• Stars
    star
    2,052
  • Rank 22,523 (Top 0.5 %)
  • Language
    Python
  • License
    Apache License 2.0
  • Created over 2 years ago
  • Updated 9 months ago

Reviews

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

Repository Details

EasyNLP: A Comprehensive and Easy-to-use NLP Toolkit



EasyNLP is a Comprehensive and Easy-to-use NLP Toolkit

website online Open in PAI-DSW open issues GitHub pull-requests GitHub latest commit PRs Welcome

EasyNLP 中文介绍

EasyNLP is an easy-to-use NLP development and application toolkit in PyTorch, first released inside Alibaba in 2021. It is built with scalable distributed training strategies and supports a comprehensive suite of NLP algorithms for various NLP applications. EasyNLP integrates knowledge distillation and few-shot learning for landing large pre-trained models, together with various popular multi-modality pre-trained models. It provides a unified framework of model training, inference, and deployment for real-world applications. It has powered more than 10 BUs and more than 20 business scenarios within the Alibaba group. It is seamlessly integrated to Platform of AI (PAI) products, including PAI-DSW for development, PAI-DLC for cloud-native training, PAI-EAS for serving, and PAI-Designer for zero-code model training.

Main Features

  • Easy to use and highly customizable: In addition to providing easy-to-use and concise commands to call cutting-edge models, it also abstracts certain custom modules such as AppZoo and ModelZoo to make it easy to build NLP applications. It is equipped with the PAI PyTorch distributed training framework TorchAccelerator to speed up distributed training.
  • Compatible with open-source libraries: EasyNLP has APIs to support the training of models from Huggingface/Transformers with the PAI distributed framework. It also supports the pre-trained models in EasyTransfer ModelZoo.
  • Knowledge-injected pre-training: The PAI team has a lot of research on knowledge-injected pre-training, and builds a knowledge-injected model that wins first place in the CCF knowledge pre-training competition. EasyNLP integrates these cutting-edge knowledge pre-trained models, including DKPLM and KGBERT.
  • Landing large pre-trained models: EasyNLP provides few-shot learning capabilities, allowing users to finetune large models with only a few samples to achieve good results. At the same time, it provides knowledge distillation functions to help quickly distill large models to a small and efficient model to facilitate online deployment.
  • Multi-modality pre-trained models: EasyNLP is not about NLP only. It also supports various popular multi-modality pre-trained models to support vision-language tasks that require visual knowledge. For example, it is equipped with CLIP-style models for text-image matching and DALLE-style models for text-to-image generation.

Technical Articles

We have a series of technical articles on the functionalities of EasyNLP.

Installation

You can setup from the source:

$ git clone https://github.com/alibaba/EasyNLP.git
$ cd EasyNLP
$ python setup.py install

This repo is tested on Python 3.6, PyTorch >= 1.8.

Quick Start

Now let's show how to use just a few lines of code to build a text classification model based on BERT.

from easynlp.appzoo import ClassificationDataset
from easynlp.appzoo import get_application_model, get_application_evaluator
from easynlp.core import Trainer
from easynlp.utils import initialize_easynlp, get_args
from easynlp.utils.global_vars import parse_user_defined_parameters
from easynlp.utils import get_pretrain_model_path

initialize_easynlp()
args = get_args()
user_defined_parameters = parse_user_defined_parameters(args.user_defined_parameters)
pretrained_model_name_or_path = get_pretrain_model_path(user_defined_parameters.get('pretrain_model_name_or_path', None))

train_dataset = ClassificationDataset(
    pretrained_model_name_or_path=pretrained_model_name_or_path,
    data_file=args.tables.split(",")[0],
    max_seq_length=args.sequence_length,
    input_schema=args.input_schema,
    first_sequence=args.first_sequence,
    second_sequence=args.second_sequence,
    label_name=args.label_name,
    label_enumerate_values=args.label_enumerate_values,
    user_defined_parameters=user_defined_parameters,
    is_training=True)

valid_dataset = ClassificationDataset(
    pretrained_model_name_or_path=pretrained_model_name_or_path,
    data_file=args.tables.split(",")[-1],
    max_seq_length=args.sequence_length,
    input_schema=args.input_schema,
    first_sequence=args.first_sequence,
    second_sequence=args.second_sequence,
    label_name=args.label_name,
    label_enumerate_values=args.label_enumerate_values,
    user_defined_parameters=user_defined_parameters,
    is_training=False)

model = get_application_model(app_name=args.app_name,
    pretrained_model_name_or_path=pretrained_model_name_or_path,
    num_labels=len(valid_dataset.label_enumerate_values),
    user_defined_parameters=user_defined_parameters)

trainer = Trainer(model=model, train_dataset=train_dataset,user_defined_parameters=user_defined_parameters,
    evaluator=get_application_evaluator(app_name=args.app_name, valid_dataset=valid_dataset,user_defined_parameters=user_defined_parameters,
    eval_batch_size=args.micro_batch_size))
    
trainer.train()

The complete example can be found here.

You can also use AppZoo Command Line Tools to quickly train an App model. Take text classification on SST-2 dataset as an example. First you can download the train.tsv, and dev.tsv, then start training:

$ easynlp \
   --mode=train \
   --worker_gpu=1 \
   --tables=train.tsv,dev.tsv \
   --input_schema=label:str:1,sid1:str:1,sid2:str:1,sent1:str:1,sent2:str:1 \
   --first_sequence=sent1 \
   --label_name=label \
   --label_enumerate_values=0,1 \
   --checkpoint_dir=./classification_model \
   --epoch_num=1  \
   --sequence_length=128 \
   --app_name=text_classify \
   --user_defined_parameters='pretrain_model_name_or_path=bert-small-uncased'

And then predict:

$ easynlp \
  --mode=predict \
  --tables=dev.tsv \
  --outputs=dev.pred.tsv \
  --input_schema=label:str:1,sid1:str:1,sid2:str:1,sent1:str:1,sent2:str:1 \
  --output_schema=predictions,probabilities,logits,output \
  --append_cols=label \
  --first_sequence=sent1 \
  --checkpoint_path=./classification_model \
  --app_name=text_classify

To learn more about the usage of AppZoo, please refer to our documentation.

ModelZoo

EasyNLP currently provides the following models in ModelZoo:

  1. PAI-BERT-zh (from Alibaba PAI): pre-trained BERT models with a large Chinese corpus.
  2. DKPLM (from Alibaba PAI): released with the paper DKPLM: Decomposable Knowledge-enhanced Pre-trained Language Model for Natural Language Understanding by Taolin Zhang, Chengyu Wang, Nan Hu, Minghui Qiu, Chengguang Tang, Xiaofeng He and Jun Huang.
  3. KGBERT (from Alibaba Damo Academy & PAI): pre-train BERT models with knowledge graph embeddings injected.
  4. BERT (from Google): released with the paper BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding by Jacob Devlin, Ming-Wei Chang, Kenton Lee and Kristina Toutanova.
  5. RoBERTa (from Facebook): released with the paper RoBERTa: A Robustly Optimized BERT Pretraining Approach by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer and Veselin Stoyanov.
  6. Chinese RoBERTa (from HFL): the Chinese version of RoBERTa.
  7. MacBERT (from HFL): released with the paper Revisiting Pre-trained Models for Chinese Natural Language Processing by Yiming Cui, Wanxiang Che, Ting Liu, Bing Qin, Shijin Wang and Guoping Hu.
  8. WOBERT (from ZhuiyiTechnology): the word-based BERT for the Chinese language.
  9. FashionBERT (from Alibaba PAI & ICBU): in progress.
  10. GEEP (from Alibaba PAI): in progress.
  11. Mengzi (from Langboat): released with the paper Mengzi: Towards Lightweight yet Ingenious Pre-trained Models for Chinese by Zhuosheng Zhang, Hanqing Zhang, Keming Chen, Yuhang Guo, Jingyun Hua, Yulong Wang and Ming Zhou.
  12. Erlangshen (from IDEA): released from the repo.

Please refer to this readme for the usage of these models in EasyNLP. Meanwhile, EasyNLP supports to load pretrained models from Huggingface/Transformers, please refer to this tutorial for details.

EasyNLP Goes Multi-modal

EasyNLP also supports various popular multi-modality pre-trained models to support vision-language tasks that require visual knowledge. For example, it is equipped with CLIP-style models for text-image matching and DALLE-style models for text-to-image generation.

  1. Text-image Matching
  2. Text-to-image Generation
  3. Image-to-text Generation

Landing Large Pre-trained Models

EasyNLP provide few-shot learning and knowledge distillation to help land large pre-trained models.

  1. PET (from LMU Munich and Sulzer GmbH): released with the paper Exploiting Cloze Questions for Few Shot Text Classification and Natural Language Inference by Timo Schick and Hinrich Schutze. We have made some slight modifications to make the algorithm suitable for the Chinese language.
  2. P-Tuning (from Tsinghua University, Beijing Academy of AI, MIT and Recurrent AI, Ltd.): released with the paper GPT Understands, Too by Xiao Liu, Yanan Zheng, Zhengxiao Du, Ming Ding, Yujie Qian, Zhilin Yang and Jie Tang. We have made some slight modifications to make the algorithm suitable for the Chinese language.
  3. CP-Tuning (from Alibaba PAI): released with the paper Making Pre-trained Language Models End-to-end Few-shot Learners with Contrastive Prompt Tuning by Ziyun Xu, Chengyu Wang, Minghui Qiu, Fuli Luo, Runxin Xu, Songfang Huang and Jun Huang.
  4. Vanilla KD (from Alibaba PAI): distilling the logits of large BERT-style models to smaller ones.
  5. Meta KD (from Alibaba PAI): released with the paper Meta-KD: A Meta Knowledge Distillation Framework for Language Model Compression across Domains by Haojie Pan, Chengyu Wang, Minghui Qiu, Yichang Zhang, Yaliang Li and Jun Huang.
  6. Data Augmentation (from Alibaba PAI): augmentating the data based on the MLM head of pre-trained language models.

CLUE Benchmark

EasyNLP provides a simple toolkit to benchmark clue datasets. You can simply use just this command to benchmark CLUE dataset.

# Format: bash run_clue.sh device_id train/predict dataset
# e.g.: 
bash run_clue.sh 0 train csl

We've tested chiese bert and roberta modelson the datasets, the results of dev set are:

(1) bert-base-chinese:

Task AFQMC CMNLI CSL IFLYTEK OCNLI TNEWS WSC
P 72.17% 75.74% 80.93% 60.22% 78.31% 57.52% 75.33%
F1 52.96% 75.74% 81.71% 60.22% 78.30% 57.52% 80.82%

(2) chinese-roberta-wwm-ext:

Task AFQMC CMNLI CSL IFLYTEK OCNLI TNEWS WSC
P 73.10% 80.75% 80.07% 60.98% 80.75% 57.93% 86.84%
F1 56.04% 80.75% 81.50% 60.98% 80.75% 57.93% 89.58%

Here is the detailed CLUE benchmark example.

Tutorials

License

This project is licensed under the Apache License (Version 2.0). This toolkit also contains some code modified from other repos under other open-source licenses. See the NOTICE file for more information.

ChangeLog

  • EasyNLP v0.0.3 was released in 01/04/2022. Please refer to tag_v0.0.3 for more details and history.

Contact Us

Scan the following QR codes to join Dingtalk discussion group. The group discussions are mostly in Chinese, but English is also welcomed.

Reference

We have an arxiv paper for you to cite for the EasyNLP library:

@article{easynlp,
  doi = {10.48550/ARXIV.2205.00258},  
  url = {https://arxiv.org/abs/2205.00258},  
  author = {Wang, Chengyu and Qiu, Minghui and Zhang, Taolin and Liu, Tingting and Li, Lei and Wang, Jianing and Wang, Ming and Huang, Jun and Lin, Wei},
  title = {EasyNLP: A Comprehensive and Easy-to-use Toolkit for Natural Language Processing},
  publisher = {arXiv},  
  year = {2022}
}

More Repositories

1

arthas

Alibaba Java Diagnostic Tool Arthas/Alibaba Java诊断利器Arthas
Java
35,294
star
2

easyexcel

快速、简洁、解决大文件内存溢出的java处理Excel工具
Java
32,157
star
3

p3c

Alibaba Java Coding Guidelines pmd implements and IDE plugin
Kotlin
30,344
star
4

nacos

an easy-to-use dynamic service discovery, configuration and service management platform for building cloud native applications.
Java
30,212
star
5

canal

阿里巴巴 MySQL binlog 增量订阅&消费组件
Java
28,441
star
6

druid

阿里云计算平台DataWorks(https://help.aliyun.com/document_detail/137663.html) 团队出品,为监控而生的数据库连接池
Java
27,950
star
7

spring-cloud-alibaba

Spring Cloud Alibaba provides a one-stop solution for application development for the distributed solutions of Alibaba middleware.
Java
27,866
star
8

fastjson

FASTJSON 2.0.x has been released, faster and more secure, recommend you upgrade.
Java
25,716
star
9

flutter-go

flutter 开发者帮助 APP,包含 flutter 常用 140+ 组件的demo 演示与中文文档
Dart
23,629
star
10

Sentinel

A powerful flow control component enabling reliability, resilience and monitoring for microservices. (面向云原生微服务的高可用流控防护组件)
Java
22,352
star
11

weex

A framework for building Mobile cross-platform UI
C++
18,271
star
12

ice

🚀 ice.js: The Progressive App Framework Based On React(基于 React 的渐进式应用框架)
TypeScript
17,841
star
13

DataX

DataX是阿里云DataWorks数据集成的开源版本。
Java
15,692
star
14

lowcode-engine

An enterprise-class low-code technology stack with scale-out design / 一套面向扩展设计的企业级低代码技术体系
TypeScript
14,512
star
15

ARouter

💪 A framework for assisting in the renovation of Android componentization (帮助 Android App 进行组件化改造的路由框架)
Java
14,228
star
16

hooks

A high-quality & reliable React Hooks library. https://ahooks.pages.dev/
TypeScript
14,005
star
17

tengine

A distribution of Nginx with some advanced features
C
12,807
star
18

formily

📱🚀 🧩 Cross Device & High Performance Normal Form/Dynamic(JSON Schema) Form/Form Builder -- Support React/React Native/Vue 2/Vue 3
TypeScript
11,318
star
19

vlayout

Project vlayout is a powerfull LayoutManager extension for RecyclerView, it provides a group of layouts for RecyclerView. Make it able to handle a complicate situation when grid, list and other layouts in the same recyclerview.
Java
10,800
star
20

COLA

🥤 COLA: Clean Object-oriented & Layered Architecture
Java
9,964
star
21

MNN

MNN is a blazing fast, lightweight deep learning framework, battle-tested by business-critical use cases in Alibaba
C++
8,656
star
22

ali-dbhub

已迁移新仓库,此版本将不再维护
8,318
star
23

atlas

A powerful Android Dynamic Component Framework.
Java
8,127
star
24

otter

阿里巴巴分布式数据库同步系统(解决中美异地机房)
Java
8,069
star
25

rax

🐰 Rax is a progressive framework for building universal application. https://rax.js.org
JavaScript
7,994
star
26

anyproxy

A fully configurable http/https proxy in NodeJS
JavaScript
7,851
star
27

fish-redux

An assembled flutter application framework.
Dart
7,333
star
28

x-render

🚴‍♀️ 阿里 - 很易用的中后台「表单 / 表格 / 图表」解决方案
TypeScript
7,035
star
29

flutter_boost

FlutterBoost is a Flutter plugin which enables hybrid integration of Flutter for your existing native apps with minimum efforts
Dart
6,966
star
30

AndFix

AndFix is a library that offer hot-fix for Android App.
C++
6,954
star
31

transmittable-thread-local

📌 TransmittableThreadLocal (TTL), the missing Java™ std lib(simple & 0-dependency) for framework/middleware, provide an enhanced InheritableThreadLocal that transmits values between threads even using thread pooling components.
Java
6,750
star
32

jvm-sandbox

Real - time non-invasive AOP framework container based on JVM
Java
6,739
star
33

BizCharts

Powerful data visualization library based on G2 and React.
TypeScript
6,066
star
34

freeline

A super fast build tool for Android, an alternative to Instant Run
Java
5,497
star
35

UltraViewPager

UltraViewPager is an extension for ViewPager to provide multiple features in a single ViewPager.
Java
5,003
star
36

jetcache

JetCache is a Java cache framework.
Java
4,774
star
37

AliSQL

AliSQL is a MySQL branch originated from Alibaba Group. Fetch document from Release Notes at bottom.
C++
4,705
star
38

AliOS-Things

面向IoT领域的、高可伸缩的物联网操作系统,可去官网了解更多信息https://www.aliyun.com/product/aliosthings
C
4,583
star
39

dexposed

dexposed enable 'god' mode for single android application.
Java
4,483
star
40

butterfly

🦋Butterfly,A JavaScript/React/Vue2 Diagramming library which concentrate on flow layout field. (基于JavaScript/React/Vue2的流程图组件)
JavaScript
4,445
star
41

QLExpress

QLExpress is a powerful, lightweight, dynamic language for the Java platform aimed at improving developers’ productivity in different business scenes.
Java
4,361
star
42

BeeHive

🐝 BeeHive is a solution for iOS Application module programs, it absorbed the Spring Framework API service concept to avoid coupling between modules.
Objective-C
4,288
star
43

HandyJSON

A handy swift json-object serialization/deserialization library
Swift
4,233
star
44

x-deeplearning

An industrial deep learning framework for high-dimension sparse data
PureBasic
4,185
star
45

Tangram-Android

Tangram is a modular UI solution for building native page dynamically including Tangram for Android, Tangram for iOS and even backend CMS. This project provides the sdk on Android.
Java
4,110
star
46

coobjc

coobjc provides coroutine support for Objective-C and Swift. We added await method、generator and actor model like C#、Javascript and Kotlin. For convenience, we added coroutine categories for some Foundation and UIKit API in cokit framework like NSFileManager, JSON, NSData, UIImage etc. We also add tuple support in coobjc.
Objective-C
4,025
star
47

jstorm

Enterprise Stream Process Engine
Java
3,914
star
48

dragonwell8

Alibaba Dragonwell8 JDK
Java
3,826
star
49

LuaViewSDK

A cross-platform framework to build native, dynamic and swift user interface - 强大轻巧灵活的客户端动态化解决方案
Objective-C
3,707
star
50

fastjson2

🚄 FASTJSON2 is a Java JSON library with excellent performance.
Java
3,673
star
51

Alink

Alink is the Machine Learning algorithm platform based on Flink, developed by the PAI team of Alibaba computing platform.
Java
3,572
star
52

f2etest

F2etest是一个面向前端、测试、产品等岗位的多浏览器兼容性测试整体解决方案。
JavaScript
3,564
star
53

GGEditor

A visual graph editor based on G6 and React
TypeScript
3,414
star
54

GraphScope

🔨 🍇 💻 🚀 GraphScope: A One-Stop Large-Scale Graph Computing System from Alibaba | 一站式图计算系统
C++
3,277
star
55

designable

🧩 Make everything designable 🧩
TypeScript
3,266
star
56

cobar

a proxy for sharding databases and tables
Java
3,210
star
57

macaca

Automation solution for multi-platform. 多端自动化解决方案
3,171
star
58

lightproxy

💎 Cross platform Web debugging proxy
TypeScript
3,111
star
59

pont

🌉数据服务层解决方案
TypeScript
3,035
star
60

higress

🤖 AI Gateway | AI Native API Gateway
Go
2,918
star
61

euler

A distributed graph deep learning framework.
C++
2,849
star
62

sentinel-golang

Sentinel Go enables reliability and resiliency for Go microservices
Go
2,763
star
63

beidou

🌌 Isomorphic framework for server-rendered React apps
JavaScript
2,735
star
64

ChatUI

The UI design language and React library for Conversational UI
TypeScript
2,602
star
65

pipcook

Machine learning platform for Web developers
TypeScript
2,539
star
66

kiwi

🐤 Kiwi-国际化翻译全流程解决方案
TypeScript
2,533
star
67

yugong

阿里巴巴去Oracle数据迁移同步工具(全量+增量,目标支持MySQL/DRDS)
Java
2,504
star
68

jvm-sandbox-repeater

A Java server-side recording and playback solution based on JVM-Sandbox
Java
2,503
star
69

tsar

Taobao System Activity Reporter
C
2,446
star
70

tidevice

tidevice can be used to communicate with iPhone device
Python
2,411
star
71

TProfiler

TProfiler是一个可以在生产环境长期使用的性能分析工具
Java
2,377
star
72

tair

A distributed key-value storage system developed by Alibaba Group
C++
2,179
star
73

dubbo-spring-boot-starter

Dubbo Spring Boot Starter
Java
2,097
star
74

RedisShake

redis-shake is a tool for synchronizing data between two redis databases. Redis-shake 是一个用于在两个 redis之 间同步数据的工具,满足用户非常灵活的同步、迁移需求。
Go
2,077
star
75

uirecorder

UI Recorder is a multi-platform UI test recorder.
JavaScript
2,061
star
76

AliceMind

ALIbaba's Collection of Encoder-decoders from MinD (Machine IntelligeNce of Damo) Lab
Python
1,967
star
77

LVS

A distribution of Linux Virtual Server with some advanced features. It introduces a new packet forwarding method - FULLNAT other than NAT/Tunneling/DirectRouting, and defense mechanism against synflooding attack - SYNPROXY.
C
1,947
star
78

GCanvas

A lightweight cross-platform graphics rendering engine. (超轻量的跨平台图形引擎) https://alibaba.github.io/GCanvas
C
1,873
star
79

alpha

Alpha是一个基于PERT图构建的Android异步启动框架,它简单,高效,功能完善。 在应用启动的时候,我们通常会有很多工作需要做,为了提高启动速度,我们会尽可能让这些工作并发进行。但这些工作之间可能存在前后依赖的关系,所以我们又需要想办法保证他们执行顺序的正确性。Alpha就是为此而设计的,使用者只需定义好自己的task,并描述它依赖的task,将它添加到Project中。框架会自动并发有序地执行这些task,并将执行的结果抛出来。
HTML
1,873
star
80

Tangram-iOS

Tangram is a modular UI solution for building native page dynamically, including Tangram for Android, Tangram for iOS and even backend CMS. This project provides the sdk on iOS platform.
Objective-C
1,863
star
81

testable-mock

换种思路写Mock,让单元测试更简单
Java
1,827
star
82

compileflow

🎨 core business process engine of Alibaba Halo platform, best process engine for trade scenes. | 一个高性能流程编排引擎
Java
1,793
star
83

SREWorks

Cloud Native DataOps & AIOps Platform | 云原生数智运维平台
Java
1,792
star
84

EasyCV

An all-in-one toolkit for computer vision
Python
1,780
star
85

LazyScrollView

An iOS ScrollView to resolve the problem of reusability in views.
Objective-C
1,774
star
86

EasyRec

A framework for large scale recommendation algorithms.
Python
1,764
star
87

ilogtail

Fast and Lightweight Observability Data Collector
C++
1,740
star
88

MongoShake

MongoShake is a universal data replication platform based on MongoDB's oplog. Redundant replication and active-active replication are two most important functions. 基于mongodb oplog的集群复制工具,可以满足迁移和同步的需求,进一步实现灾备和多活功能。
Go
1,714
star
89

xquic

XQUIC Library released by Alibaba is a cross-platform implementation of QUIC and HTTP/3 protocol.
C
1,687
star
90

lowcode-demo

An enterprise-class low-code technology stack with scale-out design / 一套面向扩展设计的企业级低代码技术体系
TypeScript
1,683
star
91

async_simple

Simple, light-weight and easy-to-use asynchronous components
C++
1,662
star
92

havenask

C++
1,586
star
93

clusterdata

cluster data collected from production clusters in Alibaba for cluster management research
Jupyter Notebook
1,554
star
94

mdrill

for千亿数据即席分析
Java
1,538
star
95

kt-connect

A toolkit for Integrating with your kubernetes dev environment more efficiently
Go
1,519
star
96

Virtualview-Android

A light way to build UI in custom XML.
Java
1,455
star
97

yalantinglibs

A collection of modern C++ libraries, include coro_rpc, struct_pack, struct_json, struct_xml, struct_pb, easylog, async_simple
C++
1,431
star
98

tb_tddl

1,410
star
99

react-intl-universal

Internationalize React apps. Not only for Component but also for Vanilla JS.
JavaScript
1,337
star
100

data-juicer

A one-stop data processing system to make data higher-quality, juicier, and more digestible for LLMs! 🍎 🍋 🌽 ➡️ ➡️🍸 🍹 🍷为大语言模型提供更高质量、更丰富、更易”消化“的数据!
Python
1,292
star