Is there a way to load and inference a onnx with libtorch?

Hi, I’m wondering if it’s possible to load onnx in C++ (libtorch) and inference with PyTorch’s ops?