Paper
28 October 2024 Evaluation on parameter-efficient continual instruction tuning of large language models
TianXiang Ren, XingShen Song, JiangYong Shi, JinSheng Deng, JiHao Cao
Author Affiliations +
Proceedings Volume 13404, Fifth International Conference on Control, Robotics, and Intelligent System (CCRIS 2024); 134040C (2024) https://doi.org/10.1117/12.3050469
Event: Fifth International Conference on Control, Robotics, and Intelligent System (2024), 2024, Macau, China
Abstract
Recent years have witnessed a spurt in large language models for its strong generalization performance, many continual instruction tuning methods based on parameter-efficient tuning have been proposed to further push large language models towards artificial general intelligence. However, comprehensive assessments of recent advancements in these methods are notably lacking. To fill the gap, a three-dimensional (average performance, continual learning proficiency and general ability delta) evaluation protocol are introduced based on two task streams, which include wide-span tasks and long sequence tasks. Meanwhile, three continual instruction tuning methods following different strategies are thoroughly evaluated on three distinct language models. The empirical analysis reveals that regularization-based methods are well-suited for wide-range task streams, while replay-based methods excel in long sequence task streams, particularly for tasks of rich logic reasoning in maintaining general ability. Simultaneously, the study underscores the need for new continual instruction tuning methods based on parameter-efficient tuning that balance performance on new tasks with the preservation of general capabilities, especially for more sophisticated architecture-based method.
(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.
TianXiang Ren, XingShen Song, JiangYong Shi, JinSheng Deng, and JiHao Cao "Evaluation on parameter-efficient continual instruction tuning of large language models", Proc. SPIE 13404, Fifth International Conference on Control, Robotics, and Intelligent System (CCRIS 2024), 134040C (28 October 2024); https://doi.org/10.1117/12.3050469
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Education and training

Positron emission tomography

Performance modeling

Data modeling

Fast wavelet transforms

Machine learning

3D modeling

Back to Top