@article{softmaxgradienttamperingdecouplingthe1,
  title = {Altering Backward Pass Gradients improves Convergence},
  author = {Bishshoy Das and Milton Mondal and Brejesh lall and Shiv Dutt Joshi and Sumantra Dutta Roy},
  year = {2021},
  eprint = {2111.12495},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2111.12495v3},
}