@misc{indiciaedb6e41820639,
  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},
  url = {https://arxiv.org/abs/2502.11895},
  note = {Source identifier: 2502.11895}
}