Exploring Neural Models for Parsing Natural Language into First-Order Logic