[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-1568":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":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":10,"archived":22,"fork":22,"defaultBranch":23,"hasWiki":24,"hasPages":22,"topics":25,"createdAt":10,"pushedAt":10,"updatedAt":26,"readmeContent":27,"aiSummary":28,"trendingCount":16,"starSnapshotCount":16,"syncStatus":29,"lastSyncTime":30,"discoverSource":31},1568,"dive-into-llms","Lordog\u002Fdive-into-llms","Lordog","《动手学大模型Dive into LLMs》系列编程实践教程","",null,"Jupyter Notebook",40730,4966,251,8,0,43,528,3365,274,45,false,"main",true,[],"2026-06-12 02:00:29","\u003Cp align=\"center\">\n\u003Ch1 align=\"center\">《动手学大模型》系列编程实践教程\u003C\u002Fh1>\n\u003C\u002Fp>\n\u003Cp align=\"center\">\n  \t\u003Ca href=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fversion-v0.1.0-blue\">\n      \u003Cimg alt=\"version\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fversion-v0.1.0-blue?color=FF8000?color=009922\" \u002F>\n    \u003C\u002Fa>\n  \u003Ca >\n       \u003Cimg alt=\"Status-building\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FStatus-building-blue\" \u002F>\n  \t\u003C\u002Fa>\n  \u003Ca >\n       \u003Cimg alt=\"PRs-Welcome\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPRs-Welcome-red\" \u002F>\n  \t\u003C\u002Fa>\n   \t\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fstargazers\">\n       \u003Cimg alt=\"stars\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FLordog\u002Fdive-into-llms\" \u002F>\n  \t\u003C\u002Fa>\n  \t\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fnetwork\u002Fmembers\">\n       \u003Cimg alt=\"FORK\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fforks\u002FLordog\u002Fdive-into-llms?color=FF8000\" \u002F>\n  \t\u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fissues\">\n      \u003Cimg alt=\"Issues\" src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fissues\u002FLordog\u002Fdive-into-llms?color=0088ff\"\u002F>\n    \u003C\u002Fa>\n    \u003Cbr \u002F>\n\u003C\u002Fp>\n\n\n\u003Cdiv align=\"center\">\n\u003Cp align=\"center\">\n  \u003Ca href=\"#项目动机\">项目动机\u003C\u002Fa>\u002F\n  \u003Ca href=\"#教程目录\">教程目录\u003C\u002Fa>\u002F\n  \u003Ca href=\"#贡献者列表\">贡献者列表\u003C\u002Fa>\n\u003C\u002Fp>\n\u003C\u002Fdiv>\n\n## 💡 Updates\n\n2025\u002F06\u002F06  感谢各位朋友们的关注和积极反馈！我们从以下两个方面对本教程进行了更新：\n\n- [x] 上线国产化《大模型开发全流程》公益教程（含PPT、实验手册和视频），此处特别感谢华为昇腾社区的支持！\n- [x] 在原系列编程实践教程的基础上进行内容更新，并增加了新的主题（数学推理、GUI Agent、大模型对齐、隐写术等）！\n\n## 🎯 项目动机\n\n《动手学大模型》系列编程实践教程，由上海交通大学《自然语言处理前沿技术》（NIS8021）、《人工智能安全技术》课程（NIS3353）讲义拓展而来（教师：[张倬胜](https:\u002F\u002Fbcmi.sjtu.edu.cn\u002Fhome\u002Fzhangzs\u002F)），旨在提供大模型相关的入门编程参考。本教程属公益性质、完全免费。通过简单实践，帮助同学们快速入门大模型，更好地开展课程设计或学术研究。\n\n## 📚 教程目录\n\n| 教程内容         | 简介                                                         | 地址                                                         |\n| ---------------- | ------------------------------------------------------------ | ------------------------------------------------------------ |\n| 微调与部署       | 预训练模型微调与部署指南：想提升预训练模型在指定任务上的性能？让我们选择合适的预训练模型，在特定任务上进行微调，并将微调后的模型部署成方便使用的Demo！ | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter1\u002Fdive-into-llm.pdf)] [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter1\u002FREADME.md)] [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter1\u002Fdive-tuning.ipynb)] |\n| 提示学习与思维链 | 大模型的API调用与推理指南：“AI在线求鼓励？大模型对一些问题的回答令人大跌眼镜，但它可能只是想要一句「鼓励」” | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter2\u002Fdive-into-prompting.pdf)] [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter2\u002FREADME.md)] [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter2\u002Fdive-prompting.ipynb)] |\n| 知识编辑         | 语言模型的编辑方法和工具：想操控语言模型在对指定知识的记忆？让我们选择合适的编辑方法，对特定知识进行编辑，并将对编辑后的模型进行验证！ | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fblob\u002Fmain\u002Fdocuments\u002Fchapter3\u002Fdive_edit_0410.pdf)] [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter3\u002FREADME.md)]  [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter3\u002Fdive_edit.ipynb)] |\n| 数学推理         | 如何让大模型学会数学推理？让我们快速蒸馏一个迷你R1！         | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fblob\u002Fmain\u002Fdocuments\u002Fchapter4\u002Fmath.pdf)] [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter4\u002FREADME.md)]  [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter4\u002Fsft_math.ipynb)] |\n| 模型水印         | 语言模型的文本水印：在语言模型生成的内容中嵌入人类不可见的水印 | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fblob\u002Fmain\u002Fdocuments\u002Fchapter5\u002Fwatermark.pdf)] [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter5\u002FREADME.md)]  [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter5\u002Fwatermark.ipynb)] |\n| 越狱攻击         | 想要得到更好的安全，要先从学会攻击开始。让我们了解越狱攻击如何撬开大模型的嘴！ | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fblob\u002Fmain\u002Fdocuments\u002Fchapter6\u002Fdive-Jailbreak.pdf)] [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter6\u002FREADME.md)] [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter6\u002Fdive-jailbreak.ipynb)] |\n| 大模型隐写       | “看不见的墨水”！想让大模型在流畅回答的同时，悄悄携带只有“自己人”能识别的信息吗？大模型隐写告诉你！ | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fblob\u002Fmain\u002Fdocuments\u002Fchapter7\u002Fstega.pdf)] [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter7\u002FREADME.md)] [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter7\u002Fllm_stega.ipynb)] |\n| 多模态模型       | 作为能够更充分模拟真实世界的多模态大语言模型，其如何实现更强大的多模态理解和生成能力？多模态大语言模型是否能够帮助实现AGI？ | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fblob\u002Fmain\u002Fdocuments\u002Fchapter8\u002Fmllms.pdf)]  [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter8\u002FREADME.md)] [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter8\u002Fmllms.ipynb)] |\n| GUI智能体        | 想要饭来张口、解放双手？那么让我们一起来让AI Agent替你点外卖、回消息、购物比价吧！ | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fblob\u002Fmain\u002Fdocuments\u002Fchapter9\u002FGUIagent.pdf)]  [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter9\u002FREADME.md)] [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter9\u002FGUIagent.ipynb)] |\n| 智能体安全       | 大模型智能体迈向了未来操作系统之旅。然而，大模型在开放智能体场景中能意识到风险威胁吗？ | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fblob\u002Fmain\u002Fdocuments\u002Fchapter10\u002Fdive-into-safety.pdf)] [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter10\u002FREADME.md)] [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter10\u002Fagent.ipynb)] |\n| RLHF安全对齐     | 基于PPO的RLHF实验指南：本教程”十分危险“，阅读后请检查你的大模型是否在冷笑。 | [[课件](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Fblob\u002Fmain\u002Fdocuments\u002Fchapter11\u002FRLHF.pdf)] [[教程](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter11\u002FREADME.md)] [[脚本](https:\u002F\u002Fgithub.com\u002FLordog\u002Fdive-into-llms\u002Ftree\u002Fmain\u002Fdocuments\u002Fchapter11\u002FRLHF.ipynb)] |\n\n\n\n## 🔥 新上线：国产化《大模型开发全流程》\n\n- **✨ 我们联合华为昇腾推出的《大模型开发全流程》公益教程正式上线！前沿技术+代码实践，手把手带你玩转AI大模型 ✨**: \n\n  在《动手学大模型》原系列教程的基础上，我们联合华为开发了《大模型开发全流程》系列课程。本系列教程基于昇腾基础软硬件开发，覆盖PPT、实验手册、视频等教程形式。该教程分为初级、中级、高级系列，面向不同的大模型实践需求，旨在将前沿技术通过代码实践的方式，为相关研究者、开发者由浅入深地提供快速上手、应用昇腾已支持模型和全新模型迁移调优的全流程开发指南。\n  \n- **🚀 前往昇腾社区探索《大模型开发全流程》系列课程**： \n  \n  👉《[大模型开发学习专区](https:\u002F\u002Fwww.hiascend.com\u002Fedu\u002Fgrowth\u002Flm-development#classification-floor-1)》@ 昇腾社区 👈 \n  \n- **✨ 课程内容展示 ✨**\n\n  \u003C!-- \u003Cimg src=\".\u002Fpics\u002Ficon\u002Ftitle.jpg\" width=\"300\"\u002F>\n  \u003Cimg src=\".\u002Fpics\u002Ficon\u002Fcover.png\" width=\"300\"\u002F>\n  \u003Cimg src=\".\u002Fpics\u002Ficon\u002Fteam.png\" width=\"300\"\u002F>\n  \u003Cimg src=\".\u002Fpics\u002Ficon\u002Fagent.png\" width=\"300\"\u002F> -->\n\n\u003Cp align = \"center\">\n  \u003Cimg src=\".\u002Fpics\u002Ficon\u002Ftitle.jpg\" width=\"48%\"\u002F>\n  \u003Cimg src=\".\u002Fpics\u002Ficon\u002Fcover.png\" width=\"48%\"\u002F>\n  \u003Cimg src=\".\u002Fpics\u002Ficon\u002Fteam.png\" width=\"48%\"\u002F>\n  \u003Cimg src=\".\u002Fpics\u002Ficon\u002Fagent.png\" width=\"48%\"\u002F>\n\u003C\u002Fp>\n\n## 🙏 免责声明\n\n本教程所有内容仅仅来自于贡献者的个人经验、互联网数据、日常科研工作中的相关积累。所有技巧仅供参考，不保证百分百正确。若有任何问题，欢迎提交 Issue 或 PR。另本项目所用徽章来自互联网，如侵犯了您的图片版权请联系我们删除，谢谢。\n\n## 🤝 欢迎贡献\n\n本教程目前是一个正在进行中的项目，如有疏漏在所难免，欢迎任何的PR及issue讨论。\n\n## ❤️ 贡献者列表\n\n感谢以下老师和同学对本项目的支持与贡献：\n\n**《动手学大模型》系列教程开发团队**：\n\n- 上海交通大学：[张倬胜](https:\u002F\u002Fbcmi.sjtu.edu.cn\u002Fhome\u002Fzhangzs\u002F)、[袁童鑫](https:\u002F\u002Fgithub.com\u002FLordog)、[马欣贝](https:\u002F\u002Fscholar.google.com\u002Fcitations?user=LpUi3EgAAAAJ&hl=zh-CN&oi=ao)、 [何志威](https:\u002F\u002Fzwhe99.github.io)、[杜巍](https:\u002F\u002Fscholar.google.com\u002Fcitations?user=tFYUBLkAAAAJ&hl=en)、[赵皓东](https:\u002F\u002Fdongdongzhaoup.github.io\u002F)、[吴宗儒](https:\u002F\u002Fzrw00.github.io\u002F)、[吴铮](https:\u002F\u002Fwuzheng02.github.io\u002F)、[董凌众](https:\u002F\u002Fgithub.com\u002FLZ-Dong)、[张玉龙](https:\u002F\u002Faslan-yulong.github.io\u002F)\n\n- 新加坡国立大学：[费豪](http:\u002F\u002Fhaofei.vip\u002F)\n\n**《大模型开发全流程》系列教程开发团队：**\n\n- 上海交通大学：[张倬胜](https:\u002F\u002Fbcmi.sjtu.edu.cn\u002Fhome\u002Fzhangzs\u002F)、[刘功申](https:\u002F\u002Finfosec.sjtu.edu.cn\u002FDirectoryDetail.aspx?id=75)、[陈星宇](https:\u002F\u002Fscholar.google.com\u002Fcitations?user=d-dNtjrMJ5YC&hl=en)、[程彭洲](https:\u002F\u002Fscholar.google.com\u002Fcitations?user=qxnwzDUAAAAJ&hl=en)、[董凌众](https:\u002F\u002Fgithub.com\u002FLZ-Dong)、 [何志威](https:\u002F\u002Fzwhe99.github.io)、[鞠天杰](https:\u002F\u002Fscholar.google.com\u002Fcitations?user=f8PPcnoAAAAJ&hl=en)、[马欣贝](https:\u002F\u002Fscholar.google.com\u002Fcitations?user=LpUi3EgAAAAJ&hl=zh-CN&oi=ao)、 [吴铮](https:\u002F\u002Fscholar.google.com\u002Fcitations?hl=zh-CN&user=qBM1UbUAAAAJ&view_op=list_works&gmla=AIfU4H6PG9JyjRub6BYIIZ4isQE7MBAM3Eoec6OJfX4z_8-pOE8bI1Wgdo3XL5qOZWR3U-h-lIP2q0zXt5gzyFKMSg7MNnBBWLv5d1IVG30UANczTP0)、[吴宗儒](https:\u002F\u002Fzrw00.github.io\u002F)、[闫子赫](https:\u002F\u002Fscholar.google.com\u002Fcitations?user=O2YfSHoAAAAJ&hl=zh-CN)、[姚杳](https:\u002F\u002Fscholar.google.com\u002Fcitations?user=tLMP3IkAAAAJ)、[袁童鑫](https:\u002F\u002Fgithub.com\u002FLordog)、[赵皓东](https:\u002F\u002Fdongdongzhaoup.github.io\u002F);\n\n- 华为昇腾社区：ZOMI、谢乾、程黎明、楼梨华、焦泽昱\n\n## 🌟 Star History\n\n[![Star History Chart](https:\u002F\u002Fapi.star-history.com\u002Fsvg?repos=Lordog\u002Fdive-into-llms&type=Date)](https:\u002F\u002Fstar-history.com\u002F#Lordog\u002Fdive-into-llms&Date)\n","《动手学大模型Dive into LLMs》系列编程实践教程旨在为初学者提供全面的大模型编程指导。该项目通过Jupyter Notebook形式，详细介绍了大模型的微调与部署、提示学习与思维链、知识编辑以及数学推理等核心功能和技术特点。教程内容涵盖了从基础概念到实际操作的各个环节，并且提供了丰富的示例代码和实验手册，便于用户快速上手。该教程适用于对大模型感兴趣的学生、研究人员及开发者，尤其适合需要进行课程设计或学术研究的场景使用。",2,"2026-06-11 02:44:43","top_all"]