Neuromorphic spike-based large language model
National Science Review · 05.02.2026 · 13 (4)
DOI: 10.1093/nsr/nwaf551 ↗Kurz gefasst
Fachartikel in der Zeitschrift National Science Review, online erschienen am 04.12.2025. Verfasst von Han Xu und 19 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 52 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Distinguished Young Scholars of China, National Natural Science Foundation of China, Beijing Science and Technology Plan Project.
Abstract (Originalsprache)
This work proposes a unified neuromorphic spike-based large-language-model (NSLLM) framework to simultaneously address the challenges of high energy consumption and low interpretability in LLMs. Our framework transforms LLMs into efficient NSLLMs by converting their behaviors into neural dynamics—such as spike trains—through rigorous mathematical modeling and complemented by advanced techniques including quantization and sparsification. This transformation also enables the analysis of information encoding processes using computational neuroscience tools, thereby offering a novel neuroscientific perspective that conceptualizes LLMs as neural populations to enhance their interpretability. Leveraging a hardware-algorithm co-design paradigm, an NSLLM can completely eliminate matrix multiplication (MatMul) while maintaining high performance. We designed a custom MatMul-free hardware core on the VCK190 field-programmable gate array to validate the 1.5-billion-parameter NSLLM, achieving a dynamic power consumption of only 13.849 W and an inference throughput of 161.8 tokens per second. Compared with the A800 GPU, this implementation improves energy efficiency, memory usage and inference throughput by 19.8$\times$, 21.3$\times$ and 2.2$\times$, respectively. This work provides a novel perspective within a unified framework to enhance both the energy efficiency and interpretability of LLMs, offering valuable insights for future neuromorphic chip designs tailored for large models.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China.
- School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049, China.
- Brain-Inspired Models Group, Beijing Academy of Artificial Intelligence, Beijing 100862, China.
- School of Future Technology, University of Chinese Academy of Sciences, Beijing 101408, China.
- Zhongguancun Academy, Beijing 100094, China.
- Department of Psychology, Tsinghua University, Beijing 100084, China.
Förderung
- National Distinguished Young Scholars of China
- National Natural Science Foundation of China
- Beijing Science and Technology Plan Project
- China Postdoctoral Science Foundation
Zitieren
APA
Xu, H., Qiu, X., Xu, Y., Elbtity, M. E., Zhou, P., Tian, Y., Zhu, R. J., Zhang, J., Gu, S., Pan, Y., Chou, Y., Wen, Q., Yao, M., Qian, J., Tian, Y., Ma, L., Huang, T., Eshraghian, J. K., Xu, B., & Li, G. (2026). Neuromorphic spike-based large language model. National Science Review, 13(4). https://doi.org/10.1093/nsr/nwaf551
Vancouver
Xu H, Qiu X, Xu Y, Elbtity ME, Zhou P, Tian Y, et al. Neuromorphic spike-based large language model. National Science Review. 2026;13(4). doi:10.1093/nsr/nwaf551
BibTeX
@article{Xu2026,
author = {Xu, Han and Qiu, Xuerui and Xu, Yunhui and Elbtity, Mohammed E and Zhou, Peng and Tian, Yang and Zhu, Rui-Jie and Zhang, Jiahong and Gu, Shaowei and Pan, Yuqi and Chou, Yuhong and Wen, Qinghao and Yao, Man and Qian, Jiangbo and Tian, Yonghong and Ma, Lei and Huang, Tiejun and Eshraghian, Jason K and Xu, Bo and Li, Guoqi},
title = {Neuromorphic spike-based large language model},
journal = {National Science Review},
year = {2026},
volume = {13},
number = {4},
doi = {10.1093/nsr/nwaf551}
}
EUVIMED