Huawei_Technology
/
mindspore
Not watched
1
0
Fork 0
Code
Releases 13 Wiki evaluate Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain HPC
Browse Source

delete nodetype in model.fbs

tags/v1.2.0-rc1
zhaodezan 4 years ago
parent
81aeaf2546
commit
2ca6b7fbde
30 changed files with 167 additions and 168 deletions
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +0
    -1
      mindspore/lite/include/lite_utils.h
  2. +1
    -0
      mindspore/lite/include/model.h
  3. +2
    -2
      mindspore/lite/micro/coder/graph.cc
  4. +2
    -8
      mindspore/lite/schema/model.fbs
  5. +7
    -1
      mindspore/lite/src/common/utils.h
  6. +0
    -1
      mindspore/lite/src/lite_model.h
  7. +2
    -2
      mindspore/lite/src/ops/compat/attr_transfer_common.cc
  8. +3
    -2
      mindspore/lite/src/tensor.h
  9. +1
    -1
      mindspore/lite/test/common/import_from_meta_graphT.cc
  10. +18
    -18
      mindspore/lite/test/st/control_flow_test.cc
  11. +16
    -16
      mindspore/lite/test/st/sub_graph_test.cc
  12. +9
    -9
      mindspore/lite/test/ut/src/infer_test.cc
  13. +15
    -15
      mindspore/lite/test/ut/src/runtime/kernel/arm/fp32_grad/network_test.cc
  14. +0
    -1
      mindspore/lite/test/ut/src/runtime/kernel/arm/int8/deconv_int8_tests.cc
  15. +0
    -1
      mindspore/lite/test/ut/src/runtime/kernel/arm/int8/pad_int8_tests.cc
  16. +19
    -19
      mindspore/lite/test/ut/src/scheduler_test.cc
  17. +17
    -17
      mindspore/lite/test/ut/tools/optimizer/fusion/constant_folding_fusion_test.cc
  18. +4
    -4
      mindspore/lite/test/ut/tools/optimizer/fusion/conv_activation_fusion_test.cc
  19. +5
    -5
      mindspore/lite/test/ut/tools/optimizer/fusion/conv_biasadd_fusion_test.cc
  20. +15
    -15
      mindspore/lite/test/ut/tools/optimizer/fusion/conv_bn_fusion_test.cc
  21. +7
    -7
      mindspore/lite/test/ut/tools/optimizer/fusion/conv_scale_fusion_test.cc
  22. +10
    -11
      mindspore/lite/tools/anf_exporter/anf_exporter.cc
  23. +2
    -2
      mindspore/lite/tools/common/graph_util.cc
  24. +3
    -2
      mindspore/lite/tools/common/tensor_util.h
  25. +1
    -1
      mindspore/lite/tools/converter/legacy_optimizer/fusion/matmul_biasadd_fusion_pass.cc
  26. +2
    -2
      mindspore/lite/tools/converter/legacy_optimizer/fusion/mul_add_fusion_pass.cc
  27. +2
    -2
      mindspore/lite/tools/converter/legacy_optimizer/graph/batchnorm_convert_scale_pass.cc
  28. +1
    -1
      mindspore/lite/tools/converter/legacy_optimizer/graph/switch_pass.cc
  29. +1
    -1
      mindspore/lite/tools/converter/legacy_optimizer/graph/topological_sort_pass.cc
  30. +2
    -1
      mindspore/lite/tools/converter/parser/caffe/caffe_node_parser.cc

+ 0
- 1
mindspore/lite/include/lite_utils.h View File

@@ -38,7 +38,6 @@ using TensorPtrVector = std::vector<mindspore::schema::Tensor *>;
using DeviceContextVector = std::vector<DeviceContext>;
using Uint32Vector = std::vector<uint32_t>;
using String = std::string;
using NodeType = int; /**< 0 : NodeType_ValueNode, 1 : NodeType_Parameter, 2 : NodeType_CNode. */
using AllocatorPtr = std::shared_ptr<Allocator>;

/// \brief Set data of MSTensor from string vector.


+ 1
- 0
mindspore/lite/include/model.h View File

@@ -17,6 +17,7 @@
#define MINDSPORE_LITE_INCLUDE_MODEL_H_
#include <vector>
#include "include/lite_utils.h"
#include "src/common/utils.h"

namespace mindspore::lite {
struct MS_API Model {


+ 2
- 2
mindspore/lite/micro/coder/graph.cc View File

@@ -59,7 +59,7 @@ int CoderGraph::ConvertTensors() {
MS_CHECK_PTR_WITH_EXE(origin_tensor, clear_tensors());
// tensor dims
std::vector<int> shape;
if (origin_tensor->nodeType() == schema::NodeType_ValueNode) {
if (origin_tensor->nodeType() == NodeType_ValueNode) {
MS_CHECK_PTR_WITH_EXE(origin_tensor->dims(), clear_tensors());
for (uint32_t j = 0; j < origin_tensor->dims()->size(); j++) {
MS_CHECK_PTR(origin_tensor->dims()->data());
@@ -73,7 +73,7 @@ int CoderGraph::ConvertTensors() {
Tensor *dstTensor = new (std::nothrow)
lite::Tensor(TypeId(origin_data_type), shape, origin_tensor->format(), TensorCategory(origin_tensor));
MS_CHECK_PTR(dstTensor);
if (origin_tensor->nodeType() == schema::NodeType_ValueNode && origin_tensor->data() != nullptr &&
if (origin_tensor->nodeType() == NodeType_ValueNode && origin_tensor->data() != nullptr &&
origin_tensor->data()->size() > 0) {
if (shape.empty()) {
shape.push_back(1);


+ 2
- 8
mindspore/lite/schema/model.fbs View File

@@ -23,12 +23,6 @@ file_identifier "MSL2";
// File extension of any written files.
file_extension "ms";

enum NodeType: int {
ValueNode, // const
Parameter, // var
CNode // op
}

table QuantParam {
scale: double;
zeroPoint: int;
@@ -45,7 +39,7 @@ table QuantParam {
}

table Tensor {
nodeType: NodeType;
nodeType: int;
// data type
dataType: int;
// shape
@@ -73,7 +67,7 @@ table Primitive {

table CNode {
name: string;
nodeType: NodeType = CNode;
nodeType: int (deprecated);
primitive: Primitive;
inputIndex: [uint];
outputIndex: [uint];


+ 7
- 1
mindspore/lite/src/common/utils.h View File

@@ -29,6 +29,12 @@

namespace mindspore {
namespace lite {
enum NodeType {
NodeType_ValueNode, // const
NodeType_Parameter, // var
NodeType_CNode // op
};

const int USEC = 1000000;
const int MSEC = 1000;
std::vector<std::string> StringSplit(std::string str, const std::string &pattern);
@@ -154,7 +160,7 @@ std::vector<std::string> Tokenize(const std::string &src, const std::string &del

enum RemoveSubStrMode { PREFIX, SUFFIX, ANY };

// remove redundant charactor
// remove redundant character
std::string RemoveSubStr(const std::string &from, const std::string &sub_str, RemoveSubStrMode mode = ANY);

template <typename T>


+ 0
- 1
mindspore/lite/src/lite_model.h View File

@@ -66,7 +66,6 @@ class LiteModel : public Model {
node->primitive_ = c_node->primitive();
node->quant_type_ = c_node->quantType();
node->name_ = c_node->name()->c_str();
node->node_type_ = static_cast<NodeType>(c_node->nodeType());
auto count = c_node->inputIndex()->size();
for (uint32_t j = 0; j < count; ++j) {
node->input_indices_.push_back(size_t(c_node->inputIndex()->template GetAs<uint32_t>(j)));


+ 2
- 2
mindspore/lite/src/ops/compat/attr_transfer_common.cc View File

@@ -39,8 +39,8 @@ schema::Tensor *AttrToTensor(void *data, int data_size, bool is_array, TypeId ty
memcpy(uint8_data.data(), data, dst_tensor->Size());
auto shape = dst_tensor->shape();
flatbuffers::FlatBufferBuilder fbb(1024);
auto tensor_offset = schema::CreateTensorDirect(fbb, schema::NodeType_ValueNode, type_id, &shape, schema::Format_NHWC,
0, 0, &uint8_data);
auto tensor_offset =
schema::CreateTensorDirect(fbb, NodeType_ValueNode, type_id, &shape, schema::Format_NHWC, 0, 0, &uint8_data);
fbb.Finish(tensor_offset);
delete dst_tensor;
auto buf = fbb.GetBufferPointer();


+ 3
- 2
mindspore/lite/src/tensor.h View File

@@ -27,6 +27,7 @@

#include "src/common/log_adapter.h"
#include "schema/model_generated.h"
#include "src/common/utils.h"

namespace mindspore {
namespace lite {
@@ -246,9 +247,9 @@ inline size_t DataTypeSize(const TypeId type) {
}
}

inline Tensor::Category TensorCategory(const schema::NodeType node_type, const size_t shape_num, const TypeId data_type,
inline Tensor::Category TensorCategory(const int node_type, const size_t shape_num, const TypeId data_type,
const size_t data_size) {
return (node_type == schema::NodeType::NodeType_ValueNode)
return (node_type == NodeType_ValueNode)
? (shape_num == 0 && data_size == DataTypeSize(data_type) ? Tensor::Category::CONST_SCALAR
: Tensor::Category::CONST_TENSOR)
: Tensor::Category::VAR;


+ 1
- 1
mindspore/lite/test/common/import_from_meta_graphT.cc View File

@@ -42,7 +42,7 @@ int AnfImporterFromMetaGraphT::ConverterConstTensor() {
for (size_t i = 0; i < meta_graph_->allTensors.size(); i++) {
auto &tensor = meta_graph_->allTensors.at(i);
MS_ASSERT(tensor != nullptr);
if (tensor->nodeType != schema::NodeType::NodeType_ValueNode) {
if (tensor->nodeType != NodeType_ValueNode) {
continue;
}
auto parameter = func_graph_->add_parameter();


+ 18
- 18
mindspore/lite/test/st/control_flow_test.cc View File

@@ -221,7 +221,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {
// ------- tensor ---------
// tensor: 0 before-add input0 <main graph input>
auto tensor_0 = std::make_unique<schema::TensorT>();
tensor_0->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_0->nodeType = lite::NodeType_ValueNode;
tensor_0->format = schema::Format_NHWC;
tensor_0->dataType = TypeId::kNumberTypeFloat32;
tensor_0->dims = {1};
@@ -231,7 +231,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor: 1 before-add input1 <const>
auto tensor_1 = std::make_unique<schema::TensorT>();
tensor_1->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_1->nodeType = lite::NodeType_ValueNode;
tensor_1->format = schema::Format_NHWC;
tensor_1->dataType = TypeId::kNumberTypeFloat32;
tensor_1->dims = {1};
@@ -244,7 +244,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor: 2 before-add output/partial input
auto tensor_2 = std::make_unique<schema::TensorT>();
tensor_2->nodeType = schema::NodeType::NodeType_Parameter;
tensor_2->nodeType = lite::NodeType_Parameter;
tensor_2->format = schema::Format_NHWC;
tensor_2->dataType = TypeId::kNumberTypeFloat32;
tensor_2->dims = {1};
@@ -254,7 +254,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor: 3 before-add input1 <const>
auto tensor_3 = std::make_unique<schema::TensorT>();
tensor_3->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_3->nodeType = lite::NodeType_ValueNode;
tensor_3->format = schema::Format_NHWC;
tensor_3->dataType = TypeId::kNumberTypeFloat32;
tensor_3->dims = {1};
@@ -266,7 +266,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {
MS_LOG(DEBUG) << "tensor 3";

auto tensor_4 = std::make_unique<schema::TensorT>();
tensor_4->nodeType = schema::NodeType::NodeType_Parameter;
tensor_4->nodeType = lite::NodeType_Parameter;
tensor_4->format = schema::Format_NHWC;
tensor_4->dataType = TypeId::kNumberTypeFloat32;
tensor_4->dims = {1};
@@ -276,7 +276,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor :5 partial output <bool>
auto tensor_5 = std::make_unique<schema::TensorT>();
tensor_5->nodeType = schema::NodeType::NodeType_Parameter;
tensor_5->nodeType = lite::NodeType_Parameter;
tensor_5->format = schema::Format_NHWC;
tensor_5->dataType = TypeId::kNumberTypeBool;
tensor_5->dims = {1};
@@ -286,7 +286,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor: 6 switch true output
auto tensor_6 = std::make_unique<schema::TensorT>();
tensor_6->nodeType = schema::NodeType::NodeType_Parameter;
tensor_6->nodeType = lite::NodeType_Parameter;
tensor_6->format = schema::Format_NHWC;
tensor_6->dataType = TypeId::kNumberTypeFloat32;
tensor_6->dims = {1};
@@ -296,7 +296,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor: 5 switch False output
auto tensor_7 = std::make_unique<schema::TensorT>();
tensor_7->nodeType = schema::NodeType::NodeType_Parameter;
tensor_7->nodeType = lite::NodeType_Parameter;
tensor_7->format = schema::Format_NHWC;
tensor_7->dataType = TypeId::kNumberTypeFloat32;
tensor_7->dims = {1};
@@ -306,7 +306,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor: 6 body-add input ,other input is switch true output
auto tensor_8 = std::make_unique<schema::TensorT>();
tensor_8->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_8->nodeType = lite::NodeType_ValueNode;
tensor_8->format = schema::Format_NHWC;
tensor_8->dataType = TypeId::kNumberTypeFloat32;
tensor_8->dims = {1};
@@ -318,7 +318,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {
MS_LOG(DEBUG) << "tensor_8";

auto tensor_9 = std::make_unique<schema::TensorT>();
tensor_9->nodeType = schema::NodeType::NodeType_Parameter;
tensor_9->nodeType = lite::NodeType_Parameter;
tensor_9->format = schema::Format_NHWC;
tensor_9->dataType = TypeId::kNumberTypeFloat32;
tensor_9->dims = {1};
@@ -328,7 +328,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor: 7 after-add input ,other input is switch false output
auto tensor_10 = std::make_unique<schema::TensorT>();
tensor_10->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_10->nodeType = lite::NodeType_ValueNode;
tensor_10->format = schema::Format_NHWC;
tensor_10->dataType = TypeId::kNumberTypeFloat32;
tensor_10->dims = {1};
@@ -341,7 +341,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor: 8 main graph output
auto tensor_11 = std::make_unique<schema::TensorT>();
tensor_11->nodeType = schema::NodeType::NodeType_Parameter;
tensor_11->nodeType = lite::NodeType_Parameter;
tensor_11->format = schema::Format_NHWC;
tensor_11->dataType = TypeId::kNumberTypeFloat32;
tensor_11->dims = {1};
@@ -351,7 +351,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {

// tensor: 9 cond-Less input, other input is tensor 2
auto tensor_12 = std::make_unique<schema::TensorT>();
tensor_12->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_12->nodeType = lite::NodeType_ValueNode;
tensor_12->format = schema::Format_NHWC;
tensor_12->dataType = TypeId::kNumberTypeFloat32;
tensor_12->dims = {1};
@@ -363,7 +363,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {
MS_LOG(DEBUG) << "tensor_12";

auto tensor_13 = std::make_unique<schema::TensorT>();
tensor_13->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_13->nodeType = lite::NodeType_ValueNode;
tensor_13->format = schema::Format_NHWC;
tensor_13->dataType = TypeId::kNumberTypeFloat32;
tensor_13->dims = {1};
@@ -375,7 +375,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {
MS_LOG(DEBUG) << "tensor_13";

auto tensor_14 = std::make_unique<schema::TensorT>();
tensor_14->nodeType = schema::NodeType::NodeType_Parameter;
tensor_14->nodeType = lite::NodeType_Parameter;
tensor_14->format = schema::Format_NHWC;
tensor_14->dataType = TypeId::kNumberTypeFloat32;
tensor_14->dims = {1};
@@ -384,7 +384,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {
MS_LOG(DEBUG) << "tensor 14";

auto tensor_15 = std::make_unique<schema::TensorT>();
tensor_15->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_15->nodeType = lite::NodeType_ValueNode;
tensor_15->format = schema::Format_NHWC;
tensor_15->dataType = TypeId::kNumberTypeFloat32;
tensor_15->dims = {1};
@@ -396,7 +396,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {
MS_LOG(DEBUG) << "tensor_15";

auto tensor_16 = std::make_unique<schema::TensorT>();
tensor_16->nodeType = schema::NodeType::NodeType_Parameter;
tensor_16->nodeType = lite::NodeType_Parameter;
tensor_16->format = schema::Format_NHWC;
tensor_16->dataType = TypeId::kNumberTypeFloat32;
tensor_16->dims = {1};
@@ -405,7 +405,7 @@ TEST_F(ControlFlowTest, TestMergeWhileModel) {
MS_LOG(DEBUG) << "tensor_16";

auto tensor_17 = std::make_unique<schema::TensorT>();
tensor_17->nodeType = schema::NodeType::NodeType_Parameter;
tensor_17->nodeType = lite::NodeType_Parameter;
tensor_17->format = schema::Format_NHWC;
tensor_17->dataType = TypeId::kNumberTypeFloat32;
tensor_17->dims = {1};


+ 16
- 16
mindspore/lite/test/st/sub_graph_test.cc View File

@@ -50,12 +50,12 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
add_0->primitive->value.value = add_0_prim;
add_0->name = "Add0";
auto tensor_0 = std::make_unique<schema::TensorT>();
tensor_0->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_0->nodeType = lite::NodeType_ValueNode;
tensor_0->format = schema::Format_NHWC;
tensor_0->dataType = TypeId::kNumberTypeFloat32;
tensor_0->dims = {1};
auto tensor_1 = std::make_unique<schema::TensorT>();
tensor_1->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_1->nodeType = lite::NodeType_ValueNode;
tensor_1->format = schema::Format_NHWC;
tensor_1->dataType = TypeId::kNumberTypeFloat32;
tensor_1->dims = {1};
@@ -64,7 +64,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
ASSERT_NE(data1, nullptr);
data1[0] = 1;
auto tensor_2 = std::make_unique<schema::TensorT>();
tensor_2->nodeType = schema::NodeType::NodeType_Parameter;
tensor_2->nodeType = lite::NodeType_Parameter;
tensor_2->format = schema::Format_NHWC;
tensor_2->dataType = TypeId::kNumberTypeFloat32;
meta_graph->nodes.emplace_back(std::move(add_0));
@@ -83,7 +83,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
add_1->primitive->value.value = add_1_prim;
add_1->name = "Add1";
auto tensor_3 = std::make_unique<schema::TensorT>();
tensor_3->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_3->nodeType = lite::NodeType_ValueNode;
tensor_3->format = schema::Format_NHWC;
tensor_3->dataType = TypeId::kNumberTypeFloat32;
tensor_3->dims = {1};
@@ -92,7 +92,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
ASSERT_NE(data3, nullptr);
data3[0] = 1;
auto tensor_4 = std::make_unique<schema::TensorT>();
tensor_4->nodeType = schema::NodeType::NodeType_Parameter;
tensor_4->nodeType = lite::NodeType_Parameter;
tensor_4->format = schema::Format_NHWC;
tensor_4->dataType = TypeId::kNumberTypeFloat32;
meta_graph->nodes.emplace_back(std::move(add_1));
@@ -122,7 +122,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
add_5->primitive->value.value = add_5_prim;
add_5->name = "Add5";
auto tensor_13 = std::make_unique<schema::TensorT>();
tensor_13->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_13->nodeType = lite::NodeType_ValueNode;
tensor_13->format = schema::Format_NHWC;
tensor_13->dataType = TypeId::kNumberTypeFloat32;
tensor_13->dims = {1};
@@ -131,7 +131,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
ASSERT_NE(data13, nullptr);
data13[0] = 1;
auto tensor_14 = std::make_unique<schema::TensorT>();
tensor_14->nodeType = schema::NodeType::NodeType_Parameter;
tensor_14->nodeType = lite::NodeType_Parameter;
tensor_14->format = schema::Format_NHWC;
tensor_14->dataType = TypeId::kNumberTypeFloat32;
meta_graph->nodes.emplace_back(std::move(add_5));
@@ -158,7 +158,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
add_2->primitive->value.value = add_2_prim;
add_2->name = "Add2";
auto tensor_5 = std::make_unique<schema::TensorT>();
tensor_5->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_5->nodeType = lite::NodeType_ValueNode;
tensor_5->format = schema::Format_NHWC;
tensor_5->dataType = TypeId::kNumberTypeFloat32;
tensor_5->dims = {1};
@@ -167,7 +167,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
ASSERT_NE(data5, nullptr);
data5[0] = 1;
auto tensor_6 = std::make_unique<schema::TensorT>();
tensor_6->nodeType = schema::NodeType::NodeType_Parameter;
tensor_6->nodeType = lite::NodeType_Parameter;
tensor_6->format = schema::Format_NHWC;
tensor_6->dataType = TypeId::kNumberTypeFloat32;
meta_graph->nodes.emplace_back(std::move(add_2));
@@ -184,7 +184,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
less->primitive->value.value = less_prim;
less->name = "less";
auto tensor_15 = std::make_unique<schema::TensorT>();
tensor_15->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_15->nodeType = lite::NodeType_ValueNode;
tensor_15->format = schema::Format_NHWC;
tensor_15->dataType = TypeId::kNumberTypeFloat32;
tensor_15->dims = {1};
@@ -193,7 +193,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
ASSERT_NE(data15, nullptr);
data15[0] = 1;
auto tensor_7 = std::make_unique<schema::TensorT>();
tensor_7->nodeType = schema::NodeType::NodeType_Parameter;
tensor_7->nodeType = lite::NodeType_Parameter;
tensor_7->format = schema::Format_NHWC;
tensor_7->dataType = TypeId::kNumberTypeFloat32;
meta_graph->nodes.emplace_back(std::move(less));
@@ -210,11 +210,11 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
switchop->primitive->value.value = switch_prim;
switchop->name = "switch";
auto tensor_8 = std::make_unique<schema::TensorT>();
tensor_8->nodeType = schema::NodeType::NodeType_Parameter;
tensor_8->nodeType = lite::NodeType_Parameter;
tensor_8->format = schema::Format_NHWC;
tensor_8->dataType = TypeId::kNumberTypeFloat32;
auto tensor_9 = std::make_unique<schema::TensorT>();
tensor_9->nodeType = schema::NodeType::NodeType_Parameter;
tensor_9->nodeType = lite::NodeType_Parameter;
tensor_9->format = schema::Format_NHWC;
tensor_9->dataType = TypeId::kNumberTypeFloat32;
meta_graph->nodes.emplace_back(std::move(switchop));
@@ -253,7 +253,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
add_3->primitive->value.value = add_3_prim;
add_3->name = "Add3";
auto tensor_10 = std::make_unique<schema::TensorT>();
tensor_10->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_10->nodeType = lite::NodeType_ValueNode;
tensor_10->format = schema::Format_NHWC;
tensor_10->dataType = TypeId::kNumberTypeFloat32;
tensor_10->dims = {1};
@@ -262,7 +262,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
ASSERT_NE(data10, nullptr);
data10[0] = 1;
auto tensor_11 = std::make_unique<schema::TensorT>();
tensor_11->nodeType = schema::NodeType::NodeType_Parameter;
tensor_11->nodeType = lite::NodeType_Parameter;
tensor_11->format = schema::Format_NHWC;
tensor_11->dataType = TypeId::kNumberTypeFloat32;
meta_graph->nodes.emplace_back(std::move(add_3));
@@ -280,7 +280,7 @@ TEST_F(SubGraphTest, RecursiveSubGraphTest) {
add_4->primitive->value.value = add_4_prim;
add_4->name = "Add4";
auto tensor_12 = std::make_unique<schema::TensorT>();
tensor_12->nodeType = schema::NodeType::NodeType_ValueNode;
tensor_12->nodeType = lite::NodeType_ValueNode;
tensor_12->format = schema::Format_NHWC;
tensor_12->dataType = TypeId::kNumberTypeFloat32;
tensor_12->dims = {1};


+ 9
- 9
mindspore/lite/test/ut/src/infer_test.cc View File

@@ -56,7 +56,7 @@ TEST_F(InferTest, TestConvNode) {
meta_graph->outputIndex = {2};

auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 28, 28, 3};
@@ -64,7 +64,7 @@ TEST_F(InferTest, TestConvNode) {
meta_graph->allTensors.emplace_back(std::move(input0));

auto weight = std::make_unique<schema::TensorT>();
weight->nodeType = schema::NodeType::NodeType_ValueNode;
weight->nodeType = lite::NodeType_ValueNode;
weight->format = schema::Format_KHWC;
weight->dataType = TypeId::kNumberTypeFloat32;
weight->dims = {32, 3, 3, 3};
@@ -85,7 +85,7 @@ TEST_F(InferTest, TestConvNode) {
meta_graph->allTensors.emplace_back(std::move(weight));

auto output = std::make_unique<schema::TensorT>();
output->nodeType = schema::NodeType::NodeType_Parameter;
output->nodeType = lite::NodeType_Parameter;
output->format = schema::Format_NHWC;
output->dataType = TypeId::kNumberTypeFloat32;
output->dims = {1, 28, 28, 32};
@@ -169,7 +169,7 @@ TEST_F(InferTest, TestAddNode) {
meta_graph->outputIndex = {2};

auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 28, 28, 3};
@@ -177,7 +177,7 @@ TEST_F(InferTest, TestAddNode) {
meta_graph->allTensors.emplace_back(std::move(input0));

auto weight = std::make_unique<schema::TensorT>();
weight->nodeType = schema::NodeType::NodeType_ValueNode;
weight->nodeType = lite::NodeType_ValueNode;
weight->format = schema::Format_KHWC;
weight->dataType = TypeId::kNumberTypeFloat32;
weight->dims = {1, 28, 28, 3};
@@ -186,7 +186,7 @@ TEST_F(InferTest, TestAddNode) {
meta_graph->allTensors.emplace_back(std::move(weight));

auto output = std::make_unique<schema::TensorT>();
output->nodeType = schema::NodeType::NodeType_Parameter;
output->nodeType = lite::NodeType_Parameter;
output->format = schema::Format_NHWC;
output->dataType = TypeId::kNumberTypeFloat32;
output->offset = -1;
@@ -260,7 +260,7 @@ TEST_F(InferTest, TestParallelExecutor) {
meta_graph->outputIndex = {2};

auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 28, 28, 3};
@@ -268,7 +268,7 @@ TEST_F(InferTest, TestParallelExecutor) {
meta_graph->allTensors.emplace_back(std::move(input0));

auto weight = std::make_unique<schema::TensorT>();
weight->nodeType = schema::NodeType::NodeType_ValueNode;
weight->nodeType = lite::NodeType_ValueNode;
weight->format = schema::Format_NHWC;
weight->dataType = TypeId::kNumberTypeFloat32;
weight->dims = {1, 28, 28, 3};
@@ -277,7 +277,7 @@ TEST_F(InferTest, TestParallelExecutor) {
meta_graph->allTensors.emplace_back(std::move(weight));

auto output = std::make_unique<schema::TensorT>();
output->nodeType = schema::NodeType::NodeType_Parameter;
output->nodeType = lite::NodeType_Parameter;
output->format = schema::Format_NHWC;
output->dataType = TypeId::kNumberTypeFloat32;
output->offset = -1;


+ 15
- 15
mindspore/lite/test/ut/src/runtime/kernel/arm/fp32_grad/network_test.cc View File

@@ -187,7 +187,7 @@ TEST_F(NetworkTest, tuning_layer) {
meta_graph->outputIndex = {5, 14};

auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {BATCH_SIZE, FEATURE_SIZE};
@@ -195,7 +195,7 @@ TEST_F(NetworkTest, tuning_layer) {
meta_graph->allTensors.emplace_back(std::move(input0));
// tensor 1 - relu
auto relu_out = std::make_unique<schema::TensorT>();
relu_out->nodeType = schema::NodeType::NodeType_Parameter;
relu_out->nodeType = lite::NodeType_Parameter;
relu_out->format = schema::Format_NHWC;
relu_out->dataType = TypeId::kNumberTypeFloat32;
relu_out->dims = {BATCH_SIZE, FEATURE_SIZE};
@@ -203,7 +203,7 @@ TEST_F(NetworkTest, tuning_layer) {
meta_graph->allTensors.emplace_back(std::move(relu_out));
// tensor 2 - matmul weights
auto weight = std::make_unique<schema::TensorT>();
weight->nodeType = schema::NodeType::NodeType_ValueNode;
weight->nodeType = lite::NodeType_ValueNode;
weight->format = schema::Format_KHWC;
weight->dataType = TypeId::kNumberTypeFloat32;
weight->dims = {NUM_CLASSES, FEATURE_SIZE};
@@ -218,7 +218,7 @@ TEST_F(NetworkTest, tuning_layer) {
delete[] buf;
// tensor 3 - matmul
auto input3 = std::make_unique<schema::TensorT>();
input3->nodeType = schema::NodeType::NodeType_Parameter;
input3->nodeType = lite::NodeType_Parameter;
input3->format = schema::Format_NHWC;
input3->dataType = TypeId::kNumberTypeFloat32;
input3->dims = {BATCH_SIZE, NUM_CLASSES};
@@ -226,7 +226,7 @@ TEST_F(NetworkTest, tuning_layer) {
meta_graph->allTensors.emplace_back(std::move(input3));
// tensor 4 - fc bias
auto bias = std::make_unique<schema::TensorT>();
bias->nodeType = schema::NodeType::NodeType_ValueNode;
bias->nodeType = lite::NodeType_ValueNode;
bias->format = schema::Format_NHWC;
bias->dataType = TypeId::kNumberTypeFloat32;
bias->dims = {NUM_CLASSES};
@@ -242,7 +242,7 @@ TEST_F(NetworkTest, tuning_layer) {

// tensor 5 - bias_add
auto input5 = std::make_unique<schema::TensorT>();
input5->nodeType = schema::NodeType::NodeType_Parameter;
input5->nodeType = lite::NodeType_Parameter;
input5->format = schema::Format_NHWC;
input5->dataType = TypeId::kNumberTypeFloat32;
input5->dims = {BATCH_SIZE, NUM_CLASSES};
@@ -251,7 +251,7 @@ TEST_F(NetworkTest, tuning_layer) {
// tensor 6 - Label
{
auto label = std::make_unique<schema::TensorT>();
label->nodeType = schema::NodeType::NodeType_ValueNode;
label->nodeType = lite::NodeType_ValueNode;
label->format = schema::Format_NHWC;
label->dataType = TypeId::kNumberTypeFloat32;
label->dims = {BATCH_SIZE * NUM_CLASSES};
@@ -260,7 +260,7 @@ TEST_F(NetworkTest, tuning_layer) {
}
// tensor 7 - Softmaxentropy
auto input7 = std::make_unique<schema::TensorT>();
input7->nodeType = schema::NodeType::NodeType_Parameter;
input7->nodeType = lite::NodeType_Parameter;
input7->format = schema::Format_NHWC;
input7->dataType = TypeId::kNumberTypeFloat32;
input7->dims = {BATCH_SIZE, NUM_CLASSES};
@@ -268,7 +268,7 @@ TEST_F(NetworkTest, tuning_layer) {
meta_graph->allTensors.emplace_back(std::move(input7));
// tensor 8 - biasGrad
auto input8 = std::make_unique<schema::TensorT>();
input8->nodeType = schema::NodeType::NodeType_Parameter;
input8->nodeType = lite::NodeType_Parameter;
input8->format = schema::Format_NHWC;
input8->dataType = TypeId::kNumberTypeFloat32;
input8->dims = {NUM_CLASSES};
@@ -276,7 +276,7 @@ TEST_F(NetworkTest, tuning_layer) {
meta_graph->allTensors.emplace_back(std::move(input8));
// tensor 9 - matmul2
auto input9 = std::make_unique<schema::TensorT>();
input9->nodeType = schema::NodeType::NodeType_Parameter;
input9->nodeType = lite::NodeType_Parameter;
input9->format = schema::Format_NHWC;
input9->dataType = TypeId::kNumberTypeFloat32;
input9->dims = {NUM_CLASSES, FEATURE_SIZE};
@@ -284,7 +284,7 @@ TEST_F(NetworkTest, tuning_layer) {
meta_graph->allTensors.emplace_back(std::move(input9));
// tensor 10 weights accumulate
auto input10 = std::make_unique<schema::TensorT>();
input10->nodeType = schema::NodeType::NodeType_ValueNode;
input10->nodeType = lite::NodeType_ValueNode;
input10->format = schema::Format_NHWC;
input10->dataType = TypeId::kNumberTypeFloat32;
input10->dims = {NUM_CLASSES, FEATURE_SIZE};
@@ -296,7 +296,7 @@ TEST_F(NetworkTest, tuning_layer) {
// tensor 11 - lr
{
auto lr = std::make_unique<schema::TensorT>();
lr->nodeType = schema::NodeType::NodeType_ValueNode;
lr->nodeType = lite::NodeType_ValueNode;
lr->format = schema::Format_NHWC;
lr->dataType = TypeId::kNumberTypeFloat32;
lr->dims = {1};
@@ -309,7 +309,7 @@ TEST_F(NetworkTest, tuning_layer) {
// tensor 12 - momentum
{
auto input12 = std::make_unique<schema::TensorT>();
input12->nodeType = schema::NodeType::NodeType_ValueNode;
input12->nodeType = lite::NodeType_ValueNode;
input12->format = schema::Format_NHWC;
input12->dataType = TypeId::kNumberTypeFloat32;
input12->dims = {1};
@@ -321,7 +321,7 @@ TEST_F(NetworkTest, tuning_layer) {
}
// tensor 13 - bias accumulate
auto input13 = std::make_unique<schema::TensorT>();
input13->nodeType = schema::NodeType::NodeType_ValueNode;
input13->nodeType = lite::NodeType_ValueNode;
input13->format = schema::Format_NHWC;
input13->dataType = TypeId::kNumberTypeFloat32;
input13->dims = {NUM_CLASSES};
@@ -334,7 +334,7 @@ TEST_F(NetworkTest, tuning_layer) {
// tensor 14 - loss
{
auto loss14 = std::make_unique<schema::TensorT>();
loss14->nodeType = schema::NodeType::NodeType_ValueNode;
loss14->nodeType = lite::NodeType_ValueNode;
loss14->format = schema::Format_NHWC;
loss14->dataType = TypeId::kNumberTypeFloat32;
loss14->dims = {1};


+ 0
- 1
mindspore/lite/test/ut/src/runtime/kernel/arm/int8/deconv_int8_tests.cc View File

@@ -31,7 +31,6 @@ namespace mindspore {
using mindspore::lite::QuantArg;
using mindspore::lite::Tensor;
using mindspore::schema::Format_NHWC;
using mindspore::schema::NodeType_Parameter;
class TestDeconvInt8 : public mindspore::CommonTest {
public:
TestDeconvInt8() {}


+ 0
- 1
mindspore/lite/test/ut/src/runtime/kernel/arm/int8/pad_int8_tests.cc View File

@@ -26,7 +26,6 @@
namespace mindspore {
using mindspore::lite::QuantArg;
using mindspore::lite::Tensor;
using mindspore::schema::NodeType_Parameter;
class TestPadInt8 : public mindspore::CommonTest {
public:
TestPadInt8() {}


+ 19
- 19
mindspore/lite/test/ut/src/scheduler_test.cc View File

@@ -96,49 +96,49 @@ TEST_F(SchedulerTest, TestConstructSubGraphsTwoBranch) {
concat->name = "concat";

auto tensor0 = std::make_unique<mindspore::schema::TensorT>();
tensor0->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor0->nodeType = mindspore::lite::NodeType_ValueNode;
tensor0->format = mindspore::schema::Format_NHWC;
tensor0->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor0->dims = {1, 16, 16, 4};
tensor0->offset = -1;
auto tensor1 = std::make_unique<mindspore::schema::TensorT>();
tensor1->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor1->nodeType = mindspore::lite::NodeType_ValueNode;
tensor1->format = mindspore::schema::Format_NHWC;
tensor1->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor1->dims = {1, 16, 16, 2};
tensor1->offset = -1;
auto tensor2 = std::make_unique<mindspore::schema::TensorT>();
tensor2->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor2->nodeType = mindspore::lite::NodeType_ValueNode;
tensor2->format = mindspore::schema::Format_NHWC;
tensor2->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor2->dims = {1, 16, 16, 2};
tensor2->offset = -1;
auto tensor3 = std::make_unique<mindspore::schema::TensorT>();
tensor3->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor3->nodeType = mindspore::lite::NodeType_ValueNode;
tensor3->format = mindspore::schema::Format_NHWC;
tensor3->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor3->dims = {1, 16, 16, 2};
tensor3->offset = -1;
auto tensor4 = std::make_unique<mindspore::schema::TensorT>();
tensor4->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor4->nodeType = mindspore::lite::NodeType_ValueNode;
tensor4->format = mindspore::schema::Format_NHWC;
tensor4->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor4->dims = {1, 16, 16, 2};
tensor4->offset = -1;
auto tensor5 = std::make_unique<mindspore::schema::TensorT>();
tensor5->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor5->nodeType = mindspore::lite::NodeType_ValueNode;
tensor5->format = mindspore::schema::Format_NHWC;
tensor5->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor5->dims = {1, 16, 16, 2};
tensor5->offset = -1;
auto tensor6 = std::make_unique<mindspore::schema::TensorT>();
tensor6->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor6->nodeType = mindspore::lite::NodeType_ValueNode;
tensor6->format = mindspore::schema::Format_NHWC;
tensor6->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor6->dims = {1, 16, 16, 2};
tensor6->offset = -1;
auto tensor7 = std::make_unique<mindspore::schema::TensorT>();
tensor7->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor7->nodeType = mindspore::lite::NodeType_ValueNode;
tensor7->format = mindspore::schema::Format_NHWC;
tensor7->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor7->dims = {1, 16, 16, 4};
@@ -257,67 +257,67 @@ TEST_F(SchedulerTest, TestConstructSubGraphsThreeBranch) {
concat->name = "concat";

auto tensor0 = std::make_unique<mindspore::schema::TensorT>();
tensor0->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor0->nodeType = mindspore::lite::NodeType_ValueNode;
tensor0->format = mindspore::schema::Format_NHWC;
tensor0->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor0->dims = {1, 16, 16, 3};
tensor0->offset = -1;
auto tensor1 = std::make_unique<mindspore::schema::TensorT>();
tensor1->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor1->nodeType = mindspore::lite::NodeType_ValueNode;
tensor1->format = mindspore::schema::Format_NHWC;
tensor1->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor1->dims = {1, 16, 16, 1};
tensor1->offset = -1;
auto tensor2 = std::make_unique<mindspore::schema::TensorT>();
tensor2->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor2->nodeType = mindspore::lite::NodeType_ValueNode;
tensor2->format = mindspore::schema::Format_NHWC;
tensor2->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor2->dims = {1, 16, 16, 1};
tensor2->offset = -1;
auto tensor3 = std::make_unique<mindspore::schema::TensorT>();
tensor3->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor3->nodeType = mindspore::lite::NodeType_ValueNode;
tensor3->format = mindspore::schema::Format_NHWC;
tensor3->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor3->dims = {1, 16, 16, 1};
tensor3->offset = -1;
auto tensor4 = std::make_unique<mindspore::schema::TensorT>();
tensor4->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor4->nodeType = mindspore::lite::NodeType_ValueNode;
tensor4->format = mindspore::schema::Format_NHWC;
tensor4->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor4->dims = {1, 16, 16, 1};
tensor4->offset = -1;
auto tensor5 = std::make_unique<mindspore::schema::TensorT>();
tensor5->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor5->nodeType = mindspore::lite::NodeType_ValueNode;
tensor5->format = mindspore::schema::Format_NHWC;
tensor5->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor5->dims = {1, 16, 16, 1};
tensor5->offset = -1;
auto tensor6 = std::make_unique<mindspore::schema::TensorT>();
tensor6->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor6->nodeType = mindspore::lite::NodeType_ValueNode;
tensor6->format = mindspore::schema::Format_NHWC;
tensor6->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor6->dims = {1, 16, 16, 1};
tensor6->offset = -1;
auto tensor7 = std::make_unique<mindspore::schema::TensorT>();
tensor7->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor7->nodeType = mindspore::lite::NodeType_ValueNode;
tensor7->format = mindspore::schema::Format_NHWC;
tensor7->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor7->dims = {1, 16, 16, 1};
tensor7->offset = -1;
auto tensor8 = std::make_unique<mindspore::schema::TensorT>();
tensor8->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor8->nodeType = mindspore::lite::NodeType_ValueNode;
tensor8->format = mindspore::schema::Format_NHWC;
tensor8->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor8->dims = {1, 16, 16, 1};
tensor8->offset = -1;
auto tensor9 = std::make_unique<mindspore::schema::TensorT>();
tensor9->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor9->nodeType = mindspore::lite::NodeType_ValueNode;
tensor9->format = mindspore::schema::Format_NHWC;
tensor9->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor9->dims = {1, 16, 16, 1};
tensor9->offset = -1;
auto tensor10 = std::make_unique<mindspore::schema::TensorT>();
tensor10->nodeType = mindspore::schema::NodeType::NodeType_ValueNode;
tensor10->nodeType = mindspore::lite::NodeType_ValueNode;
tensor10->format = mindspore::schema::Format_NHWC;
tensor10->dataType = mindspore::TypeId::kNumberTypeFloat32;
tensor10->dims = {1, 16, 16, 3};


+ 17
- 17
mindspore/lite/test/ut/tools/optimizer/fusion/constant_folding_fusion_test.cc View File

@@ -56,7 +56,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType op_type, void *op_node) {

// input 0: data1
auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 2, 2, 3};
@@ -72,7 +72,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType op_type, void *op_node) {

// input 1: data2
auto input1 = std::make_unique<schema::TensorT>();
input1->nodeType = schema::NodeType::NodeType_ValueNode;
input1->nodeType = lite::NodeType_ValueNode;
input1->format = schema::Format_NHWC;
input1->dataType = TypeId::kNumberTypeFloat32;
input1->dims = {1, 2, 2, 3};
@@ -88,7 +88,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType op_type, void *op_node) {

// final add output
auto add_output = std::make_unique<schema::TensorT>();
add_output->nodeType = schema::NodeType::NodeType_Parameter;
add_output->nodeType = lite::NodeType_Parameter;
add_output->format = schema::Format_NHWC;
add_output->dataType = TypeId::kNumberTypeFloat32;
add_output->dims = {1, 2, 2, 3};
@@ -115,7 +115,7 @@ MetaGraphTptr BuildGraphForOneInput(schema::PrimitiveType op_type, void *op_node

// input 0: data1
auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 2, 2, 3};
@@ -131,7 +131,7 @@ MetaGraphTptr BuildGraphForOneInput(schema::PrimitiveType op_type, void *op_node

// final add output
auto add_output = std::make_unique<schema::TensorT>();
add_output->nodeType = schema::NodeType::NodeType_Parameter;
add_output->nodeType = lite::NodeType_Parameter;
add_output->format = schema::Format_NHWC;
add_output->dataType = TypeId::kNumberTypeFloat32;
add_output->dims = {1, 2, 2, 3};
@@ -168,7 +168,7 @@ MetaGraphTptr BuildMixGraph() {

// input 0: data1
auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 2, 2, 3};
@@ -184,7 +184,7 @@ MetaGraphTptr BuildMixGraph() {

// input 1: data2
auto input1 = std::make_unique<schema::TensorT>();
input1->nodeType = schema::NodeType::NodeType_ValueNode;
input1->nodeType = lite::NodeType_ValueNode;
input1->format = schema::Format_NHWC;
input1->dataType = TypeId::kNumberTypeFloat32;
input1->dims = {1, 2, 2, 3};
@@ -200,7 +200,7 @@ MetaGraphTptr BuildMixGraph() {

// addoutput
auto add_output = std::make_unique<schema::TensorT>();
add_output->nodeType = schema::NodeType::NodeType_Parameter;
add_output->nodeType = lite::NodeType_Parameter;
add_output->format = schema::Format_NHWC;
add_output->dataType = TypeId::kNumberTypeFloat32;
add_output->dims = {1, 2, 2, 3};
@@ -213,7 +213,7 @@ MetaGraphTptr BuildMixGraph() {

// input 2: data3
auto input2 = std::make_unique<schema::TensorT>();
input2->nodeType = schema::NodeType::NodeType_ValueNode;
input2->nodeType = lite::NodeType_ValueNode;
input2->format = schema::Format_NHWC;
input2->dataType = TypeId::kNumberTypeFloat32;
input2->dims = {1, 2, 2, 3};
@@ -229,7 +229,7 @@ MetaGraphTptr BuildMixGraph() {

// final mul output
auto mul_output = std::make_unique<schema::TensorT>();
mul_output->nodeType = schema::NodeType::NodeType_Parameter;
mul_output->nodeType = lite::NodeType_Parameter;
mul_output->format = schema::Format_NHWC;
mul_output->dataType = TypeId::kNumberTypeFloat32;
mul_output->dims = {1, 2, 2, 3};
@@ -278,7 +278,7 @@ MetaGraphTptr BuildSplitGraph() {

// input 0: data1
auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 2, 2, 3};
@@ -294,7 +294,7 @@ MetaGraphTptr BuildSplitGraph() {

// split output1
auto split_output1 = std::make_unique<schema::TensorT>();
split_output1->nodeType = schema::NodeType::NodeType_Parameter;
split_output1->nodeType = lite::NodeType_Parameter;
split_output1->format = schema::Format_NHWC;
split_output1->dataType = TypeId::kNumberTypeFloat32;
split_output1->dims = {1, 1, 2, 3};
@@ -307,7 +307,7 @@ MetaGraphTptr BuildSplitGraph() {

// split output2
auto split_output2 = std::make_unique<schema::TensorT>();
split_output2->nodeType = schema::NodeType::NodeType_Parameter;
split_output2->nodeType = lite::NodeType_Parameter;
split_output2->format = schema::Format_NHWC;
split_output2->dataType = TypeId::kNumberTypeFloat32;
split_output2->dims = {1, 1, 2, 3};
@@ -320,7 +320,7 @@ MetaGraphTptr BuildSplitGraph() {

// input 1: data2
auto input1 = std::make_unique<schema::TensorT>();
input1->nodeType = schema::NodeType::NodeType_ValueNode;
input1->nodeType = lite::NodeType_ValueNode;
input1->format = schema::Format_NHWC;
input1->dataType = TypeId::kNumberTypeFloat32;
input1->dims = {1, 1, 2, 3};
@@ -336,7 +336,7 @@ MetaGraphTptr BuildSplitGraph() {

// input 2: data3
auto input2 = std::make_unique<schema::TensorT>();
input2->nodeType = schema::NodeType::NodeType_ValueNode;
input2->nodeType = lite::NodeType_ValueNode;
input2->format = schema::Format_NHWC;
input2->dataType = TypeId::kNumberTypeFloat32;
input2->dims = {1, 1, 2, 3};
@@ -352,7 +352,7 @@ MetaGraphTptr BuildSplitGraph() {

// final mul output1
auto mul_output = std::make_unique<schema::TensorT>();
mul_output->nodeType = schema::NodeType::NodeType_Parameter;
mul_output->nodeType = lite::NodeType_Parameter;
mul_output->format = schema::Format_NHWC;
mul_output->dataType = TypeId::kNumberTypeFloat32;
mul_output->dims = {1, 1, 2, 3};
@@ -360,7 +360,7 @@ MetaGraphTptr BuildSplitGraph() {

// final mul output2
auto mul_output2 = std::make_unique<schema::TensorT>();
mul_output2->nodeType = schema::NodeType::NodeType_Parameter;
mul_output2->nodeType = lite::NodeType_Parameter;
mul_output2->format = schema::Format_NHWC;
mul_output2->dataType = TypeId::kNumberTypeFloat32;
mul_output2->dims = {1, 1, 2, 3};


+ 4
- 4
mindspore/lite/test/ut/tools/optimizer/fusion/conv_activation_fusion_test.cc View File

@@ -100,7 +100,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, schema::ActivationType

// input 0: data
auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 5, 5, 3};
@@ -109,7 +109,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, schema::ActivationType

// input 1: weight
auto input1 = std::make_unique<schema::TensorT>();
input1->nodeType = schema::NodeType::NodeType_ValueNode;
input1->nodeType = lite::NodeType_ValueNode;
input1->format = schema::Format_KHWC;
input1->dataType = TypeId::kNumberTypeFloat32;
input1->dims = {8, 3, 3, 3};
@@ -118,7 +118,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, schema::ActivationType

// conv output
auto conv_output = std::make_unique<schema::TensorT>();
conv_output->nodeType = schema::NodeType::NodeType_Parameter;
conv_output->nodeType = lite::NodeType_Parameter;
conv_output->format = schema::Format_NHWC;
conv_output->dataType = TypeId::kNumberTypeFloat32;
conv_output->dims = {1, 5, 5, 8};
@@ -126,7 +126,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, schema::ActivationType

// final output
auto output = std::make_unique<schema::TensorT>();
output->nodeType = schema::NodeType::NodeType_Parameter;
output->nodeType = lite::NodeType_Parameter;
output->format = schema::Format_NHWC;
output->dataType = TypeId::kNumberTypeFloat32;
output->dims = {1, 5, 5, 8};


+ 5
- 5
mindspore/lite/test/ut/tools/optimizer/fusion/conv_biasadd_fusion_test.cc View File

@@ -100,7 +100,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, schema::PrimitiveType

// input 0: data
auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 5, 5, 3};
@@ -109,7 +109,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, schema::PrimitiveType

// input 1: weight
auto input1 = std::make_unique<schema::TensorT>();
input1->nodeType = schema::NodeType::NodeType_ValueNode;
input1->nodeType = lite::NodeType_ValueNode;
input1->format = schema::Format_KHWC;
input1->dataType = TypeId::kNumberTypeFloat32;
input1->dims = {8, 3, 3, 3};
@@ -118,7 +118,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, schema::PrimitiveType

// conv output
auto conv_output = std::make_unique<schema::TensorT>();
conv_output->nodeType = schema::NodeType::NodeType_Parameter;
conv_output->nodeType = lite::NodeType_Parameter;
conv_output->format = schema::Format_NHWC;
conv_output->dataType = TypeId::kNumberTypeFloat32;
conv_output->dims = {1, 5, 5, 8};
@@ -126,7 +126,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, schema::PrimitiveType

// input2: bias
auto input2 = std::make_unique<schema::TensorT>();
input2->nodeType = schema::NodeType::NodeType_ValueNode;
input2->nodeType = lite::NodeType_ValueNode;
input2->format = schema::Format_NHWC;
input2->dataType = TypeId::kNumberTypeFloat32;
input2->dims = {1, 5, 5, 8};
@@ -135,7 +135,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, schema::PrimitiveType

// final output
auto output = std::make_unique<schema::TensorT>();
output->nodeType = schema::NodeType::NodeType_Parameter;
output->nodeType = lite::NodeType_Parameter;
output->format = schema::Format_NHWC;
output->dataType = TypeId::kNumberTypeFloat32;
output->dims = {1, 5, 5, 8};


+ 15
- 15
mindspore/lite/test/ut/tools/optimizer/fusion/conv_bn_fusion_test.cc View File

@@ -98,7 +98,7 @@ MetaGraphTptr BuildCaffeGraph(schema::PrimitiveType conv_type) {

// input 0: data
auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 5, 5, 3};
@@ -107,7 +107,7 @@ MetaGraphTptr BuildCaffeGraph(schema::PrimitiveType conv_type) {

// input 1: weight
auto input1 = std::make_unique<schema::TensorT>();
input1->nodeType = schema::NodeType::NodeType_ValueNode;
input1->nodeType = lite::NodeType_ValueNode;
input1->format = schema::Format_KHWC;
input1->dataType = TypeId::kNumberTypeFloat32;
input1->dims = {8, 3, 3, 3};
@@ -116,7 +116,7 @@ MetaGraphTptr BuildCaffeGraph(schema::PrimitiveType conv_type) {

// conv output
auto conv_output = std::make_unique<schema::TensorT>();
conv_output->nodeType = schema::NodeType::NodeType_Parameter;
conv_output->nodeType = lite::NodeType_Parameter;
conv_output->format = schema::Format_NHWC;
conv_output->dataType = TypeId::kNumberTypeFloat32;
conv_output->dims = {1, 5, 5, 8};
@@ -124,7 +124,7 @@ MetaGraphTptr BuildCaffeGraph(schema::PrimitiveType conv_type) {

// caffe bn : mean
auto input2 = std::make_unique<schema::TensorT>();
input2->nodeType = schema::NodeType::NodeType_ValueNode;
input2->nodeType = lite::NodeType_ValueNode;
input2->format = schema::Format_NHWC;
input2->dataType = TypeId::kNumberTypeFloat32;
input2->dims = {1, 5, 5, 8};
@@ -133,7 +133,7 @@ MetaGraphTptr BuildCaffeGraph(schema::PrimitiveType conv_type) {

// caffe bn : var
auto input3 = std::make_unique<schema::TensorT>();
input3->nodeType = schema::NodeType::NodeType_ValueNode;
input3->nodeType = lite::NodeType_ValueNode;
input3->format = schema::Format_NHWC;
input3->dataType = TypeId::kNumberTypeFloat32;
input3->dims = {1, 5, 5, 8};
@@ -142,7 +142,7 @@ MetaGraphTptr BuildCaffeGraph(schema::PrimitiveType conv_type) {

// final bn output
auto output = std::make_unique<schema::TensorT>();
output->nodeType = schema::NodeType::NodeType_Parameter;
output->nodeType = lite::NodeType_Parameter;
output->format = schema::Format_NHWC;
output->dataType = TypeId::kNumberTypeFloat32;
output->dims = {1, 5, 5, 8};
@@ -179,7 +179,7 @@ MetaGraphTptr BuildTFGraph(schema::PrimitiveType conv_type) {

// input 0: data
auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 5, 5, 3};
@@ -188,7 +188,7 @@ MetaGraphTptr BuildTFGraph(schema::PrimitiveType conv_type) {

// input 1: conv_bias
auto input11 = std::make_unique<schema::TensorT>();
input11->nodeType = schema::NodeType::NodeType_ValueNode;
input11->nodeType = lite::NodeType_ValueNode;
input11->format = schema::Format_KHWC;
input11->dataType = TypeId::kNumberTypeFloat32;
input11->dims = {8, 3, 3, 3};
@@ -197,7 +197,7 @@ MetaGraphTptr BuildTFGraph(schema::PrimitiveType conv_type) {

// input 1: weight
auto input1 = std::make_unique<schema::TensorT>();
input1->nodeType = schema::NodeType::NodeType_ValueNode;
input1->nodeType = lite::NodeType_ValueNode;
input1->format = schema::Format_KHWC;
input1->dataType = TypeId::kNumberTypeFloat32;
input1->dims = {8, 3, 3, 3};
@@ -206,7 +206,7 @@ MetaGraphTptr BuildTFGraph(schema::PrimitiveType conv_type) {

// conv output
auto conv_output = std::make_unique<schema::TensorT>();
conv_output->nodeType = schema::NodeType::NodeType_Parameter;
conv_output->nodeType = lite::NodeType_Parameter;
conv_output->format = schema::Format_NHWC;
conv_output->dataType = TypeId::kNumberTypeFloat32;
conv_output->dims = {1, 5, 5, 8};
@@ -214,7 +214,7 @@ MetaGraphTptr BuildTFGraph(schema::PrimitiveType conv_type) {

// tflite bn : scale
auto input2 = std::make_unique<schema::TensorT>();
input2->nodeType = schema::NodeType::NodeType_ValueNode;
input2->nodeType = lite::NodeType_ValueNode;
input2->format = schema::Format_NHWC;
input2->dataType = TypeId::kNumberTypeFloat32;
input2->dims = {1, 5, 5, 8};
@@ -223,7 +223,7 @@ MetaGraphTptr BuildTFGraph(schema::PrimitiveType conv_type) {

// tflite bn : bias
auto input3 = std::make_unique<schema::TensorT>();
input3->nodeType = schema::NodeType::NodeType_ValueNode;
input3->nodeType = lite::NodeType_ValueNode;
input3->format = schema::Format_NHWC;
input3->dataType = TypeId::kNumberTypeFloat32;
input3->dims = {1, 5, 5, 8};
@@ -232,7 +232,7 @@ MetaGraphTptr BuildTFGraph(schema::PrimitiveType conv_type) {

// tflite bn : mean
auto input4 = std::make_unique<schema::TensorT>();
input4->nodeType = schema::NodeType::NodeType_ValueNode;
input4->nodeType = lite::NodeType_ValueNode;
input4->format = schema::Format_NHWC;
input4->dataType = TypeId::kNumberTypeFloat32;
input4->dims = {1, 5, 5, 8};
@@ -241,7 +241,7 @@ MetaGraphTptr BuildTFGraph(schema::PrimitiveType conv_type) {

// tflite bn : var
auto input5 = std::make_unique<schema::TensorT>();
input5->nodeType = schema::NodeType::NodeType_ValueNode;
input5->nodeType = lite::NodeType_ValueNode;
input5->format = schema::Format_NHWC;
input5->dataType = TypeId::kNumberTypeFloat32;
input5->dims = {1, 5, 5, 8};
@@ -250,7 +250,7 @@ MetaGraphTptr BuildTFGraph(schema::PrimitiveType conv_type) {

// final output
auto output = std::make_unique<schema::TensorT>();
output->nodeType = schema::NodeType::NodeType_Parameter;
output->nodeType = lite::NodeType_Parameter;
output->format = schema::Format_NHWC;
output->dataType = TypeId::kNumberTypeFloat32;
output->dims = {1, 5, 5, 8};


+ 7
- 7
mindspore/lite/test/ut/tools/optimizer/fusion/conv_scale_fusion_test.cc View File

@@ -115,7 +115,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, bool conv_with_bias) {

// input 0: data
auto input0 = std::make_unique<schema::TensorT>();
input0->nodeType = schema::NodeType::NodeType_ValueNode;
input0->nodeType = lite::NodeType_ValueNode;
input0->format = schema::Format_NHWC;
input0->dataType = TypeId::kNumberTypeFloat32;
input0->dims = {1, 5, 5, 3};
@@ -124,7 +124,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, bool conv_with_bias) {

// input 1: weight
auto input1 = std::make_unique<schema::TensorT>();
input1->nodeType = schema::NodeType::NodeType_ValueNode;
input1->nodeType = lite::NodeType_ValueNode;
input1->format = schema::Format_KHWC;
input1->dataType = TypeId::kNumberTypeFloat32;
input1->dims = {8, 3, 3, 3};
@@ -134,7 +134,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, bool conv_with_bias) {
if (conv_with_bias) {
// input 00: bias
auto input00 = std::make_unique<schema::TensorT>();
input00->nodeType = schema::NodeType::NodeType_ValueNode;
input00->nodeType = lite::NodeType_ValueNode;
input00->format = schema::Format_NHWC;
input00->dataType = TypeId::kNumberTypeFloat32;
input00->dims = {1, 5, 5, 3};
@@ -144,7 +144,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, bool conv_with_bias) {

// conv output
auto conv_output = std::make_unique<schema::TensorT>();
conv_output->nodeType = schema::NodeType::NodeType_Parameter;
conv_output->nodeType = lite::NodeType_Parameter;
conv_output->format = schema::Format_NHWC;
conv_output->dataType = TypeId::kNumberTypeFloat32;
conv_output->dims = {1, 5, 5, 8};
@@ -152,7 +152,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, bool conv_with_bias) {

// scale weight input
auto input2 = std::make_unique<schema::TensorT>();
input2->nodeType = schema::NodeType::NodeType_ValueNode;
input2->nodeType = lite::NodeType_ValueNode;
input2->format = schema::Format_NHWC;
input2->dataType = TypeId::kNumberTypeFloat32;
input2->dims = {1, 5, 5, 8};
@@ -161,7 +161,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, bool conv_with_bias) {

// scale bias input
auto input3 = std::make_unique<schema::TensorT>();
input3->nodeType = schema::NodeType::NodeType_ValueNode;
input3->nodeType = lite::NodeType_ValueNode;
input3->format = schema::Format_NHWC;
input3->dataType = TypeId::kNumberTypeFloat32;
input3->dims = {1, 5, 5, 8};
@@ -170,7 +170,7 @@ MetaGraphTptr BuildGraph(schema::PrimitiveType conv_type, bool conv_with_bias) {

// final scale output
auto output = std::make_unique<schema::TensorT>();
output->nodeType = schema::NodeType::NodeType_Parameter;
output->nodeType = lite::NodeType_Parameter;
output->format = schema::Format_NHWC;
output->dataType = TypeId::kNumberTypeFloat32;
output->dims = {1, 5, 5, 8};


+ 10
- 11
mindspore/lite/tools/anf_exporter/anf_exporter.cc View File

@@ -268,8 +268,8 @@ int AnfExporter::SetGraphInputIndex(const std::unique_ptr<schema::MetaGraphT> &m
for (auto &node : subgraph_input_nodes) {
for (auto input : node->inputIndex) {
auto tensor = meta_graphT->allTensors[input].get();
if (tensor->nodeType != schema::NodeType_CNode && tensor->data.empty()) {
tensor->nodeType = schema::NodeType_ValueNode;
if (tensor->nodeType != NodeType_CNode && tensor->data.empty()) {
tensor->nodeType = NodeType_ValueNode;
tensor->format = schema::Format_NHWC;
if (!IsContain(subgraph->inputIndices, input)) {
if (subgraph_index == kMainGraphIndex) {
@@ -386,7 +386,6 @@ int AnfExporter::Anf2Fb(const FuncGraphPtr &func_graph, const std::unique_ptr<sc
if (primT == nullptr) {
primT = GetPrimitiveT(cnode->input(0));
}
node->nodeType = schema::NodeType_CNode;
node->name = cnode->fullname_with_scope();
node->primitive = std::unique_ptr<schema::PrimitiveT>(primT);
ret = SetOpInputNode(cnode, meta_graphT, node.get());
@@ -622,7 +621,7 @@ int AnfExporter::ProcessTensor(const ValueNodePtr &valueNode, std::unique_ptr<sc
[](const int64_t &value) { return static_cast<int32_t>(value); });
(*paramTensor)->dims = dims;
if (train_flag && (*paramTensor)->dims.empty()) (*paramTensor)->dims = {1};
(*paramTensor)->nodeType = schema::NodeType::NodeType_ValueNode;
(*paramTensor)->nodeType = NodeType_ValueNode;
auto data = value->cast<tensor::TensorPtr>();
(*paramTensor)->data.resize(data->Size());
ret = memcpy_s((*paramTensor)->data.data(), data->Size(), data->data_c(), data->Size());
@@ -642,7 +641,7 @@ int AnfExporter::ProcessInt32OrInt64Imm(const ValueNodePtr &valueNode, std::uniq
// data of int64 is converted to int32 here.
(*paramTensor)->dataType = kNumberTypeInt32;
(*paramTensor)->dims = {1};
(*paramTensor)->nodeType = schema::NodeType::NodeType_ValueNode;
(*paramTensor)->nodeType = NodeType_ValueNode;
int real_data = opt::CastToInt(value).front();
(*paramTensor)->data.resize(sizeof(int32_t));
ret = memcpy_s((*paramTensor)->data.data(), sizeof(int32_t), &real_data, sizeof(int32_t));
@@ -663,7 +662,7 @@ void AnfExporter::ProcessBoolImm(const ValueNodePtr &valueNode, std::unique_ptr<
auto typePtr = abstractScalar->GetTypeTrack();
(*paramTensor)->dataType = typePtr->type_id();
(*paramTensor)->dims = {1};
(*paramTensor)->nodeType = schema::NodeType_ValueNode;
(*paramTensor)->nodeType = NodeType_ValueNode;
auto data = value->cast<mindspore::BoolImmPtr>();
(*paramTensor)->data.emplace_back(data->value());
node_id_map_[valueNode->fullname_with_scope()] = meta_graphT->allTensors.size();
@@ -681,7 +680,7 @@ int AnfExporter::ProcessNumber(const ValueNodePtr &valueNode, schema::TensorT *p
}
paramTensor->dataType = kNumberTypeInt32;
paramTensor->dims = {1};
paramTensor->nodeType = schema::NodeType_ValueNode;
paramTensor->nodeType = NodeType_ValueNode;
node_id_map_[valueNode->fullname_with_scope()] = meta_graphT->allTensors.size();
output_cnode->inputIndex.emplace_back(meta_graphT->allTensors.size());
meta_graphT->allTensors.emplace_back(paramTensor);
@@ -691,7 +690,7 @@ void AnfExporter::ProcessInt(const ValueNodePtr &valueNode, std::unique_ptr<sche
schema::CNodeT *output_cnode, const std::unique_ptr<schema::MetaGraphT> &meta_graphT) {
(*paramTensor)->dataType = kNumberTypeInt32;
(*paramTensor)->dims = {1};
(*paramTensor)->nodeType = schema::NodeType_ValueNode;
(*paramTensor)->nodeType = NodeType_ValueNode;
(*paramTensor)->data.emplace_back(kNumberTypeInt32);
node_id_map_[valueNode->fullname_with_scope()] = meta_graphT->allTensors.size();
output_cnode->inputIndex.emplace_back(meta_graphT->allTensors.size());
@@ -721,7 +720,7 @@ int AnfExporter::ProcessValueSequence(const ValueNodePtr &valueNode, std::unique
}
(*paramTensor)->dataType = kNumberTypeInt32;
(*paramTensor)->dims = {static_cast<int32_t>(shape.size())};
(*paramTensor)->nodeType = schema::NodeType_ValueNode;
(*paramTensor)->nodeType = NodeType_ValueNode;
(*paramTensor)->data.resize(shape.size() * sizeof(int));
ret = memcpy_s((*paramTensor)->data.data(), shape.size() * sizeof(int32_t), shape.data(),
shape.size() * sizeof(int32_t));
@@ -862,7 +861,7 @@ void AnfExporter::SetOpOutputNode(const CNodePtr &cnode, const std::unique_ptr<s
MS_LOG(ERROR) << "new msTensor failed";
return;
}
msTensor->nodeType = schema::NodeType_CNode;
msTensor->nodeType = NodeType_CNode;
fb_node->outputIndex.emplace_back(meta_graphT->allTensors.size());
if (train_flag) {
std::string name = cnode_name + "_o:" + std::to_string(i);
@@ -912,7 +911,7 @@ void AnfExporter::SetOpOutputNode(const CNodePtr &cnode, const std::unique_ptr<s
type = typePtr->type_id();
}
ms_tensor->dataType = type;
ms_tensor->nodeType = schema::NodeType_CNode;
ms_tensor->nodeType = NodeType_CNode;
ms_tensor->name = cnode_name;
fb_node->outputIndex.emplace_back(meta_graphT->allTensors.size());
node_id_map_[cnode_name] = meta_graphT->allTensors.size();


+ 2
- 2
mindspore/lite/tools/common/graph_util.cc View File

@@ -445,7 +445,7 @@ NodeIter InsertNodeBefore(schema::MetaGraphT *graphT, NodeIter existNodeIter, si
MS_LOG(ERROR) << "Copy Tensor failed";
return graphT->nodes.end();
}
toAddTensor->nodeType = schema::NodeType_CNode;
toAddTensor->nodeType = NodeType_CNode;
toAddTensor->refCount = 0;
toAddTensor->data.clear();
MS_ASSERT(toAddNodeIn->primitive != nullptr);
@@ -517,7 +517,7 @@ NodeIter InsertNodeAfter(schema::MetaGraphT *graphT, NodeIter existNodeIter, siz
*errorCode = RET_NULL_PTR;
return graphT->nodes.end();
}
toAddTensor->nodeType = schema::NodeType_CNode;
toAddTensor->nodeType = NodeType_CNode;
MS_ASSERT(toAddNodeIn->primitive != nullptr);
if (toAddNodeIn->primitive->value.type == schema::PrimitiveType_QuantDTypeCast) {
auto prim = toAddNodeIn->primitive->value.AsQuantDTypeCast();


+ 3
- 2
mindspore/lite/tools/common/tensor_util.h View File

@@ -26,6 +26,7 @@
#include "schema/inner/model_generated.h"
#include "src/common/log_adapter.h"
#include "ir/dtype/type_id.h"
#include "src/common/utils.h"

namespace mindspore {
namespace lite {
@@ -69,9 +70,9 @@ class TensorCache {
index++;
if (Category == CONST || Category == TF_CONST || Category == GRAPH_INPUT) {
tensor->refCount = 1;
tensor->nodeType = schema::NodeType_ValueNode;
tensor->nodeType = NodeType_ValueNode;
} else {
tensor->nodeType = schema::NodeType_Parameter;
tensor->nodeType = NodeType_Parameter;
}
tensor->name = name;
tensors.push_back(tensor);


+ 1
- 1
mindspore/lite/tools/converter/legacy_optimizer/fusion/matmul_biasadd_fusion_pass.cc View File

@@ -80,7 +80,7 @@ STATUS MatMulBiasAddFusionPass::DoFusion(MetaGraphT *graph, const std::string &p
MS_ASSERT(graph->allTensors.size() > baNodeInputIndex.at(BIASADD_OP_BIAS_INDEX));
const auto &baNodeBiasTensor = graph->allTensors.at(baNodeInputIndex.at(BIASADD_OP_BIAS_INDEX));
MS_ASSERT(baNodeBiasTensor != nullptr);
if (baNodeBiasTensor->refCount != schema::NodeType::NodeType_ValueNode) {
if (baNodeBiasTensor->refCount != NodeType_ValueNode) {
// dont fusion, return
return RET_OK;
}


+ 2
- 2
mindspore/lite/tools/converter/legacy_optimizer/fusion/mul_add_fusion_pass.cc View File

@@ -94,7 +94,7 @@ STATUS MulAddFusionPass::DoFusion(MetaGraphT *graph, const std::string &patternN
MS_ASSERT(graph->allTensors.size() > mulNodeInputIndex.at(MUL_OP_BIAS_INDEX));
const auto &mulNodeBiasTensor = graph->allTensors.at(mulNodeInputIndex.at(MUL_OP_BIAS_INDEX));
MS_ASSERT(mulNodeBiasTensor != nullptr);
if (mulNodeBiasTensor->nodeType != schema::NodeType::NodeType_ValueNode) {
if (mulNodeBiasTensor->nodeType != NodeType_ValueNode) {
// dont fusion, return
return RET_OK;
}
@@ -111,7 +111,7 @@ STATUS MulAddFusionPass::DoFusion(MetaGraphT *graph, const std::string &patternN
MS_ASSERT(graph->allTensors.size() > addNodeInputIndex.at(ADD_OP_BIAS_INDEX));
const auto &addNodeBiasTensor = graph->allTensors.at(addNodeInputIndex.at(ADD_OP_BIAS_INDEX));
MS_ASSERT(addNodeBiasTensor != nullptr);
if (addNodeBiasTensor->nodeType != schema::NodeType::NodeType_ValueNode) {
if (addNodeBiasTensor->nodeType != NodeType_ValueNode) {
// dont fusion, return
return RET_OK;
}


+ 2
- 2
mindspore/lite/tools/converter/legacy_optimizer/graph/batchnorm_convert_scale_pass.cc View File

@@ -108,7 +108,7 @@ STATUS BatchNormConvertScalePass::GenNewScaleTensor(MetaGraphT *graph, const std
}
newScaleWeightTensor->dataType = bnMeanTensor->dataType;
newScaleWeightTensor->format = bnMeanTensor->format;
newScaleWeightTensor->refCount = schema::NodeType::NodeType_ValueNode;
newScaleWeightTensor->refCount = NodeType_ValueNode;
newScaleWeightTensor->dims = bnMeanTensor->dims;
auto weightShapeSize = GetShapeSize(*bnMeanTensor);
newScaleWeightTensor->data.resize(weightShapeSize * sizeof(float));
@@ -131,7 +131,7 @@ STATUS BatchNormConvertScalePass::GenNewScaleTensor(MetaGraphT *graph, const std
newScaleBiasTensor->dataType = bnMeanTensor->dataType;
newScaleBiasTensor->format = bnMeanTensor->format;

newScaleBiasTensor->refCount = schema::NodeType::NodeType_ValueNode;
newScaleBiasTensor->refCount = NodeType_ValueNode;
newScaleBiasTensor->dims = bnMeanTensor->dims;
weightShapeSize = GetShapeSize(*bnMeanTensor);
newScaleBiasTensor->data.resize(weightShapeSize * sizeof(float));


+ 1
- 1
mindspore/lite/tools/converter/legacy_optimizer/graph/switch_pass.cc View File

@@ -215,7 +215,7 @@ std::unique_ptr<schema::CNodeT> SingleSwitchPass::MakeMergeNode(const std::strin
merge_node->inputIndex.push_back(old_merge_input[i]);
} else {
auto tensor = NewTensor(in_tensor);
tensor->nodeType = schema::NodeType_CNode;
tensor->nodeType = NodeType_CNode;
graph_->allTensors.push_back(std::move(tensor));
merge_node->inputIndex.push_back(graph_->allTensors.size() - 1);
}


+ 1
- 1
mindspore/lite/tools/converter/legacy_optimizer/graph/topological_sort_pass.cc View File

@@ -31,7 +31,7 @@ STATUS TopologicalSortPass::Run(schema::MetaGraphT *graph) {
std::vector<size_t> sinked_tensor_idxes;
// put all const tensor index into sinked_tensor_idxes
for (size_t i = 0; i < graph->allTensors.size(); i++) {
if (graph->allTensors.at(i)->nodeType == schema::NodeType::NodeType_ValueNode) {
if (graph->allTensors.at(i)->nodeType == NodeType_ValueNode) {
sinked_tensor_idxes.insert(sinked_tensor_idxes.end(), i);
}
}


+ 2
- 1
mindspore/lite/tools/converter/parser/caffe/caffe_node_parser.cc View File

@@ -18,6 +18,7 @@
#include <memory>
#include "securec/include/securec.h"
#include "ir/dtype/type_id.h"
#include "src/common/utils.h"

namespace mindspore {
namespace lite {
@@ -33,7 +34,7 @@ schema::TensorT *ConvertWeight(const caffe::BlobProto &proto) {
ConvertShape(proto, &shapeVec);
weight->dims = shapeVec;
weight->dataType = kNumberTypeFloat32;
weight->nodeType = schema::NodeType::NodeType_ValueNode;
weight->nodeType = NodeType_ValueNode;

// cal Weight num
int count = 1;


Write Preview
Loading…
Cancel
Save