[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-93429":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":8,"htmlUrl":8,"language":8,"languages":8,"totalLinesOfCode":8,"stars":9,"forks":10,"watchers":11,"openIssues":12,"contributorsCount":12,"subscribersCount":12,"size":12,"stars1d":13,"stars7d":14,"stars30d":14,"stars90d":12,"forks30d":12,"starsTrendScore":15,"compositeScore":16,"rankGlobal":8,"rankLanguage":8,"license":8,"archived":17,"fork":17,"defaultBranch":18,"hasWiki":19,"hasPages":17,"topics":20,"createdAt":8,"pushedAt":8,"updatedAt":21,"readmeContent":22,"aiSummary":23,"trendingCount":12,"starSnapshotCount":12,"syncStatus":11,"lastSyncTime":24,"discoverSource":25},93429,"Awesome_ai_learning","h9-tec\u002FAwesome_ai_learning","h9-tec",null,229,30,2,0,6,67,44,77.13,false,"main",true,[],"2026-07-22 04:02:09","# The No-Hype AI Learning Guide (2026 Edition)\n\nA curated map of AI learning resources that actually deliver. No \"make money with AI\" courses, no influencer funnels, no recycled 2021 lists. Every resource here was checked in mid-2026, is made by people who build AI for a living, and has proven real impact on real learners.\n\n**\"Learning AI\" means two completely different things, so this guide has two tracks:**\n\n| Track | Who it's for | What you'll do |\n|---|---|---|\n| **Track 1: Study AI as a Field** | Future ML\u002FAI engineers and researchers | Write code, touch math, build models from scratch |\n| **Track 2: Use AI Effectively** | Professionals, managers, students, creators | Master AI tools for real work, no coding required |\n\n**Difficulty legend:**\n\n- 🟢 **Easy**: no prerequisites, start today\n- 🟡 **Medium**: needs some Python\u002Fmath, or solid hands-on AI tool experience\n- 🔴 **Hard**: needs strong coding + math foundations, or deep prior context\n\n**Selection criteria (why you can trust this list):**\n\n1. Free, or clearly worth the price (most are 100% free)\n2. Built or taught by practitioners with verifiable work (Stanford, MIT, Hugging Face, Anthropic, OpenAI, Google, fast.ai, Karpathy)\n3. Still maintained and relevant in 2026\n4. Zero income promises, zero fear marketing\n\n---\n\n## Track 1: Study AI as a Field\n\nThe goal of this track: go from zero to someone who can build, train, evaluate, and ship models. It is ordered as a path. Do not skip Stage 0.\n\n### Stage 0: Foundations (Math + Code + Intuition)\n\n| Resource | Level | Cost | Why it earns its place |\n|---|---|---|---|\n| [Elements of AI](https:\u002F\u002Fwww.elementsofai.com) (Univ. of Helsinki) | 🟢 | Free | The cleanest conceptual intro to AI ever made. Built by a university, used by 1M+ learners, zero hype. Do this first to know what you are getting into. |\n| [Kaggle Learn](https:\u002F\u002Fwww.kaggle.com\u002Flearn) | 🟢 | Free | Bite-size, hands-on micro courses: Python, pandas, intro ML. Runs in the browser, no setup. The fastest way to get coding for ML. |\n| [3Blue1Brown: Neural Networks](https:\u002F\u002Fwww.3blue1brown.com\u002Ftopics\u002Fneural-networks) | 🟢 | Free | Visual intuition for what a neural network actually is, plus his linear algebra and calculus series. Watch before touching any framework. |\n| [Mathematics for Machine Learning](https:\u002F\u002Fmml-book.github.io) (Deisenroth, Faisal, Ong) | 🟡 | Free PDF | The one math book written specifically for ML learners. Linear algebra, calculus, probability, optimization, all with ML motivation. Official free PDF. |\n| [StatQuest (Josh Starmer, YouTube)](https:\u002F\u002Fwww.youtube.com\u002F@statquest) | 🟢 | Free | Statistics and classical ML explained so clearly it feels illegal. Perfect companion when a concept refuses to click. |\n\n**Exit criteria for Stage 0:** you can write basic Python, you know what a matrix multiplication and a gradient are, and you can explain what \"training a model\" means to a friend.\n\n### Stage 1: Core Machine Learning\n\n| Resource | Level | Cost | Why it earns its place |\n|---|---|---|---|\n| [Machine Learning Specialization](https:\u002F\u002Fwww.coursera.org\u002Fspecializations\u002Fmachine-learning-introduction) (Andrew Ng, Stanford + DeepLearning.AI) | 🟢→🟡 | Free to audit | Still the gold-standard first ML course. Regression, classification, neural nets, decision trees, recommenders. Calm, rigorous, beginner-safe. |\n| [Google Machine Learning Crash Course](https:\u002F\u002Fdevelopers.google.com\u002Fmachine-learning\u002Fcrash-course) | 🟡 | Free | Fast, practical, recently refreshed with modern content. Great as a second pass to consolidate Ng's material with hands-on exercises. |\n| [Harvard CS50's Introduction to AI with Python](https:\u002F\u002Fcs50.harvard.edu\u002Fai\u002F) | 🟡 | Free | The best coverage of the \"non-LLM\" side of AI: search, logic, probability, optimization. Makes you a broader engineer than deep-learning-only paths. |\n| [Kaggle Competitions](https:\u002F\u002Fwww.kaggle.com\u002Fcompetitions) | 🟡 | Free | Not a course. This is where theory becomes skill. Enter a beginner competition (Titanic, House Prices), read winning solutions, iterate. |\n\n**Exit criteria for Stage 1:** you have trained several models yourself, you understand overfitting, train\u002Ftest splits, and evaluation metrics, and you have at least one Kaggle notebook you're not ashamed of.\n\n### Stage 2: Deep Learning\n\nTwo schools, both excellent. Top-down (build things first, understand later) = fast.ai. Bottom-up (understand first, build later) = Ng's Deep Learning Specialization. Pick one as your spine, use the rest as support.\n\n| Resource | Level | Cost | Why it earns its place |\n|---|---|---|---|\n| [Practical Deep Learning for Coders](https:\u002F\u002Fcourse.fast.ai) (fast.ai, Jeremy Howard) | 🟡 | Free | The top-down classic. You train a state-of-the-art image model in lesson 1, then peel back the layers. Produces builders, not textbook-quoters. |\n| [Deep Learning Specialization](https:\u002F\u002Fwww.coursera.org\u002Fspecializations\u002Fdeep-learning) (Andrew Ng) | 🟡 | Free to audit | The bottom-up classic. CNNs, RNNs, transformers, and the debugging mindset (bias\u002Fvariance, error analysis) that separates engineers from tutorial-followers. |\n| [MIT 6.S191: Introduction to Deep Learning](http:\u002F\u002Fintrotodeeplearning.com) | 🟡 | Free | Refreshed every January. The 2026 edition added expanded LLM and agentic AI coverage. Dense, modern, with Colab labs. |\n| [Dive into Deep Learning (d2l.ai)](https:\u002F\u002Fd2l.ai) | 🟡 | Free | Interactive book with math + runnable code (PyTorch\u002FJAX), adopted at 500+ universities in 70 countries. The best \"read and run\" reference. |\n| [Neural Networks and Deep Learning](http:\u002F\u002Fneuralnetworksanddeeplearning.com) (Michael Nielsen) | 🟡 | Free | Short, beautiful online book. Builds a digit-recognizing network from scratch and makes backpropagation genuinely intuitive. |\n\n**Exit criteria for Stage 2:** you can build and train a CNN and a small transformer in PyTorch, and explain backpropagation without hand-waving.\n\n### Stage 3: LLMs and the Modern Era (Build It From Scratch)\n\nThis is where most 2026 jobs live. The theme of this stage: stop using black boxes, build the box.\n\n| Resource | Level | Cost | Why it earns its place |\n|---|---|---|---|\n| [Neural Networks: Zero to Hero](https:\u002F\u002Fkarpathy.ai\u002Fzero-to-hero.html) (Andrej Karpathy) | 🟡→🔴 | Free | The single best resource on this entire list. Backprop from scratch, then makemore, then GPT from scratch, then a tokenizer from scratch. Widely considered the clearest explanation of model internals that exists. |\n| [Hugging Face LLM Course](https:\u002F\u002Fhuggingface.co\u002Flearn\u002Fllm-course) | 🟡 | Free + certificate | The practical pipeline: tokenizers, transformers library, fine-tuning, sharing models. The industry-standard toolkit, taught by the people who built it. |\n| [Hugging Face Agents Course](https:\u002F\u002Fhuggingface.co\u002Flearn\u002Fagents-course) | 🟡 | Free + certificate | From LLM to agent: tools, reasoning loops, smolagents \u002F LlamaIndex \u002F LangGraph, plus a final benchmarked project. |\n| [Speech and Language Processing, 3rd ed.](https:\u002F\u002Fweb.stanford.edu\u002F~jurafsky\u002Fslp3\u002F) (Jurafsky & Martin) | 🔴 | Free draft | THE NLP textbook, continuously updated with transformer\u002FLLM chapters. Free online. If you work in NLP, you will return to it for years. |\n| [Stanford CS336: Language Modeling from Scratch](https:\u002F\u002Fstanford-cs336.github.io) (Percy Liang, Tatsunori Hashimoto) | 🔴 | Free materials | Build a full LLM pipeline yourself: data, tokenization, architecture, training, alignment, inference. The most complete \"no black boxes\" university course, Spring 2026 edition available. |\n| [Stanford CS25: Transformers United](https:\u002F\u002Fweb.stanford.edu\u002Fclass\u002Fcs25\u002F) | 🔴 | Free lectures | Rotating seminar with speakers from OpenAI, Anthropic, DeepMind, Meta. The best free source for frontier topics that textbooks haven't caught up to. |\n| [Build a Large Language Model (From Scratch)](https:\u002F\u002Fgithub.com\u002Frasbt\u002FLLMs-from-scratch) (Sebastian Raschka) | 🔴 | Book paid, repo free | Pretrain and fine-tune a GPT-style model end to end in plain PyTorch. The full companion code is open source. |\n| [Understanding Deep Learning](https:\u002F\u002Fudlbook.github.io\u002Fudlbook\u002F) (Simon Prince) | 🔴 | Free PDF | The modern theory book: transformers, diffusion, RL, all with exceptional figures. What Goodfellow's classic [Deep Learning](https:\u002F\u002Fwww.deeplearningbook.org) was for 2016, this is for now. Keep both as references. |\n\n**Bonus for speech\u002FArabic NLP people:** the [Hugging Face Audio Course](https:\u002F\u002Fhuggingface.co\u002Flearn\u002Faudio-course) covers ASR and TTS with transformers, and the [Deep RL Course](https:\u002F\u002Fhuggingface.co\u002Flearn\u002Fdeep-rl-course) covers the RL foundations behind RLHF. Both free with certificates.\n\n**Exit criteria for Stage 3:** you have trained a small GPT from scratch, fine-tuned an open model on your own dataset, and can read a modern paper without drowning.\n\n### Stage 4: Production AI Engineering\n\nModels in notebooks are hobbies. Models in production are careers.\n\n| Resource | Level | Cost | Why it earns its place |\n|---|---|---|---|\n| [Designing Machine Learning Systems](https:\u002F\u002Fgithub.com\u002Fchiphuyen\u002Fdmls-book) (Chip Huyen) | 🔴 | Book paid, notes free | The system-design bible for classical ML in production: data, features, deployment, monitoring, drift. |\n| [AI Engineering](https:\u002F\u002Fgithub.com\u002Fchiphuyen\u002Faie-book) (Chip Huyen, 2025) | 🔴 | Book paid, resources free | The follow-up for the foundation-model era: evals, RAG, fine-tuning decisions, latency\u002Fcost, AI-as-judge. The most-read book on O'Reilly in 2025, and the repo's free resource list is a goldmine by itself. |\n| [Made With ML](https:\u002F\u002Fmadewithml.com) (Goku Mohandas) | 🔴 | Free | End-to-end MLOps: design, develop, deploy, iterate, with production-grade code. Battle-tested and open. |\n| [Anthropic: Building Effective Agents](https:\u002F\u002Fwww.anthropic.com\u002Fresearch\u002Fbuilding-effective-agents) | 🟡 | Free | Short, sober engineering guidance on when you need an agent and when a simple workflow beats one. Rare anti-hype writing from a frontier lab. |\n\n### Staying Current (Without Drowning)\n\n| Resource | Level | Cost | Why it earns its place |\n|---|---|---|---|\n| [Hugging Face Daily Papers](https:\u002F\u002Fhuggingface.co\u002Fpapers) | 🔴 | Free | Community-curated daily paper feed. The practical successor to Papers with Code. |\n| [Ahead of AI](https:\u002F\u002Fmagazine.sebastianraschka.com) (Sebastian Raschka) | 🟡 | Free tier | Careful, technical breakdowns of what actually changed each month. No breathless takes. |\n| [Interconnects](https:\u002F\u002Fwww.interconnects.ai) (Nathan Lambert) | 🔴 | Free tier | The best writing on RLHF, post-training, and open models, from someone doing the work. |\n\n---\n\n## Track 2: Use AI Effectively (No Code Required)\n\nThe goal of this track: become the person in your team who gets 10x more out of AI tools than everyone else, and knows when NOT to trust them.\n\n### Level 1: AI Literacy (Understand What You're Using)\n\n| Resource | Level | Cost | Why it earns its place |\n|---|---|---|---|\n| [Elements of AI](https:\u002F\u002Fwww.elementsofai.com) | 🟢 | Free | Yes, it's in both tracks. Understanding what AI can and cannot do is the foundation of using it well. A few hours, life-long payoff. |\n| [Google AI Essentials](https:\u002F\u002Fgrow.google\u002Fai-essentials\u002F) | 🟢 | Free to learn | Under 10 hours, hands-on, built for people with zero technical background. You leave with skills you use the next day. |\n| [AI for Everyone](https:\u002F\u002Fwww.coursera.org\u002Flearn\u002Fai-for-everyone) (Andrew Ng) | 🟢 | Free to audit | The classic non-technical intro: what AI can do, what it can't, and how to spot real opportunities in your organization. No equations. |\n| [Anthropic AI Fluency: Framework & Foundations](https:\u002F\u002Fanthropic.skilljar.com) | 🟢 | Free + certificate | Co-developed with university professors. Teaches the 4D framework (Delegation, Description, Discernment, Diligence): judgment, not just prompts. Works for any AI tool, not just Claude. |\n\n### Level 2: Daily Work With AI (Prompting + Judgment)\n\n| Resource | Level | Cost | Why it earns its place |\n|---|---|---|---|\n| [Anthropic Prompt Engineering docs](https:\u002F\u002Fdocs.claude.com\u002Fen\u002Fdocs\u002Fbuild-with-claude\u002Fprompt-engineering\u002Foverview) | 🟡 | Free | The most systematic official prompting guide: clear structure, examples, roles, chain-of-thought. Techniques transfer to every major model. |\n| [OpenAI Academy](https:\u002F\u002Facademy.openai.com) | 🟢→🟡 | Free | Official self-paced courses and live workshops, from AI basics to working with agents and ChatGPT at work. |\n| [Prompt Engineering Guide (DAIR.AI)](https:\u002F\u002Fwww.promptingguide.ai) | 🟡 | Free | Vendor-neutral, research-grounded catalog of prompting techniques with papers behind each one. The reference you keep open in a tab. |\n| [One Useful Thing](https:\u002F\u002Fwww.oneusefulthing.org) (Prof. Ethan Mollick, Wharton) | 🟢 | Free | The most trusted research-based newsletter on AI's real impact on work and education. If you subscribe to one AI newsletter, make it this one. |\n| Co-Intelligence (Ethan Mollick, book) | 🟢 | Paid (~$20) | NYT bestseller and the sanest book on working WITH AI: treat it as a co-worker, stay the human in the loop, always invite it to the table. |\n| [More Useful Things: Prompt Library](https:\u002F\u002Fwww.moreusefulthings.com) (Wharton Generative AI Lab) | 🟢 | Free | Free, tested prompts for teaching, learning, and work from Mollick's lab. Steal shamelessly, then adapt. |\n\n### Level 3: Power User (Agents, Automation, Building Without Code)\n\n| Resource | Level | Cost | Why it earns its place |\n|---|---|---|---|\n| [Anthropic Academy](https:\u002F\u002Fwww.anthropic.com\u002Flearn) | 🟡 | Free + certificates | 16 self-paced courses launched in 2026: Claude 101, working with AI tools, Claude Cowork for non-developers, up to MCP and Claude Code for those who go further. |\n| [Microsoft Generative AI for Beginners](https:\u002F\u002Fmicrosoft.github.io\u002Fgenerative-ai-for-beginners\u002F) | 🟡 | Free | 21 structured lessons from prompting fundamentals to building simple AI apps. The gentlest bridge from \"user\" toward \"builder\". |\n| [DeepLearning.AI Short Courses](https:\u002F\u002Fwww.deeplearning.ai\u002Fshort-courses\u002F) | 🟡 | Free | Dozens of 1-2 hour courses (prompting, RAG, agents, evals) taught jointly with the companies that built the tools. Pick only what your work needs. |\n| [n8n Learning Path](https:\u002F\u002Fdocs.n8n.io\u002Flearning-path\u002F) | 🟡 | Free | Learn workflow automation properly, then plug AI into it. This is how AI stops being a chat window and starts saving hours weekly: email flows, reports, summaries, support. |\n\n**A rule that beats any course:** pick ONE real task from your actual job this week (a report, an analysis, an email backlog), and force yourself to do it with AI end to end. One finished real task teaches more than five certificates.\n\n---\n\n## Suggested 90-Day Paths\n\n**Track 1 (studying AI), if you can invest ~10 hrs\u002Fweek:**\n\n- Weeks 1-3: Elements of AI + Kaggle Python + 3Blue1Brown\n- Weeks 4-8: Andrew Ng ML Specialization + first Kaggle competition\n- Weeks 9-12: fast.ai part 1 OR Deep Learning Specialization\n- Then: Karpathy Zero to Hero, and let it hurt a little. That pain is the field entering your brain.\n\n**Track 2 (using AI), if you can invest ~3 hrs\u002Fweek:**\n\n- Weeks 1-2: Google AI Essentials + Elements of AI\n- Weeks 3-4: Anthropic AI Fluency + subscribe to One Useful Thing\n- Weeks 5-8: Anthropic prompting docs + rebuild 3 real work tasks with AI\n- Weeks 9-12: OpenAI Academy or Anthropic Academy track for your role + one small automation in n8n\n\n---\n\n## How to Spot AI Course Hype (Save Your Money)\n\nSkip anything that:\n\n1. Promises income (\"$10k\u002Fmonth with AI\") instead of skills\n2. Sells fear (\"AI will replace you unless you buy today\")\n3. Teaches only tool clicks with zero concepts (tool UIs change monthly, concepts don't)\n4. Has no named instructor with verifiable real-world work\n5. Shows screenshots of outputs but never explains methods\n6. Costs hundreds of dollars for content this list gives you free\n\nThe best AI education in the world is currently free. The scarce resource is not money. It is your consistency.\n\n---\n\n## Contributing\n\nFound a resource that meets the criteria above? Open an issue or PR. Resources that are paid, hype-driven, or unmaintained will not be accepted.\n\n## License\n\nMIT. Share it, translate it, teach with it.\n\n---\n\n*Last verified: July 2026. Links and course editions change; if something is dead, open an issue.*\n","这是一个面向2026年更新的AI学习资源导航指南，旨在为不同背景的学习者提供经过实践验证、无营销噱头的高质量AI学习路径。项目分为两条主线：一是面向AI\u002FML工程师与研究者的“学习AI作为学科”路径，涵盖数学基础、编程实践、模型构建等硬技能；二是面向职场人士、管理者与创作者的“有效使用AI”路径，强调零代码掌握主流AI工具。所有资源均经人工筛选，来源包括斯坦福、MIT、Hugging Face等一线机构与从业者，强调免费性、时效性与教学实效。适用于自学规划、教育机构课程参考及团队AI能力培养场景。","2026-07-18 02:30:05","CREATED_QUERY"]