@article{speculativedecodingviahybriddraftingand, title = {Speculative Decoding via Hybrid Drafting and Rollback-Aware Branch Parallelism}, author = {Yuhao Shen and Junyi Shen and Quan Kong and Tianyu Liu and Yao Lu and Cong Wang}, year = {2025}, eprint = {2506.01979}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2506.01979v1}, }