@misc{indiciae86e3d04d000f,
  title = {SPIRe: Boosting LLM Inference Throughput with Speculative Decoding},
  author = {Sanjit Neelam and Daniel Heinlein and Vaclav Cvicek and Akshay Mishra and Reiner Pope},
  year = {2025},
  url = {https://arxiv.org/abs/2504.06419},
  note = {Source identifier: 2504.06419}
}