无需训练即可快速构建高效大模型草稿模型，显著降低推理延迟。
SDFP: Speculative Decoding with FIT-Pruned Models for Training-Free and Plug-and-Play LLM Acceleration
- 通过弗希信息迹剪枝法自动压缩大模型，生成轻量草稿模型。
- 在多个基准上实现1.32倍至1.5倍的解码加速，输出分布不变。
- 完全免训练、即插即用，适合低延迟多媒体应用部署。
大语言模型支撑着图像字幕、检索、推荐和内容生成等交互式多媒体应用，但其自回归解码带来显著延迟。推测解码利用轻量草稿模型降低延迟，但部署常受限于获取、调优和维护有效草稿模型的成本与复杂性。现有方法通常需额外训练或专用化，即使免训练方法也需昂贵搜索或优化。我们提出SDFP，一种完全免训练且即插即用的框架，通过基于弗希信息迹（FIT）的层剪枝构建草稿模型。以层敏感度为输出扰动代理，SDFP移除低影响层，得到紧凑草稿模型，同时保持与原模型的兼容性，支持标准推测验证。SDFP无需额外训练、超参数调优或独立维护草稿模型，实现快速、可部署的草稿构建。在多个基准测试中，SDFP实现1.32x-1.5x的解码速度提升，且不改变目标模型的输出分布，适用于低延迟多媒体应用。
原文摘要 · Abstract (English)
Large language models (LLMs) underpin interactive multimedia applications such as captioning, retrieval, recommendation, and creative content generation, yet their autoregressive decoding incurs substantial latency. Speculative decoding reduces latency using a lightweight draft model, but deployment is often limited by the cost and complexity of acquiring, tuning, and maintaining an effective draft model. Recent approaches usually require auxiliary training or specialization, and even training-free methods incur costly search or optimization. We propose SDFP, a fully training-free and plug-and-play framework that builds the draft model via Fisher Information Trace (FIT)-based layer pruning of a given LLM. Using layer sensitivity as a proxy for output perturbation, SDFP removes low-impact layers to obtain a compact draft while preserving compatibility with the original model for standard speculative verification. SDFP needs no additional training, hyperparameter tuning, or separately maintained drafts, enabling rapid, deployment-friendly draft construction. Across benchmarks, SDFP delivers 1.32x-1.5x decoding speedup without altering the target model's output distribution, supporting low-latency multimedia applications.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。