[graph] Add node the graph of layers

Create class for the graph of layers which will represent nodes
This class also represents the object which will be created with APIs from now.
The actual class object resides inside this layer node object.

Properties for this layer node will be intercepted by the layer node object,
and properties of the graph node will be set in the layer node.
Others will be passed down to the layer node.

See Aso #986

**Self evaluation:**
1. Build test: [x]Passed [ ]Failed [ ]Skipped
2. Run test: [x]Passed [ ]Failed [ ]Skipped

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>

diff --git a/nntrainer/layers/layer_internal.h b/nntrainer/layers/layer_internal.h
index eef9ac3a98d253cde764441d3214c7f34df55297..b3e9489f9a4f38dde2fb02aa388e43160c018053 100644 (file)
--- a/nntrainer/layers/layer_internal.h
+++ b/nntrainer/layers/layer_internal.h
@@ -515,6 +515,7 @@ protected:
    */
   float loss;
 
+  // TODO: remove this from here
   ActivationType activation_type;
 
   WeightRegularizer weight_regularizer;
@@ -525,6 +526,7 @@ protected:
 
   WeightInitializer bias_initializer; /** initializer for bias */
 
+  // TODO: remove this from here
   /**
    * @brief   Output of this layer should be flattened
    */

diff --git a/nntrainer/layers/layer_node.h b/nntrainer/layers/layer_node.h
new file mode 100644 (file)
index 0000000..cc2be3a
--- /dev/null
+++ b/nntrainer/layers/layer_node.h
@@ -0,0 +1,122 @@
+// SPDX-License-Identifier: Apache-2.0
+/**
+ * Copyright (C) 2021 Parichay Kapoor <pk.kapoor@samsung.com>
+ *
+ * @file       graph_node.h
+ * @date       1 April 2021
+ * @see                https://github.com/nnstreamer/nntrainer
+ * @author     Parichay Kapoor <pk.kapoor@samsung.com>
+ * @bug                No known bugs except for NYI items
+ * @brief      This is the graph node interface for c++ API
+ */
+
+#ifndef __LAYER_NODE_H__
+#define __LAYER_NODE_H__
+
+#include <graph_node.h>
+#include <layer.h>
+#include <layer_internal.h>
+
+namespace nntrainer {
+
+/**
+ * @class   LayerNode class
+ * @brief   layer node class for the graph
+ */
+class LayerNode : public ml::train::Layer, public GraphNode {
+public:
+  /**
+   * @brief     Constructor of LayerNode class
+   *
+   */
+  LayerNode(std::shared_ptr<nntrainer::Layer> l, size_t idx) :
+    layer(l),
+    index(idx) {}
+
+  /**
+   * @brief     Destructor of LayerNode Class
+   *
+   */
+  ~LayerNode() = default;
+
+  /**
+   * Support all the interface requirements by ml::train::Layer
+   */
+
+  /**
+   * @brief Get the layer type
+   *
+   * @return const std::string type representation
+   */
+  const std::string getType() const { return layer->getType(); }
+
+  /**
+   * @brief     set Property of layer
+   *
+   * @param[in] properties values of property
+   * @retval #ML_ERROR_NONE Successful.
+   * @retval #ML_ERROR_INVALID_PARAMETER invalid parameter.
+   * @details   This function accepts vector of properties in the format -
+   *  { std::string property_name, void * property_val, ...}
+   */
+  int setProperty(std::vector<std::string> properties) {
+    return layer->setProperty(properties);
+  }
+
+  /**
+   * @brief     Get name of the layer
+   *
+   * @retval    name of the layer
+   * @note      This name is unique to this layer in a model
+   * @Note      This name might be changed once this layer is added to the model
+   * to keep the name unique to the model
+   */
+  std::string getName() noexcept { return layer->getName(); }
+
+  /**
+   * Support all the interface requirements by GraphNode<nntrainer::Layer>
+   */
+
+  /**
+   * @brief     Get underlying object
+   *
+   */
+  std::shared_ptr<nntrainer::Layer> &getObject() { return layer; }
+
+  /**
+   * @brief     Get underlying object
+   *
+   */
+  const std::shared_ptr<nntrainer::Layer> &getObject() const { return layer; }
+
+  /**
+   * @brief     Get index of the node
+   *
+   */
+  size_t getIndex() { return index; }
+
+  /**
+   * @brief     Get the trainable property of the underlying object
+   *
+   * @return boolean true if trainable, else false
+   */
+  bool getTrainable() noexcept { return layer->getTrainable(); }
+
+#ifdef PROFILE
+  int event_key;
+#endif
+
+private:
+  std::shared_ptr<nntrainer::Layer>
+    layer;      /**< The actual object in the graph node */
+  size_t index; /**< index of each node */
+
+  std::vector<std::string> input_layers;  /**< input layer names */
+  std::vector<std::string> output_layers; /**< output layer names */
+  bool flatten; /**< flatten the output of this node */
+  ActivationType
+    activation_type; /**< activation applied to the output of this node */
+};
+
+} // namespace nntrainer
+#endif // __LAYER_NODE_H__