Spike-driven Large Language Model
Han Xu, Xuerui Qiu, Baiyu Chen, Xinhao Luo, Xingrun Xing, Jiahong Zhang, Bo Lei, Tiejun Huang, Bo Xu, Guoqi Li · 2026
Read Spike-driven Large Language Model by Han Xu, Xuerui Qiu, Baiyu Chen, Xinhao Luo, Xingrun Xing, Jiahong Zhang, Bo Lei, Tiejun Huang, Bo Xu, Guoqi Li (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-16475.txt.
Description
Xu and colleagues build a large language model driven by spikes instead of dense matrix multiplication. Their SDLLM replaces dense products with sparse additions, using a two-step gamma-SQP spike encoding that keeps quantization aligned with meaning, plus bidirectional encoding under symmetric quantization and membrane clipping. Spike trains with few or no firings dominate, cutting firing rates and halving time steps. Compared with earlier spike-based language models, it cut inference costs and improved accuracy by about 4 percent, pointing toward event-driven neuromorphic chips.
Catalog record
| CREATOR | Han Xu, Xuerui Qiu, Baiyu Chen, Xinhao Luo, Xingrun Xing, Jiahong Zhang, Bo Lei, Tiejun Huang, Bo Xu, Guoqi Li |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 78231 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.16475