@inproceedings{lee2026polestar,
  title={Polestar: Drift-Aware Cache Calibration and Token Commitment for Efficient Inference of Diffusion {LLM}s},
  author={Mingyu Lee and Akshat Ramachandran and Souvik Kundu and Tushar Krishna},
  booktitle={The Fortieth Annual Conference on Neural Information Processing Systems},
  year={2026},
  url={https://arxiv.org/abs/2607.14107}
}