103 links · newest first
A study guide for learning large language model engineering. It appears to cover core LLM concepts, architecture, tokenization, pretraining, inference, prompting, fine-tuning, alignment, evaluation, optimization, and related production practices. The available page text confirms the Dropbox file an…
Visit originalThis video explains how an AI agent harness—the code surrounding a model—controls tools, context, memory, execution loops, permissions, and approvals. It covers the five core harness components, the step-by-step task loop, context management for long tasks, prompt-injection and permission concerns,…
Visit originalThe XDA Developers article explores using self-hosted AI models for research, coding, and image generation with a GPU the author already owns. It highlights running AI locally as a way to avoid recurring cloud-service costs, with the only stated expense being a small payment to a savings account. T…
Visit originalSyntax’s CJ explains how chatbots and large language models work, combining historical context with demonstrations of working code. The video covers neural networks, tokenization, embeddings, vectors, transformers, self-attention, feed-forward networks, sampling methods, context windows, autoregres…
Visit originalThis short introduces Calibri, a project described as running the 744-billion-parameter GLM 5.2 model on a laptop without a dedicated GPU. It explains that Calibri keeps roughly 10 GB of the model in RAM and streams the rest from storage as needed, enabling local, offline use. The video also notes …
Visit originalA practical guide to using AI to understand how existing apps or software work and then build something of your own. The page focuses on reverse engineering familiar digital products with AI assistance.
Visit originalA practical guide to Grok Bot’s Team Bots, focusing on shared memory, skills, and routines that help an AI teammate learn and work with a team. It also discusses mistakes that can spread across the team.
Visit originalThis video explains how to use reusable AI skills instead of building a separate agent for every task. It introduces the SKILL.md format and demonstrates three ways to create skills: writing one manually, having Claude generate it through an interview, or recording a workflow for Claude to learn. T…
Visit originalOfficial documentation for the Gemini API from Google AI for Developers. It explains how to integrate Gemini models into applications for text and image generation, multimodal analysis, and conversational agents, with guides and API references for getting started and using supported libraries.
Visit originalA freeCodeCamp tutorial and video about building, pre-training, and fine-tuning a 25-million-parameter language model from scratch. It presents the model architecture, tokenizer, data pipeline, and training process, emphasizing that the code can run on a CPU, though full training is substantially s…
Visit originalSam Witteveen explores using image decision models for robotic process automation, focusing on forms, scans, screenshots, and photos. The video discusses why RPA systems automate steps but struggle with decisions, introduces the ImaJev 4B and Jev Omni open models, and demonstrates a form decision i…
Visit originalA Substack note from Frank Andrade (@thepycoach) recommending a 47-minute lecture about AI. The post quotes the idea that if the lecture leaves you sleeping well, you may not have understood it, and says it explains how AI actually works and what may come next. It presents the lecture as highly wor…
Visit originalA video about Cactus Needle 3, a compact model for function calling that runs on a CPU in 66 milliseconds without a GPU or JSON parser. It shows a live home automation demo, explains the model’s simple attention-based architecture and slicing approach, and walks through a Colab notebook viewers can…
Visit originalA YouTube course by Nate Herk that shows how to build and sell Grok Bot systems as AI teammates for business automation. The video walks through his four Cs framework—context, connections, capabilities, and cadence—while demonstrating agents for email, ClickUp, content creation, meeting transcripts…
Visit originalA video by Sam Witteveen about managed agents being rolled out by various providers and the tradeoffs they create. It explains the idea of managed agents, their timeline, three layers and two kinds of managed agents, and includes a look at Claude’s managed-agent concepts. The video also discusses t…
Visit originalThis DeepSeek API documentation page introduces how to make a first API call. It explains that the DeepSeek API is compatible with the OpenAI and Anthropic API formats, so developers can access it by adjusting configuration in OpenAI/Anthropic SDKs or other compatible software. The page appears to …
Visit originalThis 1Password support article explains how to use 1Password with Claude. It appears to be a help page for connecting or working with 1Password in the Claude environment, with guidance intended for users who need setup or usage instructions. The provided text does not include the full article body,…
Visit originalAn announcement post for a 10-part Substack series about the physics and engineering of frontier large language model inference. The page identifies the series as part of a 2026 foundation model inference series by AI researchers at DistributedApps.ai. It appears to serve as an introduction and ent…
Visit originalA Dave’s Garage video in which Dave uses a 1979 PDP-11 to train a real neural network, showing Transformers and Attention in their most basic form. The page includes the video metadata, publication date, runtime, and links to the episode code, a related recent episode, and a themed shirt.
Visit originalThis video explains how to build a home AI lab for running local models without cloud services, focusing on the three core pieces: hardware, the model itself, and the software harness. It covers the RAM/VRAM tradeoffs, hardware choices such as different GPU configurations, power costs, and what kin…
Visit original