Yes, this solves my problem. Thank you. To me, the most confusing part is the python object hierarchy: so you write a Relax module (tvm.ir.module.IRModule), get a VMExecutable (tvm.relax.vm_build.VMExecutable) after relax.build() the module, the returned object has a method as_text() to show some kindof VM-instr representation and can be emulated on the host machine. Yet, there is a “lowered module” (tvm.runtime.module.Module) that can be either retrieved by VMExecutable.mod or directly from hand-written TE/TIR script, and only the lower Module can be built by tvm.build().
P.S. Would appreciate if these info is updated in the official tutorial sample code and well-documented. --- [Visit Topic](https://discuss.tvm.apache.org/t/tvm0-20-0-relax-how-to-codegen-llvm-ir-with-relax/18409/5) to respond. You are receiving this because you enabled mailing list mode. To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/1510cc2272ead0e1b276a71f58129c2274ce597c30fc645779a96ef78bafca55).
