[nnc] Prepare the interpreter backend for the future changes related to adding of...

* Add `getInputTensors` and `setOutputTensors` methods in place of the `var` method.
* Reformat the code according to coding style.

Signed-off-by: Sergei Barannikov <s.barannikov@samsung.com>

diff --git a/contrib/nnc/include/passes/interpreter/Interpreter.h b/contrib/nnc/include/passes/interpreter/Interpreter.h
index dcc499c..f70d56b 100644 (file)
--- a/contrib/nnc/include/passes/interpreter/Interpreter.h
+++ b/contrib/nnc/include/passes/interpreter/Interpreter.h
@@ -17,26 +17,26 @@
 #ifndef _NNC_BACKEND_INTERPRETER_CORE_INTERPRETER_
 #define _NNC_BACKEND_INTERPRETER_CORE_INTERPRETER_
 
-#include <functional>
-#include <map>
-#include <string>
-#include <unordered_map>
-#include <set>
-
 #include "core/modelIR/Visitor.h"
 #include "core/modelIR/Operation.h"
+#include "core/modelIR/TensorVariant.h"
+#include <string>
+#include <unordered_map>
+#include <vector>
 
-#include "core/modelIR/Tensor.h"
-
-namespace nnc
-{
-namespace mir
-{
+namespace nnc {
+namespace mir {
 
 class NNInterpreter : public IVisitor {
 public:
   explicit NNInterpreter() = default;
 
+  ~NNInterpreter() override = default;
+
+  void setInput(const std::string& name, const TensorVariant& data);
+
+  TensorVariant getResult(IODescriptor tensor);
+
   void visit(ops::BatchNormOp& op) override;
   void visit(ops::BiasAddOp& op) override;
   void visit(ops::CappedReluOp& op) override;
@@ -69,18 +69,20 @@ public:
   void visit(ops::TanhOp& op) override;
   void visit(ops::TransposeOp& op) override;
 
-  void setInput(const std::string &name, const TensorVariant& data);
-  TensorVariant getResult(IODescriptor tensor);
   void dump(Operation& op, bool all = false);
 
-  ~NNInterpreter() override = default;
-
 private:
-  std::vector<TensorVariant> &var(size_t id);
+  /// @brief Gets the computed inputs for the operation.
+  std::vector<std::reference_wrapper<const TensorVariant>> getInputTensors(const Operation& op);
 
-private:
-  std::map<size_t, std::vector<TensorVariant>> vars;
-  std::unordered_map<std::string, TensorVariant> data;
+  /// @brief Saves the computed outputs for the operation.
+  void setOutputTensors(const Operation& op, std::vector<TensorVariant>&& outputs);
+
+  /// @brief Mapping of graph named inputs to their values.
+  std::unordered_map<std::string, TensorVariant> _inputTensors;
+
+  /// @brief Mapping of operations to their computed results.
+  std::unordered_map<std::size_t, std::vector<TensorVariant>> _opResults;
 };
 
 } // namespace mir

diff --git a/contrib/nnc/passes/interpreter/Interpreter.cpp b/contrib/nnc/passes/interpreter/Interpreter.cpp
index fe54c50..80286d4 100644 (file)
--- a/contrib/nnc/passes/interpreter/Interpreter.cpp
+++ b/contrib/nnc/passes/interpreter/Interpreter.cpp
@@ -78,9 +78,19 @@ namespace nnc {
 
 using namespace nnc::mir;
 
-std::vector<TensorVariant> &NNInterpreter::var(size_t id) { return vars[id]; }
+std::vector<std::reference_wrapper<const TensorVariant>>
+NNInterpreter::getInputTensors(const Operation& op) {
+  std::vector<std::reference_wrapper<const TensorVariant>> tensors;
+  for (IODescriptor ir_tensor : op.getPrevNodes())
+    tensors.emplace_back(_opResults.at(ir_tensor.op->getId()).at(ir_tensor.index));
+  return tensors;
+}
+
+void NNInterpreter::setOutputTensors(const Operation& op, std::vector<TensorVariant>&& outputs) {
+  _opResults.emplace(op.getId(), std::move(outputs));
+}
 
-static void dumpIndex (Index ndx) {
+static void dumpIndex(Index ndx) {
   for (int i = 0; i < ndx.rank(); i++) {
     std::cout << (i ? "," : "(") << ndx.at(i);
   }
@@ -88,19 +98,19 @@ static void dumpIndex (Index ndx) {
 }
 
 #if(0)
-  #define DUMP(x, y) dump(x, (y))
+#define DUMP(x, y) dump(x, (y))
 #else
-  #define DUMP(x, y)
+#define DUMP(x, y)
 #endif
 
 void NNInterpreter::dump(Operation& op, bool all) {
   // TODO: in theory there could be several outputs from the given 'op'.
-  TensorVariant tensor = var(op.getId())[0];
+  TensorVariant tensor = _opResults.at(op.getId()).at(0);
   auto shape = tensor.getShape();
   std::cout << "Tensor '" <<
-               (op.getNextNodes().size() ? op.getNextNodes()[0]->getName() : "output") <<
-                "' DType = " << (int)tensor.getDataType()  << ", ElementSize = " <<
-                tensor.getElementSize() << ", Shape" << shape;
+            (op.getNextNodes().size() ? op.getNextNodes()[0]->getName() : "output") <<
+            "' DType = " << (int)tensor.getDataType() << ", ElementSize = " <<
+            tensor.getElementSize() << ", Shape" << shape;
   std::cout << " ElementsNumber " << shape.numElements() << "\n";
   static bool do_it = false;
   if (do_it || all) {
@@ -118,208 +128,183 @@ void NNInterpreter::dump(Operation& op, bool all) {
   }
 }
 
-void NNInterpreter::setInput(const std::string &name, const TensorVariant& t) {
-// TODO: our tests are failed with fe enable exception
-//  feenableexcept(FE_INVALID | FE_OVERFLOW);
-//  |
-//                 FE_DIVBYZERO |
-//                 FE_OVERFLOW  |
-//                 FE_UNDERFLOW);
-//  feenableexcept(FE_ALL_EXCEPT);
+void NNInterpreter::setInput(const std::string& name, const TensorVariant& t) {
+  _inputTensors.emplace(name, t);
+}
 
-  data.emplace(name, t);
+TensorVariant NNInterpreter::getResult(IODescriptor tensor) {
+  return _opResults.at(tensor.op->getId()).at(tensor.index);
 }
 
 void NNInterpreter::visit(ops::InputOp& op) {
-  (void)op;
-  auto it = data.find(op.getName());
-  if( it == data.end() )
-  {
+  auto it = _inputTensors.find(op.getName());
+  if (it == _inputTensors.end())
     throw std::runtime_error("Can't find data for node \"" + op.getName() +
                              ". Input data was not set correctly?");
-  }
-  var(op.getId()) = {it->second};
+  setOutputTensors(op, {it->second});
 }
 
 void NNInterpreter::visit(ops::ConstantOp& op) {
-  assert(data.find(op.getName()) == data.end());
-  var(op.getId()) = {op.getValue()};
-}
-
-TensorVariant NNInterpreter::getResult(IODescriptor tensor) {
-  return vars.at(tensor.op->getId()).at(tensor.index);
+  assert(_inputTensors.find(op.getName()) == _inputTensors.end());
+  setOutputTensors(op, {op.getValue()});
 }
 
 void NNInterpreter::visit(ops::ConcatOp& op) {
-  auto &operands = op.getPrevNodes();
-  std::vector<TensorVariant> ins;
-  for (auto &in : operands)
-  {
-    ins.push_back(var(in.op->getId())[in.index]);
-  }
-  var(op.getId()) = Concat<float>(ins, op.getOutputShape(0), op.getAxis())();
+  auto inputs = getInputTensors(op);
+  auto outputs = Concat<float>(inputs, op.getOutputShape(0), op.getAxis())();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::Conv2DOp& op) {
-  auto input = op.getPrevNodes()[0];
-  auto kernel = op.getPrevNodes()[1];
-  auto input_tensor = var(input.op->getId())[input.index];
-  auto kernel_tensor = var(kernel.op->getId())[kernel.index];
-  var(op.getId()) = Conv2D(input_tensor, kernel_tensor, op)();
+  auto inputs = getInputTensors(op);
+  auto outputs = Conv2D(inputs[0], inputs[1], op)();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, true);
 }
 
 void NNInterpreter::visit(ops::ReshapeOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  auto input = var(operand.op->getId())[operand.index];
-  var(op.getId()) = Reshape<float>(input, op.getOutputShape(0))();
+  auto inputs = getInputTensors(op);
+  auto outputs = Reshape<float>(inputs[0], op.getOutputShape(0))();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::ReluOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  Tensor<float> input(var(operand.op->getId())[operand.index]);
-  var(op.getId()) = Fill<float>(
-      op.getOutputShape(0), [&input](const Index &id) { return std::max(input.at(id), 0.0f); })();
+  auto inputs = getInputTensors(op);
+  Tensor<float> input(inputs[0]);
+  auto outputs = Fill<float>(op.getOutputShape(0),
+                             [&input](const Index& id) { return std::max(input.at(id), 0.0f); })();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::SigmoidOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  Tensor<float> input(var(operand.op->getId())[operand.index]);
-  var(op.getId()) = Fill<float>(op.getOutputShape(0), [&input](const Index& id) {
+  auto inputs = getInputTensors(op);
+  Tensor<float> input(inputs[0]);
+  auto outputs = Fill<float>(op.getOutputShape(0), [&input](const Index& id) {
     return 1.f / (1.f + std::exp(-input.at(id)));
   })();
+  setOutputTensors(op, std::move(outputs));
 }
 
 void NNInterpreter::visit(ops::SoftmaxOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  auto input = var(operand.op->getId())[operand.index];
-  var(op.getId()) = Softmax(op.getInputShape(0), input, op.getAxis())();
+  auto inputs = getInputTensors(op);
+  auto outputs = Softmax(op.getInputShape(0), inputs[0], op.getAxis())();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::PoolOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  auto input = var(operand.op->getId())[operand.index];
-  var(op.getId()) = Pool(input, op)();
+  auto inputs = getInputTensors(op);
+  auto outputs = Pool(inputs[0], op)();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::FullyConnectedOp& op) {
-  auto operand1 = op.getPrevNodes()[0];
-  auto operand2 = op.getPrevNodes()[1];
-  TensorVariant input1 = var(operand1.op->getId())[operand1.index];
-  TensorVariant input2 = var(operand2.op->getId())[operand2.index];
-  var(op.getId()) = FullyConnected(input1, input2, op)();
+  auto inputs = getInputTensors(op);
+  auto outputs = FullyConnected(inputs[0], inputs[1], op)();
+  setOutputTensors(op, std::move(outputs));
 }
 
 void NNInterpreter::visit(ops::GemmOp& op) {
-  auto operand_a = op.getPrevNodes()[0];
-  auto operand_b = op.getPrevNodes()[1];
-  auto operand_c = op.getPrevNodes()[2];
-  const TensorVariant input_a = var(operand_a.op->getId())[operand_a.index];
-  const TensorVariant input_b = var(operand_b.op->getId())[operand_b.index];
-  const TensorVariant input_c = var(operand_c.op->getId())[operand_c.index];
-  var(op.getId()) = Gemm<float>(input_a, input_b, input_c, op)();
+  auto inputs = getInputTensors(op);
+  auto outputs = Gemm<float>(inputs[0], inputs[1], inputs[2], op)();
+  setOutputTensors(op, std::move(outputs));
 }
 
 void NNInterpreter::visit(ops::CappedReluOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  Tensor<float> input(var(operand.op->getId())[operand.index]);
-  var(op.getId()) = Fill<float>(op.getOutputShape(0), [&input, &op](const Index &id) {
+  auto inputs = getInputTensors(op);
+  Tensor<float> input(inputs[0]);
+  auto outputs = Fill<float>(op.getOutputShape(0), [&input, &op](const Index& id) {
     return std::min(std::max(input.at(id), 0.0f), op.getCap());
   })();
+  setOutputTensors(op, std::move(outputs));
 }
 
-void NNInterpreter::visit(ops::DepthwiseConv2DOp& op){
-  auto input = op.getPrevNodes()[0];
-  auto kernel = op.getPrevNodes()[1];
-  auto input_tensor(var(input.op->getId())[input.index]);
-  auto kernel_tensor(var(kernel.op->getId())[kernel.index]);
-  var(op.getId()) = DepthwiseConv2D(input_tensor, kernel_tensor, op)();
+void NNInterpreter::visit(ops::DepthwiseConv2DOp& op) {
+  auto inputs = getInputTensors(op);
+  auto outputs = DepthwiseConv2D(inputs[0], inputs[1], op)();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, true);
 }
 
 void NNInterpreter::visit(ops::BiasAddOp& op) {
-  auto operand1 = op.getPrevNodes()[0];
-  auto operand2 = op.getPrevNodes()[1];
-  auto input1 = var(operand1.op->getId())[operand1.index];
-  auto input2 = var(operand2.op->getId())[operand2.index];
-  var(op.getId()) = BiasAdd(input1, input2)();
+  auto inputs = getInputTensors(op);
+  auto outputs = BiasAdd(inputs[0], inputs[1])();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::BatchNormOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  TensorVariant input(var(operand.op->getId())[operand.index]);
-  // TODO implement this
-    var(op.getId()) = BatchNorm<float>(input, op)();
+  auto inputs = getInputTensors(op);
+  auto outputs = BatchNorm<float>(inputs[0], op)();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::ScaleOp& op) {
-  auto operand1 = op.getPrevNodes()[0];
-  auto operand2 = op.getPrevNodes()[1];
-  auto input1 = var(operand1.op->getId())[operand1.index];
-  auto input2 = var(operand2.op->getId())[operand2.index];
-  var(op.getId()) = Scale(input1, input2)();
+  auto inputs = getInputTensors(op);
+  auto outputs = Scale(inputs[0], inputs[1])();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
-
 void NNInterpreter::visit(ops::SliceOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  auto input = Tensor<float>(var(operand.op->getId())[operand.index]);
-  var(op.getId()) = Fill<float>(op.getOutputShape(0), [&input, &op](const Index& id) {
+  auto inputs = getInputTensors(op);
+  auto input = Tensor<float>(inputs[0]);
+  auto outputs = Fill<float>(op.getOutputShape(0), [&input, &op](const Index& id) {
     Index idx = nnc::shift(id, op.getStarts());
     return input.at(idx);
   })();
+  setOutputTensors(op, std::move(outputs));
 }
 
 void NNInterpreter::visit(ops::DropoutOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  TensorVariant input(var(operand.op->getId())[operand.index]);
+  auto inputs = getInputTensors(op);
+  TensorVariant input(inputs[0]);
   // TODO implement this
-   var(op.getId()) = Dropout<float>(input, op)();
+  auto outputs = Dropout<float>(input, op)();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::TanhOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  Tensor<float> input(var(operand.op->getId())[operand.index]);
-  var(op.getId()) = Fill<float>(op.getOutputShape(0), [&input](const Index &id) {
+  auto inputs = getInputTensors(op);
+  Tensor<float> input(inputs[0]);
+  auto outputs = Fill<float>(op.getOutputShape(0), [&input](const Index& id) {
     return std::tanh(input.at(id));
   })();
+  setOutputTensors(op, std::move(outputs));
 }
 
 void NNInterpreter::visit(ops::ElementwiseOp& op) {
-  auto operands = op.getPrevNodes();
+  auto inputs = getInputTensors(op);
+
   std::vector<Tensor<float>> ins;
   // Reserve space for tensor variants to avoid reference invalidation when pushing into vector
   std::vector<TensorVariant> broadcasted{};
   broadcasted.reserve(op.getNumInputs());
 
-  for (auto &in : operands) {
-    auto& tmp = var(in.op->getId())[in.index];
+  for (auto in : inputs) {
     if (op.getBroadcast()) {
-      broadcasted.emplace_back(tmp, op.getOutputShape(0));
+      broadcasted.emplace_back(in, op.getOutputShape(0));
       ins.emplace_back(broadcasted.back());
     } else {
-      ins.emplace_back(tmp);
+      ins.emplace_back(in);
     }
   }
-  float (*func)(float,float); // Another dirty hack
+  float (* func)(float, float); // Another dirty hack
   switch (op.getOpType()) {
     case ops::ElementwiseOp::OpType::add:
       func = [](float a, float b) { return a + b; };
       break;
     case ops::ElementwiseOp::OpType::mul:
-      func = [](float a, float b) { return a * b;};
+      func = [](float a, float b) { return a * b; };
       break;
     case ops::ElementwiseOp::OpType::max:
-      func = [](float a, float b) { return std::max(a,b);};
+      func = [](float a, float b) { return std::max(a, b); };
       break;
     case ops::ElementwiseOp::OpType::div:
       func = [](float a, float b) { return a / b; };
@@ -330,136 +315,119 @@ void NNInterpreter::visit(ops::ElementwiseOp& op) {
     default:
       assert(false && "Unsupported Optype");
   }
-  var(op.getId()) = Fill<float>(
-    op.getOutputShape(0),
-    [&func, &ins](const Index& id) {
-      float acc = ins[0].at(id);
-      for (size_t i = 1; i < ins.size(); i++)
-        acc = func(acc, ins[i].at(id));
-      return acc;
-    })();
+  auto outputs = Fill<float>(op.getOutputShape(0), [&func, &ins](const Index& id) {
+    float acc = ins[0].at(id);
+    for (size_t i = 1; i < ins.size(); i++)
+      acc = func(acc, ins[i].at(id));
+    return acc;
+  })();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::DeConv2DOp& op) {
-  auto input = op.getPrevNodes()[0];
-  auto kernel = op.getPrevNodes()[1];
-  auto input_tensor = var(input.op->getId())[input.index];
-  auto kernel_tensor = var(kernel.op->getId())[kernel.index];
-  var(op.getId()) = DeConv2D(input_tensor, kernel_tensor, op)();
+  auto inputs = getInputTensors(op);
+  auto outputs = DeConv2D(inputs[0], inputs[1], op)();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::EluOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  Tensor<float> input(var(operand.op->getId())[operand.index]);
-  var(op.getId()) = Fill<float>(op.getOutputShape(0), [&input, &op](const Index &id) {
+  auto inputs = getInputTensors(op);
+  Tensor<float> input(inputs[0]);
+  auto outputs = Fill<float>(op.getOutputShape(0), [&input, &op](const Index& id) {
     if (input.at(id) >= 0)
       return input.at(id);
     else
-      return op.getAlpha()*(expf(input.at(id))-1);
+      return op.getAlpha() * (expf(input.at(id)) - 1);
   })();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::SqueezeOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  auto& input = var(operand.op->getId())[operand.index];
-  //Squeeze is just a special case of reshape
-  var(op.getId()) = Reshape<float>(input, op.getOutputShape(0))();
+  auto inputs = getInputTensors(op);
+  // Squeeze is just a special case of reshape.
+  auto outputs = Reshape<float>(inputs[0], op.getOutputShape(0))();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::PadOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  auto& input = var(operand.op->getId())[operand.index];
-  var(op.getId()) = Pad(input, op)();
+  auto inputs = getInputTensors(op);
+  auto outputs = Pad(inputs[0], op)();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::SqrtOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  auto input = Tensor<float>(var(operand.op->getId())[operand.index]);
-  var(op.getId()) = Fill<float>(op.getOutputShape(0), [&input](const Index id) {
+  auto inputs = getInputTensors(op);
+  Tensor<float> input(inputs[0]);
+  auto outputs = Fill<float>(op.getOutputShape(0), [&input](const Index id) {
     return sqrt(input.at(id));
   })();
+  setOutputTensors(op, std::move(outputs));
 }
 
 void NNInterpreter::visit(ops::ResizeOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  Tensor<float> input(var(operand.op->getId())[operand.index]);
-  switch (op.getMode()) {
-    case ops::ResizeOp::ResizeMethod::nearestNeighbor: {
-      auto scales = op.getScales();
-      var(op.getId()) = Fill<float>(op.getOutputShape(0), [&scales, &input](const Index& id) {
-        Index in_idx;
-        in_idx.resize(4);
-        for (int i = 0; i < input.getShape().rank(); i++) {
-          in_idx.at(i) = static_cast<int> (floorf(id.at(i) / scales[i]));
-        }
-        return input.at(in_idx);
-      })();
-      break;
+  auto inputs = getInputTensors(op);
+  Tensor<float> input(inputs[0]);
+  assert(op.getMode() == ops::ResizeOp::ResizeMethod::nearestNeighbor);
+  auto scales = op.getScales();
+  auto outputs = Fill<float>(op.getOutputShape(0), [&scales, &input](const Index& id) {
+    Index in_idx;
+    in_idx.resize(4);
+    for (int i = 0; i < input.getShape().rank(); i++) {
+      in_idx.at(i) = static_cast<int> (floorf(id.at(i) / scales[i]));
     }
-    default:
-      assert(false && "Not supported Optype");
-  }
+    return input.at(in_idx);
+  })();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 
 }
 
 void NNInterpreter::visit(ops::ReduceFOp& op) {
+  auto inputs = getInputTensors(op);
+
+  assert(op.getFuncType() == ops::ReduceFOp::FuncType::mean);
   // should always be an integer in a float
-  const float reduction_area =
-    static_cast<float>(op.getInputShape(0).numElements() / op.getOutputShape(0).numElements());
-
-  auto operand = op.getPrevNodes()[0];
-  auto& input = var(operand.op->getId())[operand.index];
-
-  std::function<float(float, float)> func;
-  switch (op.getFuncType()) {
-    case ops::ReduceFOp::FuncType::mean: {
-      func = [](float running_sum, float item) { return running_sum + item; };
-      var(op.getId()) = ReduceN<float>(op.getInputShape(0),
-                                       op.getOutputShape(0), input, op.getReductionDims(), func)();
-      Tensor<float> out_t = Tensor<float>(var(op.getId())[0]); // for numerical  stability
-      var(op.getId()) = Fill<float>(op.getOutputShape(0),
-                                    [&out_t, reduction_area](const Index& id) {
-                                      return out_t.at(id) / reduction_area;
-                                    })();
-    }
-      break;
-    default:
-      assert(false && "Not Implemented");
-  }
+  const float reduction_area = op.getInputShape(0).numElements() /
+                               op.getOutputShape(0).numElements();
+
+  auto tmp = ReduceN<float>(op.getInputShape(0), op.getOutputShape(0), inputs[0],
+                            op.getReductionDims(),
+                            [](float running_sum, float item) { return running_sum + item; })();
+  Tensor<float> out_t(tmp[0]); // for numerical  stability
+  auto outputs = Fill<float>(op.getOutputShape(0), [&out_t, reduction_area](const Index& id) {
+    return out_t.at(id) / reduction_area;
+  })();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::TransposeOp& op) {
-  auto operand = op.getPrevNodes()[0];
-  auto& input = var(operand.op->getId())[operand.index];
-  var(op.getId()) = Transpose(input, op)();
+  auto inputs = getInputTensors(op);
+  auto outputs = Transpose(inputs[0], op)();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 
 void NNInterpreter::visit(ops::GatherOp& op) {
-  auto data_descr = op.getPrevNodes()[0];
-  auto indices_descr = op.getPrevNodes()[1];
-  const auto& data = var(data_descr.op->getId())[data_descr.index];
-  const auto& indices = var(indices_descr.op->getId())[indices_descr.index];
-  var(op.getId()) = Gather(data, indices, op)();
+  auto inputs = getInputTensors(op);
+  auto outputs = Gather(inputs[0], inputs[1], op)();
+  setOutputTensors(op, std::move(outputs));
 }
 
 void NNInterpreter::visit(ops::LeakyReluOp& op) {
-  auto operand = op.getPrevNodes()[0];
+  auto inputs = getInputTensors(op);
   float alpha = op.getAlpha();
-  Tensor<float> input(var(operand.op->getId())[operand.index]);
-  var(op.getId()) = Fill<float>(
-    op.getOutputShape(0), [&input, alpha](const Index& id) {
-      float val = input.at(id);
-      return val > 0.0f ? val : val * alpha;
-    })();
-
+  Tensor<float> input(inputs[0]);
+  auto outputs = Fill<float>(op.getOutputShape(0), [&input, alpha](const Index& id) {
+    float val = input.at(id);
+    return val > 0.0f ? val : val * alpha;
+  })();
+  setOutputTensors(op, std::move(outputs));
   DUMP(op, false);
 }
 

diff --git a/contrib/nnc/passes/interpreter/ops/Concat.h b/contrib/nnc/passes/interpreter/ops/Concat.h
index 998d363..d87d088 100644 (file)
--- a/contrib/nnc/passes/interpreter/ops/Concat.h
+++ b/contrib/nnc/passes/interpreter/ops/Concat.h
@@ -25,16 +25,15 @@ namespace nnc
 template <typename T> class Concat : public Fill<T>
 {
 public:
-  explicit Concat(const std::vector<mir::TensorVariant> &inputs, const mir::Shape &outputShape,
-                  int32_t axis)
-      : Fill<T>(outputShape, getSingleFunction(inputs, axis))
-  {
-  }
+  Concat(const std::vector<std::reference_wrapper<const mir::TensorVariant>>& inputs,
+         const mir::Shape& outputShape,
+         int32_t axis)
+      : Fill<T>(outputShape, getSingleFunction(inputs, axis)) {}
 
 private:
-  const std::function<T(const mir::Index &)> getSingleFunction(const std::vector<mir::TensorVariant> &inputs,
-                                                               int32_t axis)
-  {
+  const std::function<T(const mir::Index&)>
+  getSingleFunction(const std::vector<std::reference_wrapper<const mir::TensorVariant>>& inputs,
+                    int32_t axis) {
     std::vector<mir::Tensor<T>> inputAccessors;
     for (auto &in : inputs)
     {