Unlocking efficiency in large language model inference: A comprehensive survey of speculative decoding

H Xia, Z Yang, Q Dong, P Wang, Y Li, T Ge… - arXiv preprint arXiv …, 2024 - arxiv.org
To mitigate the high inference latency stemming from autoregressive decoding in Large
Language Models (LLMs), Speculative Decoding has emerged as a novel decoding …

Llm inference unveiled: Survey and roofline model insights

Z Yuan, Y Shang, Y Zhou, Z Dong, Z Zhou… - arXiv preprint arXiv …, 2024 - arxiv.org
The field of efficient Large Language Model (LLM) inference is rapidly evolving, presenting a
unique blend of opportunities and challenges. Although the field has expanded and is …

Medusa: Simple llm inference acceleration framework with multiple decoding heads

T Cai, Y Li, Z Geng, H Peng, JD Lee, D Chen… - arXiv preprint arXiv …, 2024 - arxiv.org
The inference process in Large Language Models (LLMs) is often limited due to the absence
of parallelism in the auto-regressive decoding process, resulting in most operations being …

Towards efficient generative large language model serving: A survey from algorithms to systems

X Miao, G Oliaro, Z Zhang, X Cheng, H Jin… - arXiv preprint arXiv …, 2023 - arxiv.org
In the rapidly evolving landscape of artificial intelligence (AI), generative large language
models (LLMs) stand at the forefront, revolutionizing how we interact with our data. However …

Distillspec: Improving speculative decoding via knowledge distillation

Y Zhou, K Lyu, AS Rawat, AK Menon… - arXiv preprint arXiv …, 2023 - arxiv.org
Speculative decoding (SD) accelerates large language model inference by employing a
faster draft model for generating multiple tokens, which are then verified in parallel by the …

Eagle-2: Faster inference of language models with dynamic draft trees

Y Li, F Wei, C Zhang, H Zhang - arXiv preprint arXiv:2406.16858, 2024 - arxiv.org
Inference with modern Large Language Models (LLMs) is expensive and time-consuming,
and speculative sampling has proven to be an effective solution. Most speculative sampling …

Break the sequential dependency of llm inference using lookahead decoding

Y Fu, P Bailis, I Stoica, H Zhang - arXiv preprint arXiv:2402.02057, 2024 - arxiv.org
Autoregressive decoding of large language models (LLMs) is memory bandwidth bounded,
resulting in high latency and significant wastes of the parallel processing power of modern …

Rephrasing the web: A recipe for compute and data-efficient language modeling

P Maini, S Seto, H Bai, D Grangier, Y Zhang… - arXiv preprint arXiv …, 2024 - arxiv.org
Large language models are trained on massive scrapes of the web, which are often
unstructured, noisy, and poorly phrased. Current scaling laws show that learning from such …

Eagle: Speculative sampling requires rethinking feature uncertainty

Y Li, F Wei, C Zhang, H Zhang - arXiv preprint arXiv:2401.15077, 2024 - arxiv.org
Auto-regressive decoding makes the inference of Large Language Models (LLMs) time-
consuming. We propose a simple framework, EAGLE (Extrapolation Algorithm for Greater …

A survey on efficient inference for large language models

Z Zhou, X Ning, K Hong, T Fu, J Xu, S Li, Y Lou… - arXiv preprint arXiv …, 2024 - arxiv.org
Large Language Models (LLMs) have attracted extensive attention due to their remarkable
performance across various tasks. However, the substantial computational and memory …