Quantitative Analysis of Performance Drop in DeepSeek Model Quantization

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhao, Enbo, Shen, Yi, Shi, Shuming, Huang, Jieyun, Chen, Zhihao, Wang, Ning, Xiao, Siqi, Zhang, Jian, Wang, Kai, Lian, Shiguo
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918057265332224
author Zhao, Enbo
Shen, Yi
Shi, Shuming
Huang, Jieyun
Chen, Zhihao
Wang, Ning
Xiao, Siqi
Zhang, Jian
Wang, Kai
Lian, Shiguo
author_facet Zhao, Enbo
Shen, Yi
Shi, Shuming
Huang, Jieyun
Chen, Zhihao
Wang, Ning
Xiao, Siqi
Zhang, Jian
Wang, Kai
Lian, Shiguo
contents Recently, there is a high demand for deploying DeepSeek-R1 and V3 locally, possibly because the official service often suffers from being busy and some organizations have data privacy concerns. While single-machine deployment offers infrastructure simplicity, the models' 671B FP8 parameter configuration exceeds the practical memory limits of a standard 8-GPU machine. Quantization is a widely used technique that helps reduce model memory consumption. However, it is unclear what the performance of DeepSeek-R1 and V3 will be after being quantized. This technical report presents the first quantitative evaluation of multi-bitwidth quantization across the complete DeepSeek model spectrum. Key findings reveal that 4-bit quantization maintains little performance degradation versus FP8 while enabling single-machine deployment on standard NVIDIA GPU devices. We further propose DQ3_K_M, a dynamic 3-bit quantization method that significantly outperforms traditional Q3_K_M variant on various benchmarks, which is also comparable with 4-bit quantization (Q4_K_M) approach in most tasks. Moreover, DQ3_K_M supports single-machine deployment configurations for both NVIDIA H100/A100 and Huawei 910B. Our implementation of DQ3\_K\_M is released at https://github.com/UnicomAI/DeepSeek-Eval, containing optimized 3-bit quantized variants of both DeepSeek-R1 and DeepSeek-V3.
format Preprint
id arxiv_https___arxiv_org_abs_2505_02390
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Quantitative Analysis of Performance Drop in DeepSeek Model Quantization
Zhao, Enbo
Shen, Yi
Shi, Shuming
Huang, Jieyun
Chen, Zhihao
Wang, Ning
Xiao, Siqi
Zhang, Jian
Wang, Kai
Lian, Shiguo
Machine Learning
Artificial Intelligence
Recently, there is a high demand for deploying DeepSeek-R1 and V3 locally, possibly because the official service often suffers from being busy and some organizations have data privacy concerns. While single-machine deployment offers infrastructure simplicity, the models' 671B FP8 parameter configuration exceeds the practical memory limits of a standard 8-GPU machine. Quantization is a widely used technique that helps reduce model memory consumption. However, it is unclear what the performance of DeepSeek-R1 and V3 will be after being quantized. This technical report presents the first quantitative evaluation of multi-bitwidth quantization across the complete DeepSeek model spectrum. Key findings reveal that 4-bit quantization maintains little performance degradation versus FP8 while enabling single-machine deployment on standard NVIDIA GPU devices. We further propose DQ3_K_M, a dynamic 3-bit quantization method that significantly outperforms traditional Q3_K_M variant on various benchmarks, which is also comparable with 4-bit quantization (Q4_K_M) approach in most tasks. Moreover, DQ3_K_M supports single-machine deployment configurations for both NVIDIA H100/A100 and Huawei 910B. Our implementation of DQ3\_K\_M is released at https://github.com/UnicomAI/DeepSeek-Eval, containing optimized 3-bit quantized variants of both DeepSeek-R1 and DeepSeek-V3.
title Quantitative Analysis of Performance Drop in DeepSeek Model Quantization
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2505.02390