUPlace, float>)
 REGISTER_OP_CPU_KERNEL(concat_grad,

paddle/operators/conditional_block_op.cc

@@ -178,8 +178,9 @@ class ConditionalBlockGradMaker : public framework::SingleGradOpDescMaker {
     grad_op->SetInput("Out", Output("Out"));
     grad_op->SetInput(framework::GradVarName("Out"), OutputGrad("Out"));
     grad_op->SetInput("Scope", Output("Scope"));
-    grad_op->SetOutput(framework::GradVarName("X"), InputGrad("X"));
-    grad_op->SetOutput(framework::GradVarName("Params"), InputGrad("Params"));
+    grad_op->SetOutput(framework::GradVarName("X"), InputGrad("X", false));
+    grad_op->SetOutput(framework::GradVarName("Params"),
+                       InputGrad("Params", false));
     grad_op->SetBlockAttr("sub_block", *this->grad_block_[0]);
     return std::unique_ptr<framework::OpDesc>(grad_op);
   }

paddle/operators/recurrent_op.cc

@@ -570,7 +570,7 @@ class RecurrentGradOpDescMaker : public framework::SingleGradOpDescMaker {
     for (auto &input_param : this->InputNames()) {
       grad->SetInput(input_param, this->Input(input_param));
       grad->SetOutput(framework::GradVarName(input_param),
-                      this->InputGrad(input_param));
+                      this->InputGrad(input_param, false));
     }
 
     for (auto &output_param : this->OutputNames()) {

paddle/operators/sequence_concat_op.cc

@@ -67,12 +67,12 @@ class SequenceConcatOpMaker : public framework::OpProtoAndCheckerMaker {
                  "The level should be less than the level number of inputs.")
         .SetDefault(0);
     AddComment(R"DOC(
-The sequence_concat operator concatenates multiple LoDTensors. 
-It only supports sequence (LoD Tensor with level number is 1) 
+The sequence_concat operator concatenates multiple LoDTensors.
+It only supports sequence (LoD Tensor with level number is 1)
 or a nested sequence (LoD tensor with level number is 2) as its input.
 - Case1:
   If the axis is other than 0(here, axis is 1 and level is 1),
-  each input should have the same LoD information and the LoD 
+  each input should have the same LoD information and the LoD
   information of the output keeps the same as the input.
 
   LoD(x0) = {{0,2,4}, {0,1,2,3,4}}; Dims(x0) = (4,3,4)
@@ -80,7 +80,7 @@ or a nested sequence (LoD tensor with level number is 2) as its input.
   LoD(Out) = {{0,2,4}, {0,1,2,3,4}}; Dims(Out) = (4,7,4)
 
 - Case2:
-  If the axis is 0(here, leve is 0), the inputs are concatenated along 
+  If the axis is 0(here, leve is 0), the inputs are concatenated along
   time steps, the LoD information of the output need to re-compute.
   The LoD information of level-1 should be same.
 
@@ -124,8 +124,9 @@ class SequenceConcatGradOp : public framework::OperatorWithKernel {
 }  // namespace paddle
 
 namespace ops = paddle::operators;
-REGISTER_OP(sequence_concat, ops::SequenceConcatOp, ops::SequenceConcatOpMaker,
-            sequence_concat_grad, ops::SequenceConcatGradOp);
+REGISTER_OP_EX(sequence_concat, ops::SequenceConcatOp,
+               ops::SequenceConcatOpMaker, sequence_concat_grad,
+               ops::SequenceConcatGradOp, false);
 REGISTER_OP_CPU_KERNEL(
     sequence_concat,
     ops::SequenceConcatOpKernel<paddle::platform::CPUDeviceContext, float>);

paddle/operators/sum_op.cc

@@ -106,8 +106,8 @@ class SumOpMaker : public framework::OpProtoAndCheckerMaker {
     AddComment(R"DOC(
 Sum operator.
 
-This operators sums the input tensors. All the inputs can carry the 
-LoD (Level of Details) information. However, the output only shares 
+This operators sums the input tensors. All the inputs can carry the
+LoD (Level of Details) information. However, the output only shares
 the LoD information with the first input.
 )DOC");
   }
@@ -170,7 +170,7 @@ class SumGradMaker : public framework::GradOpDescMakerBase {
   using framework::GradOpDescMakerBase::GradOpDescMakerBase;
 
   std::vector<std::unique_ptr<framework::OpDesc>> operator()() const override {
-    auto x_grads = InputGrad("X");
+    auto x_grads = InputGrad("X", false);
     std::vector<std::unique_ptr<framework::OpDesc>> grad_ops;
     grad_ops.reserve(x_grads.size());
     auto og = OutputGrad("Out");