BitNet a4.8: 4-bit Activations for 1-bit LLMs

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wang, Hongyu, Ma, Shuming, Wei, Furu
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916471799545856
author Wang, Hongyu
Ma, Shuming
Wei, Furu
author_facet Wang, Hongyu
Ma, Shuming
Wei, Furu
contents Recent research on the 1-bit Large Language Models (LLMs), such as BitNet b1.58, presents a promising direction for reducing the inference cost of LLMs while maintaining their performance. In this work, we introduce BitNet a4.8, enabling 4-bit activations for 1-bit LLMs. BitNet a4.8 employs a hybrid quantization and sparsification strategy to mitigate the quantization errors introduced by the outlier channels. Specifically, we utilize 4-bit activations for inputs to the attention and feed-forward network layers, while sparsifying intermediate states followed with 8-bit quantization. Extensive experiments demonstrate that BitNet a4.8 achieves performance comparable to BitNet b1.58 with equivalent training costs, while being faster in inference with enabling 4-bit (INT4/FP4) kernels. Additionally, BitNet a4.8 activates only 55% of parameters and supports 3-bit KV cache, further enhancing the efficiency of large-scale LLM deployment and inference.
format Preprint
id arxiv_https___arxiv_org_abs_2411_04965
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle BitNet a4.8: 4-bit Activations for 1-bit LLMs
Wang, Hongyu
Ma, Shuming
Wei, Furu
Computation and Language
Machine Learning
Recent research on the 1-bit Large Language Models (LLMs), such as BitNet b1.58, presents a promising direction for reducing the inference cost of LLMs while maintaining their performance. In this work, we introduce BitNet a4.8, enabling 4-bit activations for 1-bit LLMs. BitNet a4.8 employs a hybrid quantization and sparsification strategy to mitigate the quantization errors introduced by the outlier channels. Specifically, we utilize 4-bit activations for inputs to the attention and feed-forward network layers, while sparsifying intermediate states followed with 8-bit quantization. Extensive experiments demonstrate that BitNet a4.8 achieves performance comparable to BitNet b1.58 with equivalent training costs, while being faster in inference with enabling 4-bit (INT4/FP4) kernels. Additionally, BitNet a4.8 activates only 55% of parameters and supports 3-bit KV cache, further enhancing the efficiency of large-scale LLM deployment and inference.
title BitNet a4.8: 4-bit Activations for 1-bit LLMs
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2411.04965