Created by: banbishan

安卓9手机运行编译好的c++可执行文件，optimize_out文件夹里为用tools优化过的__model__.nb和param.nb文件 ./model --model_dir="optimize_out" --optimized_model_dir="op" 报错如下

terminating with uncaught exception of type std::length_error: basic_string
Aborted

程序如下：

#include <gflags/gflags.h>
#include <stdio.h>
#include <vector>
#include <iostream>
#include <time.h>
#include "paddle_api.h"          // NOLINT
#include "paddle_use_kernels.h"  // NOLINT
#include "paddle_use_ops.h"      // NOLINT
#include "paddle_use_passes.h"   // NOLINT
using namespace paddle::lite_api;  // NOLINT
using namespace std;
DEFINE_string(model_dir, "", "Model dir path.");
DEFINE_string(optimized_model_dir, "", "Optimized model dir.");
DEFINE_bool(prefer_int8_kernel, false, "Prefer to run model with int8 kernels");

int64_t ShapeProduction(const shape_t& shape) {
  int64_t res = 1;
  for (auto i : shape) res *= i;
  return res;
}

void RunModel() {
  // 1. Set CxxConfig
  CxxConfig config;
  config.set_model_dir(FLAGS_model_dir);
  std::vector<Place> valid_places{Place{TARGET(kARM), PRECISION(kFloat)}};
  if (FLAGS_prefer_int8_kernel) {
    valid_places.push_back(Place{TARGET(kARM), PRECISION(kInt8)});
    config.set_preferred_place(Place{TARGET(kARM), PRECISION(kInt8)});
  } else {
    config.set_preferred_place(Place{TARGET(kARM), PRECISION(kFloat)});
  }
  config.set_valid_places(valid_places);

  // 2. Create PaddlePredictor by CxxConfig
  std::shared_ptr<PaddlePredictor> predictor =
      CreatePaddlePredictor<CxxConfig>(config);

  // 3. Prepare input data
  std::unique_ptr<Tensor> input_tensor(std::move(predictor->GetInput(0)));
  input_tensor->Resize(shape_t({1,1,48,512}));
  auto* data = input_tensor->mutable_data<float>();
  for(int i = 0; i < ShapeProduction(input_tensor->shape()); ++i){
      data[i] = 1;
  }

  std::unique_ptr<Tensor> init_scores(std::move(predictor->GetInput(2)));
  init_scores->Resize(shape_t({1,1}));
  auto* data_scores = init_scores->mutable_data<float>();
  for(int i = 0; i < ShapeProduction(init_scores->shape()); ++i){
      data_scores[i] = 0;
  }
  lod_t lod_s{{0,1},{0,1}};
  init_scores->SetLoD(lod_s);

  std::unique_ptr<Tensor> init_ids(std::move(predictor->GetInput(1)));
  init_ids->Resize(shape_t({1,1}));
  auto* data_ids = init_ids->mutable_data<float>();
  for(int i = 0; i < ShapeProduction(init_ids->shape()); ++i){
      data_ids[i] = 0;
  }
  lod_t lod_i{{0,1},{0,1}};
  init_ids->SetLoD(lod_i);
  // 4. Run predictor
  predictor->Run();
  clock_t start = clock();
  for (int i = 0; i <1000; ++i){
      predictor->Run();
  }
  clock_t end = clock();
  std::cout<<(double)(end-start)/CLOCKS_PER_SEC<<std::endl;
  // 6. Save optimition model
  predictor->SaveOptimizedModel(FLAGS_optimized_model_dir,
                                LiteModelType::kNaiveBuffer);
}

int main(int argc, char** argv) {
  google::ParseCommandLineFlags(&argc, &argv, true);
  RunModel();
  return 0;
}