Paper Overview
Field: LLM Authors: Joan Vendrell Gallart, Russell Bent, Michael Grosskopf Published: 2026-05-28 arXiv: 2605.27703
Summary
Large Language Models are increasingly deployed inside agentic systems, where they must follow structured protocols, adapt to evolving states, and operate under memory, latency, and cost constraints. In such regimes, prompt extension is unreliable: growing contexts can push compact models outside their effective prompt domain, while deployment-time fine-tuning remains limited by scarce data and compute.
The authors propose a hierarchical control-and-learning framework in which a compact model is first distilled to learn the required output schema, then supervised online by an oracle-controller loop. The controller monitors protocol validity and semantic performance, projects accumulated histories into a feasible prompt domain, and triggers lightweight oracle-supervised fine-tuning under drift. This separates schema learning for communication compatibility from semantic adaptation for task-level correction.
The work formalizes prompt-domain feasibility and attention-induced saturation, arguing that one should control the effective prompt state rather than rely on nominal context length. Using multi-fidelity Bayesian optimization as a controlled sequential testbed, the authors characterize core deployment failure modes and demonstrate reliability and cost-efficiency improvements compared to non-hierarchical, distillation-only, and non-distillation baselines.
--- *Auto-collected on 2026-05-29*