karpathy

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.

HorseCheng

python earthquake.py %1 %2

n1snt

Note

I’ve written a newer optimized 2026 version of this guide with a faster setup using Zinit + Starship and improved plugin loading.

This version avoids slow shell startups and works better with modern Zsh setups.

Updated guide: https://gist.github.com/n1snt/2cccc8aa5f7b645a7628d3512c70deb6

Oh my zsh.

prof3ssorSt3v3

const APP = {
  deferredInstall: null,
  init() {
    if ('serviceWorker' in navigator) {
      //register our service worker
      navigator.serviceWorker
        .register('/sw.js', {
          updateViaCache: 'none',
          scope: '/',
        })

theodorosploumis

Profile	download (kb/s)	upload (kb/s)	latency (ms)
Native	0	0	0
GPRS	50	20	500
56K Dial-up	50	30	120
Mobile EDGE	240	200	840
2G Regular	250	50	300
2G Good	450	150	150
3G Slow	780	330	200

kfur

// ==UserScript==
// @name DarkReader
// @match *://*/*
// @grant none
// @run-at document-start
// ==/UserScript==

// MIT License

// Copyright (c) 2019 Alexander Shutau

kepta

In this article I will take a very simplistic approach in understanding memory leaks and I will also attempt to diagnose them.

In todays world of abundant memory, we seldom worry about memory leakages. But I hate to tell you that we live in a real world and nothing comes for free.

Oh my fancy functional programming

Disclosure: I absolutely love functional programming. Functional programming is cool and with the new ES6 syntax it becomes even cooler.

gwpantazes

How To Install Different JDK Versions on MacOS with Homebrew

Keywords: Java, JDK (Java Development Kit), MacOS, Homebrew, Specific Version

This how-to guide covers how to install different versions of the JDK on MacOS with Homebrew.

Table of Contents

• Preface
• This Guide Favors OpenJDK

KevinAst

Integrating GitBook with JSDoc to Document Your Open Source Project

Introduction

Good documentation should include two distinct elements - a Guide and an API:

1. The Guide builds concepts, providing examples, etc.

GitBook is well suited to

bvaughn

React lifecycle cheatsheet

Method	Side effects1	State updates2	Example uses
Mounting
componentWillMount		✓	Constructor equivalent for createClass
render			Create and return element(s)
componentDidMount	✓	✓	DOM manipulations, network requests, etc.
Updating
componentWillReceiveProps		✓	Update state based on changed props