Sentence-level MT evaluation without reference translations: beyond language modeling

3483

Knowledge Triples