Michal Sudwoj westernmagic

LLM Wiki

A pattern for building personal knowledge bases using LLMs.

This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.

The core idea

Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.

EDIT: check out fix-python instead, make Python run "as usual" on NixOS!

How to: make Python dependencies installed via pip work on NixOS

The issue

You are using NixOS
You start working on a Python project
You manage the dependencies in a classic Python virtual environment using pip or poetry

QNAP TS-453 Pro: LCD, LEDs, fan & buttons

QNAP NAS model TS-453 Pro has limited documentation on how to control its various peripherals, namely:

the front LCD display, a 2 row, 16 column blue & white LCD display, and its two menu buttons ("ENTER", "SELECT")
the front LEDs "STATUS" (red & green) & "USB" (blue)
the 4 "disk error" LEDs (red)
the front "USB COPY" button
the main rear fan

High-Performance Matrix Multiplication

This is a short post that explains how to write a high-performance matrix multiplication program on modern processors. In this tutorial I will use a single core of the Skylake-client CPU with AVX2, but the principles in this post also apply to other processors with different instruction sets (such as AVX512).

Intro

Matrix multiplication is a mathematical operation that defines the product of

This can reduce files to ~15% of their size (2.3M to 345K, in one case) with no obvious degradation of quality.

ghostscript -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/printer -dNOPAUSE -dQUIET -dBATCH -sOutputFile=output.pdf input.pdf

Other options for PDFSETTINGS:

/screen selects low-resolution output similar to the Acrobat Distiller "Screen Optimized" setting.
/ebook selects medium-resolution output similar to the Acrobat Distiller "eBook" setting.
/printer selects output similar to the Acrobat Distiller "Print Optimized" setting.
/prepress selects output similar to Acrobat Distiller "Prepress Optimized" setting.

	script
	var CopyOption = Java.type("java.nio.file.StandardCopyOption");
	// put all files here >>>
	var ddlPath = java.nio.file.FileSystems.getDefault().getPath('ddl')

	var fs = java.nio.file.FileSystems.getDefault();
	var f = java.nio.file.Files;

	if ( ! f.exists(ddlPath)) {
	f.createDirectory(ddlPath)

	# Paths {{{
	set folder = ~/Mail # mailbox location
	set alias_file = ~/.mutt/alias # where to store aliases
	set header_cache = ~/.mutt/cache/headers # where to store headers
	set message_cachedir = ~/.mutt/cache/bodies # where to store bodies
	set certificate_file = ~/.mutt/cerficates # where to store certs
	set mailcap_path = ~/.mutt/mailcap # entries for filetypes
	set tmpdir = ~/.mutt/temp # where to keep temp files
	set signature = ~/.mutt/sig # my signature file

	#!/bin/bash

	# Vadim Zaliva lord@crocodile.org
	# based on https://gist.github.com/hadess/6847281

	if [[ $EUID -ne 0 ]]; then
	echo "This script must be run as root" 1>&2
	exit 1
	fi

	#!/bin/bash

	# default location of theme and wallpaper
	# replace 'awtheme' with your current theme
	wallpaperDestination="$HOME/.config/awesome/themes/awtheme/wallpaper";
	# add #include "/home/arch/.Xresources_colors" to .Xresources
	colorScriptDestination="/home/arch/.Xresources_colors"

	# add [[ -f "$HOME/.bash_colors" ]] && . "$HOME/.bash_colors"; to .bashrc
	# bashColorsDestination="/home/arch/.bash_colors"

	#include <string.h>

	void __memcpy_glibc_2_2_5(void , const void *, size_t);

	asm(".symver __memcpy_glibc_2_2_5, memcpy@GLIBC_2.2.5");
	void __wrap_memcpy(void dest, const void *src, size_t n)
	{
	return __memcpy_glibc_2_2_5(dest, src, n);
	}