2023 EuroLLVM - ML-on-CPU: should vectorization happen in the LLVM backend or higher up the stack?

#LLVM