Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Spike-driven Large Language Model

About

Current Large Language Models (LLMs) are primarily based on large-scale dense matrix multiplications. Inspired by the brain's information processing mechanism, we explore the fundamental question: how to effectively integrate the brain's spiking-driven characteristics into LLM inference. Spiking Neural Networks (SNNs) possess spike-driven characteristics, and some works have attempted to combine SNNs with Transformers. However, achieving spike-driven LLMs with billions of parameters, relying solely on sparse additions, remains a challenge in the SNN field. To address the issues of limited representational capacity and sparsity in existing spike encoding schemes at the LLM level, we propose SDLLM, a spike-driven large language model that eliminates dense matrix multiplications through sparse addition operations. Specifically, we use the plug-and-play gamma-SQP two-step spike encoding method to ensure that the quantization process aligns with the model's semantic space, mitigating representation degradation caused by binary spikes. Furthermore, we introduce bidirectional encoding under symmetric quantization and membrane potential clipping mechanisms, leading to spike trains with no or low firing counts dominating, significantly reducing the model's spike firing rate, while halving the number of time steps. Experimental results show that SDLLM not only significantly reduces inference costs but also achieves state-of-the-art task performance under the spike-based paradigm. For example, compared to previous spike-based LLMs, SDLLM reduces energy consumption by 7x and improves accuracy by 4.2%. Our model provides inspiration for the architecture design of the next generation of event-driven neuromorphic chips.

Han Xu, Xuerui Qiu, Baiyu Chen, Xinhao Luo, Xingrun Xing, Jiahong Zhang, Bo Lei, Tiejun Huang, Bo Xu, Guoqi Li• 2026

Related benchmarks

TaskDatasetResultRank
Language ModelingC4
Perplexity7.15
1688
Language ModelingWikiText2
Perplexity5.18
277
Question AnsweringQA Suite Zero-shot (PIQA, ARC-E, ARC-C, BoolQ, HellaSwag, WinoGrande)
PIQA Accuracy82.48
141
Knowledge & ReasoningTriviaQA
Exact Match70.26
20
Reading ComprehensionSQuAD
EM28.32
16
Mathematical ReasoningGSM8K
String Accuracy (Str.)46.55
16
Question AnsweringDuQuant PIQA, ARC-e, ARC-c, BoolQ, HellaS, WinoG
PIQA Accuracy75.9
12
Showing 7 of 7 rows

Other info

Follow for update