Decoupling Logic from Persona: Structural Immunity of Edge LLM Agents to Context Pollution

Key points

• Small language-model agents on edge devices must hold a persona and reason correctly at once, inside one context window that fills with conversational history and persona instructions.1
• The logic path receives only the core turn and emits a verifiable structured state (Micro-State); the persona path renders it in character with the full history.1
• Separation costs one extra decode on a topic's first turn (28.2 s vs 18.2 s on Llama) and buys persona hot-swapping in 1.7 ms without re-running the logic path.1
• Code, rubric, fixtures, adapters and logs are released.1

Sources (1)

• [1]Decoupling Logic from Persona: Structural Immunity of Edge LLM Agents to Context Pollution
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 09:53 AM

  We study what happens to the logical part of such an agent when that history is long, misleading and persona-heavy (persona-logic interference), and present a Decoupling Architecture (AO-DA) that separates logical inference ("What") from persona expression ("How") into two inference paths on one INT4 base model with hot-swappable LoRA adapters.

  Small language-model agents on edge devices must hold a persona and reason correctly at once, inside one context window that fills with conversational history and persona instructions.

Extractive summary: sentences quoted from the sources.