[提交于 2026年5月12日]

标题：多智能体LLM级联中的语义语域压缩

查看论文PDF：语义语域压缩在多智能体LLM级联中的应用 HTML (实验性)

摘要：多智能体LLM系统通常将复杂任务分解为专门的职责角色。然而，这种模块化引入了一种表征风险：当中间代理在不同的语言语域间转换文本时，它们可能会系统性地压缩下游决策所需的关键语义区分。我们将这一现象称为语义语域压缩，并将其描述为多智能体级联中一种可观察的故障模式。利用三代理流水线（收集器-评估器-决策器），我们通过句子转换器嵌入空间中的标签间分离度来量化压缩。在政治事实核查（LIAR）、情感分析（SST-5）和医疗分诊（Triagegeist）中，关键评估在评估器阶段一致地将标签可分离性降低了41.7%，而身份传递几乎完整地保留了它。五种架构变体因果性地将定向语义转换隔离为主要驱动因素。一种寻求可信度的变体产生了最小的几何压缩，但其输出却转向了“大多为真”，表明转换效价独立于压缩幅度控制着分布崩溃的方向。压缩在三个领域中以不同强度泛化：事实核查中为41.7%，情感分析中为27.2%，分诊中为20.0%。提示层面的回归解释了78%的方差，操作约束与较低的压缩相关。这些结果表明，语义语域压缩是多智能体LLM系统中一种可测量且可泛化的现象，对高风险领域的安全评估具有影响。

提交历史

来自：Manuele Tele Junior Fernandez [ 查看邮箱 ] [v1] 2026年5月12日星期二 01:11:12 UTC (36 KB)

[Submitted on 12 May 2026]

Title:Semantic Register Compression in Multi-Agent LLM Cascades

View a PDF of the paper titled Semantic Register Compression in Multi-Agent LLM Cascades, by Manuele Tele Junior Fernandez

View PDF HTML (experimental)

Abstract:Multi-agent LLM systems commonly decompose complex tasks into specialized roles. However, this modularity introduces a representational risk: when intermediate agents transform text across linguistic registers, they can systematically compress the semantic distinctions needed for accurate downstream decisions. We term this phenomenon semantic register compression and characterize it as an observable failure mode in multi-agent cascades. Using a three-agent pipeline (Collector-Evaluator-Decider), we quantify compression via inter-label separation in sentence-transformer embedding space. Across political fact-checking (LIAR), sentiment analysis (SST-5), and medical triage (Triagegeist), critical evaluation consistently reduces label separability by 41.7% at the Evaluator stage, while identity passthrough preserves it nearly fully. Five architectural variants causally isolate oriented semantic transformation as the primary driver. A credibility-seeking variant produces minimal geometric compression yet shifts outputs toward mostly-true, demonstrating that transformation valence controls the direction of distributional collapse independently of compression magnitude. Compression generalizes across the three domains with varying intensity: 41.7% in fact-checking, 27.2% in sentiment, and 20.0% in triage. Prompt-level regression explains 78% of the variance, with operational constraints associated with lower compression. These results demonstrate that semantic register compression is a measurable and generalizable phenomenon in multi-agent LLM systems, with implications for safety evaluation in high-stakes domains.

Submission history

From: Manuele Tele Junior Fernandez [

view email

]

[v1]

Tue, 12 May 2026 01:11:12 UTC (36 KB)