When the Code Autopilot Breaks: Why Large Language Models Falter in Embedded Machine Learning

doi:10.1109/MC.2025.3603949

When the Code Autopilot Breaks: Why Large Language Models Falter in Embedded Machine Learning

2025-11-01

1

PRE

AI

DOI：10.1109/MC.2025.3603949

Original

Original request for help

Save

Abstract

En 中文

This article presents an empirical investigation of failure modes in large language model (LLM)-powered embedded machine learning pipeline, based on an autopilot framework that orchestrates data preprocessing, model conversion, and on-device inference code generation. Though grounded in specific devices, our study reveals broader challenges in LLM-based code generation.

Keywords:

Codes

Large language models

Computational modeling

Pipelines

Data preprocessing

Machine learning

Data models

Autopilot

When the Code Autopilot Breaks: Why Large Language Models Falter in Embedded Machine Learning

Abstract

Journal

Organization