@article{zhang2025dfloat11,
  title={70\% Size, 100\% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float},
  author={Zhang, Tianyi and Hariri, Mohsen and Zhong, Shaochen and Chaudhary, Vipin and Sui, Yang and Hu, Xia and Shrivastava, Anshumali},
  journal={arXiv preprint arXiv:2504.11651},
  year={2025}
}