---
license: apache-2.0
---

The Acceptance Prediction Head for Llama-2-chat 7B and 70B model pair trained with `weight_mismatch=6` and `resnet_num_layers=3`. It is recommended to be used with `stop_threshold=0.7`. See [arxiv: 2405.19715](https://arxiv.org/abs/2405.19715) for more details.

Usage: [GitHub](https://github.com/Kaffaljidhmah2/SpecDec_pp)


### Citation

```bibtex
@article{huang2024specdec++,
  title={SpecDec++: Boosting Speculative Decoding via Adaptive Candidate Lengths},
  author={Huang, Kaixuan and Guo, Xudong and Wang, Mengdi},
  journal={arXiv preprint arXiv:2405.19715},
  year={2024}
}
```