# open weight models (blogs) — RSS Amplifier Recent posts from the 5 feeds in the RSS Amplifier directory that cover open weight models. Page: Feed: --- ## [How does Backpropagation work?](https://bayinformationsystems.substack.com/p/how-does-backpropagation-work) _2026-08-18 · Edward Grundy · Bay Information Systems_ F(X) = Y describes what a model does once it’s trained. But how does F gets to that state in the first place? ## [What's going on with open-weight models?](https://read.technically.dev/p/whats-going-on-with-open-weight-models) _2026-08-18 · Will Raphaelson · Technically_ How open-weight models went from a curiosity to genuinely great in 2 short years. ## [I Want You for AI Safety](https://counterfactual.blog/p/welcome-to-the-pivotal-era) _2026-08-12 · Sophie Kim · The Counterfactual_ A pitch to join AI safety, or fund it – prompted by the summer AI started hacking companies. ## [The Four Types of Machine Learning](https://bayinformationsystems.substack.com/p/the-four-types-of-machine-learning) _2026-08-11 · Edward Grundy · Bay Information Systems_ Machine learning reduces to one expression: F(X) = Y. A model F takes input data X and produces output Y. From these combinations, we can learn all the main strategies to solve data problems with ML. ## [Software Eng for Vibe Coders: On Frontends + Backends](https://read.technically.dev/p/software-eng-vibe-coders-frontends-backends) _2026-08-11 · Justin · Technically_ A new series to help non-engineers build products that can handle going viral. ## [Dispatch: Kimi K3 licensing, Liquid AI, and stacked PRs on GitHub](https://read.technically.dev/p/dispatch-august-9-2026) _2026-08-09 · David Krevitt · Technically_ 3 stories you might've missed from the last 2 weeks. ## [The Unfinished Notebook](https://bayinformationsystems.substack.com/p/the-unfinished-notebook) _2026-08-07 · Edward Grundy · Bay Information Systems_ Notebooks are a great way to communicate narrative and encourage engagement in data projects. ## [How do "computer use" agents work?](https://read.technically.dev/p/how-do-computer-use-agents-work) _2026-08-06 · Tereza Tizkova · Technically_ Why it's surprisingly challenging to let an LLM use a computer for you. ## [Model Choice Determines VLM Inference Cost](https://bayinformationsystems.substack.com/p/model-choice-determines-vlm-inference) _2026-08-04 · Edward Grundy · Bay Information Systems_ Output length bias in the training process impacts the cost of model inference. ## [Happy Sysadmin Day](https://managing.blue/2026/07/31/happy-sysadmin-day-9/) _2026-07-31 · adamo · Managing organized complexity_ Projecting the current trends, future software will consist of a synthetically generated code. No one will have the foggiest idea what is in that code. Somewhere in the basement of MegaCorp will be a huge computer file with the listing of that code. The last sysadmin will sit beside the file, handcuffed to the disk drive like a Continue reading Happy Sysadmin Day ## [Stream Torrents & IPTV Instantly (Sponsored)](https://crawlproof.com/a/4rPLmJXbsprB) _2026-07-31 · **Sponsored**_ Search magnet links and stream movies, music, books, and live TV in-browser ## [Embedding-Based Relationship Discovery: Finding the Real Network in Your Data](https://bayinformationsystems.substack.com/p/embedding-based-relationship-discovery) _2026-07-28 · Edward Grundy · Bay Information Systems_ Most organisations have a formal model of their customer relationships and a real one that differs from it. Embeddings and community detection surface the real structure. ## [What is Rust?](https://read.technically.dev/p/what-is-rust) _2026-07-28 · Will Raphaelson · Technically_ The fast, safe, super difficult to write programming language that’s finally getting its flowers. ## [Engineering in the Exponential Age](https://bayinformationsystems.substack.com/p/engineering-in-the-exponential-age) _2026-07-23 · Edward Grundy · Bay Information Systems_ Capability is arriving faster than teams can absorb it. The engineering task becomes protocols and interfaces in place of bespoke services. ## [Typhoon Season](https://counterfactual.blog/p/typhoon-season) _2026-07-22 · Sophie Kim · The Counterfactual_ The CCP's Hacking Army, Critical Infra, Trump's Texts ## [What's Harness Engineering?](https://read.technically.dev/p/whats-harness-engineering) _2026-07-21 · Paul Iusztin · Technically_ If AI models are a commodity, the harness is the thing you want to own. ## [China's Technological Playbook](https://counterfactual.blog/p/chinas-technological-playbook) _2026-07-16 · Sophie Kim · The Counterfactual_ Part 1: The Theft Economy ## [The Algorithm Lottery](https://bayinformationsystems.substack.com/p/the-algorithm-lottery) _2026-07-16 · Edward Grundy · Bay Information Systems_ Recommendation systems deliberately surface under-explored content to find new audience clusters. Understanding why this happens changes how you think about content distribution. ## [What does Turbopuffer do?](https://read.technically.dev/p/what-does-turbopuffer-do) _2026-07-14 · David Krevitt · Technically_ And what's going on in the search tooling market? ## [software calligraphy – part 2](https://managing.blue/2026/07/11/software-calligraphy-part-2/) _2026-07-10 · adamo · Managing organized complexity_ As a systems administrator I am used to stitching together other people s code and systems to produce a working outcome within given constraints. Which means, the rise of agentic coding should make me happy. Alas, outside work, it seems to make me sad. Which is why I still discuss my saddening experience with friends, coaches Continue reading software calligraphy part 2 ## [How to Structure an ML Team](https://bayinformationsystems.substack.com/p/how-to-structure-an-ml-team) _2026-07-09 · Edward Grundy · Bay Information Systems_ The four horsemen of the techpocalyse! ## [Home of the Agentic Pull Request (Sponsored)](https://crawlproof.com/a/S65PWONBiLDz) _2026-07-09 · **Sponsored**_ CLI-first, TUI-ready toolkits and a moshscript installer — join the waitlist ## [What’s wrong with grantmaking?](https://counterfactual.blog/p/whats-wrong-with-grantmaking) _2026-07-08 · Sophie Kim · The Counterfactual_ And how do we fix it? ## [What Is NER and Why Does It Matter for PII?](https://bayinformationsystems.substack.com/p/what-is-ner-and-why-does-it-matter) _2026-07-07 · Edward Grundy · Bay Information Systems_ Some tasks are so well understood there are specific classes of models just for that task. Identify named entities and redacting private information are two of those specific tasks. ## [What is Rust? Part 1: What is a Programming Language?](https://read.technically.dev/p/whats-a-programming-language) _2026-07-07 · Will Raphaelson · Technically_ The fast, safe, super difficult to write programming language that’s finally getting its flowers. ## [Falling Token Prices and the Cost of Inference](https://bayinformationsystems.substack.com/p/falling-token-prices-and-the-cost) _2026-07-02 · Edward Grundy · Bay Information Systems_ How do you know if you are saving money when things get cheaper? ## [Why Private Inference Is Not Fully Private (Yet)](https://bayinformationsystems.substack.com/p/why-private-inference-is-not-fully) _2026-06-30 · Edward Grundy · Bay Information Systems_ Fully Homomorphic Encryption will bring massive change to the inference landscape, but it is not yet possible. ## [Bioweapons in the Age of AI](https://counterfactual.blog/p/bioweapons-in-the-age-of-ai) _2026-06-26 · Sophie Kim · The Counterfactual_ 11,600 words of terror - Bioweapon.AI is finally finished!! ## [software calligraphy](https://managing.blue/2026/06/25/software-calligraphy/) _2026-06-25 · adamo · Managing organized complexity_ or What objecting to LLM agents teaches you about you (sometimes) There was a small positional game that I ve wanted to implement for some years, long before the LLMs and their agents became a thing. I did not have the time. By now we know that using an LLM shrinks this time demand. So I Continue reading software calligraphy ## [Everything is a Pipeline](https://read.technically.dev/p/everything-is-a-pipeline) _2026-06-25 · Afzal Jasani · Technically_ Why technical concepts are often more alike then we might think. ## [Multi-Tenant Architecture: A Practical Spectrum](https://bayinformationsystems.substack.com/p/multi-tenant-architecture-a-practical) _2026-06-25 · Edward Grundy · Bay Information Systems_ Multi-tenancy is a spectrum from shared tables to fully separate infrastructure, and the right point on that spectrum depends on your risks. ## [Open Weight models and soft power](https://managing.blue/2026/06/21/open-weight-models-and-soft-power/) _2026-06-21 · adamo · Managing organized complexity_ I see Open-weight AI models as a form of digital soft power. When an entity releases their models for free public use, they aren t doing this operating under an altruistic motive; they re aiming at shaping global perception: users get to run the latest (or the best for their hardware) model for free, but in return, Continue reading Open Weight models and soft power ## [Ship software peacefully (Sponsored)](https://crawlproof.com/a/4uvUdUZibmSj) _2026-06-21 · **Sponsored**_ Connect your repo and deploy instantly — Railway handles config, scaling, and monitoring. ## [What are code sandboxes?](https://read.technically.dev/p/what-are-code-sandboxes) _2026-06-18 · Justin · Technically_ Why coding agents need a safe place to play, just like we all do. ## [LLM Data Formatting Choices: A Taxonomy](https://bayinformationsystems.substack.com/p/llm-data-formatting-choices-a-taxonomy) _2026-06-18 · Edward Grundy · Bay Information Systems_ The format you use to pass data to a language model affects reliability and cost more than most expect... ## [Reheat](https://bayinformationsystems.substack.com/p/reheat) _2026-06-16 · Edward Grundy · Bay Information Systems_ Our new open source tool pulls your Search Console queries, embeds them as vectors, clusters by semantic intent, and surfaces ranked content gaps and opportunities as a report. ## [GPU Market Overview for Local AI/ML Inference: VRAM Capacity and Software Stack Compatibility (Mid-2026)](https://bayinformationsystems.substack.com/p/gpu-market-overview-for-local-aiml) _2026-06-15 · Edward Grundy · Bay Information Systems_ UK focus with approximate pricing. ## [Bronze, Silver, Gold: A Layered Data Architecture in PostgreSQL](https://bayinformationsystems.substack.com/p/bronze-silver-gold-a-layered-data) _2026-06-11 · Edward Grundy · Bay Information Systems_ The medallion architecture maps cleanly onto PostgreSQL. Each layer has a single concern, with transitions being where the interesting engineering lives ## [A Measurement Framework for Machine Learning Projects](https://bayinformationsystems.substack.com/p/a-measurement-framework-for-machine) _2026-06-09 · Edward Grundy · Bay Information Systems_ A practical framework for defining, building, and maintaining an ML system -- including a free checklist!! ## [adventures with meshcore-cli](https://managing.blue/2026/06/07/adventures-with-meshcore-cli/) _2026-06-07 · adamo · Managing organized complexity_ I have a good friend who is into meshcore. Seeking to expand the network he participates in, he asked me if I could host a repeater on the rooftop, to which I agreed. He also gave me a little companion to connect so that I could also test and participate if I liked. The companion Continue reading adventures with meshcore-cli ## [When Russia Wants a Bioweapon,](https://counterfactual.blog/p/when-states-get-serious-about-biological) _2026-06-04 · Sophie Kim · The Counterfactual_ I am disturbed ## [Event-Driven Embeddings in PostgreSQL: async and LISTEN/NOTIFY](https://bayinformationsystems.substack.com/p/event-driven-embeddings-in-postgresql) _2026-06-04 · Edward Grundy · Bay Information Systems_ Postgres LISTEN/NOTIFY mechanism can trigger embedding generation on INSERT, without polling, without a separate scheduler, and without coupling services. ## [System Over Model, Tested: Reproducing Mythos's FreeBSD Find on Local Open-Weight Models](https://clearbluejar.github.io/posts/system-over-model-tested-mythos-freebsd-local-openweight/) _2026-06-04 · clearbluejar · clearbluejar_ Mythos found a 17-year-old FreeBSD RCE; AISLE reproduced it with gpt-5.4-nano via their nano-analyzer pipeline. I ran the pipeline on two local open-weight models, gpt-oss-20b and gemma-4-31b-it. The misses recovered on re-run. The real problem was the false-positive rate, and one extra system stage cut it from 30 to 5 with the CVE still standing. ## [Turns Out, You Can’t Just Bomb a Datacenter](https://counterfactual.blog/p/turns-out-you-cant-just-bomb-a-datacenter) _2026-06-03 · Sophie Kim · The Counterfactual_ Notes on kinetic MAIM & escalation constraints ## [Don’t Build Models, Build Measurements](https://bayinformationsystems.substack.com/p/dont-build-models-build-measurements) _2026-06-02 · Edward Grundy · Bay Information Systems_ If an organisation can define "what correct looks like" ML projects will likely succeed; it also answers the gnarly "Are you automating my job?" question ## [What Is Data Engineering?](https://bayinformationsystems.substack.com/p/what-is-data-engineering) _2026-05-28 · Edward Grundy · Bay Information Systems_ Two Jobs, One Title; which one do you love most? ## [Open Weight Models and the Separation of Concerns in AI](https://bayinformationsystems.substack.com/p/open-weight-models-and-the-separation) _2026-05-26 · Edward Grundy · Bay Information Systems_ Often framed as an ideological discussion, open weight models actually solve the business problem which ruins closed model providers over the long run. ## [How to Actually Spend Billions on AI Safety](https://counterfactual.blog/p/how-to-actually-spend-billions-on) _2026-05-12 · Sophie Kim · The Counterfactual_ A concrete strategy for deploying the largest wave of philanthropic capital in history ## [pyghidra-mcp Meets Ghidra GUI: Drive Project-Wide RE with Local AI](https://clearbluejar.github.io/posts/pyghidra-mcp-meets-ghidra-gui-drive-project-wide-re-with-local-ai/) _2026-05-05 · clearbluejar · clearbluejar_ pyghidra-mcp v0.2.0 ships a GUI-backed mode that lets a local LLM drive a live Ghidra CodeBrowser at full project scope. Renames, plate comments, and cross-binary pivots land in real time, with every edit tagged in Ghidra's undo history while the session is alive. ## [“The world needs five LLMs”](https://managing.blue/2026/04/25/the-world-needs-five-llms/) _2026-04-25 · adamo · Managing organized complexity_ Thomas J. Watson famously said “the world only needs five computers”. With the data center being the computer we can say that the world concentrated around five clouds and he was just about right. Now it concentrates roughly around five LLMs. ## [Stop Donating to AI Safety Research\*](https://counterfactual.blog/p/donate-less-to-ai-safety) _2026-04-25 · Sophie Kim · The Counterfactual_ \*if you want us to win ## [Should We Be Worried About Bioweapons?](https://counterfactual.blog/p/should-we-be-worried-about-bioweapons) _2026-04-17 · Sophie Kim · The Counterfactual_ An Analysis of Historical Case Studies - Part 1: Non-State Actors ## [Handy Manny and the LLMs](https://managing.blue/2026/04/13/handy-manny-and-the-llms/) _2026-04-13 · adamo · Managing organized complexity_ One of the kids was watching Handy Manny and now I cannot stop thinking that LLM agents are like Handy Manny’s tools. We are like Handy Manny discussing with the tools and directing the repairs.