Q-Sparse: All Large Language Models can be Fully Sparsely-Activated

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Hongyu, Ma, Shuming, Wang, Ruiping, Wei, Furu
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929434472218624
author Wang, Hongyu
Ma, Shuming
Wang, Ruiping
Wei, Furu
author_facet Wang, Hongyu
Ma, Shuming
Wang, Ruiping
Wei, Furu
contents We introduce, Q-Sparse, a simple yet effective approach to training sparsely-activated large language models (LLMs). Q-Sparse enables full sparsity of activations in LLMs which can bring significant efficiency gains in inference. This is achieved by applying top-K sparsification to the activations and the straight-through-estimator to the training. We also introduce Block Q-Sparse for batch training and inference. The key results from this work are, (1) Q-Sparse can achieve results comparable to those of baseline LLMs while being much more efficient at inference time; (2) We present an inference-optimal scaling law for sparsely-activated LLMs; (3) Q-Sparse is effective in different settings, including training-from-scratch, continue-training of off-the-shelf LLMs, and finetuning; (4) Q-Sparse works for both full-precision and 1-bit LLMs (e.g., BitNet b1.58). Particularly, the synergy of BitNet b1.58 and Q-Sparse (can be equipped with MoE) provides the cornerstone and a clear path to revolutionize the efficiency, including cost and energy consumption, of future LLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2407_10969
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Q-Sparse: All Large Language Models can be Fully Sparsely-Activated
Wang, Hongyu
Ma, Shuming
Wang, Ruiping
Wei, Furu
Computation and Language
Machine Learning
We introduce, Q-Sparse, a simple yet effective approach to training sparsely-activated large language models (LLMs). Q-Sparse enables full sparsity of activations in LLMs which can bring significant efficiency gains in inference. This is achieved by applying top-K sparsification to the activations and the straight-through-estimator to the training. We also introduce Block Q-Sparse for batch training and inference. The key results from this work are, (1) Q-Sparse can achieve results comparable to those of baseline LLMs while being much more efficient at inference time; (2) We present an inference-optimal scaling law for sparsely-activated LLMs; (3) Q-Sparse is effective in different settings, including training-from-scratch, continue-training of off-the-shelf LLMs, and finetuning; (4) Q-Sparse works for both full-precision and 1-bit LLMs (e.g., BitNet b1.58). Particularly, the synergy of BitNet b1.58 and Q-Sparse (can be equipped with MoE) provides the cornerstone and a clear path to revolutionize the efficiency, including cost and energy consumption, of future LLMs.
title Q-Sparse: All Large Language Models can be Fully Sparsely-Activated
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2407.10969