no code implementations • WS 2020 • Hanoz Bhathena, Angelica Willis, Nathan Dass
We evaluate the compositionality of general-purpose sentence encoders by proposing two different metrics to quantify compositional understanding capability of sentence encoders.