@article{whenwhyandhowmuchadaptivelearningrate,
  title = {Optimal Linear Decay Learning Rate Schedules and Further Refinements},
  author = {Aaron Defazio and Ashok Cutkosky and Harsh Mehta and Konstantin Mishchenko},
  year = {2023},
  eprint = {2310.07831},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2310.07831v2},
}