@article{normalizedattentionwithoutprobabilitycag,
  title = {Normalized Attention Without Probability Cage},
  author = {Oliver Richter and Roger Wattenhofer},
  year = {2020},
  eprint = {2005.09561},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2005.09561v1},
}