Deep Learning: How compilers help optimize models on hardware – Voonze

To efficiently use deep learning models on different target hardware, compilers such as TensorFlow XLA and Apache TVM use the LLVM compiler stack.

Link to Full Article: Read Here