Structural Abstraction as an Inductive Bias for Non-Stationary Language Model Training

@misc{rahmati2026structural,
  title = {Structural Abstraction as an Inductive Bias for Non-Stationary Language Model Training},
  author = {Elnaz Rahmati and Nona Ghazizadeh and Zhivar Sourati and Nina Rouhani and Morteza Dehghani},
  year = {2026},
  eprint = {2603.17198},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2603.17198},
}