Local AI
rncz.net
Welcome! I'm rncz — a software engineer who's curious about running all kinds of AI tools locally, whether for professional work or just your home lab and hobby projects. This site is the text side of things: clear explanations and commands you can copy easily. The videos live on the AI rncz YouTube channel — if you find it useful, please subscribe!
-
ComfyUI on Windows 11 with ROCm: Flux.1-dev on AMD GPUs
Run the official ROCm nightly release together with PyTorch to run ComfyUI on Windows 11, and generate images with Flux.1-dev on an AMD GPU (RX 7000, RX 8000, RX 9000, Strix Halo).
-
Muse Glimmer 30B by Meta: The Best Local Coding Agent?
Install Muse Glimmer 30B from Meta with llama.cpp, speed it up with DFlash speculative decoding, and connect it to the Pi coding agent.
-
Pi Coding Agent + Llama.cpp: The Best Local AI Setup
Install the Pi coding agent, connect it to a local LLM served by llama.cpp, and use skills and git diffs to stay in control.
-
Installing Llama.cpp on Windows 11 (AMD & Nvidia)
Install Llama.cpp on Windows, pick the right GPU build (CUDA or Vulkan), and run your first LLM with llama-server.