REST: Retrieval-Based Speculative Decoding (arXiv:2311.08252)

Fields

id

sources/pap-eff-rest-speculative

URL

https://arxiv.org/abs/2311.08252

Kind

blog

Accessed

2026-10-03

notes

arXiv:2311.08252. REST: Retrieval-Based Speculative Decoding. By Zhenyu He et al. Submitted 2023-11-14. Retrieval-based speculative decoding that uses retrieval instead of a draft model. 1.62x-2.36x speedup on 7B and 13B models. 2026-10-10: Fetched https://arxiv.org/abs/2311.08252. Title: 'REST: Retrieval-Based Speculative Decoding'. Peer-reviewed venue confirmed: NAACL 2024. Abstract confirms retrieval-based speculative decoding using datastore retrieval without a draft model.

Notes

arXiv:2311.08252. REST: Retrieval-Based Speculative Decoding. By Zhenyu He et al. Submitted 2023-11-14. Retrieval-based speculative decoding that uses retrieval instead of a draft model. 1.62x-2.36x speedup on 7B and 13B models.

2026-10-10: Fetched https://arxiv.org/abs/2311.08252. Title: 'REST: Retrieval-Based Speculative Decoding'. Peer-reviewed venue confirmed: NAACL 2024. Abstract confirms retrieval-based speculative decoding using datastore retrieval without a draft model.

Cross-references

This record references no other records.