Illegal instruction (core dumped) using forward in libtorch

Question

I'm trying to load and use a model from Python into c++ using libtorch. The program compiles correctly but I get Illegal instruction (core dumped) using forward on Input.

That's the code:

void test(vector<module_type>& model){
    //pseudo input
    vector<torch::jit::IValue> inputs;
    inputs.push_back(torch::ones({1, 3, 224, 224}));
    //ERROR IS HERE
    at::Tensor output = model[0].forward(inputs).toTensor();
    cout << output << endl;
}

int main(int argc, char *argv[]) {

    if (argc == 2){
        cout << argv[1] << endl;
    }
    else {
        cerr << "no path of model is given" << endl;
        return -1;
    }
    // test
    module_type module = torch::jit::load(argv[1]);
    vector<module_type> modul;
    modul.push_back(module);
    test(modul);

}

CMake:

cmake_minimum_required(VERSION 3.0 FATAL_ERROR)
project(main)

find_package(Torch REQUIRED)

add_executable(main main.cpp)
target_link_libraries(main "${TORCH_LIBRARIES}")
set_property(TARGET main PROPERTY CXX_STANDARD 11)

Answer 1

1) torch::jit::load return type is std::shared_ptr<torch::jit::script::Module>, so your code should rather be at::Tensor output = model[0]->forward(inputs).toTensor();

2) It is possible that for some reason the export of your Python model failed, but it's hard to tell without seeing the actual python code you used. To see what how many methods are available, try :

auto module = torch::jit::load(argv[1]);
size_t number_of_methods = module->get_methods().size();

Basically, if number_of_methods is 0, you have a problem : the serialized object contains no method (and the issue comes from your python code). Otherwise, the forward method should be available.