@article{transformersprovablylearnsparsetoken,
  title = {Transformers Provably Learn Sparse Token Selection While Fully-Connected Nets Cannot},
  author = {Zixuan Wang and Stanley Wei and Daniel Hsu and Jason D. Lee},
  year = {2024},
  eprint = {2406.06893},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2406.06893v1},
}