@inproceedings{extremelanguagemodelcompressionwith1, title = {Extremely Small BERT Models from Mixed-Vocabulary Training}, author = {Sanqiang Zhao and Raghav Gupta and Yang song and Denny Zhou}, year = {2019}, booktitle = {EACL 2021 2}, eprint = {1909.11687}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/1909.11687v2}, }