Can I Creat Model during Runtime?

C++
2

One way to achieve this is by defining the model in TorchScript and then training it in C++:

  1. Make sure your PyTorch and libtorch installation are of the same version / commit. One way to ensure it is to install the nightly builds:

    1. PyTorch: pip install torch_nightly -f https://download.pytorch.org/whl/nightly/cu90/torch_nightly.html (change “cu90” to your CUDA version)
    2. libtorch: wget https://download.pytorch.org/libtorch/nightly/cpu/libtorch-shared-with-deps-latest.zip &&
      unzip libtorch-shared-with-deps-latest.zip

  2. Run the following Python script to export a model:

    import torch
    
    class Model(torch.jit.ScriptModule):
        def __init__(self):
            super(Model, self).__init__()
            self.linear = torch.nn.Linear(2, 2)
    
        @torch.jit.script_method
        def forward(self, x):
            a = 0
            while a < 4: # While loop example
                print(a)
                a += 1
            x = self.linear(x)
            return x
    
    model = Model()
    model.save('model1.pt')
  1. Create a cpp file called example-app.cpp that contains the following:
    #include <torch/torch.h>
    #include <torch/script.h>
    #include <iostream>
    
    int main() {
      std::string module_path = "../model1.pt";
      // NOTE: depending on the libtorch version, you might need to change this line to `torch::jit::script::Module module = torch::jit::load(module_path);`
      // See https://discuss.pytorch.org/t/how-to-load-previously-saved-model-in-cpp-frontend/52336/6 for details.
      std::shared_ptr<torch::jit::script::Module> module = torch::jit::load(module_path);
    
      // TorchScript module's forward method only accepts std::vector<torch::jit::IValue> as input
      std::vector<torch::jit::IValue> inputs;
      inputs.push_back(torch::randn({2, 2}).set_requires_grad(true));
    
      std::cout << "gradient of self.linear.weight: " << "\n";
      std::cout << module->get_module("linear")->get_parameter("weight").grad() << "\n";
    
      torch::Tensor prediction = module->forward(inputs).toTensor();
      prediction.sum().backward();
    
      // Verify that backward works
      std::cout << "gradient of self.linear.weight: " << "\n";
      std::cout << module->get_module("linear")->get_parameter("weight").grad() << "\n";
    }
  1. Create CMakeLists.txt, put in the following:
    cmake_minimum_required(VERSION 3.0 FATAL_ERROR)
    project(example-app)
    
    find_package(Torch REQUIRED)
    
    add_executable(example-app example-app.cpp)
    target_link_libraries(example-app "${TORCH_LIBRARIES}")
    set_property(TARGET example-app PROPERTY CXX_STANDARD 11)
  1. mkdir build && cd build
  2. cmake -DCMAKE_PREFIX_PATH=/absolute/path/to/libtorch .. # /absolute/path/to/libtorch should be the absolute path to the unzipped LibTorch distribution
  3. make
  4. Run example-app to see output

Later on, when you want to modify the network structure of your model, you can change the model description code in Python, export the model again, and then re-train the model using the same C++ code. This way you don’t need to recompile the C++ code.

2 Likes