PaddlePaddle/PaddleSlim PaddlePaddle/PaddleSlim

  • Stars
    star
    1,507
  • Rank 29,927 (Top 0.7 %)
  • Language
    Python
  • License
    Apache License 2.0
  • Created over 4 years ago
  • Updated about 1 month ago
Twitter logo Share LinkedIn logo Share Facebook logo Share
PaddlePaddle/PaddleSlim
github

PaddlePaddle

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

PaddleSlim is an open-source library for deep model compression and architecture search.

PaddleSlim

PaddleSlim是一个专注于深度学习模型压缩的工具库，提供低比特量化、知识蒸馏、稀疏化和模型结构搜索等模型压缩策略，帮助开发者快速实现模型的小型化。

产品动态

  • 🔥 2022.01.18: 发布YOLOv8自动化压缩示例，量化预测加速2.5倍。

  • 【直播分享】2022-12-13 20:30 《自动化压缩技术详解及ViT模型实战》，微信扫码报名

2022.08.16：自动化压缩功能升级
模型 Base mAPval
0.5:0.95		 ACT量化mAPval
0.5:0.95		 模型体积压缩比 预测时延FP32
 预测时延INT8
 预测加速比
PPYOLOE-s 43.1 42.6 3.9倍 6.51ms 2.12ms 3.1倍
YOLOv5s 37.4 36.9 3.8倍 5.95ms 1.87ms 3.2倍
YOLOv6s 42.4 41.3 3.9倍 9.06ms 1.83ms 5.0倍
YOLOv7 51.1 50.9 3.9倍 26.84ms 4.55ms 5.9倍
YOLOv7-Tiny 37.3 37.0 3.9倍 5.06ms 1.68ms 3.0倍
历史更新

  • 2022.07.01: 发布v2.3.0版本

    • 发布自动化压缩功能
      • 支持代码无感知压缩：开发者只需提供推理模型文件和数据，既可进行离线量化（PTQ）、量化训练（QAT）、稀疏训练等压缩任务。
      • 支持自动策略选择，根据任务特点和部署环境特性：自动搜索合适的离线量化方法,自动搜索最佳的压缩策略组合方式。
      • 发布自然语言处理、图像语义分割、图像目标检测三个方向的自动化压缩示例。
      • 发布X2Paddle模型自动化压缩方案:YOLOv5、YOLOv6、YOLOv7、HuggingFace、MobileNet。
    • 升级量化功能
      • 统一量化模型格式；离线量化支持while op；修复BERT大模型量化训练过慢的问题。
      • 新增7种离线量化方法, 包括HIST, AVG, EMD, Bias Correction, AdaRound等。
    • 支持半结构化稀疏训练
    • 新增延时预估工具
      • 支持对稀疏化模型、低比特量化模型的性能预估；支持预估指定模型在特定部署环境下 (ARM CPU + Paddle Lite) 的推理性能；提供 SD625、SD710、RK3288 芯片 + Paddle Lite 的预估接口。
      • 提供部署环境自动扩展工具，可以自动增加在更多 ARM CPU 设备上的预估工具。

  • 2021.11.15: 发布v2.2.0版本

    • 支持动态图离线量化功能.

  • 2021.5.20: 发布V2.1.0版本

    • 扩展离线量化方法
    • 新增非结构化稀疏
    • 增强剪枝功能
    • 修复OFA功能若干bug

更多信息请参考：release note

基础压缩功能概览

PaddleSlim支持以下功能，也支持自定义量化、裁剪等功能。

Quantization Pruning NAS Distilling

注：

  • *表示仅支持静态图，**表示仅支持动态图
  • 敏感度裁剪指的是通过各个层的敏感度分析来确定各个卷积层的剪裁率，需要和其他裁剪方法配合使用。

PaddleSlim在典型视觉和自然语言处理任务上做了模型压缩，并且测试了Nvidia GPU、ARM等设备上的加速情况，这里展示部分模型的压缩效果，详细方案可以参考下面CV和NLP模型压缩方案:


表1: 部分场景模型压缩加速情况

注意事项
  • YOLOv3: 在移动端SD855上加速3.55倍。
  • PP-OCR: 体积由8.9M减少到2.9M, 在SD855上加速1.27倍。
  • BERT: 模型参数由110M减少到80M，精度提升的情况下，Tesla T4 GPU FP16计算加速1.47倍。

不同压缩方法效果

自动压缩效果

image
表3: 自动压缩效果

离线量化效果对比

image
表2: 多种离线量化方法效果对比

安装

安装发布版本：

pip install paddleslim

安装develop版本：

git clone https://github.com/PaddlePaddle/PaddleSlim.git & cd PaddleSlim
python setup.py install
  • 验证安装：安装完成后您可以使用 python 或 python3 进入 python 解释器，输入import paddleslim, 没有报错则说明安装成功。
  • 版本对齐：
PaddleSlim PaddlePaddle PaddleLite
2.0.0 2.0 2.8
2.1.0 2.1.0 2.8
2.1.1 2.1.1 >=2.8
2.3.0 2.3.0 >=2.11
2.4.0 2.4.0 >=2.11
develop develop >=2.11

文档教程

模型压缩技术

快速开始

更多教程

进阶教程详细介绍了每一步的流程，帮助您把相应方法迁移到您自己的模型上。

推理部署

CV模型压缩

多场景效果展示

本系列教程均基于Paddle官方的模型套件中模型进行压缩，若您不是模型套件用户，更推荐使用快速教程和进阶教程。

NLP模型压缩

API文档

FAQ

1. 量化训练或者离线量化后的模型体积为什么没有变小？

答：这是因为量化后保存的参数是虽然是int8范围，但是类型是float。这是因为Paddle训练前向默认的Kernel不支持INT8 Kernel实现，只有Paddle Inference TensorRT的推理才支持量化推理加速。为了方便量化后验证量化精度，使用Paddle训练前向能加载此模型，默认保存的Float32类型权重，体积没有发生变换。

2. macOS + Python3.9环境或者Windows环境下, 安装出错, "command 'swig' failed"

答: 请参考#1258

许可证书

本项目的发布受Apache 2.0 license许可认证。

贡献代码

我们非常欢迎你可以为PaddleSlim提供代码，也十分感谢你的反馈。

技术交流

  • 如果你发现任何PaddleSlim存在的问题或者是建议, 欢迎通过GitHub Issues给我们提issues。

  • 欢迎加入PaddleSlim 微信技术交流群

More Repositories

1

PaddleOCR

Awesome multilingual OCR toolkits based on PaddlePaddle (practical ultra lightweight OCR system, support 80+ languages recognition, provide data annotation and synthesis tools, support training and deployment among server, mobile, embedded and IoT devices)
Python
38,354
star
2

Paddle

PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice （『飞桨』核心框架，深度学习&机器学习高性能单机、分布式训练和跨平台部署）
C++
21,619
star
3

PaddleHub

Awesome pre-trained models toolkit based on PaddlePaddle. (400+ models including Image, Text, Audio, Video and Cross-Modal with Easy Inference & Serving)
Python
12,439
star
4

PaddleDetection

Object Detection toolkit based on PaddlePaddle. It supports object detection, instance segmentation, multiple object tracking and real-time multi-person keypoint detection.
Python
12,003
star
5

PaddleNLP

👑 Easy-to-use and powerful NLP and LLM library with 🤗 Awesome model zoo, supporting wide-range of NLP tasks from research to industrial applications, including 🗂Text Classification, 🔍 Neural Search, ❓ Question Answering, ℹ️ Information Extraction, 📄 Document Intelligence, 💌 Sentiment Analysis etc.
Python
11,233
star
6

PaddleSpeech

Easy-to-use Speech Toolkit including Self-Supervised Learning model, SOTA/Streaming ASR with punctuation, Streaming TTS with text frontend, Speaker Verification System, End-to-End Speech Translation and Keyword Spotting. Won NAACL2022 Best Demo Award.
Python
10,060
star
7

PaddleSeg

Easy-to-use image segmentation library with awesome pre-trained model zoo, supporting wide-range of practical tasks in Semantic Segmentation, Interactive Segmentation, Panoptic Segmentation, Image Matting, 3D Segmentation, etc.
Python
8,188
star
8

PaddleGAN

PaddlePaddle GAN library, including lots of interesting applications like First-Order motion transfer, Wav2Lip, picture repair, image editing, photo2cartoon, image style transfer, GPEN, and so on.
Python
7,661
star
9

models

Officially maintained, supported by PaddlePaddle, including CV, NLP, Speech, Rec, TS, big models and so on.
Python
6,868
star
10

Paddle-Lite

PaddlePaddle High Performance Deep Learning Inference Engine for Mobile and Edge (飞桨高性能深度学习端侧推理引擎）
C++
6,839
star
11

ERNIE

Official implementations for various pre-training models of ERNIE-family, covering topics of Language Understanding & Generation, Multimodal Understanding & Generation, and beyond.
Python
6,197
star
12

PaddleClas

A treasure chest for visual classification and recognition powered by PaddlePaddle
Python
5,244
star
13

VisualDL

Deep Learning Visualization Toolkit（『飞桨』深度学习可视化工具 ）
HTML
4,716
star
14

PaddleX

PaddlePaddle End-to-End Development Toolkit（『飞桨』深度学习全流程开发工具）
Python
4,564
star
15

PaddleRec

Recommendation Algorithm大规模推荐算法库，包含推荐系统经典及最新算法LR、Wide&Deep、DSSM、TDM、MIND、Word2Vec、Bert4Rec、DeepWalk、SSR、AITM，DSIN，SIGN，IPREC、GRU4Rec、Youtube_dnn、NCF、GNN、FM、FFM、DeepFM、DCN、DIN、DIEN、DLRM、MMOE、PLE、ESMM、ESCMM, MAML、xDeepFM、DeepFEFM、NFM、AFM、RALM、DMR、GateNet、NAML、DIFM、Deep Crossing、PNN、BST、AutoInt、FGCNN、FLEN、Fibinet、ListWise、DeepRec、ENSFM，TiSAS，AutoFIS等，包含经典推荐系统数据集criteo 、movielens等
Python
4,100
star
16

PARL

A high-performance distributed training framework for Reinforcement Learning
Python
3,182
star
17

awesome-DeepLearning

深度学习入门课、资深课、特色课、学术案例、产业实践案例、深度学习知识百科及面试题库The course, case and knowledge of Deep Learning and AI
Jupyter Notebook
2,752
star
18

book

Deep Learning 101 with PaddlePaddle （『飞桨』深度学习框架入门教程）
Jupyter Notebook
2,728
star
19

FastDeploy

⚡️An Easy-to-use and Fast Deep Learning Model Deployment Toolkit for ☁️Cloud 📱Mobile and 📹Edge. Including Image, Video, Text and Audio 20+ main stream scenarios and 150+ SOTA models with end-to-end optimization, multi-platform and multi-framework support.
C++
2,695
star
20

Research

novel deep learning research works with PaddlePaddle
Python
1,694
star
21

PGL

Paddle Graph Learning (PGL) is an efficient and flexible graph learning framework based on PaddlePaddle
Python
1,558
star
22

PaddleVideo

Awesome video understanding toolkits based on PaddlePaddle. It supports video data annotation tools, lightweight RGB and skeleton based action recognition model, practical applications for video tagging and sport action detection.
Python
1,392
star
23

Paddle.js

Paddle.js is a web project for Baidu PaddlePaddle, which is an open source deep learning framework running in the browser. Paddle.js can either load a pre-trained model, or transforming a model from paddle-hub with model transforming tools provided by Paddle.js. It could run in every browser with WebGL/WebGPU/WebAssembly supported. It could also run in Baidu Smartprogram and WX miniprogram.
JavaScript
928
star
24

Serving

A flexible, high-performance carrier for machine learning models（『飞桨』服务化部署框架）
C++
869
star
25

PaddleHelix

Bio-Computing Platform Featuring Large-Scale Representation Learning and Multi-Task Deep Learning “螺旋桨”生物计算工具集
Python
784
star
26

RocketQA

🚀 RocketQA, dense retrieval for information retrieval and question answering, including both Chinese and English state-of-the-art models.
Python
742
star
27

X2Paddle

Deep learning model converter for PaddlePaddle. (『飞桨』深度学习模型转换工具)
Python
713
star
28

Knover

Large-scale open domain KNOwledge grounded conVERsation system based on PaddlePaddle
Python
670
star
29

Paddle-Lite-Demo

lib, demo, model, data
C++
640
star
30

Paddle2ONNX

ONNX Model Exporter for PaddlePaddle
Python
637
star
31

Parakeet

PAddle PARAllel text-to-speech toolKIT (supporting Tacotron2, Transformer TTS, FastSpeech2/FastPitch, SpeedySpeech, WaveFlow and Parallel WaveGAN)
Python
599
star
32

FlyCV

FlyCV is a high-performance library for processing computer visual tasks.
C++
560
star
33

Anakin

High performance Cross-platform Inference-engine, you could run Anakin on x86-cpu,arm, nv-gpu, amd-gpu,bitmain and cambricon devices.
C++
530
star
34

Paddle3D

A 3D computer vision development toolkit based on PaddlePaddle. It supports point-cloud object detection, segmentation, and monocular 3D object detection models.
Python
529
star
35

Quantum

Jupyter Notebook
528
star
36

PaddleYOLO

🚀🚀🚀 YOLO series of PaddlePaddle implementation, PP-YOLOE+, RT-DETR, YOLOv5, YOLOv6, YOLOv7, YOLOv8, YOLOX, YOLOv5u, YOLOv7u, YOLOv6Lite, RTMDet and so on. 🚀🚀🚀
Python
500
star
37

PaddleFL

Federated Deep Learning in PaddlePaddle
Python
480
star
38

VIMER

视觉预训练基础模型仓库
Python
479
star
39

PaddleTS

Awesome Easy-to-Use Deep Time Series Modeling based on PaddlePaddle, including comprehensive functionality modules like TSDataset, Analysis, Transform, Models, AutoTS, and Ensemble, etc., supporting versatile tasks like time series forecasting, representation learning, and anomaly detection, etc., featured with quick tracking of SOTA deep models.
Python
444
star
40

PaddleFleetX

飞桨大模型开发套件，提供大语言模型、跨模态大模型、生物计算大模型等领域的全流程开发工具链。
Python
417
star
41

PaddleRS

Awesome Remote Sensing Toolkit based on PaddlePaddle.
Python
330
star
42

PaddleSpatial

PaddleSpatial is an open-source spatial-temporal computing tool based on PaddlePaddle.
GLSL
316
star
43

PaddleCloud

PaddlePaddle Docker images and K8s operators for PaddleOCR/Detection developers to use on public/private cloud.
Go
279
star
44

ERNIE-SDK

ERNIE Bot Agent is a Large Language Model (LLM) Agent Framework, powered by the advanced capabilities of ERNIE Bot and the platform resources of Baidu AI Studio.
Jupyter Notebook
274
star
45

MetaGym

Collection of Reinforcement Learning / Meta Reinforcement Learning Environments.
Python
267
star
46

PASSL

PASSL包含 SimCLR，MoCo v1/v2，BYOL，CLIP，PixPro，simsiam, SwAV, BEiT，MAE 等图像自监督算法以及 Vision Transformer，DEiT，Swin Transformer，CvT，T2T-ViT，MLP-Mixer，XCiT，ConvNeXt，PVTv2 等基础视觉算法
Python
257
star
47

PaddleScience

PaddleScience is SDK and library for developing AI-driven scientific computing applications based on PaddlePaddle.
Python
234
star
48

docs

Documentations for PaddlePaddle
Python
230
star
49

InterpretDL

InterpretDL: Interpretation of Deep Learning Models，基于『飞桨』的模型可解释性算法库。
Python
226
star
50

Paddle-Inference-Demo

C++
223
star
51

PaddleRobotics

PaddleRobotics is an open-source algorithm library for robots based on Paddle, including open-source parts such as human-robot interaction, complex motion control, environment perception, SLAM positioning, and navigation.
Python
210
star
52

PaddleMIX

Paddle Multimodal Integration and eXploration, supporting mainstream multi-modal tasks, including end-to-end large-scale multi-modal pretrain models and diffusion model toolbox. Equipped with high performance and flexibility.
Python
207
star
53

TrustAI

飞桨可信AI
Python
179
star
54

ElasticCTR

ElasticCTR，即飞桨弹性计算推荐系统，是基于Kubernetes的企业级推荐系统开源解决方案。该方案融合了百度业务场景下持续打磨的高精度CTR模型、飞桨开源框架的大规模分布式训练能力、工业级稀疏参数弹性调度服务，帮助用户在Kubernetes环境中一键完成推荐系统部署，具备高性能、工业级部署、端到端体验的特点，并且作为开源套件，满足二次深度开发的需求。
Python
176
star
55

PALM

a Fast, Flexible, Extensible and Easy-to-use NLP Large-scale Pretraining and Multi-task Learning Framework.
Python
174
star
56

AutoDL

Python
158
star
57

PLSC

Paddle Large Scale Classification Tools，supports ArcFace, CosFace, PartialFC, Data Parallel + Model Parallel. Model includes ResNet, ViT, Swin, DeiT, CaiT, FaceViT, MoCo, MAE, ConvMAE, CAE.
Python
142
star
58

CINN

Compiler Infrastructure for Neural Networks
C++
139
star
59

LiteKit

Off-The-Shelf AI Development Kit for APP Developers based on Paddle Lite （『飞桨』移动端开箱即用AI套件, 包含Java & Objective C接口支持）
Objective-C
131
star
60

PaddleFlow

Go
104
star
61

PaddleDTX

Paddle with Decentralized Trust based on Xuperchain
Go
87
star
62

PaddleSports

Python
86
star
63

XWorld

A C++/Python simulator package for reinforcement learning
C++
84
star
64

hapi

hapi is a High-level API that supports both static and dynamic execution modes
Jupyter Notebook
76
star
65

benchmark

Python
74
star
66

PaddleSleeve

PaddleSleeve
Python
70
star
67

Mobile

Embedded and Mobile Deployment
Python
70
star
68

community

PaddlePaddle Developer Community
Jupyter Notebook
66
star
69

PaConvert

Code Convert to PaddlePaddle Toolkit
Python
66
star
70

PaddleDepth

Python
58
star
71

PaddleCustomDevice

PaddlePaddle custom device implementaion. (『飞桨』自定义硬件接入实现)
C++
49
star
72

PaddlePaddle.org

PaddlePaddle.org is the repository for the website of the PaddlePaddle open source project.
CSS
48
star
73

PaddleTest

PaddlePaddle TestSuite
Python
43
star
74

PaDiff

Paddle Automatically Diff Precision Toolkits.
Python
42
star
75

EasyData

Python
35
star
76

epep

Easy & Effective Application Framework for PaddlePaddle
Python
34
star
77

paddle-ce-latest-kpis

Paddle Continuous Evaluation, keep updating.
Python
26
star
78

VisionTools

Python
22
star
79

Contrib

contribution works with PaddlePaddle from the third party developers
Python
20
star
80

PaddleCraft

Take neural networks as APIs for human-like AI.
Python
20
star
81

PaddleTransfer

飞桨迁移学习算法库
Python
19
star
82

recordio

An implementation of the RecordIO file format.
Go
19
star
83

continuous_evaluation

Macro Continuous Evaluation Platform for Paddle.
Python
19
star
84

Perf

SOTA benchmark
Python
17
star
85

Paddle-bot

Python
17
star
86

examples

Python
17
star
87

continuous_integration

Python
16
star
88

tape

C++
14
star
89

PaddleSOT

A Bytecode level Implementation of Symbolic OpCode Translator For PaddlePaddle
Python
14
star
90

paddle_upgrade_tool

upgrade paddle-1.x to paddle-2.0
Python
12
star
91

talks

Shell
6
star
92

CLA

5
star
93

any

Legacy Repo only for PaddlePaddle with version <= 1.3
C++
5
star