Reliable automatic evaluation of summarization systems is challenging du...
Existing data-to-text generation datasets are mostly limited to English....
With recent improvements in natural language generation (NLG) models for...
NLP researchers need more, higher-quality text datasets. Human-labeled
d...
We introduce GEM, a living benchmark for natural language Generation (NL...
This thesis studies how the segmentation results, produced by convolutio...
Confidently making progress on multilingual modeling requires challengin...