Julian Neytchev dobriak

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.

Browser Based

Web Wormhole https://webwormhole.io/ https://github.com/saljam/webwormhole
NOTE: Probably the only browser tool that streams file directly. Should transfer file of any size (hundreds of GBs etc)
Localsend https://web.localsend.org/
FilePizza https://file.pizza/
PairDrop https://pairdrop.net/ https://github.com/schlagmichdoch/pairdrop
1. ~~ShareDrop sharedrop.io https://github.com/szimek/sharedrop~~ (SOLD, not recommended, use PairDrop)
2. ~~SnapDrop snapdrop.net https://github.com/RobinLinus/snapdrop~~ (SOLD, not recommended, use PairDrop)
ToffeeShare https://toffeeshare.com/

PostgreSQL is Enough

See the new site: https://postgresisenough.dev

Autofs on Mac OS X

With autofs you can easily mount network volumes upon first access to the folder where you want to mount the volume. Autofs is available for many OS and is preinstalled on Mac OS X so I show you how I mounted my iTunes library folder using this method.

Prepare autofs to use a separate configuration file

autofs needs to be configured so that it knows where to gets its configuration. Edit the file /etc/auto_master and add the last line:

#
# Automounter master map
#

+auto_master # Use directory service

Installing Kerberos on Redhat 7

This installation is going to require 2 servers one acts as kerberos KDC server and the other machine is going to be client. Lets assume the FQDN's are (here cw.com is the domain name, make a note of the domain name here):

Kerberos KDC Server: kdc.cw.com
Kerberos Client: kclient.cw.com

Important: Make sure that both systems have their hostnames properly set and both systems have the hostnames and IP addresses of both systems in

	#!/usr/bin/env python3
	import argparse, json, sys, time
	from urllib import request

	PROMPTS = [
	{"name": "code_python", "prompt": "Write a Python function that returns the n-th Fibonacci number using memoization. Include a docstring."},
	{"name": "code_cpp", "prompt": "Write a C++ template function `clamp(x, lo, hi)` that returns x clamped to [lo, hi]. No std::clamp."},
	{"name": "explain_concept", "prompt": "Explain how speculative decoding works in large language model inference, in three short paragraphs."},
	{"name": "summarize", "prompt": "Summarize in two sentences: The Industrial Revolution began in Britain in the late 18th century, transforming manufacturing through mechanization, steam power, and the factory system. It spread to continental Europe and North America during the 19th century."},
	{"name": "qa_factual", "prompt": "Q: What are the four fundamental forces of physics?\nA:"},

	"""
	The most atomic way to train and run inference for a GPT in pure, dependency-free Python.
	This file is the complete algorithm.
	Everything else is just efficiency.

	@karpathy
	"""

	import os # os.path.exists
	import math # math.log, math.exp

	package main

	import (
	"context"
	"flag"
	"fmt"
	"log"
	"net/http"
	"os"
	"os/signal"

	import numpy as np
	import scipy.linalg
	import scipy.stats

	def ks_key(X):
	'''Estimate the key from a pitch class distribution

	Parameters
	----------
	X : np.ndarray, shape=(12,)

	package main

	import (
	"crypto/tls"
	"golang.org/x/crypto/acme/autocert"
	"log"
	"net"
	"net/http"
	)