[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-louisfb01--start-llms":3,"tool-louisfb01--start-llms":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",140436,2,"2026-04-05T23:32:43",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":80,"owner_email":81,"owner_twitter":82,"owner_website":83,"owner_url":84,"languages":81,"stars":85,"forks":86,"last_commit_at":87,"license":88,"difficulty_score":89,"env_os":90,"env_gpu":90,"env_ram":90,"env_deps":91,"category_tags":94,"github_topics":95,"view_count":10,"oss_zip_url":81,"oss_zip_packed_at":81,"status":16,"created_at":107,"updated_at":108,"faqs":109,"releases":110},961,"louisfb01\u002Fstart-llms","start-llms","A complete guide to start and improve your LLM skills in 2026 with little background in the field and stay up-to-date with the latest news and state-of-the-art techniques!","start-llms 是一份面向 2026 年的大语言模型（LLM）完全免费学习指南，由 AI 内容创作者 Louis Bouchard 维护。它专为编程和机器学习基础薄弱的学习者设计，提供从零基础到专家级别的系统化学习路径，帮助用户掌握最新的大模型技术与行业动态。\n\n这份指南解决了 LLM 学习资源分散、门槛高、更新快的问题。它将书籍、视频、在线课程、实践项目等优质资源按主题分类整理，涵盖提示工程（Prompting）、检索增强生成（RAG）、AI 伦理等核心方向，并允许用户根据自身情况灵活选择学习顺序——不喜欢读书可以跳过，不想上课也可以自学。\n\nstart-llms 特别适合以下人群：有一定 Python 基础想转行 LLM 的开发者、希望系统了解大模型技术的研究人员、以及想跟进 SOTA 进展的 AI 从业者。指南中所有核心资源均为免费，仅部分付费课程使用推广链接，完全不影响自主学习。\n\n其独特价值在于\"策展思维\"：维护者持续筛选高质量内容，整合社区讨论、新闻资讯和实战技巧，形成动态更新的知识图谱。对于零基础用户，指南还贴心地链接了前置的机器学习入门仓库，确保学习链条完整。","# Start with Large Language Models (LLMs) - Become an expert for free!\n\n## A complete guide to start and improve your LLM skills in 2026 without an advanced background in the field and stay up-to-date with the latest news and state-of-the-art techniques!\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002Fla4Sfu4.png\" width=\"512\"\u002F>](https:\u002F\u002Fwww.louisbouchard.ai\u002Ffrom-zero-to-hero-with-llms\u002F)\n\nFirst, if you have 0 programming or AI knowledge, please follow [this guide](https:\u002F\u002Fgithub.com\u002Flouisfb01\u002Fstart-machine-learning) I made for this exact purpose and come back here!\n\nThis guide is intended for anyone with a small background in programming and machine learning. There is no specific order to follow, but a classic path would be from top to bottom. If you don't like reading books, skip them. If you don't want to follow an online course, you can also skip it. There is not a single way to become a machine learning expert, and with motivation, you can absolutely achieve it.\n\nAll resources listed here are free, except some online courses and books, which are certainly recommended for a better understanding, but it is definitely possible to become an expert without them, with a little more time spent on online readings, videos, and practice. When it comes to paying courses, the links in this guide are affiliated links. Please use them if you feel like following a course, as it will support me. Thank you, and have fun learning! Remember, this is completely up to you and not necessary. I felt like it was useful to me and maybe useful to others as well.\n\nDon't be afraid to repeat videos or learn from multiple sources. Repetition is the key of success to learning!\n\nMaintainer: [louisfb01](https:\u002F\u002Fgithub.com\u002Flouisfb01), also active on [YouTube](https:\u002F\u002Fwww.youtube.com\u002F@whatsai) and as a [Podcaster](https:\u002F\u002Fopen.spotify.com\u002Fshow\u002F4rKRJXaXlClkDyInjHkxq3) if you want to see\u002Fhear more about AI & LLMs! You can also learn more twice a week in [my personal newsletter](https:\u002F\u002Flouisbouchard.substack.com\u002F)!\n\n[![Twitter](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Furl\u002Fhttps\u002Ftwitter.com\u002Fcloudposse.svg?style=social&label=Follow%20%40whats_ai)](https:\u002F\u002Ftwitter.com\u002FWhats_AI)\n\nFeel free to submit an issue for any great resources to add to this repository.\n\n\n***Tag me on Twitter [@Whats_AI](https:\u002F\u002Ftwitter.com\u002FWhats_AI) or LinkedIn [@Louis Bouchard](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fwhats-ai\u002F)  if you share the list!***\n\n### Want to know what this guide is about? Watch this video:\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002FrS1d89V.png\" width=\"512\"\u002F>](https:\u002F\u002Fyoutu.be\u002F58XmJxb1x_o)\n\n\n## Table of Contents\n- [Prerequesites](#prerequesites)\n- [Start with short YouTube video introductions as a first step](#youtubevideos)\n- [LLM Books and articles (for readers)](#readers)\n- [Follow online courses](#courses)\n- [Practice, practice, and practice!](#practice)\n- [Prompting](#prompting)\n- [Retrieval Augmented Generation (RAG)](#rag)\n- [More resources (Communities, cheat sheets, news, and more!)](#moreresources)\n- [How to find a machine learning job](#findajob)\n- [AI Ethics](#aiethics)\n- [Learn more and do more... with LLMs](#domore)\n\n## Prerequesites\u003Ca name=\"prerequesites\">\u003C\u002Fa>\n\nIf you have 0 programming or AI knowledge, please follow [this guide](https:\u002F\u002Fgithub.com\u002Flouisfb01\u002Fstart-machine-learning) I made for this exact purpose. Check out the python section mostly and then you will have a strong enough background to come back here!\n\nIf you are somewhat familiar with Python and AI, then I wish you happy learning!\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002FhLr2aQF.png\" width=\"512\"\u002F>](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbeginner-to-advanced-llm-dev?ref=1f9b29)\n\n## Start with short YouTube video introductions as a first step\u003Ca name=\"youtubevideos\">\u003C\u002Fa>\n### Start with short YouTube videos introductions\nThis is the best way to start from nothing. Here, I list a few of the best videos I found that will give you a great first introduction to the terms you need to know to get started in the LLM field.\n    \n* Understanding the terminology\n    * [Mastering AI Jargon - Your Guide to OpenAI & LLM Terms - Louis Bouchard](https:\u002F\u002Fyoutu.be\u002Fq4G6X09NEu4) - A quick introduction to the most used terms in the LLM (or GPT) world.\n* Understanding Transformers and LLMs (i.e. models behind ChatGPT)!\n    * [Foundational Knowledge for LLMs and building on top of LLMs](https:\u002F\u002Fyoutu.be\u002FR5_udqy1L4s) - 2 free sessions of 2 hours each, covering all you need to know about LLMs.\n    * [Intro to Large Language Models](https:\u002F\u002Fyoutu.be\u002FzjkBMFhNj_g) - An amazing 1-hour talk from [Andrej Karpathy](https:\u002F\u002Ftwitter.com\u002Fkarpathy).\n    * [Natural Language Processing and Large Language Models](https:\u002F\u002Fwww.youtube.com\u002Fplaylist?list=PLs8w1Cdi-zvYskDS2icIItfZgxclApVLv) - amazing video introductions to the attention mechanism, tokens, embeddings and more to better understand everything behind large language models like GPT by [Luis Serrano](https:\u002F\u002Ftwitter.com\u002FSerranoAcademy).\n    * [What are Transformer Models and how do they work?](https:\u002F\u002Fyoutu.be\u002FqaWMOYf4ri8) - [Luis Serrano](https:\u002F\u002Ftwitter.com\u002FSerranoAcademy)\n    * [The Illustrated Word2vec - A Gentle Intro to Word Embeddings in Machine Learning](https:\u002F\u002Fyoutu.be\u002FISPId9Lhc1g) - A clear explanation of word embeddings in machine learning by [Jay Alammar](https:\u002F\u002Ftwitter.com\u002FJayAlammar).\n    * [A Hackers' Guide to Language Models](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=jkrNMKz9pWU) - by [Jeremy Howard (fast.ai)](https:\u002F\u002Ftwitter.com\u002Fjeremyphoward).\n    * [Let's build GPT: from scratch, in code, spelled out.](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=kCc8FmEb1nY) - by [Andrej Karpathy](https:\u002F\u002Ftwitter.com\u002Fkarpathy).\n\nAnother easy **way to get started and keep learning is by listening to podcasts** in your spare time. Driving to work, on the bus, or having trouble falling asleep? Listen to some AI podcasts to get used to the terms and patterns, and learn about the field through inspiring stories! I invite you to follow a few of the best I personally prefer, like [Lex Fridman](https:\u002F\u002Fopen.spotify.com\u002Fshow\u002F2MAi0BvDc6GTFvKFPXnkCL), [Machine Learning Street Talk](https:\u002F\u002Fopen.spotify.com\u002Fshow\u002F02e6PZeIOdpmBGT9THuzwR), and obviously, my podcast: [Louis Bouchard Podcast](https:\u002F\u002Fopen.spotify.com\u002Fshow\u002F4rKRJXaXlClkDyInjHkxq3), where you will learn about incredibly talented people in the field with inspiring stories sharing the knowledge they worked so hard to gather. A new one I really enjoy listening to that keeps me up to date is the [ThursdAI podcast](https:\u002F\u002Fthursdai.news\u002F) by my friend [Alex Volkov](https:\u002F\u002Ftwitter.com\u002Faltryne).\n\n## Here is a list of awesome courses available on YouTube that you should definitely follow and are 100% free.\n\n* Louis Bouchard's LLM free course videos \"[Train & Fine-Tune LLMs for Production Course by Activeloop, Towards AI & Intel Disruptor](https:\u002F\u002Fyoutube.com\u002Fplaylist?list=PLO4GrDnQanVcPlQUBuMd_pwRkILfc463G&si=QbXeHeDs5RSKH3nY)\". \"A playlist for our LLM course: Gen AI 360: Foundational Model Certification!\"\n* [Create a Large Language Model from Scratch with Python – Tutorial](https:\u002F\u002Fyoutu.be\u002FUU1WVnMk4E8) - by [freeCodeCamp](https:\u002F\u002Ftwitter.com\u002FfreeCodeCamp). \"Learn how to build your own large language model, from scratch. This course goes into the data handling, math, and transformers behind large language models. You will use Python.\"\n* [LLM University (LLMU) from Cohere](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=uV1H6E8y_Sg&list=PLLalUvky4CLIpL4PkbTyf9DeXxJaZzEgU) - by [Cohere](https:\u002F\u002Ftwitter.com\u002Fcohere). LLM University (LLMU) is a set of comprehensive learning resources for anyone interested in natural language processing (NLP), from beginners to advanced learners.\n* [The Attention Mechanism in Large Language Models](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=OxCpWwDCDFQ&list=PLs8w1Cdi-zva4fwKkl9EK13siFvL9Wewf) - by Luis Serrano. In this video series, Luis explains the Transformer architecture going increasingly in depth. It is a very good overview and explanation of Transformers and the attention mechanism that I believe should be watched by all AI professionals.\n\n\n## LLM Books and articles (for readers)\u003Ca name=\"readers\">\u003C\u002Fa>\nIf you prefer the article and reading path, here are some suggestions:\n\n* [Building LLMs for Production: Enhancing LLM Abilities and Reliability with Prompting, Fine-Tuning, and RAG](https:\u002F\u002Famzn.to\u002F4bqYU9b) - by Towards AI. \"Discover the key tech stacks for adapting Large Language Models to real-world applications, including Prompt Engineering, Fine-tuning, and Retrieval Augment Generation.\" (Or get the e-book [here](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbuildingllmsforproduction?ref=1f9b29). You can DM me for a nice discount!)\n* [The LLM Engineer's Handbook](https:\u002F\u002Fwww.packtpub.com\u002Fen-us\u002Fproduct\u002Fllm-engineers-handbook-9781836200079?utm_medium=affiliate&utm_campaign=51b66f4e-29b2-68d8-f16e-67e2f9dfe6d0&utm_term=c35fe2b3-a89c-3ed8-1def-67c6a827b8eb&utm_content=B31105)—Build and refine LLMs step by step, covering data preparation, RAG, and fine-tuning.\n* [The Illustrated Transformer](https:\u002F\u002Fjalammar.github.io\u002Fillustrated-transformer\u002F) - by Jay Alammar. This is a famous article providing an amazing explanation to how current language models work.\n* [A Practical Introduction to LLMs](https:\u002F\u002Ftowardsdatascience.com\u002Fa-practical-introduction-to-llms-65194dda1148) - by [Shawhin Talebi](https:\u002F\u002Fshawhin.medium.com\u002F). \n* [Medium](https:\u002F\u002Fwhats-ai.medium.com\u002Fmembership) is pretty much the best place to find great explanations, either on [Towards AI](https:\u002F\u002Fpub.towardsai.net\u002F) or [Towards Data Science](https:\u002F\u002Ftowardsdatascience.com\u002F) publications. I also share my own articles there and I love using the platform. You can subscribe to Medium using my affiliated link [here](https:\u002F\u002Fwhats-ai.medium.com\u002Fmembership) if this sounds interesting to you and if you'd like to support me at the same time!\n* [Reading lists for new MILA students](https:\u002F\u002Fdocs.google.com\u002Fdocument\u002Fd\u002F1IXF3h0RU5zz4ukmTrVKVotPQypChscNGf5k6E25HGvA\u002Fedit#) - Anonymous\n* [A complete roadmap to master NLP in 2022](https:\u002F\u002Fwww.analyticsvidhya.com\u002Fblog\u002F2022\u002F01\u002Froadmap-to-master-nlp-in-2022\u002F)\n* NLTK Book is the free resource to learn about fundamental theories behind NLP: https:\u002F\u002Fwww.nltk.org\u002Fbook\u002F\n* [The Annotated Transformer](https:\u002F\u002Fnlp.seas.harvard.edu\u002F2018\u002F04\u002F03\u002Fattention.html) - Harvard\n\n\n## Follow online courses\u003Ca name=\"courses\">\u003C\u002Fa>\nIf you like some more guidance, I can advise checking out (optional) online courses, such as...\n* [Generative AI with Large Language Models](https:\u002F\u002Fimp.i384100.net\u002FR5WzQR) - Paid\n* [Become an NLP pro with Coursera's Natural Language Processing Specialization by deeplearning.ai](https:\u002F\u002Fcoursera.pxf.io\u002FP0vO9e) - Paid\n* [Gradio Course - Create User Interfaces for Machine Learning Models - freeCodeCamp](https:\u002F\u002Fyoutu.be\u002FRiCQzBluTxU) - Free\n* [Train & Fine-Tune LLMs for Production Course by Activeloop, Towards AI & Intel Disruptor](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Fllms\u002F?utm_source=social&utm_medium=youtube&utm_campaign=llmcourse) - Free\n* [The LLM University by Cohere](https:\u002F\u002Fdocs.cohere.com\u002Fdocs\u002Fllmu?ref=txt.cohere.com) - Free\n* [From Beginners to Advanced LLM Developer](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbeginner-to-advanced-llm-dev?ref=1f9b29) - by Towards AI. \"Build Your First Scalable Product with LLMs, Prompting, RAG, Fine-Tuning, and Agents! Master the skills top companies need and build your own advanced LLM MVP with real-world applications.\"\n* [Become an NLP pro with Coursera's Natural Language Processing Specialization by deeplearning.ai](https:\u002F\u002Fcoursera.pxf.io\u002FP0vO9e) - Paid\n _\"Break into the NLP space. Master cutting-edge NLP techniques through four hands-on courses!\"_\n* [An NLP Nano Degree!](https:\u002F\u002Fimp.i115008.net\u002FjW4K60) — Paid\n_\"Learn cutting-edge natural language processing techniques to process speech and analyze text. Build probabilistic and deep learning models, such as hidden Markov models and recurrent neural networks, to teach the computer to do tasks such as speech recognition, machine translation, and more!\"_\n* [Introduction to Large Language Models with Google Cloud](https:\u002F\u002Fimp.i115008.net\u002FeKbDLD) - Paid\n* [Learn to train, fine-tune and use LLMs in your applications.](https:\u002F\u002Fwww.wandb.courses\u002Fpages\u002Fw-b-courses) - Free by Weights & Biases\n* [Large Language Models with Semantic Search](https:\u002F\u002Fwww.deeplearning.ai\u002Fshort-courses\u002Flarge-language-models-semantic-search\u002F) - Free, Deeplearning.ai and Cohere\n\nYou can easily google for more, but after reading and watching those, I believe you already have a good enough understanding of LLMs to get into the real deal: practice.\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002FhLr2aQF.png\" width=\"512\"\u002F>](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbeginner-to-advanced-llm-dev?ref=1f9b29)\n\n## Practice, practice, and practice!\u003Ca name=\"practice\">\u003C\u002Fa>\n### Practice is key\n\nThe most important thing in programming is practice. This applies to machine learning too. It can be hard to find a personal project to practice. I strongly advise you to try to build something by yourself, but I understand it may be intimidating. What I would then suggest is to follow one or two **extremely** applied courses and use the resource to build your own project based on the code examples they provide you, and ChatGPT or GitHub Copilot to [work for you](#domore) as a code assistant for the rest of the work.\n\nHere are a few of the most applied courses I could find for LLMs:\n\n* Looking to build a quick text classification model or word vectorizer, [fasttext](https:\u002F\u002Ffasttext.cc\u002Fdocs\u002Fen\u002Fsupervised-tutorial.html) is a good library to quickly train up a model.\n* [Huggingface](https:\u002F\u002Fhuggingface.co\u002Fcourse\u002Fchapter1\u002F1) is THE place to get modern day NLP models, and they also include a whole [course](https:\u002F\u002Fhuggingface.co\u002Fcourse\u002Fchapter1\u002F1) about it.\n* [LangChain & Vector Databases in Production](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Flangchain\u002F) - An amazing free resource we built at Towards AI in partnership with Activeloop and the Intel Disruptor Initiative to learn about LangChain & Vector Databases in Production. \"Whether you are an experienced developer who's a newcomer to the AI realm or an experienced machine learning enthusiast, this course is designed for you. Our goal is to make AI accessible and practical, transforming how you approach your daily tasks and the overall impact of your work.\"\n* [Training & Fine-Tuning LLMs for Production](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Fllms\u002F) - An amazing free resource we built at Towards AI in partnership with Activeloop and the Intel Disruptor Initiative to learn about Training & Fine-Tuning LLMs for Production. \"If you want to learn how to train and fine-tune LLMs from scratch and have intermediate Python knowledge as well as access to moderate compute resources (for some cases, just a Google Colab will suffice!), you should be all set to take and complete the course. This course is designed with a wide audience in mind, including beginners in AI, current machine learning engineers, students, and professionals considering a career transition to AI. We aim to provide you with the necessary tools to apply and tailor Large Language Models across a wide range of industries to make AI more accessible and practical.\"\n* [The Real-World ML Tutorial & Community](https:\u002F\u002Frealworldmachinelearning.carrd.co\u002F) - Paid\n\nA reminder. The best way to learn is to build something! I really prone to learn by doing. Those courses are all great but optional. You can do it on your own, and most companies providing resources for working with LLMs (OpenAI, LangChain, Activeloop, Cohere, W&B...) have great tutorials to get you started and build something. Then, you can [ask ChatGPT](#domore) to help you finish it!\n\n\n## Prompting\u003Ca name=\"prompting\">\u003C\u002Fa>\nPrompting is an important new skill to learn for both using the models and building NLP-related apps.\n\n* [What is Prompting? Talking with AI Models...](https:\u002F\u002Fyoutu.be\u002FpZsJbYIFCCw) - Free\n* [ChatGPT Prompt Engineering for Developers](https:\u002F\u002Fimp.i384100.net\u002FrQBVMy) - Paid\n* [Learn Prompting](https:\u002F\u002Flearnprompting.org\u002F) - This is a great **free** course intending to teach prompting and give tips for specific models. It is **all you need** for prompting!\n* [Techniques to improve reliability](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-cookbook\u002Fblob\u002Fmain\u002Farticles\u002Ftechniques_to_improve_reliability.md#techniques-to-improve-reliability) - OpenAI Cookbook on prompting techniques.\n\n## More on Retrieval Augmented Generation (RAG) and fine-tuning\u003Ca name=\"rag\">\u003C\u002Fa>\nMost people build RAG-based apps currently. Here are a few resources that I loved to get you started and have a good understanding of it...\n\n* [A Survey of Techniques for Maximizing LLM Performance](https:\u002F\u002Fyoutu.be\u002FahnGLM-RC1Y) - Amazing video by OpenAI covering when to use prompt engineering, RAG or fine-tuning. This is a must-see for everyone in the field!\n* [RAG vs Fine-Tuning vs Deep Memory vs training LLM from Scratch: when to do what with LLMs](https:\u002F\u002Fyoutu.be\u002FpHv9SsE4Mb4) - Simlarly, this is a short video covering when you should use RAG, fine-tuning or prompt engineering in your applications.\n* [Building a Q&A Chatbot using GPT and embeddings](https:\u002F\u002Fyoutu.be\u002FLB5g-AhfPG8) - Applied YouTube Tutorial by [Jeremy Pinto](https:\u002F\u002Ftwitter.com\u002Fjerpint).\n* [How to build an AI that can answer questions about your website](https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Ftutorials\u002Fweb-qa-embeddings\u002Fhow-to-build-an-ai-that-can-answer-questions-about-your-website) - Free OpenAI tutorial.\n* [From Beginners to Advanced LLM Developer](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbeginner-to-advanced-llm-dev?ref=1f9b29) - by Towards AI. \"Build Your First Scalable Product with LLMs, Prompting, RAG, Fine-Tuning, and Agents! Master the skills top companies need and build your own advanced LLM MVP with real-world applications.\"\n* [How to Build a RAG-based ChatGPT Web App: Meet Our new AI Tutor](https:\u002F\u002Fyoutu.be\u002F7ytyK6u3aAk) - YouTube introduction on how I built a RAG-based chatbot (and how you can, too).\n* [Training & Fine-Tuning LLMs for Production](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Fllms\u002F) - Learn how to train and fine-tune LLMs from scratch.\n* [Train and Deploy a Real-Time Financial Advisor](https:\u002F\u002Fgithub.com\u002Fiusztinpaul\u002Fhands-on-llms) - Hands-on LLMs Course by [Paul Iusztin](https:\u002F\u002Fgithub.com\u002Fiusztinpaul), [Pau Labarta Bajo](https:\u002F\u002Fgithub.com\u002FPaulescu) and [Alexandru Razvant](https:\u002F\u002Fgithub.com\u002FJoywalker).\n* [Retrieval Augmented Generation for Production with LangChain & LlamaIndex](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Frag) - Whether planning to build a chat with data application for your organization or just learning how to leverage Generative AI in various industries, this course is for you. The course addresses critical issues such as increasing retrieval accuracy, reducing hallucinations in AI outputs, enhancing explainability, addressing copyright concerns, and offering more tailored, up-to-date data inputs. We go beyond basic RAG applications, equipping you with the skills to create more complex, reliable products with tools like LangChain, LlamaIndex, and Deep Memory. Emphasizing hands-on learning, this course is a gateway to mastering advanced RAG techniques and applications in real-world scenarios.\n* [Building LLMs for Production: Enhancing LLM Abilities and Reliability with Prompting, Fine-Tuning, and RAG](https:\u002F\u002Famzn.to\u002F4bqYU9b) - by Towards AI. \"Discover the key tech stacks for adapting Large Language Models to real-world applications, including Prompt Engineering, Fine-tuning, and Retrieval Augment Generation.\" (Or get the e-book [here](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbuildingllmsforproduction?ref=1f9b29). You can DM me for a nice discount!)\n\n\n[![Twitter](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Furl\u002Fhttps\u002Ftwitter.com\u002Fcloudposse.svg?style=social&label=Follow%20%40whats_ai)](https:\u002F\u002Ftwitter.com\u002FWhats_AI)\n\n\n## More Resources\u003Ca name=\"moreresources\">\u003C\u002Fa>\n### Join communities!\n\n* [A Discord server with many AI enthusiasts](https:\u002F\u002Fdiscord.gg\u002Flearnaitogether) - Learn together, ask questions, find kaggle teammates, share your projects, and more.\n* [A Discord server where you can stay up-to-date with the latest AI news](https:\u002F\u002Fws.towardsai.net\u002Fdiscord) - Stay up-to-date with the latest AI news, ask questions, share your projects, and much more.\n* [Learn Prompting Discord community](https:\u002F\u002Fdiscord.gg\u002Flearn-prompting-1046228027434086460) - Chat with fellow prompt engineers.\n\n* Follow reddit communities - Ask questions, share your projects, follow news, and more.\n    * [artificial](https:\u002F\u002Fwww.reddit.com\u002Fr\u002Fartificial\u002F) - Artificial Intelligence\n    * [MachineLearning](https:\u002F\u002Fwww.reddit.com\u002Fr\u002FMachineLearning\u002F) - Machine Learning (Biggest subreddit of the field)\n    * [DeepLearningPapers](https:\u002F\u002Fwww.reddit.com\u002Fr\u002FDeepLearningPapers\u002F) - Deep Learning Papers\n    * [ComputerVision](https:\u002F\u002Fwww.reddit.com\u002Fr\u002Fcomputervision\u002F) - Extracting useful information from images and videos\n    * [learnmachinelearning](https:\u002F\u002Fwww.reddit.com\u002Fr\u002Flearnmachinelearning\u002F) - Learn Machine Learning\n    * [ArtificialInteligence](https:\u002F\u002Fwww.reddit.com\u002Fr\u002FArtificialInteligence\u002F) - AI\n    * [LatsestInML](https:\u002F\u002Fwww.reddit.com\u002Fr\u002FLatestInML\u002F) - Game-changing developments in machine learning you shouldn't miss\n\n\n### Follow the news in the field!\n\n* Subscribe to YouTube channels that share new papers - Stay up to date with the news in the field!\n    * [Louis Bouchard](https:\u002F\u002Fwww.youtube.com\u002Fchannel\u002FUCUzGQrN-lyyc0BWTYoJM_Sg) - Weekly videos covering new papers\n    * [Two Minutes Papers](https:\u002F\u002Fwww.youtube.com\u002Fuser\u002Fkeeroyz) - Bi-weekly videos covering new papers\n    * [Bycloud](https:\u002F\u002Fwww.youtube.com\u002Fchannel\u002FUCgfe2ooZD3VJPB6aJAnuQng) - Weekly videos covering new papers\n    \n* LinkedIn Groups\n    * [Artificial Intelligence, Machine Learning and Deep Learning News](https:\u002F\u002Fwww.linkedin.com\u002Fgroups\u002F8942343\u002F) - News of the field shared by everyone in the group\n    * [Artificial Intelligence | Deep Learning  | Machine Learning](https:\u002F\u002Fwww.linkedin.com\u002Fgroups\u002F45655\u002F)\n    * [Applied Artificial Intelligence](https:\u002F\u002Fwww.linkedin.com\u002Fgroups\u002F127447\u002F)\n    \n* Facebook Groups\n    * [Artificial Intelligence & Deep Learning](https:\u002F\u002Fwww.facebook.com\u002Fgroups\u002FDeepNetGroup) - The definitive and most active FB Group on A.I., Neural Networks and Deep Learning. All things new and interesting on the frontier of A.I. and Deep Learning. Neural networks will redefine what it means to be a smart machine in the years to come.\n    * [Deep learning](https:\u002F\u002Fwww.facebook.com\u002Fgroups\u002FDeepLearnng\u002F) - Nowadays society tends to be soft and automated evolving into the 4th industrial revolution, which consequently drives the constituents into the swirl of societal upheaval. To survive or take a lead one is supposed to be equipped with associated tools. Machine is becoming smarter and more intelligent. Machine learning is inescapable skill and it requires people to be familiar with. This group is for these people who are interest in the development of their talents to fit in.\n\n* Newsletters\n   * [Synced AI TECHNOLOGY & INDUSTRY REVIEW](https:\u002F\u002Fsyncedreview.com\u002F) - China's leading media & information provider for AI & Machine Learning.\n   * [Inside AI](https:\u002F\u002Finside.com\u002Fai) - A daily roundup of stories and commentary on Artificial Intelligence, Robotics, and Neurotechnology.\n   * [AI Weekly](http:\u002F\u002Faiweekly.co\u002F) - A weekly collection of AI News and resources on Artificial Intelligence and Machine Learning.\n   * [AI Ethics Weekly](https:\u002F\u002Flighthouse3.com\u002Fnewsletter\u002F) - The latest updates in AI Ethics delivered to your inbox every week.\n   * [Louis Bouchard Weekly](https:\u002F\u002Flouisbouchard.substack.com\u002F) - One and only one paper clearly explained weekly with an article, video demo, demo, code, etc.\n   * [ThursdAI](https:\u002F\u002Fsub.thursdai.news\u002F) - Recaps of the most high-signal AI weekly spaces!\n   * [Toward's AI newsletter](https:\u002F\u002Ftowardsai.net\u002Fai\u002Fnewsletter) - Summarizing the most interesting news and learning resources weekly as well as community updates from the Learn AI Together Discord community. Perfect for ML professionals and enthusiasts.\n   * [The Batch](https:\u002F\u002Fwww.deeplearning.ai\u002Fthe-batch\u002F) - Andrew Ng \u002F Deeplearning.ai\n    \n* Follow Medium publications\n    * [Towards Data Science](https:\u002F\u002Ftowardsdatascience.com\u002F) - \"Sharing concepts, ideas, and codes\"\n    * [Towards AI](https:\u002F\u002Fmedium.com\u002Ftowards-artificial-intelligence) - \"The Best of Tech, Science, and Engineering.\"\n    * [OneZero](https:\u002F\u002Fonezero.medium.com\u002F) - \"The undercurrents of the future. A Medium publication about tech and science.\"\n    \n## Find a machine learning job\u003Ca name=\"findajob\">\u003C\u002Fa>\n\n* Read [this section from the article](https:\u002F\u002Fwww.louisbouchard.ai\u002Flearnai\u002F#how-to-find-a-job) full of interview tips and **how to prepare for them**.\n* Learn how the interview process goes and getting better at preparing for them by watching how others did it, like the [interview series](https:\u002F\u002Fyoutube.com\u002Fplaylist?list=PLO4GrDnQanVfrRIuIT_1rlLLTgQJdfXmS) I ran with experts from NVIDIA, Zoox (Self-driving company), D-ID (Generative AI Startup), etc.\n\n## AI Ethics\u003Ca name=\"aiethics\">\u003C\u002Fa>\n* [What are Ethics and Why do they Matter? Machine Learning Edition](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=F0cxzESR7ec&list=PLtmWHNX-gukIU6V33Bc8eP8OD41I4GywR&ab_channel=RachelThomas) - by Rachel Thomas, founder of fast.ai\n* [AI4People—An Ethical Framework for a Good AI Society: Opportunities, Risks, Principles, and Recommendations](https:\u002F\u002Flink.springer.com\u002Fcontent\u002Fpdf\u002F10.1007\u002Fs11023-018-9482-5.pdf) - Floridi et al., 2018, AI4People AI for a good society\n* [Ethics guidelines for trustworthy AI](https:\u002F\u002Fwayback.archive-it.org\u002F12090\u002F20210728013426\u002Fhttps:\u002F\u002Fdigital-strategy.ec.europa.eu\u002Fen\u002Flibrary\u002Fethics-guidelines-trustworthy-ai) - European Commission high-level expert group 7 points for a trustworthy AI.\n* [An Introduction to Ethics in Robotics and AI](https:\u002F\u002Flink.springer.com\u002Fbook\u002F10.1007\u002F978-3-030-51110-4) - a free e-book by Christoph Bartneck, Christoph Lütge, Alan Wagner, and Sean Welsh.\n\n\n## Learn more and do more... with LLMs\u003Ca name=\"domore\">\u003C\u002Fa>\nChatGPT, Bing, Claude... are incredible. Of course, they have limitations. Yet, you can leverage those to learn anything you want. I use it for coding or asking lots of questions in general. You need to double-check when you ask for important questions.\nStill, it is a powerful **tool**. Yes, it is a tool, not a human replacement. Use it as a _dumb_ assistant that knows about pretty much everything.\n\nHere's [a clear example](https:\u002F\u002Fchat.openai.com\u002Fshare\u002F883389c9-f0f8-4a3e-a3af-ee9860d448a8) of how I used it for a project to better understand a function from a project I was not familiar with. This is for python, but those models are extremely powerful for coding in general, understanding new platforms (like AWS, GCP, working with a virtual machine, a server, SSH connections, etc.... anything you are not familiar with that is useful in the LLM space).\n\n_p.s. I didn't mention Bing and Claude for fun. Don't be overly dependent on a single company like OpenAI. There are (and will always be) other companies in the fight for the best LLM. I wanted to create an example for the guide this morning when..._\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002FRDdL7el.png\" width=\"512\"\u002F>](https:\u002F\u002Ftwitter.com\u002Fsatourian\u002Fstatus\u002F1722257478115811421)\n\n---\n\n\n***Tag me on Twitter [@Whats_AI](https:\u002F\u002Ftwitter.com\u002FWhats_AI) or LinkedIn [@Louis Bouchard](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fwhats-ai\u002F)  if you share the list!***\n\n👀 **If you'd like to support my work**, you can check to [Sponsor](https:\u002F\u002Fgithub.com\u002Fsponsors\u002Flouisfb01) this repository or support me on [Patreon](https:\u002F\u002Fwww.patreon.com\u002Fwhatsai).\n\nThis guide is still regularly updated.\n","# 从零开始掌握大语言模型（LLMs）—— 免费成为专家！\n\n## 一份完整的 2026 年 LLM 入门与进阶指南，无需深厚的领域背景，助你紧跟最新动态与前沿技术！\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002Fla4Sfu4.png\" width=\"512\"\u002F>](https:\u002F\u002Fwww.louisbouchard.ai\u002Ffrom-zero-to-hero-with-llms\u002F)\n\n首先，如果你没有任何编程或 AI 基础，请先按照[这份指南](https:\u002F\u002Fgithub.com\u002Flouisfb01\u002Fstart-machine-learning)学习，这是专门为零基础准备的，完成后再回到这里！\n\n本指南面向具备一定编程和机器学习基础的读者。学习顺序没有严格要求，但经典路径是从上到下。如果你不喜欢读书，可以跳过书籍部分；如果不想跟在线课程，也可以跳过。成为机器学习专家没有唯一路径，只要有动力，你绝对可以实现。\n\n这里列出的所有资源都是免费的，除了部分在线课程和书籍——这些确实有助于深入理解，但即使没有它们，通过多花时间在在线阅读、视频和实践上，也完全能够成为专家。对于付费课程，本指南中的链接是联盟链接。如果你打算学习课程，请使用这些链接，这将支持我继续创作。谢谢，祝你学习愉快！请记住，这完全取决于你，并非必需。我觉得这对我有帮助，或许对其他人也有用。\n\n不要害怕重复观看视频或从多个来源学习。重复是学习成功的关键！\n\n维护者：[louisfb01](https:\u002F\u002Fgithub.com\u002Flouisfb01)，同时在 [YouTube](https:\u002F\u002Fwww.youtube.com\u002F@whatsai) 和 [播客](https:\u002F\u002Fopen.spotify.com\u002Fshow\u002F4rKRJXaXlClkDyInjHkxq3) 上活跃，如果你想了解更多 AI 和 LLM 内容！你也可以在[我的个人通讯](https:\u002F\u002Flouisbouchard.substack.com\u002F)中每周两次获取更多信息！\n\n[![Twitter](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Furl\u002Fhttps\u002Ftwitter.com\u002Fcloudposse.svg?style=social&label=Follow%20%40whats_ai)](https:\u002F\u002Ftwitter.com\u002FWhats_AI)\n\n欢迎提交 issue 来推荐优秀的资源加入本仓库。\n\n***如果你在社交媒体上分享这份清单，请在 Twitter [@Whats_AI](https:\u002F\u002Ftwitter.com\u002FWhats_AI) 或 LinkedIn [@Louis Bouchard](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fwhats-ai\u002F) 上标记我！***\n\n### 想知道这份指南讲什么？观看这个视频：\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002FrS1d89V.png\" width=\"512\"\u002F>](https:\u002F\u002Fyoutu.be\u002F58XmJxb1x_o)\n\n\n## 目录\n- [前置要求](#prerequesites)\n- [第一步：从简短的 YouTube 视频介绍开始](#youtubevideos)\n- [LLM 书籍与文章（适合阅读爱好者）](#readers)\n- [跟随在线课程学习](#courses)\n- [实践，实践，再实践！](#practice)\n- [提示工程（Prompting）](#prompting)\n- [检索增强生成（Retrieval Augmented Generation, RAG）](#rag)\n- [更多资源（社区、速查表、新闻等）](#moreresources)\n- [如何找到机器学习工作](#findajob)\n- [AI 伦理](#aiethics)\n- [利用 LLM 学习更多、做得更多](#domore)\n\n## 前置要求\u003Ca name=\"prerequesites\">\u003C\u002Fa>\n\n如果你没有任何编程或 AI 基础，请先按照[这份指南](https:\u002F\u002Fgithub.com\u002Flouisfb01\u002Fstart-machine-learning)学习，这是专门为零基础准备的。重点查看 Python 部分，之后你就具备了足够的基础回到这里继续学习！\n\n如果你对 Python 和 AI 已有一定了解，那么祝你学习愉快！\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002FhLr2aQF.png\" width=\"512\"\u002F>](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbeginner-to-advanced-llm-dev?ref=1f9b29)\n\n## 第一步：从简短的 YouTube 视频介绍开始\u003Ca name=\"youtubevideos\">\u003C\u002Fa>\n### 从简短的 YouTube 视频介绍入手\n这是从零开始的最佳方式。这里我列出了一些最优质的视频，能帮助你快速了解 LLM 领域入门所需的关键术语。\n\n* 理解术语\n    * [掌握 AI 术语 —— OpenAI 与 LLM 术语指南 - Louis Bouchard](https:\u002F\u002Fyoutu.be\u002Fq4G6X09NEu4) —— 快速介绍 LLM（或 GPT）世界中最常用的术语。\n* 理解 Transformer 和 LLM（即 ChatGPT 背后的模型）！\n    * [LLM 基础知识与基于 LLM 的构建](https:\u002F\u002Fyoutu.be\u002FR5_udqy1L4s) —— 两场免费的 2 小时课程，涵盖 LLM 所需的全部知识。\n    * [大语言模型导论](https:\u002F\u002Fyoutu.be\u002FzjkBMFhNj_g) —— [Andrej Karpathy](https:\u002F\u002Ftwitter.com\u002Fkarpathy) 的精彩 1 小时演讲。\n    * [自然语言处理与大语言模型](https:\u002F\u002Fwww.youtube.com\u002Fplaylist?list=PLs8w1Cdi-zvYskDS2icIItfZgxclApVLv) —— [Luis Serrano](https:\u002F\u002Ftwitter.com\u002FSerranoAcademy) 关于注意力机制（attention mechanism）、token、嵌入（embeddings）等的精彩视频介绍，帮助你深入理解 GPT 等大语言模型背后的原理。\n    * [什么是 Transformer 模型，它们如何工作？](https:\u002F\u002Fyoutu.be\u002FqaWMOYf4ri8) —— [Luis Serrano](https:\u002F\u002Ftwitter.com\u002FSerranoAcademy)\n    * [图解 Word2vec —— 机器学习中词嵌入（Word Embeddings）的温和介绍](https:\u002F\u002Fyoutu.be\u002FISPId9Lhc1g) —— [Jay Alammar](https:\u002F\u002Ftwitter.com\u002FJayAlammar) 对机器学习中词嵌入的清晰解释。\n    * [黑客指南：语言模型](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=jkrNMKz9pWU) —— [Jeremy Howard (fast.ai)](https:\u002F\u002Ftwitter.com\u002Fjeremyphoward)\n    * [让我们构建 GPT：从零开始，用代码，详细讲解](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=kCc8FmEb1nY) —— [Andrej Karpathy](https:\u002F\u002Ftwitter.com\u002Fkarpathy)\n\n另一个轻松**入门并持续学习的方式是听播客**——利用你的空闲时间。开车上班、坐公交，或者失眠时？听一些 AI 播客来熟悉术语和模式，通过鼓舞人心的故事了解这个领域！我推荐几个我个人最喜欢的：[Lex Fridman](https:\u002F\u002Fopen.spotify.com\u002Fshow\u002F2MAi0BvDc6GTFvKFPXnkCL)、[Machine Learning Street Talk](https:\u002F\u002Fopen.spotify.com\u002Fshow\u002F02e6PZeIOdpmBGT9THuzwR)，当然还有我的播客：[Louis Bouchard Podcast](https:\u002F\u002Fopen.spotify.com\u002Fshow\u002F4rKRJXaXlClkDyInjHkxq3)，在这里你会听到领域内杰出人才分享他们辛苦积累的知识和鼓舞人心的故事。还有一个我最近非常喜欢、能让我保持更新的播客是 [ThursdAI podcast](https:\u002F\u002Fthursdai.news\u002F)，由我的朋友 [Alex Volkov](https:\u002F\u002Ftwitter.com\u002Faltryne) 主持。\n\n## 以下是 YouTube 上绝对值得关注的优质免费课程列表，100% 免费。\n\n* Louis Bouchard 的大语言模型（Large Language Model，LLM）免费课程视频 \"[Train & Fine-Tune LLMs for Production Course by Activeloop, Towards AI & Intel Disruptor](https:\u002F\u002Fyoutube.com\u002Fplaylist?list=PLO4GrDnQanVcPlQUBuMd_pwRkILfc463G&si=QbXeHeDs5RSKH3nY)\"。 \"我们 LLM 课程的播放列表：Gen AI 360：基础模型认证！\"\n* [Create a Large Language Model from Scratch with Python – Tutorial](https:\u002F\u002Fyoutu.be\u002FUU1WVnMk4E8) - 由 [freeCodeCamp](https:\u002F\u002Ftwitter.com\u002FfreeCodeCamp) 提供。\"学习如何从零开始构建自己的大语言模型。本课程深入讲解大语言模型背后的数据处理、数学原理和 Transformer（变换器）架构。你将使用 Python。\"\n* [LLM University (LLMU) from Cohere](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=uV1H6E8y_Sg&list=PLLalUvky4CLIpL4PkbTyf9DeXxJaZzEgU) - 由 [Cohere](https:\u002F\u002Ftwitter.com\u002Fcohere) 提供。LLM 大学（LLM University，LLMU）是一套全面的学习资源，适合任何对自然语言处理（Natural Language Processing，NLP）感兴趣的人，从初学者到高级学习者。\n* [The Attention Mechanism in Large Language Models](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=OxCpWwDCDFQ&list=PLs8w1Cdi-zva4fwKkl9EK13siFvL9Wewf) - 由 Luis Serrano 提供。在这个视频系列中，Luis 深入讲解了 Transformer 架构。这是对 Transformer 和注意力机制（Attention Mechanism）非常出色的概述和解释，我认为所有 AI 从业者都应该观看。\n\n\n## LLM 书籍和文章（适合阅读者）\u003Ca name=\"readers\">\u003C\u002Fa>\n如果你更喜欢文章和阅读路径，以下是一些建议：\n\n* [Building LLMs for Production: Enhancing LLM Abilities and Reliability with Prompting, Fine-Tuning, and RAG](https:\u002F\u002Famzn.to\u002F4bqYU9b) - 由 Towards AI 提供。\"探索将大语言模型适配到实际应用的关键技术栈，包括提示工程（Prompt Engineering）、微调（Fine-tuning）和检索增强生成（Retrieval Augmented Generation，RAG）。\"（或点击[此处](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbuildingllmsforproduction?ref=1f9b29)获取电子书。可以私信我获取优惠！）\n* [The LLM Engineer's Handbook](https:\u002F\u002Fwww.packtpub.com\u002Fen-us\u002Fproduct\u002Fllm-engineers-handbook-9781836200079?utm_medium=affiliate&utm_campaign=51b66f4e-29b2-68d8-f16e-67e2f9dfe6d0&utm_term=c35fe2b3-a89c-3ed8-1def-67c6a827b8eb&utm_content=B31105)—逐步构建和优化 LLM，涵盖数据准备、RAG 和微调。\n* [The Illustrated Transformer](https:\u002F\u002Fjalammar.github.io\u002Fillustrated-transformer\u002F) - 由 Jay Alammar 提供。这是一篇著名的文章，对当前语言模型的工作原理进行了精彩的解释。\n* [A Practical Introduction to LLMs](https:\u002F\u002Ftowardsdatascience.com\u002Fa-practical-introduction-to-llms-65194dda1148) - 由 [Shawhin Talebi](https:\u002F\u002Fshawhin.medium.com\u002F) 提供。\n* [Medium](https:\u002F\u002Fwhats-ai.medium.com\u002Fmembership) 是寻找优质解释的最佳平台，无论是在 [Towards AI](https:\u002F\u002Fpub.towardsai.net\u002F) 还是 [Towards Data Science](https:\u002F\u002Ftowardsdatascience.com\u002F) 出版物上。我也在那里分享自己的文章，非常喜欢使用这个平台。如果你觉得有趣并想同时支持我，可以通过我的[推广链接](https:\u002F\u002Fwhats-ai.medium.com\u002Fmembership)订阅 Medium！\n* [Reading lists for new MILA students](https:\u002F\u002Fdocs.google.com\u002Fdocument\u002Fd\u002F1IXF3h0RU5zz4ukmTrVKVotPQypChscNGf5k6E25HGvA\u002Fedit#) - 匿名\n* [A complete roadmap to master NLP in 2022](https:\u002F\u002Fwww.analyticsvidhya.com\u002Fblog\u002F2022\u002F01\u002Froadmap-to-master-nlp-in-2022\u002F)\n* NLTK Book 是学习 NLP 基础理论的免费资源：https:\u002F\u002Fwww.nltk.org\u002Fbook\u002F\n* [The Annotated Transformer](https:\u002F\u002Fnlp.seas.harvard.edu\u002F2018\u002F04\u002F03\u002Fattention.html) - 哈佛大学\n\n\n## 关注在线课程\u003Ca name=\"courses\">\u003C\u002Fa>\n如果你需要更多指导，我建议查看以下（可选）在线课程...\n* [Generative AI with Large Language Models](https:\u002F\u002Fimp.i384100.net\u002FR5WzQR) - 付费\n* [Become an NLP pro with Coursera's Natural Language Processing Specialization by deeplearning.ai](https:\u002F\u002Fcoursera.pxf.io\u002FP0vO9e) - 付费\n* [Gradio Course - Create User Interfaces for Machine Learning Models - freeCodeCamp](https:\u002F\u002Fyoutu.be\u002FRiCQzBluTxU) - 免费\n* [Train & Fine-Tune LLMs for Production Course by Activeloop, Towards AI & Intel Disruptor](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Fllms\u002F?utm_source=social&utm_medium=youtube&utm_campaign=llmcourse) - 免费\n* [The LLM University by Cohere](https:\u002F\u002Fdocs.cohere.com\u002Fdocs\u002Fllmu?ref=txt.cohere.com) - 免费\n* [From Beginners to Advanced LLM Developer](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbeginner-to-advanced-llm-dev?ref=1f9b29) - 由 Towards AI 提供。\"使用 LLM、提示工程、RAG、微调和智能体（Agents）构建你的首个可扩展产品！掌握顶尖公司所需的技能，用真实应用场景构建自己的高级 LLM 最小可行产品（MVP）。\"\n* [Become an NLP pro with Coursera's Natural Language Processing Specialization by deeplearning.ai](https:\u002F\u002Fcoursera.pxf.io\u002FP0vO9e) - 付费\n _\"进入 NLP 领域。通过四门实践课程掌握前沿 NLP 技术！\"_\n* [An NLP Nano Degree!](https:\u002F\u002Fimp.i115008.net\u002FjW4K60) — 付费\n_\"学习前沿自然语言处理技术，处理语音并分析文本。构建概率模型和深度学习模型，如隐马尔可夫模型（Hidden Markov Models）和循环神经网络（Recurrent Neural Networks），教会计算机完成语音识别、机器翻译等任务！\"_\n* [Introduction to Large Language Models with Google Cloud](https:\u002F\u002Fimp.i115008.net\u002FeKbDLD) - 付费\n* [Learn to train, fine-tune and use LLMs in your applications.](https:\u002F\u002Fwww.wandb.courses\u002Fpages\u002Fw-b-courses) - 由 Weights & Biases 免费提供\n* [Large Language Models with Semantic Search](https:\u002F\u002Fwww.deeplearning.ai\u002Fshort-courses\u002Flarge-language-models-semantic-search\u002F) - 免费，Deeplearning.ai 和 Cohere 联合提供\n\n你可以轻松搜索到更多资源，但在阅读和学习上述内容后，我相信你已经对 LLM 有了足够的理解，可以进入真正的环节：实践。\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002FhLr2aQF.png\" width=\"512\"\u002F>](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbeginner-to-advanced-llm-dev?ref=1f9b29)\n\n## 实践，实践，再实践！\u003Ca name=\"practice\">\u003C\u002Fa>\n\n### 实践是关键\n\n编程中最重要的是实践，机器学习（Machine Learning）也不例外。找到一个个人项目来练习可能很困难。我强烈建议你尝试自己构建一些东西，但我理解这可能会让人感到畏惧。那么我建议你跟随一到两门**极其**注重实践的课程，利用这些资源，基于他们提供的代码示例来构建自己的项目，并让 ChatGPT 或 GitHub Copilot [为你工作](#domore)，作为代码助手完成其余的工作。\n\n以下是我能找到的关于大语言模型（LLMs, Large Language Models）最注重实践的几门课程：\n\n* 想要快速构建文本分类模型或词向量器（word vectorizer），[fasttext](https:\u002F\u002Ffasttext.cc\u002Fdocs\u002Fen\u002Fsupervised-tutorial.html) 是一个快速训练模型的优秀库。\n* [Huggingface](https:\u002F\u002Fhuggingface.co\u002Fcourse\u002Fchapter1\u002F1) 是获取现代自然语言处理（NLP, Natural Language Processing）模型的**首选**平台，他们还提供了一整套关于它的[课程](https:\u002F\u002Fhuggingface.co\u002Fcourse\u002Fchapter1\u002F1)。\n* [LangChain & Vector Databases in Production](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Flangchain\u002F) - 这是我们在 Towards AI 与 Activeloop 和 Intel Disruptor Initiative 合作开发的精彩免费资源，用于学习生产环境中的 LangChain 和向量数据库（Vector Databases）。\"无论你是经验丰富的开发者但刚接触 AI 领域，还是经验丰富的机器学习爱好者，这门课程都为你而设计。我们的目标是让 AI 变得易于获取且实用，改变你处理日常任务的方式以及工作的整体影响力。\"\n* [Training & Fine-Tuning LLMs for Production](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Fllms\u002F) - 这是我们在 Towards AI 与 Activeloop 和 Intel Disruptor Initiative 合作开发的精彩免费资源，用于学习生产环境中大语言模型的训练与微调（Training & Fine-Tuning）。\"如果你想学习如何从零开始训练和微调大语言模型，具备中级 Python 知识，并能访问中等计算资源（某些情况下，只需 Google Colab 就足够了！），那么你就具备了参加并完成这门课程的条件。这门课程面向广泛的受众，包括 AI 初学者、现任机器学习工程师、学生以及考虑转行到 AI 领域的专业人士。我们旨在为你提供必要的工具，以便在各个行业中应用和定制大语言模型，使 AI 更易于获取且实用。\"\n* [The Real-World ML Tutorial & Community](https:\u002F\u002Frealworldmachinelearning.carrd.co\u002F) - 付费\n\n一个提醒。学习的最佳方式是构建东西！我非常倾向于通过实践来学习。这些课程都很棒，但都是可选的。你可以自己完成，而且大多数提供大语言模型工作资源的公司（OpenAI、LangChain、Activeloop、Cohere、W&B...）都有很棒的教程来帮助你入门并构建一些东西。然后，你可以[向 ChatGPT 求助](#domore)帮你完成它！\n\n\n## 提示工程（Prompting）\u003Ca name=\"prompting\">\u003C\u002Fa>\n提示工程是使用模型和构建自然语言处理相关应用的一项重要新技能。\n\n* [What is Prompting? Talking with AI Models...](https:\u002F\u002Fyoutu.be\u002FpZsJbYIFCCw) - 免费\n* [ChatGPT Prompt Engineering for Developers](https:\u002F\u002Fimp.i384100.net\u002FrQBVMy) - 付费\n* [Learn Prompting](https:\u002F\u002Flearnprompting.org\u002F) - 这是一个很棒的**免费**课程，旨在教授提示工程并为特定模型提供技巧。对于提示工程来说，它**就是你所需要的一切**！\n* [Techniques to improve reliability](https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-cookbook\u002Fblob\u002Fmain\u002Farticles\u002Ftechniques_to_improve_reliability.md#techniques-to-improve-reliability) - OpenAI Cookbook 关于提示工程技巧的文档。\n\n## 更多关于检索增强生成（Retrieval Augmented Generation, RAG）和微调（fine-tuning）的内容\u003Ca name=\"rag\">\u003C\u002Fa>\n\n目前大多数人都在构建基于 RAG 的应用程序。以下是一些我非常喜欢的资源，可以帮助你入门并深入理解 RAG...\n\n* [最大化 LLM 性能的技术综述](https:\u002F\u002Fyoutu.be\u002FahnGLM-RC1Y) - OpenAI 出品的精彩视频，介绍了何时使用提示工程（prompt engineering）、RAG 或微调（fine-tuning）。这是该领域每个人都必看的视频！\n* [RAG vs 微调 vs 深度记忆 vs 从头训练 LLM：何时该做什么](https:\u002F\u002Fyoutu.be\u002FpHv9SsE4Mb4) - 同样，这是一个简短的视频，介绍在你的应用中何时应该使用 RAG、微调或提示工程。\n* [使用 GPT 和嵌入（embeddings）构建问答聊天机器人](https:\u002F\u002Fyoutu.be\u002FLB5g-AhfPG8) - [Jeremy Pinto](https:\u002F\u002Ftwitter.com\u002Fjerpint) 的实用 YouTube 教程。\n* [如何构建一个能回答关于你网站问题的 AI](https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Ftutorials\u002Fweb-qa-embeddings\u002Fhow-to-build-an-ai-that-can-answer-questions-about-your-website) - 免费的 OpenAI 教程。\n* [从初学者到高级 LLM 开发者](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbeginner-to-advanced-llm-dev?ref=1f9b29) - 由 Towards AI 提供。\"用 LLM、提示工程、RAG、微调和智能体（Agents）构建你的第一个可扩展产品！掌握顶尖公司所需的技能，用真实应用场景构建你自己的高级 LLM 最小可行产品（MVP）。\"\n* [如何构建基于 RAG 的 ChatGPT 网页应用：认识我们的新 AI 导师](https:\u002F\u002Fyoutu.be\u002F7ytyK6u3aAk) - YouTube 视频，介绍我如何构建基于 RAG 的聊天机器人（以及你如何也能做到）。\n* [面向生产的 LLM 训练与微调](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Fllms\u002F) - 学习如何从头开始训练和微调 LLM。\n* [训练并部署实时金融顾问](https:\u002F\u002Fgithub.com\u002Fiusztinpaul\u002Fhands-on-llms) - 由 [Paul Iusztin](https:\u002F\u002Fgithub.com\u002Fiusztinpaul)、[Pau Labarta Bajo](https:\u002F\u002Fgithub.com\u002FPaulescu) 和 [Alexandru Razvant](https:\u002F\u002Fgithub.com\u002FJoywalker) 提供的 LLM 实战课程。\n* [使用 LangChain 和 LlamaIndex 面向生产的检索增强生成](https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Frag) - 无论你是计划为组织构建数据聊天应用，还是只想学习如何在各个行业利用生成式 AI，这门课程都适合你。课程解决了关键问题，如提高检索准确性、减少 AI 输出中的幻觉（hallucinations）、增强可解释性、解决版权问题，以及提供更定制化、最新的数据输入。我们超越基础 RAG 应用，让你掌握使用 LangChain、LlamaIndex 和深度记忆（Deep Memory）等工具创建更复杂、更可靠产品的技能。强调实践学习，这门课程是掌握高级 RAG 技术和真实应用场景的入门途径。\n* [为生产构建 LLM：用提示工程、微调和 RAG 增强 LLM 能力和可靠性](https:\u002F\u002Famzn.to\u002F4bqYU9b) - 由 Towards AI 提供。\"发现将大语言模型适配到真实应用的关键技术栈，包括提示工程、微调和检索增强生成。\"（或[在此](https:\u002F\u002Facademy.towardsai.net\u002Fcourses\u002Fbuildingllmsforproduction?ref=1f9b29)获取电子书。可以私信我获取优惠！）\n\n\n[![Twitter](https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Furl\u002Fhttps\u002Ftwitter.com\u002Fcloudposse.svg?style=social&label=Follow%20%40whats_ai)](https:\u002F\u002Ftwitter.com\u002FWhats_AI)\n\n\n## 更多资源\u003Ca name=\"moreresources\">\u003C\u002Fa>\n### 加入社区！\n\n* [AI 爱好者 Discord 服务器](https:\u002F\u002Fdiscord.gg\u002Flearnaitogether) - 共同学习、提问、寻找 Kaggle 队友、分享项目等。\n* [获取最新 AI 资讯的 Discord 服务器](https:\u002F\u002Fws.towardsai.net\u002Fdiscord) - 及时了解最新 AI 新闻、提问、分享项目等。\n* [Learn Prompting Discord 社区](https:\u002F\u002Fdiscord.gg\u002Flearn-prompting-1046228027434086460) - 与 fellow 提示工程师交流。\n\n* 关注 Reddit 社区 - 提问、分享项目、关注新闻等。\n    * [artificial](https:\u002F\u002Fwww.reddit.com\u002Fr\u002Fartificial\u002F) - 人工智能\n    * [MachineLearning](https:\u002F\u002Fwww.reddit.com\u002Fr\u002FMachineLearning\u002F) - 机器学习（该领域最大的子版块）\n    * [DeepLearningPapers](https:\u002F\u002Fwww.reddit.com\u002Fr\u002FDeepLearningPapers\u002F) - 深度学习论文\n    * [ComputerVision](https:\u002F\u002Fwww.reddit.com\u002Fr\u002Fcomputervision\u002F) - 从图像和视频中提取有用信息\n    * [learnmachinelearning](https:\u002F\u002Fwww.reddit.com\u002Fr\u002Flearnmachinelearning\u002F) - 学习机器学习\n    * [ArtificialInteligence](https:\u002F\u002Fwww.reddit.com\u002Fr\u002FArtificialInteligence\u002F) - AI\n    * [LatsestInML](https:\u002F\u002Fwww.reddit.com\u002Fr\u002FLatestInML\u002F) - 你不应错过的机器学习突破性进展\n\n### 紧跟领域动态！\n\n* 订阅分享新论文的 YouTube 频道 — 及时了解领域最新动态！\n    * [Louis Bouchard](https:\u002F\u002Fwww.youtube.com\u002Fchannel\u002FUCUzGQrN-lyyc0BWTYoJM_Sg) — 每周发布新论文解读视频\n    * [Two Minutes Papers](https:\u002F\u002Fwww.youtube.com\u002Fuser\u002Fkeeroyz) — 每两周发布新论文解读视频\n    * [Bycloud](https:\u002F\u002Fwww.youtube.com\u002Fchannel\u002FUCgfe2ooZD3VJPB6aJAnuQng) — 每周发布新论文解读视频\n    \n* LinkedIn 群组\n    * [Artificial Intelligence, Machine Learning and Deep Learning News](https:\u002F\u002Fwww.linkedin.com\u002Fgroups\u002F8942343\u002F) — 群组成员分享的领域新闻\n    * [Artificial Intelligence | Deep Learning  | Machine Learning](https:\u002F\u002Fwww.linkedin.com\u002Fgroups\u002F45655\u002F)\n    * [Applied Artificial Intelligence](https:\u002F\u002Fwww.linkedin.com\u002Fgroups\u002F127447\u002F)\n    \n* Facebook 群组\n    * [Artificial Intelligence & Deep Learning](https:\u002F\u002Fwww.facebook.com\u002Fgroups\u002FDeepNetGroup) — 关于人工智能（Artificial Intelligence, AI）、神经网络（Neural Networks）和深度学习（Deep Learning）最权威、最活跃的 Facebook 群组。分享 AI 和深度学习前沿的新鲜有趣内容。神经网络将在未来几年重新定义智能机器的含义。\n    * [Deep learning](https:\u002F\u002Fwww.facebook.com\u002Fgroups\u002FDeepLearnng\u002F) — 当今社会正趋向于柔性化和自动化发展，迈入第四次工业革命，这不可避免地推动社会成员卷入社会变革的漩涡。为了生存或领先，人们需要掌握相关工具。机器正变得越来越智能。机器学习（Machine Learning）是不可或缺的技能，需要人们熟练掌握。本群组面向那些希望发展自身才能以适应时代的人。\n\n* 新闻通讯（Newsletters）\n   * [Synced AI TECHNOLOGY & INDUSTRY REVIEW](https:\u002F\u002Fsyncedreview.com\u002F) — 中国领先的 AI 与机器学习（Machine Learning）媒体及信息提供商。\n   * [Inside AI](https:\u002F\u002Finside.com\u002Fai) — 每日精选人工智能、机器人技术和神经科技的新闻与评论。\n   * [AI Weekly](http:\u002F\u002Faiweekly.co\u002F) — 每周收集人工智能和机器学习的新闻与资源。\n   * [AI Ethics Weekly](https:\u002F\u002Flighthouse3.com\u002Fnewsletter\u002F) — 每周将 AI 伦理（AI Ethics）的最新动态发送到您的邮箱。\n   * [Louis Bouchard Weekly](https:\u002F\u002Flouisbouchard.substack.com\u002F) — 每周精选一篇论文，通过文章、视频演示、演示、代码等方式清晰解读。\n   * [ThursdAI](https:\u002F\u002Fsub.thursdai.news\u002F) — 每周高质量 AI 资讯回顾！\n   * [Toward's AI newsletter](https:\u002F\u002Ftowardsai.net\u002Fai\u002Fnewsletter) — 每周总结最有趣的新闻和学习资源，以及来自 Learn AI Together Discord 社区的更新。非常适合机器学习（ML）专业人士和爱好者。\n   * [The Batch](https:\u002F\u002Fwww.deeplearning.ai\u002Fthe-batch\u002F) — Andrew Ng \u002F Deeplearning.ai\n    \n* 关注 Medium 出版物\n    * [Towards Data Science](https:\u002F\u002Ftowardsdatascience.com\u002F) — \"分享概念、想法和代码\"\n    * [Towards AI](https:\u002F\u002Fmedium.com\u002Ftowards-artificial-intelligence) — \"科技、科学与工程的最佳内容\"\n    * [OneZero](https:\u002F\u002Fonezero.medium.com\u002F) — \"未来的暗流。一个关于科技与科学的 Medium 出版物\"\n\n## 寻找机器学习工作\u003Ca name=\"findajob\">\u003C\u002Fa>\n\n* 阅读[文章中的这一部分](https:\u002F\u002Fwww.louisbouchard.ai\u002Flearnai\u002F#how-to-find-a-job)，其中包含大量面试技巧和**如何准备面试**。\n* 通过观看他人的面试经验来了解面试流程并更好地准备，例如我与 NVIDIA、Zoox（自动驾驶公司）、D-ID（生成式 AI 创业公司）等专家的[面试系列](https:\u002F\u002Fyoutube.com\u002Fplaylist?list=PLO4GrDnQanVfrRIuIT_1rlLLTgQJdfXmS)。\n\n## AI 伦理\u003Ca name=\"aiethics\">\u003C\u002Fa>\n* [What are Ethics and Why do they Matter? Machine Learning Edition](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=F0cxzESR7ec&list=PLtmWHNX-gukIU6V33Bc8eP8OD41I4GywR&ab_channel=RachelThomas) — 由 fast.ai 创始人 Rachel Thomas 制作\n* [AI4People—An Ethical Framework for a Good AI Society: Opportunities, Risks, Principles, and Recommendations](https:\u002F\u002Flink.springer.com\u002Fcontent\u002Fpdf\u002F10.1007\u002Fs11023-018-9482-5.pdf) — Floridi 等人，2018，AI4People 美好社会的 AI 伦理框架\n* [Ethics guidelines for trustworthy AI](https:\u002F\u002Fwayback.archive-it.org\u002F12090\u002F20210728013426\u002Fhttps:\u002F\u002Fdigital-strategy.ec.europa.eu\u002Fen\u002Flibrary\u002Fethics-guidelines-trustworthy-ai) — 欧盟委员会高级专家组提出的可信 AI 7 大要点。\n* [An Introduction to Ethics in Robotics and AI](https:\u002F\u002Flink.springer.com\u002Fbook\u002F10.1007\u002F978-3-030-51110-4) — Christoph Bartneck、Christoph Lütge、Alan Wagner 和 Sean Welsh 编写的免费电子书。\n\n\n## 利用大语言模型（LLMs）学习更多、做得更多\u003Ca name=\"domore\">\u003C\u002Fa>\nChatGPT、Bing、Claude……都非常出色。当然，它们有局限性。然而，你可以利用它们来学习任何你想学的东西。我用它来编程或提出各种一般性问题。当你提出重要问题时，需要仔细核实。\n尽管如此，它是一个强大的**工具**。是的，它是工具，不是人类的替代品。把它当作一个_笨拙的_助手，它几乎什么都知道。\n\n这里有一个[清晰的例子](https:\u002F\u002Fchat.openai.com\u002Fshare\u002F883389c9-f0f8-4a3e-a3af-ee9860d448a8)，展示了我如何在一个项目中使用它来更好地理解一个我不熟悉的项目中的函数。这是 Python 的例子，但这些模型在编程方面非常强大，也能帮助理解新平台（如 AWS、GCP、使用虚拟机、服务器、SSH 连接等……任何你在 LLM 领域不熟悉但有用的东西）。\n\n_p.s. 我提到 Bing 和 Claude 不是随便说说。不要过度依赖像 OpenAI 这样的单一公司。在这场最佳大语言模型（LLM）的竞争中，还有其他公司（而且永远会有）。今天早上我想为本指南创建一个例子，结果……_\n\n[\u003Cimg src=\"https:\u002F\u002Fimgur.com\u002FRDdL7el.png\" width=\"512\"\u002F>](https:\u002F\u002Ftwitter.com\u002Fsatourian\u002Fstatus\u002F1722257478115811421)\n\n---\n\n\n***如果你在分享这份清单，请在 Twitter [@Whats_AI](https:\u002F\u002Ftwitter.com\u002FWhats_AI) 或 LinkedIn [@Louis Bouchard](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Fwhats-ai\u002F) 上标记我！***\n\n👀 **如果你想支持我的工作**，可以查看[Sponsor](https:\u002F\u002Fgithub.com\u002Fsponsors\u002Flouisfb01)本仓库或在[Patreon](https:\u002F\u002Fwww.patreon.com\u002Fwhatsai)上支持我。\n\n本指南仍在定期更新。","# start-llms 快速上手指南\n\n> **注意**：start-llms 是一个**学习资源导航仓库**，而非可直接安装运行的软件工具。本指南将说明如何高效使用该资源库规划你的 LLM 学习路径。\n\n---\n\n## 环境准备\n\n### 系统要求\n- 基础编程知识（Python）\n- 了解机器学习基本概念\n- 稳定的网络连接（访问 YouTube、GitHub 等）\n\n### 前置依赖\n| 层级 | 要求 | 补充资源 |\n|:---|:---|:---|\n| 零基础 | 无编程\u002FAI 背景 | 先完成 [start-machine-learning](https:\u002F\u002Fgithub.com\u002Flouisfb01\u002Fstart-machine-learning) 指南 |\n| 有基础 | Python + 基础 ML 知识 | 可直接开始本指南 |\n\n---\n\n## 快速开始\n\n### 步骤 1：克隆仓库（可选，用于本地标记进度）\n\n```bash\n# 克隆资源库到本地\ngit clone https:\u002F\u002Fgithub.com\u002Flouisfb01\u002Fstart-llms.git\n\n# 进入目录\ncd start-llms\n```\n\n### 步骤 2：选择学习路径\n\n根据你的偏好选择入口：\n\n| 学习风格 | 推荐起点 | 资源类型 |\n|:---|:---|:---|\n| 视觉型学习者 | [YouTube 视频入门](#youtubevideos) | 免费视频 |\n| 阅读型学习者 | [书籍与文章](#readers) | 免费\u002F付费书籍、博客 |\n| 结构化学习 | [在线课程](#courses) | 免费\u002F付费课程 |\n| 实践优先 | [动手练习](#practice) | 代码实战 |\n\n### 步骤 3：核心学习路线（推荐顺序）\n\n```text\n1. 术语理解 → 2. Transformer 原理 → 3. 提示工程 → 4. RAG → 5. 微调部署\n```\n\n#### 第一阶段：快速入门（1-2 周）\n\n**推荐视频（国内可访问）：**\n- [Mastering AI Jargon](https:\u002F\u002Fyoutu.be\u002Fq4G6X09NEu4) — LLM 核心术语\n- [Intro to Large Language Models](https:\u002F\u002Fyoutu.be\u002FzjkBMFhNj_g) — Andrej Karpathy 1小时精讲\n- [Let's build GPT: from scratch](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=kCc8FmEb1nY) — 从零实现 GPT\n\n**替代方案（国内平台）：**\n- B站搜索 \"Andrej Karpathy GPT\" 有搬运版本\n- 关注 [TowardsAI 中文社区](https:\u002F\u002Ftowardsai.net\u002F) 获取本地化内容\n\n#### 第二阶段：系统学习\n\n**免费课程：**\n```bash\n# 无需安装，直接访问：\n# Cohere LLM University: https:\u002F\u002Fdocs.cohere.com\u002Fdocs\u002Fllmu\n# Activeloop 生产级 LLM 课程: https:\u002F\u002Flearn.activeloop.ai\u002Fcourses\u002Fllms\u002F\n```\n\n**推荐阅读：**\n- [The Illustrated Transformer](https:\u002F\u002Fjalammar.github.io\u002Fillustrated-transformer\u002F) — Transformer 可视化详解\n- [Building LLMs for Production](https:\u002F\u002Famzn.to\u002F4bqYU9b) — 生产环境实战（可联系作者获取折扣）\n\n#### 第三阶段：动手实践\n\n**核心技能栈：**\n\n| 技能 | 学习资源 | 实践工具 |\n|:---|:---|:---|\n| 提示工程 (Prompting) | 仓库 [Prompting](#prompting) 章节 | OpenAI API \u002F 本地模型 |\n| RAG 架构 | [RAG](#rag) 章节 | LangChain, LlamaIndex |\n| 模型微调 | Hugging Face 教程 | transformers, PEFT |\n| 部署上线 | 生产课程 | Docker, vLLM, TGI |\n\n---\n\n## 基本使用示例\n\n### 示例：按模块规划学习\n\n```markdown\n## 我的 8 周学习计划\n\n### Week 1-2: 基础\n- [ ] 观看 Andrej Karpathy 的 LLM 入门视频\n- [ ] 阅读 The Illustrated Transformer\n- [ ] 完成 Cohere LLM University 前 5 章\n\n### Week 3-4: 提示工程\n- [ ] 学习 Prompting 章节资源\n- [ ] 用 OpenAI API 实践 10 个不同场景的 prompt\n\n### Week 5-6: RAG 实战\n- [ ] 学习 RAG 章节\n- [ ] 搭建一个本地知识库问答系统\n\n### Week 7-8: 微调与部署\n- [ ] 完成 Activeloop 生产课程\n- [ ] 微调一个 7B 模型并部署到本地\n```\n\n### 示例：跟踪最新动态\n\n```bash\n# 关注维护者获取更新\n# Twitter\u002FX: @Whats_AI\n# Newsletter: https:\u002F\u002Flouisbouchard.substack.com\u002F\n# 中文替代: 关注「机器之心」「AI 科技评论」等公众号\n```\n\n---\n\n## 国内访问优化\n\n| 原资源 | 国内替代方案 |\n|:---|:---|\n| YouTube | B站搬运、知乎视频 |\n| Twitter\u002FX | 微博 AI 博主、即刻 |\n| Substack Newsletter | 竹白、小报童 AI 专栏 |\n| Hugging Face | ModelScope 魔搭社区、Wisemodel |\n| OpenAI API | 阿里云百炼、百度文心、智谱 GLM |\n\n---\n\n## 下一步\n\n1. **立即行动**：从 [YouTube 视频](#youtubevideos) 或 [The Illustrated Transformer](https:\u002F\u002Fjalammar.github.io\u002Fillustrated-transformer\u002F) 开始\n2. **加入社区**：关注维护者 [Louis Bouchard](https:\u002F\u002Fgithub.com\u002Flouisfb01) 的 [YouTube](https:\u002F\u002Fwww.youtube.com\u002F@whatsai) 和 [播客](https:\u002F\u002Fopen.spotify.com\u002Fshow\u002F4rKRJXaXlClkDyInjHkxq3)\n3. **提交贡献**：发现优质资源可通过 GitHub Issue 提交到本仓库","**场景背景**：张磊是一名有 2 年 Python 开发经验的传统后端工程师，公司 2024 年初要求他从零开始搭建内部知识库问答系统，他需要在 3 个月内掌握 LLM 相关技术并交付项目。\n\n### 没有 start-llms 时\n\n- **信息碎片化，无从下手**：张磊在搜索引擎和知乎上搜索\"LLM入门\"，结果鱼龙混杂，从 Transformer 论文到 ChatGPT 使用技巧混杂在一起，分不清学习优先级，浪费两周时间还在原地打转\n- **付费陷阱与过时内容**：误买了某平台 2999 元的\"大模型全栈课\"，发现一半内容讲的是 2022 年的 GPT-3，对 2024 年的 RAG、Agent 技术几乎没涉及，退款困难\n- **实践路径模糊**：看完几篇博客后尝试用 LangChain，但官方文档假设读者已有向量数据库基础，张磊卡在 Embedding 选型上，项目进度严重滞后\n- **技术更新焦虑**：每隔几天就有新模型、新框架发布，不知道哪些值得跟进，担心学的东西很快过时，陷入\"学不动又不敢不学\"的焦虑\n\n### 使用 start-llms 后\n\n- **结构化学习路径**：按照仓库推荐的\"先 YouTube 视频建立直觉 → 再系统课程打基础 → 最后专项突破 RAG\u002FAgent\"的顺序，张磊第一周就理清了 LLM 技术栈全貌，明确知道自己缺什么\n- **零成本获取优质资源**：仓库筛选的免费课程（如 Andrej Karpathy 的 Tokenizer 讲解）和开源书籍，质量远超之前付费课程，省下的预算用于购买 GPU 算力做实验\n- **即学即用的实践指引**：Prompting 和 RAG 章节直接提供了可运行的代码片段和常见坑点总结，张磊 2 周内就搭出了基于本地 Llama 的原型系统，Embedding 选型参考了社区讨论区的真实对比\n- **持续跟进前沿动态**：通过仓库维护者 Louis Bouchard 的 Newsletter 和 Twitter，张磊每周花 20 分钟了解真正重要的技术更新（如 2024 年 MoE 架构的进展），不再被信息噪音干扰\n\n**核心价值**：start-llms 用一套经过验证的免费学习路线图，帮传统开发者避开信息过载和付费陷阱，在有限时间内高效建立可落地的 LLM 工程能力。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Flouisfb01_start-llms_14140901.png","louisfb01","Louis-François Bouchard","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Flouisfb01_d681bed6.png","Making AI accessible on YouTube, Newsletter, Spotify, Apple podcasts.\r\n\r\nCo-Founder at Towards AI.\r\nex-PhD student at Mila, Polytechnique Montréal","Mila\u002FPolytechnique Montréal & @towardsai","montreal",null,"Whats_AI","https:\u002F\u002Fwww.louisbouchard.ai\u002F","https:\u002F\u002Fgithub.com\u002Flouisfb01",959,124,"2026-04-05T23:17:49","MIT",1,"未说明",{"notes":92,"python":90,"dependencies":93},"该仓库是一个学习资源整理指南，并非可直接运行的代码项目。README 中主要包含 LLM 相关的学习路径、视频课程、书籍文章和在线课程推荐，没有提供任何具体的运行环境配置、安装说明或代码依赖信息。如需实际运行 LLM 项目，建议参考指南中推荐的实践课程（如 Andrej Karpathy 的 'Let's build GPT' 或 freeCodeCamp 的教程）获取具体的环境要求。",[],[13,15,26,14,54],[96,97,98,99,100,101,102,103,104,105,106],"ai","fine-tuning","gpt","gpt-4","language-model","large-language-models","llama","llm","llms","rag","retrieval-augmented-generation","2026-03-27T02:49:30.150509","2026-04-06T08:45:17.162822",[],[]]