VE-KD: Vocabulary-Expansion Knowledge-Distillation for Training Smaller Domain-Specific Language Models.

Pengju Gao, Tomohiro Yamasaki, Kazunori Imoto

Browse the full EMNLP paper archive.