Energy-Efficient and Dequantization-Free Quantization of LLMs: A Spiking Neural Network Approach to Salient Value Mitigation
Open Access
- 12 April 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 5334-5345
Abstract
No abstract availableAll Related Versions
Funding Information
- National Natural Science Foundation of China (62172454)
- Guangdong Basic and Applied Basic Research Foundation (2023B1515020120)
- Guangzhou Basic and Applied Basic Research Scheme (2024A04J3517)
- Academic Research Fund TIER 1 (FY2024)
This publication has 30 references indexed in Scilit:
- Cerebras Architecture Deep Dive: First Look Inside the HW/SW Co-Design for Deep Learning : Cerebras Systems2022
- Brain-inspired computing needs a master planNature, 2022
- Opportunities for neuromorphic computing algorithms and applicationsNature Computational Science, 2022
- Pruning and quantization for deep neural network acceleration: A surveyNeurocomputing, 2021
- A system hierarchy for brain-inspired computingNature, 2020
- Towards spike-based machine intelligence with neuromorphic computingNature, 2019
- Value-Aware Quantization for Training and Inference of Neural Networks2018
- Loihi Asynchronous Neuromorphic Research Chip2018
- Networks of spiking neurons: The third generation of neural network modelsNeural Networks, 1997
- Alternatives to the Median Absolute DeviationJournal of the American Statistical Association, 1993