Merino: Entropy-driven Design for Generative Language Models on IoT Devices

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Youpeng, Lin, Ming, Tang, Huadong, Wu, Qiang, Wang, Jun
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917902878244864
author Zhao, Youpeng
Lin, Ming
Tang, Huadong
Wu, Qiang
Wang, Jun
author_facet Zhao, Youpeng
Lin, Ming
Tang, Huadong
Wu, Qiang
Wang, Jun
contents Generative Large Language Models (LLMs) stand as a revolutionary advancement in the modern era of artificial intelligence (AI). However, scaling down LLMs for resource-constrained hardware, such as Internet-of-Things (IoT) devices requires non-trivial efforts and domain knowledge. In this paper, we propose a novel information-entropy framework for designing mobile-friendly generative language models. The whole design procedure involves solving a mathematical programming (MP) problem, which can be done on the CPU within minutes, making it nearly zero-cost. We evaluate our designed models, termed MeRino, across fourteen NLP downstream tasks, showing their competitive performance against the state-of-the-art autoregressive transformer models under the mobile setting. Notably, MeRino achieves similar or better performance on both language modeling and zero-shot learning tasks, compared to the 350M parameter OPT while being 4.9x faster on NVIDIA Jetson Nano with 5.5x reduction in model size.
format Preprint
id arxiv_https___arxiv_org_abs_2403_07921
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Merino: Entropy-driven Design for Generative Language Models on IoT Devices
Zhao, Youpeng
Lin, Ming
Tang, Huadong
Wu, Qiang
Wang, Jun
Machine Learning
Artificial Intelligence
Computation and Language
Generative Large Language Models (LLMs) stand as a revolutionary advancement in the modern era of artificial intelligence (AI). However, scaling down LLMs for resource-constrained hardware, such as Internet-of-Things (IoT) devices requires non-trivial efforts and domain knowledge. In this paper, we propose a novel information-entropy framework for designing mobile-friendly generative language models. The whole design procedure involves solving a mathematical programming (MP) problem, which can be done on the CPU within minutes, making it nearly zero-cost. We evaluate our designed models, termed MeRino, across fourteen NLP downstream tasks, showing their competitive performance against the state-of-the-art autoregressive transformer models under the mobile setting. Notably, MeRino achieves similar or better performance on both language modeling and zero-shot learning tasks, compared to the 350M parameter OPT while being 4.9x faster on NVIDIA Jetson Nano with 5.5x reduction in model size.
title Merino: Entropy-driven Design for Generative Language Models on IoT Devices
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2403.07921