@article{continualquantizationawarepretraining,
  title = {Continual Quantization-Aware Pre-Training: When to transition from 16-bit to 1.58-bit pre-training for BitNet language models?},
  author = {Jacob Nielsen and Peter Schneider-Kamp and Lukas Galke},
  year = {2025},
  eprint = {2502.11895},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2502.11895v1},
}