AI RESEARCH
Closed-Loop LLM Discovery of Non-Standard Channel Priors in Vision Models
arXiv CS.CV
•
ArXi:2601.08517v2 Announce Type: replace Channel-configuration search, the optimization of layer specifications such as channel widths in deep neural networks, presents a combinatorial challenge constrained by tensor-shape compatibility and computational budgets. We investigate whether large language models (LLMs) can neural architecture search (NAS) by reasoning over architectural code structures in ways that complement traditional search heuristics.