================ ---------------- joker-eph wrote: This pass is supposed to expand arith ops to make them convertible to LLVM: can you expand how does your use of the tensor dialect fits here? https://github.com/llvm/llvm-project/pull/206249