mindir

mindspore/ccsrc/backend/common/optimizer/common_backend_optimization.h

@@ -21,7 +21,7 @@ namespace mindspore {
 namespace opt {
 void BackendCommonOptimization(const std::shared_ptr<session::KernelGraph> &kernel_graph);
 void CommonFinalOptimization(const std::shared_ptr<session::KernelGraph> &kernel_graph);
-void CommonUnifyMindIR(const std::shared_ptr<session::KernelGraph> &kernel_graph);
+void CommonUnifyMindIR(const std::shared_ptr<session::KernelGraph> &kernel_graph);  // for debug
 void AddDynamicShapeAttrPass(const std::shared_ptr<session::KernelGraph> &kernel_graph);
 void EliminateIllegalDataTypePass(const std::shared_ptr<session::KernelGraph> &kernel_graph);
 void DynamicShapeConvertPass(const std::shared_ptr<session::KernelGraph> &kernel_graph);

mindspore/ccsrc/backend/common/session/ascend_session.h

@@ -47,7 +47,7 @@ class AscendSession : public SessionBasic {
 
   static void BatchBuildKernel(const std::vector<std::shared_ptr<SessionTask>> &build_tasks);
 
- protected:
+ protected: // load graph to device related
   void UnifyMindIR(const KernelGraphPtr &graph) override;
   GraphId CompileGraphImpl(const AnfNodePtrList &lst, const AnfNodePtrList &outputs) override;
   GraphId CompileGraphImpl(NotNull<FuncGraphPtr> func_graph) override;

mindspore/ccsrc/backend/common/session/cpu_session.h

@@ -31,7 +31,7 @@ class CPUSession : public SessionBasic {
   ~CPUSession() override = default;
   void Init(uint32_t device_id) override;
 
- protected:
+ protected: // load graph
   void UnifyMindIR(const KernelGraphPtr &graph) override { SessionBasic::UnifyMindIR(graph); }
   void CreateOutputTensors(const GraphId &graph_id, const std::vector<tensor::TensorPtr> &input_tensors, VectorRef *,
                            std::map<tensor::TensorPtr, session::KernelWithIndex> *tensor_to_node,

mindspore/ccsrc/backend/common/session/session_basic.h

@@ -228,7 +228,7 @@ class BACKEND_EXPORT SessionBasic : public std::enable_shared_from_this<SessionB
   // When the device address of the node is used as the output of the graph, the device address will be passed
   // to the output tensor, and the output node will recreate a new device address. This third parameter records
   // the relationship between the new and old device address.
-  virtual void UpdateOutputTensors(const VectorRef *outputs,
+  virtual void UpdateOutputTensors(const VectorRef *outputs,  // the output of graph
                                    const std::map<tensor::TensorPtr, session::KernelWithIndex> &tensor_to_node,
                                    std::map<DeviceAddressPtr, DeviceAddressPtr> *);
   virtual void UnifyMindIR(const KernelGraphPtr &graph);

mindspore/lite/tools/converter/import/mindspore_importer.cc

@@ -42,7 +42,7 @@ constexpr size_t kDependInputNum = 3;
 constexpr size_t kDependFirstInputIdx = 1;
 constexpr size_t kTupleGetItemFirstInputIdx = 1;
 }  // namespace
-STATUS MindsporeImporter::Mindir2AnfAdjust(const FuncGraphPtr &func_graph, const converter::Flags &flag) {
+STATUS MindsporeImporter::Mindir2AnfAdjust(const FuncGraphPtr &func_graph, const converter::Flags &flag) {  // fmk: Ms
   MS_ASSERT(func_graph != nullptr);
   auto primitive_adjust_pass = std::make_shared<PrimitiveAdjust>();
   MS_CHECK_TRUE_MSG(primitive_adjust_pass != nullptr, RET_NULL_PTR, "primitive_adjust_pass is nullptr.");

mindspore/lite/tools/converter/micro/coder/generator/component/common_component.cc

@@ -77,7 +77,7 @@ void CodeMSModelBuild(std::ofstream &ofs, const Configurator *config) {
        "    return kMSStatusLiteNotSupport;\n"
        "  }\n";
   ofs << "  int ret = RET_OK;\n";
-  if (config->target() != kARM32M) {
+  if (config->target() != kARM32M) {  // only support ARM32M
     ofs << "  ret = Init((void*)model_data, data_size);\n";
   }
   if (config->support_parallel()) {

mindspore/python/mindspore/train/_utils.py

@@ -23,7 +23,7 @@ from mindspore.common.dtype import dtype_to_nptype, pytype_to_dtype
 from mindspore.common import dtype as mstype
 from mindspore import log as logger
 from mindspore.common.api import _cell_graph_executor
-from mindspore.train.mind_ir_pb2 import ModelProto as mindir_model
+from mindspore.train.mind_ir_pb2 import ModelProto as mindir_model  
 from mindspore.train.checkpoint_pb2 import Checkpoint
 from mindspore.train.node_strategy_pb2 import ParallelStrategyMap as ckpt_strategy
 
@@ -42,7 +42,7 @@ def _convert_type(types):
     Returns:
         list, list of element in dataset.
     """
-    ms_types = []
+    ms_types = []   
     for np_type in types:
         ms_type = pytype_to_dtype(np_type)
         ms_types.append(ms_type)