LLM News and Articles
| Monday, 2025-12-22 | ||||
| 18:09 | Manifesto Visi Epsilon: Mengunci Masa Depan, Berdaulat dalam Informasi https://medium.com/@farid.al.q/manifesto-visi-epsilon-mengunci-masa-depan-berdaulat-dalam-informasi-2d3b105f0c20 | |||
| 18:06 | Microsoft Azure Hackathon’s 2025 — Recap https://ajay-arunachalam08.medium.com/microsoft-azure-hackathons-2025-recap-8234c203b7a1 | |||
| 18:02 | Fine-Tune LLaMA-8B for Medical AI in Under 2 Hours on a 16GB GPU (Full Code Included) https://pub.towardsai.net/fine-tune-llama-8b-for-medical-ai-in-under-2-hours-on-a-16gb-gpu-full-code-included-17816b7a36d6 | |||
| 17:56 | Exploring Agentic AI: From Multi-Agent Systems to Collaborative Workflows — Understanding the… https://medium.com/@sathishkumar.babu89/exploring-agentic-ai-from-multi-agent-systems-to-collaborative-workflows-understanding-the-6ef78afc1bf8 | |||
| 17:56 | QwenLong‑L1.5‑30B‑A3B: Inside the 4M‑Token Memory Agent That Thinks Like GPT‑5 https://medium.com/data-science-in-your-pocket/qwenlong-l1-5-30b-a3b-inside-the-4m-token-memory-agent-that-thinks-like-gpt-5-810eac6dfa1f | |||
| 17:56 | The Era of 1-Bit LLMs: Why This Paper Changed How I Think About Large Language Models https://medium.com/@visnus12a22223/the-era-of-1-bit-llms-why-this-paper-changed-how-i-think-about-large-language-models-e9bae69071bb | |||
| 17:49 | Re-Ranking in RAG: Small Recall Loss, Improving Precision — Jina AI vs Cohere https://medium.com/@bojanmakivic_72327/re-ranking-in-rag-small-recall-loss-improving-precision-jina-ai-vs-cohere-ffc9d6349947 | |||
| 17:45 | Simpul Kesadaran: Refleksi atas Penaklukan Banjir Informasi https://medium.com/@farid.al.q/simpul-kesadaran-refleksi-atas-penaklukan-banjir-informasi-ac374cf208dd | |||
| 16:51 | Guide to GPU Requirements for Inference of LLMs https://medium.com/@shayanBemanian/guide-to-gpu-requirements-for-inference-of-llms-d2e52f1d7e6d | |||
| 16:49 | Designing the Invisible: UX in the Age of AI https://medium.com/@mandeepkaur1/designing-the-invisible-ux-in-the-age-of-ai-361a239e3b15 | |||
| 16:37 | Architectural Taxonomy of Modern AI Systems: Modular, Vertical, Agentic, and Hybrid Implementations https://medium.com/@nraman.n6/architectural-taxonomy-of-modern-ai-systems-modular-vertical-agentic-and-hybrid-implementations-5f3f0f0d95b9 | |||
| 16:26 | Your AI Will Obey Another AI — Even If It Refuses You https://pub.aimind.so/your-ai-will-obey-another-ai-even-if-it-refuses-you-52a25b94a1e1 | |||
| 16:15 | From passive AI agent to proactive personal assistant https://medium.com/@oleksandr.poberezhnyi/from-passive-ai-agent-to-proactive-personal-assistant-5ac2b9ce26e7 | |||
| 16:05 | Using an MCP Server with LangGraph: A Practical Guide to MCP Adapters https://medium.com/@termtrix/using-an-mcp-server-with-langgraph-a-practical-guide-to-mcp-adapters-3645b86f2324 | |||
| 16:02 | 21 Chunking Strategies That Will Fix Your Broken RAG System https://pub.towardsai.net/21-chunking-strategies-that-will-fix-your-broken-rag-system-14dac3f2b067 | |||
| 16:02 | Schema-First LLM Pipelines That Finally Behave https://medium.com/@connect.hashblock/schema-first-llm-pipelines-that-finally-behave-276fe326d125 | |||
| 15:50 | Vector Embeddings In Snowflake https://medium.com/@kalyankumar36952/vector-embeddings-in-snowflake-dbf82ec2d1b5 | |||
| 15:48 | Understanding the Context Window in Large Language Models https://medium.com/@kalyankumar36952/understanding-the-context-window-in-large-language-models-a0a04a1f271d | |||
| 15:46 | Why Reverse Prompting Is the Shortcut Most AI Engineers Won’t Tell You About https://jinlow.medium.com/why-reverse-prompting-is-the-shortcut-most-ai-engineers-wont-tell-you-about-c1c8b4158de7 | |||
| 15:35 | Google’s 2,000 Word Reality Check https://medium.com/@Michael38/googles-2-000-word-reality-check-06c77713c25c | |||
| 15:30 | How LLMs Are Trained (Without the Scary Stuff) https://ai.plainenglish.io/how-llms-are-trained-without-the-scary-stuff-af36846334e1 | |||
| 15:28 | LLMs Are the New Operating System https://medium.com/@pratikchaudhariworks/llms-are-the-new-operating-system-f0694da21bb1 | |||
| 15:07 | The Hybrid Fraud Detection Engine https://medium.com/@im_jatintyagi/system-design-case-study-the-hybrid-fraud-detection-engine-bd2bc59955ef | |||
| 15:02 | Philosophy of Durable Agents https://pub.towardsai.net/philosophy-of-durable-agents-233cc349a1a1 | |||
| 15:02 | Sam Altman Is Right: Wrappers Will Die. He Just Forgot He Built One. https://infusedata.io/sam-altman-is-right-wrappers-will-die-he-just-forgot-he-built-one-e665253146f0 | |||
| 14:42 | 2026: The Year AI Becomes Autonomous – LLMs, Agentic AI, GPUs & the New Semiconductor Era https://medium.com/@Intellibytes/2026-the-year-ai-becomes-autonomous-llms-agentic-ai-gpus-the-new-semiconductor-era-3b6db9c84b5f | |||
| 14:35 | Transformers: A Practical Understanding https://pub.towardsai.net/transformers-a-practical-understanding-0bf53df14746 | |||
| 14:30 | RAG Explained: Why Large Language Models Need a Bucket, Not Just an Ocean https://medium.com/@9kyuugirl/rag-explained-why-large-language-models-need-a-bucket-not-just-an-ocean-a4dbb03dd538 | |||
| 13:48 | What are GPUs made up of, and why are they so fast for machine learning workflows? https://viraajkadam.medium.com/what-are-gpus-made-up-of-and-why-are-they-so-fast-for-machine-learning-workflows-66de567b6ef1 | |||
| 13:02 | Beyond Parameter Counts: The Architectural Engineering of SLM Inference Pipelines https://sagar-awasthi.medium.com/beyond-parameter-counts-the-architectural-engineering-of-slm-inference-pipelines-7b9b3e727fbf | |||
| 12:45 | Forget the AI Leaderboards: These 4 Trends Actually Matter https://medium.com/@alexbuzunov/forget-the-ai-leaderboards-these-4-trends-actually-matter-75a509d0b58b | |||
| 12:19 | Attention vs. Memory: Why Transformers Killed the RNN https://medium.com/@satyamrai3362/attention-vs-memory-why-transformers-killed-the-rnn-58f3f705ede8 | |||
| 12:02 | The Verbification of Knowledge Tools and the Pandora’s Box of Epistemology https://medium.com/@dolphin.exe/the-verbification-of-knowledge-tools-and-the-pandoras-box-of-epistemology-d5a6cc871dda | |||
| 11:32 | 10 Graph + Vector Fusion Patterns for Hard QA https://medium.com/@sparknp1/10-graph-vector-fusion-patterns-for-hard-qa-d211ec1b9055 | |||
| 11:32 | Token Budgeting: 10 Tricks for Smarter Prompts https://medium.com/@jickpatel611/token-budgeting-10-tricks-for-smarter-prompts-2040494a8912 | |||
| 11:12 | RAG vs Fine-Tuning: Choosing the Right Architecture for AI Applications https://medium.com/@adraj5949/rag-vs-fine-tuning-choosing-the-right-architecture-for-ai-applications-ce20823add0f | |||
| 11:04 | Why We Still Need Data Science in the Age of LLMs https://siddhantjain-89608.medium.com/why-we-still-need-data-science-in-the-age-of-llms-b26abc88c26d | |||
| 10:57 | Search Interest Score (SIS): A Methodology for Estimating Semantic Visibility in Generative Search… https://medium.com/@wajih.benrissoul/search-interest-score-sis-a-methodology-for-estimating-semantic-visibility-in-generative-search-1aa6d3d71061 | |||
| 10:53 | Rethinking Prompts: Multilingual Experiments In LLM Token Optimization https://medium.com/softserve-technical-communication/rethinking-prompts-multilingual-experiments-in-llm-token-optimization-3b871f7f1ba2 | |||
| 10:46 | After AI Knowing Means Doing Detailed storytelling is real working (Sanjoy Nath's Qhenomenology… https://medium.com/@sanjoy_nath/after-ai-knowing-means-doing-detailed-storytelling-is-real-working-sanjoy-naths-qhenomenology-3b949d33a406 | |||
| 10:42 | The AI-Bias 10:10 https://medium.com/@balapriya1801/the-ai-bias-10-10-c6d2c5295273 | |||
| 10:33 | Build & Test LLM Apps Locally with Ollama https://yehancha.medium.com/build-test-llm-apps-locally-with-ollama-4d99c5405ed6 | |||
| 10:01 | Google Introduces A2UI (Agent-to-User Interface): An Open Sourc Protocol for Agent Driven Interfaces https://www.marktechpost.com/2025/12/22/google-introduces-a2ui-agent-to-user-interface-an-open-sourc-protocol-for-agent-driven-interfaces/ | |||
| 09:52 | HERKES İÇİN BİR TUTAM VLM SERİSİ — 4 https://medium.com/@kasim.yildirimm10/herkes-i%CC%87%C3%A7i%CC%87n-bi%CC%87r-tutam-vlm-seri%CC%87si%CC%87-4-85ea94f033fe | |||
| 09:48 | Graph RAG for Legal Reasoning: Multi-Hop Insights with Knowledge Graphs, Vector Search, and LLMs https://medium.com/47billion/graph-rag-for-legal-reasoning-multi-hop-insights-with-knowledge-graphs-vector-search-and-llms-ae30032b0f06 | |||
| 09:33 | Are You a Luddite? https://ed-burton.medium.com/are-you-a-luddite-e7986910401c | |||
| 09:10 | Introduction to Retrieval-Augmented Generation (RAG) Chatbots: A Simple Guide https://medium.com/@sathishkumar.babu89/introduction-to-retrieval-augmented-generation-rag-chatbots-a-simple-guide-924dbb1fee28 | |||
| 08:28 | SuperAI 2025: From Buzzwords to Business Reality https://medium.com/@ashishbodla/superai-2025-from-buzzwords-to-business-reality-df6f2cfa74ed | |||
| 08:18 | Everything you need to know to sound like an LLM expert https://medium.com/@baronlior/everything-you-need-to-know-to-sound-like-an-llm-expert-bfb4075e60ae | |||
| 08:10 | From Bag of Words to Language Models: How NLP Really Started https://medium.com/@goktugdagi/from-bag-of-words-to-language-models-how-nlp-really-started-9a04c5844d1c | |||
| 08:08 | Revolutionizing LLM Prompting: Self-Supervised Optimization Saves Time and Money https://iamdgarcia.medium.com/revolutionizing-llm-prompting-self-supervised-optimization-saves-time-and-money-58c458d0b91b | |||
| 08:07 | Meet Your AI Coworker: Friend or Competitor? https://ethical-hacking-kolkata.medium.com/meet-your-ai-coworker-friend-or-competitor-f2c1661098cd | |||
| 08:03 | Show HN: LLM Politeness Study (hostile and effusive tones boost LLM creativity) https://aklodhi98.github.io/llm-politeness-study/ | |||
| 08:01 | The Ghost Town in the Machine: Why I Redesigned the Way AI “Thinks” https://medium.com/@imsarthakshrma/the-ghost-town-in-the-machine-why-i-redesigned-the-way-ai-thinks-195d08356164 | |||
| 07:57 | Simple LLM Tool Calling in Laravel using Prism https://medium.com/@brice_hartmann/simple-llm-tool-calling-in-laravel-using-prism-92cc64b0d69c | |||
| 07:45 | Day 5: Transformers- The Architecture That Changed AI https://medium.com/@SomJaiswal/day-5-transformers-the-architecture-that-changed-ai-ec12221b5f83 | |||
| 07:32 | The Pragmatic Engineer Survey Exposes Failure of First-Wave AI https://medium.com/@artiquare/the-pragmatic-engineer-survey-exposes-failure-of-first-wave-ai-93c86b2303c1 | |||
| 07:28 | Day 4: Neural Networks-The Brain Behind AI https://medium.com/@SomJaiswal/day-4-neural-networks-the-brain-behind-ai-2a5a25899f3e | |||
| 07:24 | Neural Networks Intuitions: 21. Reasoning in LLMs https://raghul-719.medium.com/neural-networks-intuitions-21-reasoning-in-llms-917aafd1eead | |||
| 07:16 | vLLM, Paged Attention and KV Cache — Optimizing LLM Serving for Modern AI Systems https://medium.com/@ppartha39/vllm-paged-attention-and-kv-cache-optimizing-llm-serving-for-modern-ai-systems-f9101a9a981b | |||
| 06:42 | How AI Overviews and LLM Search Are Rewriting Click-Through Dynamics? https://medium.com/@buriedagency/how-ai-overviews-and-llm-search-are-rewriting-click-through-dynamics-9d37a8bcc2c5 | |||
| 06:42 | DeepSeek Breakthrough In 2025 https://medium.com/mlworks/deepseek-breakthrough-in-2025-5981e0d21999 | |||
| 06:39 | What It Costs to Annotate 1,000 Full Papers with LLM: Tokens vs. Electricity Cost https://medium.com/@frederickpi1969/what-it-costs-to-annotate-1-000-full-papers-with-llm-tokens-vs-electricity-cost-f8ee9b8af09b | |||
| 06:33 | How LLMs Decide What to Cite (Clear, Practical Explanation) https://medium.com/@zeeshanhaiderjhang01/how-llms-decide-what-to-cite-clear-practical-explanation-f6cfce5e7a6c | |||
| 06:25 | Ram prices skyrocketing https://medium.com/coding-nexus/ram-prices-skyrocketing-0728f9a1e806 | |||
| 06:18 | Jobs that (soon) no longer exist https://medium.com/@im-sanka/jobs-that-soon-no-longer-exist-63247beb4439 | |||
| 06:07 | Why Million-Token AI Is So Expensive (And How Mamba Fixes It) https://ai.plainenglish.io/why-million-token-ai-is-so-expensive-and-how-mamba-fixes-it-a69c849d580c | |||
| 05:20 | LLM Communications in the Wild https://medium.com/@maspinwall22/llm-communications-in-the-wild-1f57a06642ae | |||
| 05:02 | AI Agent Learning Roadmap : From Zero to Multi-Agent Systems https://manalisomani099.medium.com/ai-agent-learning-roadmap-from-zero-to-multi-agent-systems-7d20888a039f | |||
| 04:49 | Open‑Source AI’s New Dawn: Nemotron, DeepSeek, and the Quiet War for the Future of Intelligence https://medium.com/@rogt.x1997/open-source-ais-new-dawn-nemotron-deepseek-and-the-quiet-war-for-the-future-of-intelligence-f3681dc1307c | |||
| 04:24 | Thermodynamic Detection of Irreversible Phase Transitions for No-Meta Agents (Instrumentation… https://medium.com/@omanyuk/thermodynamic-detection-of-irreversible-phase-transitions-for-no-meta-agents-instrumentation-d4a04e48f7d2 | |||
| 04:15 | The Psychology of AI: New Study Reveals “Synthetic Trauma” In ChatGPT And Gemini https://medium.com/@babarranjha/the-psychology-of-ai-new-study-reveals-synthetic-trauma-in-chatgpt-and-gemini-71438dcdc23b | |||
| 04:02 | Building a Self-Hosted LLM Server You’ll Actually Use https://medium.com/@contactabhinaav/building-a-self-hosted-llm-server-youll-actually-use-dae13111447a | |||
| 03:59 | Transformers v5 Tokenizers: From Black Boxes to Customizable Architectures https://medium.com/coding-nexus/transformers-v5-tokenizers-from-black-boxes-to-customizable-architectures-6b92fd50ae3f | |||
| 03:55 | Anthropic’s Bloom: A Practical Way to Measure AI Misalignment (Without Hand-Labelling Everything) https://medium.com/coding-nexus/anthropics-bloom-a-practical-way-to-measure-ai-misalignment-without-hand-labelling-everything-8a30c7974a21 | |||
| 03:43 | I Spent ,000 to Replace AI Coding Subscriptions. I Was (Mostly) Wrong. https://medium.com/coding-nexus/i-spent-4-000-to-replace-ai-coding-subscriptions-i-was-mostly-wrong-6afd8ddc569a | |||
| 03:41 | Day 14: 21 Days of Building a Small Language Model: Positional Encodings https://devopslearning.medium.com/day-14-21-days-of-building-a-small-language-model-positional-encodings-db0ae45e0b8e | |||
| 03:22 | Local LLMs 101: What Actually Happens When You Run a Model on Your Machine https://medium.com/coding-nexus/local-llms-101-what-actually-happens-when-you-run-a-model-on-your-machine-8efccc0cd0ac | |||
| 03:09 | Hinton and 3 students https://medium.com/@achernomorov/hinton-and-3-students-1f412ec6316b | |||
| 03:03 | LLM-Judge, BLEU, ROUGE, and Perplexity https://medium.com/@SuriNaren/llm-judge-bleu-rouge-and-perplexity-dc200ff03102 | |||
| 02:53 | Before We Forget What We Are https://medium.com/@ktiyab_42514/before-we-forget-what-we-are-de1e8764eab0 | |||
| 02:38 | Building Smarter Knowledge Graphs with Tree-KG https://medium.com/ai-exploration-journey/building-smarter-knowledge-graphs-with-tree-kg-7dc93b9e8dc6 | |||
| 02:32 | LLMs vs Traditional Search Engines: What’s the Real Difference? https://medium.com/@itsamanyadav/llms-vs-traditional-search-engines-whats-the-real-difference-b9830857366f | |||
| 02:09 | How to Convert Excel to CSV — and Use LLMs to Extract Hidden Information from a Single Column https://sumtsui.medium.com/how-to-convert-excel-to-csv-and-use-llms-to-extract-hidden-information-from-a-single-column-72343aac8af3 | |||
| 01:50 | A scientific guide to no-meta epistemic irreversibility under finite memory https://medium.com/@omanyuk/a-scientific-guide-to-no-meta-epistemic-irreversibility-under-finite-memory-bba07f6671f1 | |||
| 00:39 | Ankara, Sıradan Bir Belediye Midir? Sıfırdan Eğittiğim Yapay Zekanın Yolculuğu https://medium.com/@ersingorun/ankara-s%C4%B1radan-bir-belediye-midir-s%C4%B1f%C4%B1rdan-e%C4%9Fitti%C4%9Fim-yapay-zekan%C4%B1n-yolculu%C4%9Fu-2b557634c4d6 | |||
| 00:09 | LLMs and Semantic Meaning https://medium.com/@jallenswrx2016/llms-and-semantic-meaning-754b90cee063 | |||
| Sunday, 2025-12-21 | ||||
| 23:22 | Master’s Degree Didn’t Teach Me About AI https://medium.com/tech-and-me/masters-degree-didn-t-teach-me-about-ai-3a3c1020a655 | |||
| 23:21 | The Great Frictionless Lie: Why We Are Losing Our Grip on Reality https://mycelialmirror.medium.com/the-great-frictionless-lie-why-we-are-losing-our-grip-on-reality-a87aa224ea09 | |||
| 23:08 | How a Brain-Inspired Spreading Activation Algorithm Boosts AI Knowledge Retrieval https://medium.com/@cs_maverick/how-a-brain-inspired-spreading-activation-algorithm-boosts-ai-knowledge-retrieval-c7f7ebb9286d | |||
| 22:51 | The Ultimate DevOps & AI Manifesto: From Hype to Production Reality https://medium.com/@bozyol/the-ultimate-devops-ai-manifesto-from-hype-to-production-reality-3ede8a65b434 | |||
| 22:42 | TPU: The Specialized Heart of the Generative AI Revolution https://ai.plainenglish.io/tpu-the-specialized-heart-of-the-generative-ai-revolution-f8649dbe7023 | |||
| 22:39 | Under the Hood of MCP — The Architecture of AI’s New Standard https://medium.com/@puneeth01062002/under-the-hood-of-mcp-the-architecture-of-ais-new-standard-bbfb0afefd07 | |||
| 22:21 | Enterprise Version—Train a Large Language Model https://shilpathota.medium.com/enterprise-version-train-a-large-language-model-1a537e64866e | |||
| 21:58 | The Real AI Risk Isn’t Hallucination. It’s Verified Bias. https://medium.com/@basilpuglisi/the-real-ai-risk-isnt-hallucination-it-s-verified-bias-8455a8c31473 | |||
| 21:51 | AI Rewind 2025 https://medium.com/@hamzamlwh/ai-rewind-2025-8d7c8bdce2c5 | |||
| 21:28 | From MVVM to RAG: A Senior Architect’s Journey into Generative AI https://medium.com/@ktlint/from-mvvm-to-rag-a-senior-architects-journey-into-generative-ai-ff8859a53bc3 | |||
| 21:22 | Talkspace ($TALK): The Rise of Clinical-Grade AI Supervision https://startupsapience.medium.com/talkspace-talk-the-rise-of-clinical-grade-ai-supervision-f0acba01b134 | |||
| 21:17 | I Built a Language Model in C So I Couldn’t Hide Behind Vibes https://medium.com/@dipghoshraj/i-built-a-language-model-in-c-so-i-couldnt-hide-behind-vibes-f60b6caa7e67 | |||
| 20:50 | OpenAI's profit margins surge to 70% as enterprise grows https://www.perplexity.ai/page/openai-s-profit-margins-surge-u2XDj2J8Sc6Pdc1.H32EyA | |||
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241124