awesome-model-quantization

awesome-model-quantization

全面的模型量化研究资源

此项目汇集了关于模型量化的各类论文、文档和代码,为研究者提供丰富的参考资源。内容包括二值化和量化方法的调研、基准测试,以及生成模型的压缩和加速技术。项目持续更新,并欢迎对未收录研究成果的贡献。感谢所有已作出贡献的研究者。

Awesome Model QuantizationBiBenchMQBenchSurvey of QuantizationEfficient_AIGC_RepoGithub开源项目

Awesome Model Quantization Awesome

This repo collects papers, docs, codes about model quantization for anyone who wants to do research on it. We are continuously improving the project. Welcome to PR the works (papers, repositories) that are missed by the repo. Special thanks to Xingyu Zheng, Yifu Ding, Xudong Ma, Yuxuan Wen, and all researchers who have contributed to this project!

Table of Contents

Efficient_AIGC_Repo

We highlight our newly released awesome open-source project "Awesome Efficient AIGC". Specifically, this project focuses on recent methods for compression and acceleration of generative models, such as large language models and diffusion models. Welcome to Star the Repo or PR any work you like!

https://github.com/htqin/awesome-efficient-aigc AwesomeGitHub Repo stars

Benchmark

BiBench

The paper BiBench: Benchmarking and Analyzing Network Binarization (ICML 2023) a rigorously designed benchmark with in-depth analysis for network binarization. For details, please refer to:

BiBench: Benchmarking and Analyzing Network Binarization [Paper] [Project]

Haotong Qin, Mingyuan Zhang, Yifu Ding, Aoyu Li, Zhongang Cai, Ziwei Liu, Fisher Yu, Xianglong Liu.

<details><summary>Bibtex</summary><pre><code>@inproceedings{qin2023bibench, title={BiBench: Benchmarking and Analyzing Network Binarization}, author={Qin, Haotong and Zhang, Mingyuan and Ding, Yifu and Li, Aoyu and Cai, Zhongang and Liu, Ziwei and Yu, Fisher and Liu, Xianglong}, booktitle={International Conference on Machine Learning (ICML)}, year={2023} }</code></pre></details>

survey

MQBench

The paper MQBench: Towards Reproducible and Deployable Model Quantization Benchmark (NeurIPS 2021) is a benchmark and framework for evluating the quantization algorithms under real world hardware deployments. For details, please refer to:

MQBench: Towards Reproducible and Deployable Model Quantization Benchmark [Paper] [Project]

Yuhang Li, Mingzhu Shen, Jian Ma, Yan Ren, Mingxin Zhao, Qi Zhang, Ruihao Gong, Fengwei Yu, Junjie Yan.

<details><summary>Bibtex</summary><pre><code>@article{2021MQBench, title = "MQBench: Towards Reproducible and Deployable Model Quantization Benchmark", author= "Yuhang Li* and Mingzhu Shen* and Jian Ma* and Yan Ren* and Mingxin Zhao* and Qi Zhang* and Ruihao Gong and Fengwei Yu and Junjie Yan", journal = "https://openreview.net/forum?id=TUplOmF8DsM", year = "2021" }</code></pre></details>

survey

Survey_Papers

Survey_of_Binarization

Our survey paper Binary Neural Networks: A Survey (Pattern Recognition) is a comprehensive survey of recent progress in binary neural networks. For details, please refer to:

Binary Neural Networks: A Survey [Paper] [Blog]

Haotong Qin, Ruihao Gong, Xianglong Liu*, Xiao Bai, Jingkuan Song, and Nicu Sebe.

<details><summary>Bibtex</summary><pre><code>@article{Qin:pr20_bnn_survey, title = "Binary neural networks: A survey", author = "Haotong Qin and Ruihao Gong and Xianglong Liu and Xiao Bai and Jingkuan Song and Nicu Sebe", journal = "Pattern Recognition", volume = "105", pages = "107281", year = "2020" }</code></pre></details>

survey

Survey_of_Quantization

The survey paper A Survey of Quantization Methods for Efficient Neural Network Inference (ArXiv) is a comprehensive survey of recent progress in quantization. For details, please refer to:

A Survey of Quantization Methods for Efficient Neural Network Inference [Paper]

Amir Gholami* , Sehoon Kim* , Zhen Dong* , Zhewei Yao* , Michael W. Mahoney, Kurt Keutzer. (* Equal contribution)

<details><summary>Bibtex</summary><pre><code>@misc{gholami2021survey, title={A Survey of Quantization Methods for Efficient Neural Network Inference}, author={Amir Gholami and Sehoon Kim and Zhen Dong and Zhewei Yao and Michael W. Mahoney and Kurt Keutzer}, year={2021}, eprint={2103.13630}, archivePrefix={arXiv}, primaryClass={cs.CV} }</code></pre></details>

Papers

Keywords: qnn: quantized neural networks | bnn: binarized neural networks | hardware: hardware deployment | snn: spiking neural networks | other


2024

  • [arXiv] How Good Are Low-bit Quantized LLaMA3 Models? An Empirical Study [code]GitHub Repo stars [HuggingFace]
  • [arXiv] Accurate LoRA-Finetuning Quantization of LLMs via Information Retention [code]GitHub Repo stars
  • [arXiv] BiLLM: Pushing the Limit of Post-Training Quantization for LLMs [code]GitHub Repo stars
  • [arXiv] BinaryDM: Towards Accurate Binarization of Diffusion Model [code]GitHub Repo stars
  • [arXiv] DB-LLM: Accurate Dual-Binarization for Efficient LLMs
  • [arXiv] BinaryDM: Towards Accurate Binarization of Diffusion Model [code]GitHub Repo stars
  • [arXiv] OHQ: On-chip Hardware-aware Quantization
  • [arXiv] Post-Training Quantization with Low-precision Minifloats and Integers on FPGAs [code][hardware]
  • [arXiv] Extreme Compression of Large Language Models via Additive Quantization
  • [arXiv] Quantized Side Tuning: Fast and Memory-Efficient Tuning of Quantized Large Language Models
  • [arXiv] FP6-LLM: Efficiently Serving Large Language Models Through FP6-Centric Algorithm-System Co-Design
  • [arXiv] KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization
  • [arXiv] EdgeQAT: Entropy and Distribution Guided Quantization-Aware Training for the Acceleration of Lightweight LLMs on the Edge [code] GitHub Repo stars
  • [arXiv] Any-Precision LLM: Low-Cost Deployment of Multiple, Different-Sized LLMs
  • [arXiv] LQER: Low-Rank Quantization Error Reconstruction for LLMs
  • [arXiv] KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache [code] GitHub Repo stars
  • [arXiv] QuIP#: Even Better LLM Quantization with Hadamard Incoherence and Lattice Codebooks [code] GitHub Repo stars
  • [arXiv] L4Q: Parameter Efficient Quantization-Aware Training on Large Language Models via LoRA-wise LSQ
  • [arXiv] TP-Aware Dequantization
  • [arXiv] ApiQ: Finetuning of 2-Bit Quantized Large Language Model
  • [arXiv] Any-Precision LLM: Low-Cost Deployment of Multiple, Different-Sized LLMs
  • [arXiv] BitDistiller: Unleashing the Potential of Sub-4-Bit LLMs via Self-Distillation [code] GitHub Repo stars
  • [arXiv] OneBit: Towards Extremely Low-bit Large Language Models
  • [arXiv] WKVQuant: Quantising Weight and Key/Value Cache for Large Language Models Gains More
  • [arXiv] GPTVQ: The Blessing of Dimensionality for LLM Quantization [code] GitHub Repo stars
  • [DAC] APTQ: Attention-aware Post-Training Mixed-Precision Quantization for Large Language Models
  • [DAC] A Comprehensive Evaluation of Quantization Strategies for Large Language Models
  • [arXiv] No Token Left Behind: Reliable KV Cache Compression via Importance-Aware Mixed Precision Quantization
  • [arXiv] Evaluating Quantized Large Language Models
  • [arXiv] FlattenQuant: Breaking Through the Inference Compute-bound for Large Language Models with Per-tensor Quantization
  • [arXiv] LLM-PQ: Serving LLM on Heterogeneous Clusters with Phase-Aware Partition and Adaptive Quantization
  • [arXiv] IntactKV: Improving Large Language Model Quantization by Keeping Pivot Tokens Intact
  • [arXiv] On the Compressibility of Quantized Large Language Models
  • [arXiv] EasyQuant: An Efficient Data-free Quantization Algorithm for LLMs
  • [arXiv] QAQ: Quality Adaptive Quantization for LLM KV Cache [code] GitHub Repo stars
  • [arXiv] GEAR: An Efficient KV Cache Compression Recipefor Near-Lossless Generative Inference of LLM
  • [arXiv] What Makes Quantization for Large Language Models Hard? An Empirical Study from the Lens of Perturbation
  • [arXiv] SVD-LLM: Truncation-aware Singular Value Decomposition for Large Language Model Compression [code] GitHub Repo stars
  • [ICLR] AffineQuant: Affine Transformation Quantization for Large Language Models [code] GitHub Repo stars
  • [ICLR Practical ML for Low Resource Settings Workshop] Oh! We Freeze: Improving Quantized Knowledge Distillation via Signal Propagation Analysis for Large Language Models
  • [arXiv] Accurate Block Quantization in LLMs with Outliers
  • [arXiv] QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs [code] GitHub Repo stars
  • [arXiv] Minimize Quantization Output Error with Bias Compensation [code] ![GitHub Repo

编辑推荐精选

AEE

AEE

AI Excel全自动制表工具

AEE 在线 AI 全自动 Excel 编辑器,提供智能录入、自动公式、数据整理、图表生成等功能,高效处理 Excel 任务,提升办公效率。支持自动高亮数据、批量计算、不规则数据录入,适用于企业、教育、金融等多场景。

UI-TARS-desktop

UI-TARS-desktop

基于 UI-TARS 视觉语言模型的桌面应用,可通过自然语言控制计算机进行多模态操作。

UI-TARS-desktop 是一款功能强大的桌面应用,基于 UI-TARS(视觉语言模型)构建。它具备自然语言控制、截图与视觉识别、精确的鼠标键盘控制等功能,支持跨平台使用(Windows/MacOS),能提供实时反馈和状态显示,且数据完全本地处理,保障隐私安全。该应用集成了多种大语言模型和搜索方式,还可进行文件系统操作。适用于需要智能交互和自动化任务的场景,如信息检索、文件管理等。其提供了详细的文档,包括快速启动、部署、贡献指南和 SDK 使用说明等,方便开发者使用和扩展。

Wan2.1

Wan2.1

开源且先进的大规模视频生成模型项目

Wan2.1 是一个开源且先进的大规模视频生成模型项目,支持文本到图像、文本到视频、图像到视频等多种生成任务。它具备丰富的配置选项,可调整分辨率、扩散步数等参数,还能对提示词进行增强。使用了多种先进技术和工具,在视频和图像生成领域具有广泛应用前景,适合研究人员和开发者使用。

爱图表

爱图表

全流程 AI 驱动的数据可视化工具,助力用户轻松创作高颜值图表

爱图表(aitubiao.com)就是AI图表,是由镝数科技推出的一款创新型智能数据可视化平台,专注于为用户提供便捷的图表生成、数据分析和报告撰写服务。爱图表是中国首个在图表场景接入DeepSeek的产品。通过接入前沿的DeepSeek系列AI模型,爱图表结合强大的数据处理能力与智能化功能,致力于帮助职场人士高效处理和表达数据,提升工作效率和报告质量。

Qwen2.5-VL

Qwen2.5-VL

一款强大的视觉语言模型,支持图像和视频输入

Qwen2.5-VL 是一款强大的视觉语言模型,支持图像和视频输入,可用于多种场景,如商品特点总结、图像文字识别等。项目提供了 OpenAI API 服务、Web UI 示例等部署方式,还包含了视觉处理工具,有助于开发者快速集成和使用,提升工作效率。

HunyuanVideo

HunyuanVideo

HunyuanVideo 是一个可基于文本生成高质量图像和视频的项目。

HunyuanVideo 是一个专注于文本到图像及视频生成的项目。它具备强大的视频生成能力,支持多种分辨率和视频长度选择,能根据用户输入的文本生成逼真的图像和视频。使用先进的技术架构和算法,可灵活调整生成参数,满足不同场景的需求,是文本生成图像视频领域的优质工具。

WebUI for Browser Use

WebUI for Browser Use

一个基于 Gradio 构建的 WebUI,支持与浏览器智能体进行便捷交互。

WebUI for Browser Use 是一个强大的项目,它集成了多种大型语言模型,支持自定义浏览器使用,具备持久化浏览器会话等功能。用户可以通过简洁友好的界面轻松控制浏览器智能体完成各类任务,无论是数据提取、网页导航还是表单填写等操作都能高效实现,有利于提高工作效率和获取信息的便捷性。该项目适合开发者、研究人员以及需要自动化浏览器操作的人群使用,在 SEO 优化方面,其关键词涵盖浏览器使用、WebUI、大型语言模型集成等,有助于提高网页在搜索引擎中的曝光度。

xiaozhi-esp32

xiaozhi-esp32

基于 ESP32 的小智 AI 开发项目,支持多种网络连接与协议,实现语音交互等功能。

xiaozhi-esp32 是一个极具创新性的基于 ESP32 的开发项目,专注于人工智能语音交互领域。项目涵盖了丰富的功能,如网络连接、OTA 升级、设备激活等,同时支持多种语言。无论是开发爱好者还是专业开发者,都能借助该项目快速搭建起高效的 AI 语音交互系统,为智能设备开发提供强大助力。

olmocr

olmocr

一个用于 OCR 的项目,支持多种模型和服务器进行 PDF 到 Markdown 的转换,并提供测试和报告功能。

olmocr 是一个专注于光学字符识别(OCR)的 Python 项目,由 Allen Institute for Artificial Intelligence 开发。它支持多种模型和服务器,如 vllm、sglang、OpenAI 等,可将 PDF 文件的页面转换为 Markdown 格式。项目还提供了测试框架和 HTML 报告生成功能,方便用户对 OCR 结果进行评估和分析。适用于科研、文档处理等领域,有助于提高工作效率和准确性。

飞书多维表格

飞书多维表格

飞书多维表格 ×DeepSeek R1 满血版

飞书多维表格联合 DeepSeek R1 模型,提供 AI 自动化解决方案,支持批量写作、数据分析、跨模态处理等功能,适用于电商、短视频、影视创作等场景,提升企业生产力与创作效率。关键词:飞书多维表格、DeepSeek R1、AI 自动化、批量处理、企业协同工具。

下拉加载更多