You should evaluate your language model on marginal likelihood over tokenisations Kris Cao author Laura Rimell author 2021-11 text Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing Marie-Francine Moens editor Xuanjing Huang editor Lucia Specia editor Scott Wen-tau Yih editor Association for Computational Linguistics Online and Punta Cana, Dominican Republic conference publication cao-rimell-2021-evaluate 10.18653/v1/2021.emnlp-main.161 https://aclanthology.org/2021.emnlp-main.161/ 2021-11 2104 2114