2023 AcceleratingLlmInferencewithSta

From GM-RKB
(Redirected from Spector & Ré, 2023)
Jump to navigation Jump to search

Subject Headings: LLM Inference, Staged Speculative Decoding, Small-Batch LLM Inference

Notes

Cited By

Quotes

Abstract

Recent advances with large language models (LLM) illustrate their diverse capabilities. We propose a novel algorithm, staged speculative decoding, to accelerate LLM inference in small-batch, on-device scenarios. We address the low arithmetic intensity of small-batch inference by improving upon previous work in speculative decoding. First, we restructure the speculative batch as a tree, which reduces generation costs and increases the expected tokens per batch. Second, we add a second stage of speculative decoding. Taken together, we reduce single-batch decoding latency by 3.16x with a 762M parameter GPT-2-L model while perfectly preserving output quality.

References

;

 AuthorvolumeDate ValuetitletypejournaltitleUrldoinoteyear
2023 AcceleratingLlmInferencewithStaChristopher Ré
Benjamin Spector
Accelerating Llm Inference with Staged Speculative Decoding10.48550/arXiv.2308.046232023