Saved in:
Bibliographic Details
Main Authors: Chen, Long, Song, Xiaotian, Sun, Yanan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.09659
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918021578096640
author Chen, Long
Song, Xiaotian
Sun, Yanan
author_facet Chen, Long
Song, Xiaotian
Sun, Yanan
contents Spiking Large Language Models (LLMs) have emerged as an energy-efficient alternative to conventional LLMs through their event-driven computation. To effectively obtain spiking LLMs, researchers develop different ANN-to-SNN conversion methods by leveraging pre-trained ANN parameters while inheriting the energy efficiency of SNN. However, existing conversion methods struggle with extreme activation outliers and incompatible nonlinear operations of ANN-based LLMs. To address this, we propose a loss-less ANN-SNN conversion for fully spike-driven LLMs, termed LAS. Specifically, LAS introduces two novel neurons to convert the activation outlier and nonlinear operation of ANN-based LLMs. Moreover, LAS tailors the spike-equivalent Transformer components for spiking LLMs, which can ensure full spiking conversion without any loss of performance. Experimental results on six language models and two vision-language models demonstrate that LAS achieves loss-less conversion. Notably, on OPT-66B, LAS even improves the accuracy of 2\% on the WSC task. In addition, the parameter and ablation studies further verify the effectiveness of LAS. The source code is available at https://github.com/lc783/LAS
format Preprint
id arxiv_https___arxiv_org_abs_2505_09659
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LAS: Loss-less ANN-SNN Conversion for Fully Spike-Driven Large Language Models
Chen, Long
Song, Xiaotian
Sun, Yanan
Machine Learning
Computation and Language
Spiking Large Language Models (LLMs) have emerged as an energy-efficient alternative to conventional LLMs through their event-driven computation. To effectively obtain spiking LLMs, researchers develop different ANN-to-SNN conversion methods by leveraging pre-trained ANN parameters while inheriting the energy efficiency of SNN. However, existing conversion methods struggle with extreme activation outliers and incompatible nonlinear operations of ANN-based LLMs. To address this, we propose a loss-less ANN-SNN conversion for fully spike-driven LLMs, termed LAS. Specifically, LAS introduces two novel neurons to convert the activation outlier and nonlinear operation of ANN-based LLMs. Moreover, LAS tailors the spike-equivalent Transformer components for spiking LLMs, which can ensure full spiking conversion without any loss of performance. Experimental results on six language models and two vision-language models demonstrate that LAS achieves loss-less conversion. Notably, on OPT-66B, LAS even improves the accuracy of 2\% on the WSC task. In addition, the parameter and ablation studies further verify the effectiveness of LAS. The source code is available at https://github.com/lc783/LAS
title LAS: Loss-less ANN-SNN Conversion for Fully Spike-Driven Large Language Models
topic Machine Learning
Computation and Language
url https://arxiv.org/abs/2505.09659