解决结构化序列生成中约束难满足的问题，提升输出忠实度。
LatticeBridge: Rare-Event Sequential Inference for Faithful Structured Sequence Synthesis
- 用动态构建的自动机和蒙特卡洛采样，精准追踪多个约束条件。
- 在2610个任务上，精确满足约束的比例显著高于传统方法。
- 适合需要高忠实度的文本生成场景，如数据到文本、知识库生成。
结构化序列生成常需在单个输出中满足多个由输入导出的约束。标准解码方法可能赋予流畅延续高概率，却忽略同时实现所有锚点的路径。本文将此问题视为罕见事件序列推断问题。LatticeBridge结合紧凑前缀语言模型、实例编译的表面自动机，以及具有重采样、多级分裂和源自实例短语的源支持提议项的扭曲序列蒙特卡洛（SMC）解码器。约束表示从每个输入实例编译而来，无需人工标注词类。在涵盖CommonGen、E2E NLG和WikiBio的2,610个可达成验证任务上，粒子解码器在共享提议模型下，相比贪婪、束滤波和最佳k祖先基线，在精确锚点满足率和平均锚点覆盖率上均有提升。由于精确锚点满足不排除未支持属性替换，评估报告了所需锚点覆盖率、源覆盖率、源侵入诊断、重叠度、运行时间与粒子统计量。该基准刻画了固定提议模型下的忠实度-重叠-延迟权衡边界。
原文摘要 · Abstract (English)
Structured sequence generation often requires a model to satisfy several input-derived constraints in a single output. Standard decoding methods may assign high probability to fluent continuations while placing low mass on continuations that realize all required anchors jointly. We study this regime as a rare-event sequential inference problem. LatticeBridge combines a compact prefix language model, instance-compiled surface automata, and a twisted sequential Monte Carlo (SMC) decoder with resampling, multilevel splitting, and a source-support proposal term derived from instance-provided phrases. The constraint representation is compiled from each input instance and does not rely on manually curated lexical classes. On 2,610 attainable validation tasks spanning CommonGen, E2E NLG, and WikiBio, the particle decoder improves exact anchor satisfaction and mean anchor coverage over greedy, beam-filtered, and best-of-k ancestral baselines under a shared proposal model. Since exact anchor satisfaction alone does not rule out unsupported attribute substitutions, the evaluation reports required-anchor coverage, source coverage, source-intrusion diagnostics, overlap, runtime, and particle statistics jointly. The benchmark characterizes the faithfulness-overlap-latency frontier under a fixed proposal model.
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。