i want to convert my pytorch model to Keras, use tensorrt and then run the inference with c++. I read some tutorials like https://learnopencv.com/how-to-convert-a-model-from-pytorch-to-tensorrt- where it looks very easy, but i have some questions. When i create an onnx file i have a binary structure of my neural network with the weights or? I learned from torchscript there are to ways to convert a model (tracing, scripting) and i read the same here torch.onnx — PyTorch 1.8.1 documentation. When i try to export my model to onnx, will be it “traced” or “script”? I think as default it will be traced but then these tutorials aren’t helpful for me… Maybe someone can give me a short conclusion about my questions and an advice what i can do next, that would be very nice.