What if the result is not correct?

Sure, you can see an example of how to use the debug_executor in this part of the docs.

The tensors will be dumped to a file, and you can then load this data using code such as:

data = relay.load_param_dict(bytearray(open("./_tvmdbg_device_CPU_0/output_tensors.params", "rb").read()))