HASS-Llama3-8B-Instruct-Reproduce

HASS Baseline Checkpoint (Used in PosS)

This repository provides a reproduced HASS model checkpoint that is used as a baseline in PosS (Position Specialist) experiments.

PosS is a speculative decoding method proposed in the paper:

PosS: Position Specialist Generates Better Draft for Speculative Decoding

In our experiments, this HASS checkpoint serves as the baseline draft model for comparison with the proposed position-specialized draft models.

---

🔗 Code

The full implementation of PosS, along with training details and evaluation scripts (including EAGLE-2 and HASS baselines), is available at:

👉 GitHub: https://github.com/shrango/PosS

---

📦 Files

If the model is not automatically downloaded by your framework, you may manually download the following files from this repository:

• pytorch_model.bin — model weights
• config.json — model configuration

---

📖 Citation

If you use this checkpoint in the context of PosS or refer to the PosS method, please cite:

@misc{huang2025posspositionspecialistgenerates,
  title        = {POSS: Position Specialist Generates Better Draft for Speculative Decoding},
  author       = {Langlin Huang and Chengsong Huang and Jixuan Leng and Di Huang and Jiaxin Huang},
  year         = {2025},
  eprint       = {2506.03566},
  archivePrefix= {arXiv},
  primaryClass = {cs.CL},
  url          = {https://arxiv.org/abs/2506.03566}
}