Skip to main content

Model Compilation

Once you have exported your trained PyTorch model to a onnx file, you will need to compile the model into C binaries in order to run on the board as a CCS project.

Option 1: Use our TI-TVM Docker container​

  • Launch the container and run custom commands

    make tvm

  • Shortcut for launching the container and run compilation

    make tvm-compile MODEL_PATH=./your_model_name.onnx OUT_DIR=./model_artifacts

Example terminal output

Terminal Output

Output files

Output Files

Option 2: Setup your own environment​

The 2024-mmWaveRadarSensors team has written a detailed guide on model compilation, you can follow the instructions step by step to achieve the same result.

-> Model Compilation Guide