Agentic Neural Architecture Search

2026-07-08

AI Summary

This work proposes a novel paradigm that synergizes large language models (LLMs) with neural architecture search (NAS) to overcome the limited generalizability of conventional NAS methods, which rely on handcrafted search spaces. The approach begins by leveraging an LLM to generate high-quality initial architectures, which are then transformed into “slot-based” architectures containing replaceable modular components. This enables the automatic construction of task-adaptive, structured search spaces that balance open-ended generation with efficient search. Implemented through a modular three-stage pipeline without any human intervention, the method achieves state-of-the-art performance on 11 out of 17 cross-modal tasks, significantly outperforming existing baselines and expert-designed architectures, thereby demonstrating its generality and effectiveness.

Abstract

Neural architecture search (NAS) methods have grown increasingly efficient, yet they remain bounded by manually engineered search spaces that require substantial domain expertise and must be rebuilt for every new task. Large language models (LLMs) can generate architectures in an open-ended space, but how to optimally divide the labor between LLM-driven design and NAS-driven search remains unexplored. We propose a mechanism that bridges these two paradigms: an LLM produces a high-quality seed architecture, then decomposes it into a "slotted architecture", a scaffold with named, interchangeable module slots that automatically defines a bounded, task-specific search space for conventional NAS to explore, without manual engineering. We instantiate this mechanism in AgentNAS, a modular three-phase pipeline in which each component's contribution can be measured independently. On 17 tasks spanning classification, dense regression, segmentation, and multi-label tagging across diverse modalities (NAS-Bench-360 and Unseen NAS), AgentNAS establishes a new state of the art on 11 tasks, outperforming published baselines including task-specific expert designs. Ablation studies show that the two search mechanisms are broadly complementary: the LLM-generated seed already surpasses published baselines on the majority of tasks, and NAS delivers additional gains in most cases through combinatorial recombination across slots, a mode of search that independent LLM samples cannot replicate. These patterns hold across three LLMs of different capability levels, confirming that the division of labor is robust. Our code is available at https://github.com/alroimfebruary/AgentNAS.

Problem

• Neural Architecture Search
• Large Language Models
• Search Space Design
• Architecture Generation
• Task-specific Optimization

Innovation

• Neural Architecture Search
• Large Language Models
• Slotted Architecture
• Automated Search Space Design
• Agentic NAS