LLM News and Articles
| Thursday, 2025-12-11 | ||||
| 07:35 | External Reasoning Drift in Enterprise Finance Platforms: A Governance Risk Hidden in Plain Sight https://medium.com/@tim_62250/external-reasoning-drift-in-enterprise-finance-platforms-a-governance-risk-hidden-in-plain-sight-38a9ee4e40df | |||
| 07:32 | The “Trust Wall” for AI Agents https://medium.com/@bhagyarana80/the-trust-wall-for-ai-agents-8bb2f19d9fc6 | |||
| 07:20 | The GenAI Coffee Break: Beyond the Hype [Part-2] https://generativeai.pub/the-genai-coffee-break-beyond-the-hype-part-2-b09e8a1b6d43 | |||
| 07:12 | Understanding AI Hallucinations: The Geometrical Distance of a Thought https://medium.com/@skanga/understanding-ai-hallucinations-the-geometrical-distance-of-a-thought-9b8328a2b8c7 | |||
| 07:11 | From Neurons to Neural Networks https://medium.com/@panData/from-neurons-to-neural-networks-87364d9ebc3a | |||
| 07:02 | Building LLMs From Scratch (Part 8): Causal Attention https://soloshun.medium.com/building-llms-from-scratch-part-8-causal-attention-6e4a0578c88c | |||
| 06:48 | Harvard Loop: AI-Powered Lost & Found https://medium.com/institute-for-applied-computational-science/harvard-loop-ai-powered-lost-found-048f58a0a0c2 | |||
| 06:41 | Agent, Context, and Data Platform We Need https://generativeai.pub/agent-context-and-data-platform-we-need-b86809c9f40a | |||
| 06:21 | Inside LLM Agents: How AI Workers Plan, Remember, and Act in the Real World https://medium.com/@manjunatha.inti/inside-llm-agents-how-ai-workers-plan-remember-and-act-in-the-real-world-c68be59836f1 | |||
| 05:34 | Why the Model Context Protocol (MCP) is the “USB-C” moment for AI and why every student needs to… https://medium.com/@vijay.balaji/why-the-model-context-protocol-mcp-is-the-usb-c-moment-for-ai-and-why-every-student-needs-to-881659dff469 | |||
| 04:48 | PredictBG: A New Era of Diabetes Care With Real-Time Predictions and Personalized Guidance https://medium.com/institute-for-applied-computational-science/predictbg-a-new-era-of-diabetes-care-with-real-time-predictions-and-personalized-guidance-ca24fb07e1c8 | |||
| 04:45 | 11 ChatGPT Prompts That Helped Me Make My First 00 Online https://medium.com/everyday-ai/11-chatgpt-prompts-that-helped-me-make-my-first-1000-online-839d27c4d74c | |||
| 04:37 | OpenAI (2015) https://openai.com/index/introducing-openai/ | |||
| 04:32 | Why You Need a Retrieval Bill of Materials (RBOM) https://medium.com/@bhagyarana80/why-you-need-a-retrieval-bill-of-materials-rbom-ba036c5a4a8d | |||
| 04:32 | How to Ship Multi-Agent Systems Without Chaos https://medium.com/@Quaxel/how-to-ship-multi-agent-systems-without-chaos-0fcfb0e83548 | |||
| 04:20 | LLM Cache Invalidation Patterns in Java (Token-Aware Caching) https://keerthana-13.medium.com/llm-cache-invalidation-patterns-in-java-token-aware-caching-bccaa10ff7c0 | |||
| 04:00 | Why AI Seems Like Magic Right Now (And Why That Could Be a Little Scary) https://medium.com/@kisalaykisu/why-ai-seems-like-magic-right-now-and-why-that-could-be-a-little-scary-4fddc77c58cf | |||
| 03:49 | How to choose the right VectorDB for your RAG application? — Part 1 https://levelup.gitconnected.com/how-to-choose-the-right-vectordb-for-your-rag-application-part-1-4aa416f5e5a5 | |||
| 03:23 | Por que tratar LLMs como software tradicional é um risco de governança https://medium.com/@danilorogerio_62409/por-que-tratar-llms-como-software-tradicional-%C3%A9-um-risco-de-governan%C3%A7a-015c03c7dfdf | |||
| 03:14 | Your LLM Isn’t Boring — It’s Collapsing: How Verbalized Sampling Unlocks Real AI Creativity https://medium.com/@RamPrakashD/your-llm-isnt-boring-it-s-collapsing-how-verbalized-sampling-unlocks-real-ai-creativity-815e856b5fad | |||
| 03:04 | Traditional RAG : Embedding And VectorStoreDB [Part 2] https://medium.com/@patilprasanna73/traditional-rag-embedding-and-vectorstoredb-part-2-5e702620efbb | |||
| 02:59 | Beyond Probabilistic JSON: The Rise of Neuro Symbolic AI https://generativeai.pub/beyond-probabilistic-json-the-rise-of-neuro-symbolic-ai-6fc64d9dded1 | |||
| 02:59 | LLM Council Response https://medium.com/@ericfrayer/llm-council-response-02f8b9689ebb | |||
| 02:53 | Stop Marrying Your Vector Database: The Case for Agnostic RAG https://medium.com/@gdbfgphjsd/stop-marrying-your-vector-database-the-case-for-agnostic-rag-770d30b982bd | |||
| 02:51 | LLM Council App https://medium.com/@ericfrayer/llm-council-app-f91ca689116f | |||
| 02:49 | Beyond AI Hype: Race is Just Starting (Part 2) https://medium.com/@dmik/beyond-ai-hype-race-is-just-starting-part-2-cd31c8251e06 | |||
| 02:47 | All Agentic Architectures: A Hands-On Masterclass for Building AI Agents https://medium.com/coding-nexus/all-agentic-architectures-a-hands-on-masterclass-for-building-ai-agents-242841c4cd97 | |||
| 02:43 | Modular MAX (Powered by Mojo) vs. Ollama: A Local AI Performance on M4 Review https://toyboy2.medium.com/modular-max-powered-by-mojo-vs-ollama-a-local-ai-performance-on-m4-review-98f82238da3c | |||
| 00:57 | When AI Says “Yes” But Does “No”: The Troubling Gap Between What Large Language Models Preach and… https://medium.com/@pns00911/when-ai-says-yes-but-does-no-the-troubling-gap-between-what-large-language-models-preach-and-4703df6d33b6 | |||
| 00:31 | Day 3: 21 Days of Building a Small Language Model:10 Critical PyTorch Operations for Building… https://devopslearning.medium.com/day-3-21-days-of-building-a-small-language-model-10-critical-pytorch-operations-for-building-215e1d9ecbf5 | |||
| 00:22 | AI Memory Management: The Database Imperative https://medium.com/majordigest/ai-memory-management-the-database-imperative-17de9c5a41cc | |||
| 00:13 | Reality Is Messy https://realityismessy.medium.com/reality-is-messy-3f5aed389b48 | |||
| 00:00 | Codex is Open Sourcing AI models https://huggingface.co/blog/hf-skills-training-codex | |||
| Wednesday, 2025-12-10 | ||||
| 23:00 | The rage to write https://medium.com/@i_10525/the-rage-to-write-e35a4f4e6efb | |||
| 22:55 | AgroVision+: An Agentic AI System for Real-World Plant Disease Diagnosis https://medium.com/@jembleton6/agrovision-an-agentic-ai-system-for-real-world-plant-disease-diagnosis-9caa8b75ba95 | |||
| 22:44 | OpenAI Will Devour Your Startup. Where to Hide? https://georgesalapa.medium.com/openai-will-devour-your-startup-where-to-hide-cc9834cf68ec | |||
| 22:36 | The Hidden Bottlenecks: Why Latency is the New Security Vulnerability in Agentic AI https://medium.com/@ag0612202/the-hidden-bottlenecks-why-latency-is-the-new-security-vulnerability-in-agentic-ai-26735122272a | |||
| 22:17 | PPO on Large Language Models https://medium.com/@bormartirosyan/ppo-on-large-language-models-13839b38c8e2 | |||
| 22:09 | Introducing llm-metrics-lite: A Lightweight Toolkit for Evaluating LLM Outputs Without Heavy… https://medium.com/@supriyabachal/introducing-llm-metrics-lite-a-lightweight-toolkit-for-evaluating-llm-outputs-without-heavy-54ead2220dce | |||
| 21:59 | [OpenAI] Training LLMs for Honesty via Confessions https://medium.com/@mdpman/openai-training-llms-for-honesty-via-confessions-65f0342ef5d4 | |||
| 21:56 | The Missing Reasoning Layer: A Compositional Architecture Beneath Language Models https://medium.com/@milamba/the-missing-reasoning-layer-a-compositional-architecture-beneath-language-models-21fd345cce83 | |||
| 21:53 | Chatbots Lack Transparency, So What Are We Doing To Solve This? https://medium.com/@rylanberry/chatbots-lack-transparency-so-what-are-we-doing-to-solve-this-9cd7c5a107f5 | |||
| 21:47 | I Built an AI Agent. It Ignored Me! https://medium.com/@nomannayeem/i-built-an-ai-agent-it-ignored-me-d407336fab29 | |||
| 20:37 | New OpenAI models likely pose "high" cybersecurity risk, company says https://www.axios.com/2025/12/10/openai-new-models-cybersecurity-risks | |||
| 20:26 | AI Inference Will Consume Enough Energy to Power 22% of US Households by 2028 https://medium.com/@tensormesh/ai-inference-will-consume-enough-energy-to-power-22-of-us-households-by-2028-146dd0692381 | |||
| 20:11 | Por qué no veremos “GPT, pero cuántico” (ni siquiera cuando la computación cuántica madure) https://medium.com/@alejandro-mata-ali/por-qu%C3%A9-no-veremos-gpt-pero-cu%C3%A1ntico-ni-siquiera-cuando-la-computaci%C3%B3n-cu%C3%A1ntica-madure-0ba0444d124f | |||
| 20:10 | India proposes charging OpenAI, Google for AI training; lobbying group protests https://techcrunch.com/2025/12/09/india-proposes-charging-openai-google-for-training-ai-on-copyrighted-content/ | |||
| 20:02 | I Built a Distributed AI Search Engine to Kill SEO. Turn Your Website Into an Agent. https://pub.towardsai.net/i-built-a-distributed-ai-search-engine-to-kill-seo-turn-your-website-into-an-agent-d8b80ab8ce52 | |||
| 19:59 | Building a Practical and Efficient RAG Pipeline: Design Choices, Trade-offs, and Architecture… https://medium.com/@luismfsilva40/building-a-practical-and-efficient-rag-pipeline-design-choices-trade-offs-and-architecture-862aa2c68176 | |||
| 19:53 | How to get started in AI in 2026 — a practical guide https://medium.com/@var.786/how-to-get-started-in-ai-in-2026-a-practical-guide-f0dcc943c52f | |||
| 19:53 | What Are LLMs? Understanding Advanced Language Models https://medium.com/@4MATT/what-are-llms-9ed3157d2850 | |||
| 19:53 | AI’s cultural code-switch: when language, context and geography matter more than you think https://medium.com/enrique-dans/ais-cultural-code-switch-when-language-context-and-geography-matter-more-than-you-think-4ba6a639c21d | |||
| 19:47 | Which LLM should you use for what use case? The Simplest Way to Choose the Right LLM https://medium.com/@aryadav.2810/which-llm-should-you-use-for-what-use-case-the-simplest-way-to-choose-the-right-llm-3eb35a309bfb | |||
| 19:39 | Can AI deprive of my jobs? https://medium.com/@keisuganodev/can-ai-deprive-of-my-jobs-53cad9576e7e | |||
| 19:35 | Writing an LLM Plugin System From Scratch https://medium.com/@thekzgroupllc/writing-an-llm-plugin-system-from-scratch-a365d5f39fd7 | |||
| 19:24 | Topology vs Trajectory — The Missing Dimension in AI Evaluation https://medium.com/@kimounbo38/topology-vs-trajectory-the-missing-dimension-in-ai-evaluation-bd5677e0fd55 | |||
| 19:19 | Mistral Devstral 2 is For Agentic Coding https://medium.com/@leucopsis/mistral-devstral-2-is-for-agentic-coding-92c8c7c5fe60 | |||
| 19:02 | Show HN: Recall – open-source local file organizer using Llama 3.2 and Ollama https://github.com/a1k7/Corporate-Brain | |||
| 19:02 | DeepSeek-V3.2: How an Open-Source Model Won Gold at the Math Olympics https://pub.towardsai.net/deepseek-v3-2-how-an-open-source-model-won-gold-at-the-math-olympics-020b4ac8f0c5 | |||
| 18:51 | The Retry Storm That Bankrupted Our LLM Budget https://medium.com/@ketanrapariya/the-retry-storm-that-bankrupted-our-llm-budget-afa49c28dea9 | |||
| 18:48 | The Architecture of Possibility. Industrializing Human Contemplation https://medium.com/@nikolay.niko.nikolov/the-architecture-of-possibility-industrializing-human-contemplation-88c84aafea0a | |||
| 18:37 | Predicting Production Outages Using k6 + LSTM + LLM: 2026 DevOps Superpowers https://skakarh.medium.com/predicting-production-outages-using-k6-lstm-llm-2026-devops-superpowers-1ed8d0d94ba0 | |||
| 18:35 | I Reverse Engineered ChatGPT's Memory System, and Here's What I Found https://manthanguptaa.in/posts/chatgpt_memory/ | |||
| 18:35 | Day 10/31 — Understanding LLMs, GPT, Transformers & A Fun Break with Google Arcade https://medium.com/@valaarpan05/day-10-31-understanding-llms-gpt-transformers-a-fun-break-with-google-arcade-73496d1f6405 | |||
| 18:34 | E31 : Mixed Precision Training https://medium.com/papers-i-found/e31-mixed-precision-training-f3133f3a9f42 | |||
| 18:30 | I Don’t Need a Chatbot, I Need a Staff Engineer https://medium.com/@jengas/i-dont-need-a-chatbot-i-need-a-staff-engineer-3e1ea2255014 | |||
| 18:21 | Building an Enterprise-Grade LLM Platform with vLLM: Real-World Lessons from Large-Scale… https://medium.com/@robi.tomar72/building-an-enterprise-grade-llm-platform-with-vllm-real-world-lessons-from-large-scale-053dd26fb4bd | |||
| 18:02 | Kullback-Leibler (KL) Divergence for LLMs https://pub.towardsai.net/kullback-leibler-kl-divergence-for-llms-0ca3996639c0 | |||
| 17:49 | Inside Microsoft VibeVoice-Realtime-0.5B: https://medium.com/data-science-in-your-pocket/inside-microsoft-vibevoice-realtime-0-5b-c3059aceeb0c | |||
| 17:13 | 5 ChatGPT Tricks That Most People Still Do Not Use https://medium.com/@preronawrites/5-chatgpt-tricks-that-most-people-still-do-not-use-43c6e5e6806f | |||
| 17:08 | Debugging Deep Agents with LangSmith https://blog.langchain.com/debugging-deep-agents-with-langsmith/ | |||
| 17:07 | Introducing LangSmith Fetch: Debug agents from your terminal https://blog.langchain.com/introducing-langsmith-fetch/ | |||
| 17:07 | Introducing Polly: Your AI Agent Engineer https://blog.langchain.com/introducing-polly-your-ai-agent-engineer/ | |||
| 17:04 | Study: ~250 documents is all it takes to backdoor an LLM https://www.searchenginejournal.com/ai-poisoning-black-hat-seo-is-back/561217/ | |||
| 17:04 | How to Build Multi-Step LLM Agents That Don’t Hallucinate https://medium.com/@faryalriz9/how-to-build-multi-step-llm-agents-that-dont-hallucinate-b45b33baa043 | |||
| 16:47 | How I Built an Internal AI Chatbot That Reduced HR Support Tickets by 32% https://medium.com/@purnimap8/how-i-built-an-internal-ai-chatbot-that-reduced-hr-support-tickets-by-32-8d7c42049980 | |||
| 16:39 | What’s happening inside an AI model as it thinks? https://medium.com/@mhn048c/whats-happening-inside-an-ai-model-as-it-thinks-4d5da3acfec9 | |||
| 16:32 | All 16 Fine-Tuning Techniques in 2026 To Create Your Frontier AI Model https://medium.com/coding-nexus/all-16-fine-tuning-techniques-in-2026-to-create-your-frontier-ai-model-96f8570836f3 | |||
| 16:32 | LLM Guarded Function Execution: JSON-Schema Plans with Automatic Rollbacks https://medium.com/@hjparmar1944/llm-guarded-function-execution-json-schema-plans-with-automatic-rollbacks-46b676bf01ae | |||
| 16:29 | The Light Speed Heist: How Physics Just Made Every AI Chip Obsolete https://medium.com/@nraman.n6/the-light-speed-heist-how-physics-just-made-every-ai-chip-obsolete-4a330f1461ae | |||
| 16:23 | Gemini AI: Deep Dive into Architecture, Deployment & Generation Flow https://medium.com/@nraman.n6/gemini-ai-deep-dive-into-architecture-deployment-generation-flow-aa1c7e5cf115 | |||
| 16:22 | I Built a Multi-Search RAG Agent Using LangChain — And It Felt Like Creating My Own Mini Google https://medium.com/@visnus12a22223/i-built-a-multi-search-rag-agent-using-langchain-and-it-felt-like-creating-my-own-mini-google-b150054046cd | |||
| 16:22 | I Built a Multi-Search RAG Agent Using LangChain — And It Felt Like Creating My Own Mini Google https://generativeai.pub/i-built-a-multi-search-rag-agent-using-langchain-and-it-felt-like-creating-my-own-mini-google-b150054046cd | |||
| 16:18 | Getting the Most Out of Coding Agents through Advanced Context Engineering https://medium.com/@dhruvgnk.work/getting-the-most-out-of-coding-agents-through-advanced-context-engineering-d1f0366af0d8 | |||
| 16:12 | Bigger Isn’t Better https://medium.com/@thasvithu/bigger-isnt-better-6c1944168ebd | |||
| 16:09 | RAG in Production: The Data Pipeline Nobody Talks About https://medium.com/@dataenthusiast.io/rag-in-production-the-data-pipeline-nobody-talks-about-059106ded910 | |||
| 16:04 | OpenAI Just Proved Why You Keep Lying to Yourself (And How to Fix It) https://medium.com/data-and-beyond/openai-just-proved-why-you-keep-lying-to-yourself-and-how-to-fix-it-e58beaa9fa46 | |||
| 16:02 | How AI Consulting Leaders Drive ROI With Enterprise AI in 2025 https://pub.towardsai.net/how-ai-consulting-leaders-drive-roi-with-enterprise-ai-in-2025-4b8eda9b2bda | |||
| 16:02 | Why We Should Stop Comparing AI to Humans https://tomasi-wright.medium.com/why-we-should-stop-comparing-ai-to-humans-47c5ed0345ac | |||
| 15:59 | Devstral Small 2 Changes Everything: The First Cloud-Grade Coding Model You Can Truly Run on Your… https://medium.com/coding-nexus/devstral-small-2-changes-everything-the-first-cloud-grade-coding-model-you-can-truly-run-on-your-c7aa78ba49fd | |||
| 15:55 | RAG is not dead. Agentic RAG is just better https://medium.com/@realguantum/rag-is-not-dead-agentic-rag-is-just-better-966ccd7e2ebc | |||
| 15:48 | Adoption & Usage of Open-Web AI Agents https://cobusgreyling.medium.com/adoption-usage-of-open-web-ai-agents-04f368ba88e8 | |||
| 15:48 | Context engineering for building football AI https://soccermatics.medium.com/context-engineering-for-building-football-ai-66f7d378b109 | |||
| 15:40 | LLM Training Estimator―predicts training compute, time, and validation loss based on the Chinchilla… https://medium.com/@rikkabotan/llm-training-estimator-predicts-training-compute-time-and-validation-loss-based-on-the-chinchilla-88e9f08b5ad9 | |||
| 15:35 | Large language model programming frameworks: Part 2 https://billtcheng2013.medium.com/large-language-model-programming-frameworks-part-2-4bb34a5e4dc2 | |||
| 15:02 | Deploying a Hugging Face Pipeline via Snowsight https://medium.com/snowflake/deploying-a-hugging-face-pipeline-via-snowsight-d77595e49060 | |||
| 14:59 | Building a Scalable Batch Architecture for LLM Workloads https://medium.com/@thiagosalvatore/building-a-scalable-batch-architecture-for-llm-workloads-429aa5652f12 | |||
| 14:57 | AI Skills Employers Will Expect You to Know by 2026 https://medium.com/@genai.works/ai-skills-employers-will-expect-you-to-know-by-2026-f7021a1debaf | |||
| 14:52 | BLUF: a short prompt that made ChatGPT work better for me https://medium.com/dont-code-me-on-that/bluf-a-short-prompt-that-made-chatgpt-work-better-for-me-8520cc1d2fa6 | |||
| 14:45 | The Three Architectures That Made AI Reliable: RAG, ReACT, and the Future of Control Flow https://pub.towardsai.net/the-three-architectures-that-made-ai-reliable-rag-react-and-the-future-of-control-flow-8e5e91246fc0 | |||
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241124