arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2404.08856cs.CLcs.AIcs.LG

多模态大语言模型的投机解码研究

On Speculative Decoding for Multimodal Large Language Models

Qualcomm AI Research(高通人工智能研究中心);Qualcomm Technologies, Inc.(高通技术公司)

Mukul Gagrani, Raghavv Goel, Wonseok Jeon, Junyoung Park, Mingu Lee, Christopher Lott

AI总结:

针对多模态大语言模型推理慢的问题,研究将投机解码应用于LLaVA 7B,用纯语言模型作草稿模型实现最高2.37倍加速,还提出带图像适配器的紧凑型草稿模型小幅提升图像字幕性能。

摘要:

多模态大语言模型(MLLMs)的推理速度较慢,原因在于其大语言模型骨干存在内存带宽瓶颈,且采用自回归方式生成token。本文探索了投机解码在提升MLLM推理效率方面的应用,具体针对LLaVA 7B模型展开研究。研究表明,纯语言模型可作为LLaVA 7B投机解码的优质草稿模型,无需在草稿模型中设置图像token及其相关处理组件。我们在三项不同任务上开展的实验显示,使用从头训练的1.15亿参数语言模型,投机解码可实现最高2.37倍的内存受限加速比。此外,我们提出了一种集成图像适配器的紧凑型LLaVA草稿模型,该模型在图像字幕任务中实现了小幅性能提升,同时在其他任务上保持了相当的结果。

英文摘要:

Inference with Multimodal Large Language Models (MLLMs) is slow due to their large-language-model backbone which suffers from memory bandwidth bottleneck and generates tokens auto-regressively. In this paper, we explore the application of speculative decoding to enhance the inference efficiency of MLLMs, specifically the LLaVA 7B model. We show that a language-only model can serve as a good draft model for speculative decoding with LLaVA 7B, bypassing the need for image tokens and their associated processing components from the draft model. Our experiments across three different tasks show that speculative decoding can achieve a memory-bound speedup of up to 2.37$\times$ using a 115M parameter language model that we trained from scratch. Additionally, we introduce a compact LLaVA draft model incorporating an image adapter, which shows marginal performance gains in image captioning while maintaining comparable results in other tasks.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑