How to deploy two different tvm compiled model in c++ statically?

I foud a workaround for the truncsfhf2 problem. Patch target/llvm/codegen_llvm.cc void CodeGenLLVM::EmitFloat16ConversionBuiltins(bool use_float16_abi) { … //auto builtins_module = llvm_target->GetInstance().ParseIR(module_ir); //link_modules.push_back(std::move(builtins_module)); } My model probably do not use half floats, so this should be OK, and allows me to link under windows.