Installing TVM

• Install from Source
  • Developers: Get Source from Github
  • Build the Shared Library
  • Python Package Installation
  • Install Contrib Libraries
  • Enable C++ Tests
• Docker Images
  • Docker Source
• NNPACK Contrib Installation
  • Conditions
  • Build/Install NNPACK
  • Build TVM with NNPACK support

Visit the install TVM from source page to install TVM from the source code. Installing from source gives you the maximum flexibility to configure the build effectively from the official source releases. If you are interested in deploying to mobile or embedded devices, you do not need to install the entire TVM stack on your device. Instead, you only need the runtime and can install using the deployment and integration guide.

If you would like to quickly try out TVM or run some demo and tutorials, you can install from Docker. You can also use TVM locally through pip.

# Linux/MacOS CPU build only!
# See tlcpack.ai for other pre-built binaries including CUDA
pip install apache-tvm

For more details on installation of pre-built binaries, visit tlcpack.ai.