通过自适应检索与分支复用，提升无模型推测解码速度
AdaPLD: Adaptive Retrieval and Reuse for Efficient Model-Free Speculative Decoding
- 自适应融合词法与语义匹配，扩大复用范围
- 构建分支复用假设，应对生成不确定性
- 无需训练，实测最高提速3.10倍
推测解码通过单次目标模型前向传播验证多个草稿标记，减少序列解码迭代次数。无模型变体不依赖辅助草稿模型，而是复用生成过程中已有的文本和模型状态，但其加速效果依赖草稿的可靠性。现有复用方法存在两大局限：词法锚定检索在表面形式变化下召回率低；确定性跨度复制在检索上下文无法唯一决定后续内容时易失效。我们提出AdaPLD，一种无需训练的方法，可自适应优化检索与草稿构建。AdaPLD保持高精度词法复用的同时，利用语义相似性在词法匹配失败时恢复额外复用机会；进一步构建分支复用假设以应对后续生成的不确定性，而非依赖单一复制片段。在多个基准测试中，AdaPLD减少了目标模型前向传播次数，实现最高达3.10倍的解码加速。
原文摘要 · Abstract (English)
Speculative decoding accelerates generation by verifying multiple drafted tokens in a single target-model forward pass, reducing sequential decoding iterations. Model-free variants avoid auxiliary draft models by reusing text and model states already available during generation, but their speedup depends on the reliability of the constructed drafts. We identify two limitations of existing reuse-based methods: lexically anchored retrieval has limited recall under surface-form variation, and deterministic span copying can be brittle when the retrieved context does not uniquely determine the continuation. We propose \emph{AdaPLD}, a training-free method that adaptively improves both retrieval and draft construction. AdaPLD preserves high-precision lexical reuse while using semantic similarity to recover additional reuse opportunities when lexical matching fails. It further constructs branched reuse hypotheses to account for continuation uncertainty, rather than relying on a single copied span. Across diverse benchmarks, AdaPLD reduces target-model forward passes and achieves up to $3.10\times$ decoding speedup.
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。