@inproceedings{improvingtransformeroptimizationthrough, title = {Improving Transformer Optimization Through Better Initialization}, author = {Xiao Shi Huang and Felipe Perez and Jimmy Ba and Maksims Volkovs}, year = {2020}, booktitle = {ICML 2020 1}, url = {https://proceedings.icml.cc/static/paper_files/icml/2020/5691-Paper.pdf}, }