[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-4273":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":16,"stars7d":16,"stars30d":17,"stars90d":16,"forks30d":16,"starsTrendScore":16,"compositeScore":18,"rankGlobal":10,"rankLanguage":10,"license":19,"archived":20,"fork":20,"defaultBranch":21,"hasWiki":22,"hasPages":20,"topics":23,"createdAt":10,"pushedAt":10,"updatedAt":32,"readmeContent":33,"aiSummary":34,"trendingCount":16,"starSnapshotCount":16,"syncStatus":17,"lastSyncTime":35,"discoverSource":36},4273,"angel","Angel-ML\u002Fangel","Angel-ML","A Flexible and Powerful Parameter Server for large-scale machine learning","",null,"Java",6785,1591,437,128,0,2,40.61,"Other",false,"master",true,[24,25,26,27,28,29,30,31],"high-dimensional","machine-learning","model","online-learning","parameter-server","scala","spark","spark-streaming","2026-06-12 02:01:01","![](assets\u002Fangel_logo.png)\n\n[![license](http:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-Apache2.0-brightgreen.svg?style=flat)](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002Fangel\u002Fblob\u002Fbranch-3.2.0\u002FLICENSE.TXT)\n[![Release Version](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Frelease-3.2.0-red.svg)](https:\u002F\u002Fgithub.com\u002Ftencent\u002Fangel\u002Freleases)\n[![PRs Welcome](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPRs-welcome-brightgreen.svg)](https:\u002F\u002Fgithub.com\u002Ftencent\u002Fangel\u002Fpulls)\n[![Download Code](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fdownload-zip-green.svg)](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002Fangel\u002Farchive\u002Frefs\u002Fheads\u002Fbranch-3.2.0.zip)\n\n[(ZH-CN Version)](.\u002FREADME_CN.md)\n\n**Angel** is a high-performance distributed machine learning and graph computing platform based on the philosophy of Parameter Server. It is tuned for performance with big data from Tencent and has a wide range of applicability and stability, demonstrating increasing advantage in handling higher dimension model. Angel is jointly developed by Tencent and Peking University, taking account of both high availability  in industry and innovation in academia.\n\nWith model-centered core design concept, **Angel** partitions parameters of complex models into multiple parameter-server nodes, and implements a variety of machine learning algorithms and graph algorithms using efficient model-updating interfaces and functions, as well as flexible consistency model for synchronization.\n\n**Angel** is developed with **Java** and **Scala**.  It supports running on **Yarn**. With **PS Service** abstraction, it supports **Spark on Angel**.  Graph computing and deep learning frameworks support is under development and will be released in the future.\n\nWe welcome everyone interested in machine learning or graph computing to contribute code, create issues or pull requests. Please refer to  [Angel Contribution Guide](https:\u002F\u002Fgithub.com\u002FTencent\u002Fangel\u002Fblob\u002Fmaster\u002FCONTRIBUTING.md) for more detail.\n\n## Introduction to Angel\n\n* [Architecture](.\u002Fdocs\u002Foverview\u002Farchitecture_en.md)\n* [Code Framework](.\u002Fdocs\u002Foverview\u002Fcode_framework_en.md)\n* [Design](.\u002Fdocs\u002Foverview\u002Fdesign_philosophy_en.md)\n* [Spark on Angel](.\u002Fdocs\u002Foverview\u002Fspark_on_angel_en.md)\n  * [Machine Learning](.\u002Fdocs\u002Foverview\u002Fspark_on_angel_en.md)\n  * [Graph Computing](.\u002Fdocs\u002Foverview\u002Fangel_graph_sona_en.md)\n\n## Design\n\n- [Model Partitioner](.\u002Fdocs\u002Fdesign\u002Fmodel_partitioner_en.md)\n- [SyncController](.\u002Fdocs\u002Fdesign\u002Fsync_controller_en.md)\n- [psFunc](.\u002Fdocs\u002Fdesign\u002FpsfFunc_en.md)\n- [Core API](.\u002Fdocs\u002Fapis\u002Fcore_api_en.md)\n\n\n## Quick Start\n\n* [Quick Start Example](.\u002Fdocs\u002Ftutorials\u002Fspark_on_angel_quick_start_en.md)\n\n## Deployment\n\n* [Compilation Guide](.\u002Fdocs\u002Fdeploy\u002Fsource_compile_en.md)\n* [Running on Local](.\u002Fdocs\u002Fdeploy\u002Flocal_run_en.md)\n* [Running on Yarn](.\u002Fdocs\u002Fdeploy\u002Frun_on_yarn_en.md)\n* [Configuration Details](.\u002Fdocs\u002Fdeploy\u002Fconfig_details_en.md)\n* [Resource Configuration Guide](.\u002Fdocs\u002Fdeploy\u002Fresource_config_guide_en.md)\n\n## Programming Guide\n\n* [Spark on Angel Programming Guide](.\u002Fdocs\u002Fprogrammers_guide\u002Fspark_on_angel_programing_guide_en.md)\n\n## Algorithm\n\n- [**Angel or Spark On Angel？**](.\u002Fdocs\u002Falgo\u002Fangel_or_spark_on_angel.md)\n- [**Algorithm Parameter Description**](.\u002Fdocs\u002Falgo\u002Fmodel_config_details.md)\n- **Angel**\n  - **Traditional Machine Learning Methods**\n    - [Logistic Regression(LR)](.\u002Fdocs\u002Falgo\u002Flr_on_angel_en.md)\n    - [Support Vector Machine(SVM)](.\u002Fdocs\u002Falgo\u002Fsvm_on_angel_en.md)\n    - [Factorization Machine(FM)](.\u002Fdocs\u002Falgo\u002Ffm_on_angel.md)\n    - [Linear Regression](.\u002Fdocs\u002Falgo\u002Flinear_on_angel_en.md)\n    - [Robust Regression](.\u002Fdocs\u002Falgo\u002Frobust_on_angel_en.md)\n    - [Softmax Regression](.\u002Fdocs\u002Falgo\u002Fsoftmax_on_angel_en.md)\n    - [KMeans](.\u002Fdocs\u002Falgo\u002Fkmeans_on_angel_en.md)\n    - [GBDT](.\u002Fdocs\u002Falgo\u002Fgbdt_on_angel_en.md)\n    - [LDA\\*](.\u002Fdocs\u002Falgo\u002Flda_on_angel_en.md) ([WarpLDA](.\u002Fdocs\u002Falgo\u002Fwarp_lda_on_angel.md))\n- **Spark on Angel**\n  - **Angel Mllib**\n    - [FM](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Frecommendation.md)\n    - [DeepFM](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Frecommendation.md)\n    - [DeepAndWide](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Frecommendation.md)\n    - [DCN](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Frecommendation.md)\n    - [XDeepFM](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Frecommendation.md)\n    - [AttentionFM](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Frecommendation.md)\n    - [PNN](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Frecommendation.md)\n    - [FTRL](.\u002Fdocs\u002Falgo\u002Fftrl_lr_spark.md)\n    - [Logistic Regression(LR)](.\u002Fdocs\u002Falgo\u002Fsona\u002Flr_sona.md)\n    - [FTRLFM](.\u002Fdocs\u002Falgo\u002Fftrl_fm_spark_en.md)\n    - [GBDT](.\u002Fdocs\u002Falgo\u002Fsona\u002Ffeature_gbdt_sona.md)\n  - **Angel Graph**\n    - [PageRank](.\u002Fdocs\u002Falgo\u002Fsona\u002Fpagerank_on_sona_en.md)\n    - [KCORE](.\u002Fdocs\u002Falgo\u002Fsona\u002Fkcore_sona_en.md)\n    - [HIndex](.\u002Fdocs\u002Falgo\u002Fsona\u002Fhindex_sona_en.md)\n    - [Closeness](.\u002Fdocs\u002Falgo\u002Fsona\u002Fcloseness_sona_en.md)\n    - [CommonFriends](.\u002Fdocs\u002Falgo\u002Fsona\u002Fcommonfriends_sona_en.md)\n    - [ConnectedComponents](.\u002Fdocs\u002Falgo\u002Fsona\u002FCC_sona_en.md)\n    - [TriangleCountingUndirected](.\u002Fdocs\u002Falgo\u002Fsona\u002Ftriangle_count_undirected_en.md)\n    - [Louvain](.\u002Fdocs\u002Falgo\u002Fsona\u002Flouvain_sona_en.md)\n    - [LPA](.\u002Fdocs\u002Falgo\u002Fsona\u002FLPA_sona_en.md)\n    - [LINE](.\u002Fdocs\u002Falgo\u002Fsona\u002Fline_sona_en.md)\n    - [Word2Vec](.\u002Fdocs\u002Falgo\u002Fsona\u002Fword2vec_sona_en.md)\n    - [GraphSage](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Fgraph.md)\n    - [GCN](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Fgraph.md)\n    - [DGI](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002FPyTorch-On-Angel\u002Fblob\u002Fbranch-0.2.0\u002Fdocs\u002Fgraph.md)\n\n## Community\n* Mailing list: angel-tsc@lists.deeplearningfoundation.org\n* Angel homepage in Linux FD: https:\u002F\u002Fangelml.ai\u002F\n* [Committers & Contributors](.\u002FCOMMITTERS.md)\n* [Contributing to Angel](.\u002FCONTRIBUTING.md)\n* [Roadmap](https:\u002F\u002Fgithub.com\u002FAngel-ML\u002Fangel\u002Fwiki\u002FRoadmap)\n\n## FAQ\n* [Angel FAQ](https:\u002F\u002Fgithub.com\u002FTencent\u002Fangel\u002Fwiki\u002FAngel%E5%B8%B8%E8%A7%81%E9%97%AE%E9%A2%98)\n\n## Papers\n  1. [PaSca: A Graph Neural Architecture Search System under the Scalable Paradigm](https:\u002F\u002Fdl.acm.org\u002Fdoi\u002Fpdf\u002F10.1145\u002F3485447.3511986). WWW, 2022\n  2. [Graph Attention Multi-Layer Perceptron](https:\u002F\u002Fdl.acm.org\u002Fdoi\u002Fpdf\u002F10.1145\u002F3534678.3539121). KDD, 2022\n  3. [Node Dependent Local Smoothing for Scalable Graph Learning](https:\u002F\u002Fproceedings.neurips.cc\u002Fpaper\u002F2021\u002Ffile\u002Fa9eb812238f753132652ae09963a05e9-Paper.pdf). NeurlPS, 2021\n  4. [PSGraph: How Tencent trains extremely large-scale graphs with Spark?](https:\u002F\u002Fconferences.computer.org\u002Ficde\u002F2020\u002Fpdfs\u002FICDE2020-5acyuqhpJ6L9P042wmjY1p\u002F290300b549\u002F290300b549.pdf).ICDE, 2020.\n  5. [DimBoost: Boosting Gradient Boosting Decision Tree to Higher Dimensions](https:\u002F\u002Fdl.acm.org\u002Fcitation.cfm?id=3196892). SIGMOD, 2018.\n  6. [LDA*: A Robust and Large-scale Topic Modeling System](http:\u002F\u002Fwww.vldb.org\u002Fpvldb\u002Fvol10\u002Fp1406-yu.pdf). VLDB, 2017\n  7. [Heterogeneity-aware Distributed Parameter Servers](http:\u002F\u002Fnet.pku.edu.cn\u002F~cuibin\u002FPapers\u002F2017%20sigmod.pdf). SIGMOD, 2017\n  8. [Angel: a new large-scale machine learning system](http:\u002F\u002Fnet.pku.edu.cn\u002F~cuibin\u002FPapers\u002F2017NSRangel.pdf). National Science Review (NSR), 2017\n  9. [TencentBoost: A Gradient Boosting Tree System with Parameter Server](http:\u002F\u002Fnet.pku.edu.cn\u002F~cuibin\u002FPapers\u002F2017%20ICDE%20boost.pdf).\tICDE, 2017\n","Angel-ML\u002Fangel是一个高性能的分布式机器学习和图计算平台，基于参数服务器架构设计。其核心功能包括将复杂模型的参数分割到多个参数服务器节点上，并通过高效的模型更新接口和灵活的一致性模型实现多种机器学习算法与图算法，特别适用于处理高维度模型。该项目采用Java和Scala开发，支持在Yarn上运行，并通过PS Service抽象层支持Spark on Angel。适合需要进行大规模数据处理、在线学习以及需要高效模型训练的企业级应用或学术研究场景。","2026-06-11 02:59:23","top_language"]