survival-llm

Free download for Mac and Windows: brangus.io Code (MIT): anonnerdfrenzy/survival-llm

Came across this project that puts a local AI model and about 1.5 million medical, survival and how-to articles on your own computer, so it keeps working even with the internet off. Love seeing the self-sovereignty mindset from bitcoin show up in AI anonnerdfrenzy/survival-llm

台风天停电断网，手机没信号，想查个急救方法都查不到，那种慌你懂的 Survival LLM 把模型和资料库提前装进你电脑，离线也能问：维基、急救、FEMA 求生手册等约 150 万篇，回答都附原文出处可核对。Mac 和 Windows 都有，记得先联网装好。 当不了医生，但能让你少慌一点 anonnerdfrenzy/survival-llm

Someone packed ~930,000 Wikipedia articles, FEMA and CDC guides, and a local LLM into one free app that runs on a 16 GB laptop with no internet, and every answer shows its source the internet's down and you need to know if the food in the fridge is still safe, or how to treat a burn. Survival LLM is a free, MIT-licensed app built for that. The model and library live on your computer, so nothing you ask leaves it. Set it up online once, then it runs offline. The full library is about a 9 GB download, with ~930,000 of Wikipedia's most-read articles plus medical, FEMA and CDC guides. A 1.3 GB version leaves out Wikipedia and iFixit. It checks your RAM and suggests a model that fits (see card). Scores are from 22 blackout and first-aid questions graded blind by AI judges, on an earlier library build. The part we liked: every answer lists the articles it read, and clicking one opens it with the exact passage highlighted. The README admits some models cited sources they never opened or even saw, so the app now swaps in the articles it actually read. Repo: anonnerdfrenzy/survival-llm What would you add to the library?

断网状态下也能在自己电脑上查急救、供水、食物储存这类生存问题，并给出可核对的原文出处。 anonnerdfrenzy/survival-llm 硬盘上带着模型和整套资料过日子，装完拔了网线照样能问：手边的 Ollama 负责生成，约 150 万篇（精简版 50 万）文章进 SQLite FTS5 索引，BM25 排序时给官方来源加分。 t.co

ИИ-помощник для работы без интернета Появилась бесплатная модель, которая работает офлайн и помогает в ситуациях, когда связи нет. Survival LLM знает, как очистить воду, оказать первую помощь и сохранить продукты без холодильника. Модель обучена... anonnerdfrenzy/survival-llm

Survival LLM: Your Offline AI Lifeline for Any Emergency! --- title: Survival LLM: Your Offline AI Lifeline for Any Emergency! published: True tags: offline, emergency, llm, open --- Quick Summary: 📝 Survival LLM is a free, offline AI assistant and survival library that runs locally on your computer. It answers questions by searching an extensive offline library of medical, emergency, and how-to articles, providing cited sources and detailed reasoning for its responses. Key Takeaways: 💡 - ✅ Survival LLM is a free, offline AI assistant with a vast local library for emergency and survival information. - ✅ It operates entirely on your computer, ensuring privacy and functionality without an internet connection. - ✅ Answers are provided with direct source citations, allowing users to verify information easily. - ✅ Users can add their own documents (PDFs, manuals) to its knowledge base and benefit from its personalized memory function. - ✅ It's an essential tool for emergency preparedness, offering reliable information when traditional sources might fail. Project Statistics: 📊 - ⭐ Stars: 691 - 🍴 Forks: 56 - ❗ Open Issues: 6 Tech Stack: 💻 - ✅ Python Hey fellow developers! I just had to share something truly remarkable I found on GitHub that I think you'll all be incredibly excited about. Imagine a scenario: the power's out, the internet's down, and you desperately need to know how to keep food safe, treat a burn, or purify water. What do you do when your usual online resources are gone? This is where Survival LLM swoops in as an absolute game-changer! Survival LLM is a free, open-source AI assistant and an entire survival library that runs 100% on your own computer, requiring absolutely no internet connection after its initial setup. Think about that for a second – a powerful AI, ready to assist you with critical information, completely offline and private. It's like having a knowledgeable expert and an entire emergency encyclopedia at your fingertips, even in the middle of nowhere. So, how does it work? You simply ask it questions, just like you would with any modern AI. But instead of querying the cloud, it searches an extensive offline library packed with medical, emergency, how-to guides, and encyclopedia articles. It then gives you clear answers and, here's the best part for us discerning developers, it shows you the exact passage each answer came from. No more guessing or wondering about sources – you can click right through to the original article, complete with its license. You can even ask it to 'show its reasoning' to see the AI's step-by-step thought process! The benefits for anyone, especially developers who appreciate robust, self-contained solutions, are immense. First, it's incredibly private; your questions and data never leave your machine. Second, its reliability in emergencies is unmatched because it's internet-independent. Third, you can customize it! Drop in your own PDFs, Word documents, or web pages – like your generator's manual or your family's emergency plan – and the AI will incorporate them into its searchable knowledge base. It even has a 'memory' feature, remembering details you tell it for more personalized advice. This isn't just a cool tech demo; it's a practical, life-saving tool built for real-world scenarios. From a curated library of 1.5 million articles covering everything from Wikipedia's most-read to FEMA survival guides, Survival LLM is meticulously designed for preparedness. The setup takes a bit of time to download the models (which can be quite large, up to 80GB for the biggest model), but once it's done, you're set for good. This project embodies the spirit of open-source innovation, providing a vital resource that empowers individuals with knowledge when it matters most. Seriously, go check it out – it's a must-have for anyone concerned about preparedness or simply fascinated by the power of local AI! Learn More: 🔗 View the Project on GitHub --- 🌟 Stay Connected with GitHub Open Source! > 📱 Join us on Telegram > Get daily updates on the best open-source projects > GitHub Open Source > 👥 Follow us on Facebook > Connect with our community and never miss a discovery > GitHub Open Source

پروژه Survival LLM یک دستیار هوش مصنوعی رایگان و کاملاً آفلاین برای بقاست: کتابخانه‌ای از مقالات پزشکی، اورژانسی و دایره‌المعارفی که روی کامپیوتر خودت اجرا میشه، بدون نیاز به اینترنت. پاسخها با منبع، امکان افزودن PDF/Word/EPUB شخصی و آپدیت 🔗گیتهاب : anonnerdfrenzy/survival-llm t.co

The readme on Github does say the system produces hallucinated sources anonnerdfrenzy/survival-llm, but really you're correct, for this kind of highly targeted use case the LLM's summary could probably be ignored entirely and the system could work extremely well even just as a search engine: index all of the content up front, chunked by paragraph and maybe enriched with a HyDE step, have the LLM facilitate running a handful of hybrid keyword/semantic searches based on the user's question and reranking the results, and show the exact pieces of text in the corpus that are most responsive to their question. No summary needed.

70% of Philippine AI Adopters Are Still on Basic Use More than 396,000 businesses in the Philippines now use AI, 32 percent of firms, up from 21 percent a year earlier (Source: Strand Partners for AWS, 2026). Seventy percent of those adopters are still limited to public chatbots, routine productivity apps, and off-the-shelf AI products (Source: Strand Partners for AWS, 2026). The adoption number is real. The depth is not. !Infographic The Adoption Number Is Real The AWS-commissioned study surveyed 1,000 Philippine business leaders and 1,000 members of the public (Source: Strand Partners for AWS, 2026). A Philippine Institute for Development Studies study found only 14.9 percent of firms use AI tools, concentrated among large urban firms in ICT and BPO (Source: PIDS, 2024). The figures use different samples and definitions. The hardware is already in place. 90.8 percent of establishments own computers and 81 percent have internet access (Source: PIDS, 2024). MSMEs were 99.63 percent of registered establishments in 2023 and 66.97 percent of employment (Source: DTI, 2024). Hardware is not the bottleneck. The Depth Problem Among businesses already using AI, adoption splits three ways: 70 percent sit at basic use, 18 percent have integrated AI across multiple functions, and 12 percent run custom AI systems, agentic AI, or autonomous technology (Source: Strand Partners for AWS, 2026). Among those adopters, 74 percent report productivity gains and 68 percent say revenue rose, with average revenue growth of 20 percent (Source: AWS press release, 2026). Yet 57 percent are still experimenting, only 17 percent have a formal scaling roadmap, and just 12 percent have a comprehensive AI governance framework (Source: AWS press release, 2026). Digital skills shortages topped the barrier list at 47 percent (Source: AWS press release, 2026). Compare a scaled deployment. Store Genie, built by PLDT Home and Smart Communications with Amdocs, supports more than 750 frontline agents across 149 locations and cut some order workflows from hours to minutes (Source: AWS via AI4Philippines, 2026). That is the shape of the 12 percent. What the First AI Bill Looks Like Most Philippine SMEs will start with a subscription. Google Workspace Business Starter runs $7.00 per user per month on a one-year commitment, Business Standard is $14.00 per user per month, and Business Plus is $22.00 per user per month, billed in US dollars with tax excluded (Source: Google, 2026). A five-person microenterprise on Starter spends $35 a month, or $420 a year, roughly PHP 26,700 at the Bangko Sentral reference rate of PHP 63.60 to the dollar (Source: BSP, 2026). The DTI classifies micro enterprises as those holding up to PHP 3 million in assets with one to nine employees (Source: DTI, 2024), so this is the typical budget for the country's smallest businesses. Spending money is easy. Deciding which workflow to hand to a tool, and how to tell whether it worked, is the part that needs a plan. Local Models and the Zero-Subscription Route Some AI tools now run on hardware you already own. Survival LLM is an open-source assistant and reference library that runs on your own computer and keeps working with the internet off, so questions never leave the machine (Source: Survival LLM, 2026). Its library carries roughly 1.5 million articles from Wikipedia, WikiMed, MedlinePlus, CDC, FEMA, and WHO (Source: Survival LLM, 2026). Model choice follows available memory. 16 to 31 GB of RAM gets an IBM Granite 4.2 8B model on a 6 GB download, while 128 GB gets Nemotron 3.5 Super on 80 GB (Source: Survival LLM, 2026). In the project's blind evaluation against USDA, CDC, FEMA, EPA, NHS and WHO guidance, Nemotron Super scored 86 percent accuracy with zero harmful answers, Qwen3.8 27B scored 80 percent, and Granite 4.2 8B scored 55 percent (Source: Survival LLM, 2026). The business case is narrow. Subscription cost falls to zero and customer data never leaves the building, but the cost moves to hardware and attention: a 16 GB machine gives you a 55 percent accurate model on a narrow question set, not a general assistant. The Part Nobody Plans: Personal Data The National Privacy Commission issued Advisory No. 2024-04 on December 19, 2024, setting out how the Data Privacy Act of 2012 applies to AI systems that process personal data (Source: NPC, 2024). NPC Circular No. 2022-04 covers registration of data processing systems, notification of automated decision-making or profiling, and designation of a data protection officer (Source: NPC, 2022). For an SME the practical reading is short. Put a customer's name, phone number, or order history into a cloud chatbot and you are processing personal data with an AI system, with registration and notification obligations whether or not the tool is sold as "just a chatbot". Local models shrink that exposure, though they do not replace compliance. The Training That Reaches MSMEs The ILO's AI for Digital Novice MSMEs program under the Digital PINAS joint programme reached TESDA Region 2, where staff adapted the content with 25 trainers and curriculum developers from Cagayan, Isabela, and Nueva Vizcaya (Source: ILO, 2026). Tested between July and August 2026, Cookery NC II trainees used AI to develop menus, calculate ingredient costs, and prepare social media posts, and some began selling homemade food and catering services while still in training (Source: ILO, 2026). Electrical Installation and Maintenance NC II trainees used the same tools to draft client proposals, work out prices, and package their services more professionally (Source: ILO, 2026). Look at what those tasks are: pricing, proposals, costing. Revenue-adjacent work with a measurable before and after. FAQ Q: Is 32 percent AI adoption good or bad for the Philippines? A: Measured against the 14.9 percent of firms reporting AI tool use in the PIDS study, the gap is narrowing (Source: PIDS, 2024). Measured against the 70 percent of adopters still on basic use, the country is early. Q: What does the Data Privacy Act require of a small business using AI? A: If an AI system processes personal data, NPC Advisory No. 2024-04 says the Data Privacy Act of 2012 applies to it (Source: NPC, 2024). Registration and notification of automated decision-making or profiling are covered by NPC Circular No. 2022-04 (Source: NPC, 2022). Key Takeaway Thirty-two percent of Philippine businesses use AI. Seventy percent of them are still on basic use, 57 percent are still experimenting (Source: AWS press release, 2026), and 47 percent blame skills rather than software (Source: AWS press release, 2026). The economics are not the obstacle. A five-person shop spends about PHP 26,700 a year on a full AI seat bundle, or nothing at all on a local model (Source: Google, 2026). The gap is between owning a tool and owning a workflow worth measuring. It closes where it always closes: a person who knows which task to hand over, a number that shows whether it worked, and training in the local language and business context. TESDA Region 2 put AI into a cookery classroom in Northern Luzon, and its trainees began selling food. So pick one recurring task, write down what it costs you now, and run a single month of one tool against it. Which task in your business would you measure first? Sources - PH businesses' AI adoption up in 2026 (Strand Partners for AWS via Philippine News Agency, 2026) - AWS Report: Philippines' AI Adoption Jumps 52% in One Year (AWS press release via SDN Chronicle, 2026) - AI adoption among PH businesses jumps 52% in a year (Newsbytes.PH, 2026) - AWS Report Finds Philippine AI Adoption Up to 32 Percent as Skills Gaps Persist (AWS study coverage via AI4Philippines, 2026) - Readiness for AI Adoption of Philippine Business and Industry, Discussion Paper 2024-35 (PIDS, 2024) - 2023 Philippine MSME Statistics in Brief (DTI, 2024) - Compare Flexible Pricing Plan Options (Google, 2026) - Survival LLM (Survival LLM project, 2026) - AI training helps turn technical skills into enterprise opportunities (ILO, 2026) - Advisories and Circulars (National Privacy Commission, 2026) - Bangko Sentral ng Pilipinas Exchange Rate (BSP, 2026)