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.

karpathy

"""
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

0wwafa

To print a GROK chat, paste this code in javascript console:

(function() {
    var a = document.querySelector("body > div > div").children[0].children[1].children[0];
    if (a) {
        const b = window.open("", "_blank", `width=${.85 * window.screen.width},height=${.85 * window.screen.height},left=${.075 * window.screen.width},top=${.045 * window.screen.height}`)
          , e = Array.from(document.styleSheets).map(c => {
            try {
                return Array.from(c.cssRules).map(d => d.cssText).join("\n")
            } catch (d) {

Wingysam

// ==UserScript==
// @name         Git Archive
// @namespace    http://wingysam.xyz/
// @version      2.4
// @description  Mirror every git repo you look at to gitea
// @author       Wingy <git@wingysam.xyz>
// @include      *
// @grant        GM_xmlhttpRequest
// @grant        GM_notification
// @grant        GM_openInTab

oofnikj

KEYMAPOPTS="us us"
HOSTNAMEOPTS="-n alpine"
INTERFACESOPTS="auto lo
iface lo inet loopback

auto eth0
iface eth0 inet dhcp
    hostname alpine
"
TIMEZONEOPTS="-z UTC"

adbertram

#region Ensure the WinRm service is running
Set-Service -Name "WinRM" -StartupType Automatic
Start-Service -Name "WinRM"
#endregion

#region Enable PS remoting
if (-not (Get-PSSessionConfiguration) -or (-not (Get-ChildItem WSMan:\localhost\Listener))) {
    Enable-PSRemoting -SkipNetworkProfileCheck -Force
}
#endregion

kuttor

#!/data/data/com.termux/files/usr/bin/bash


yes | pkg upgrade && \
yes | pkg install \
    python \
    python-dev \
    libffi \
    libffi-dev \
    openssl \

mikepruett3

<#
.SYNOPSIS
    Script to Initialize my custom powershell setup.
.DESCRIPTION
    Script uses scoop
.NOTES
    **NOTE** Will configure the Execution Policy for the "CurrentUser" to Unrestricted.

    Author: Mike Pruett
    Date: October 18th, 2018

matthewjberger

1.) Download a Nerd Font

2.) Unzip and copy to ~/.fonts

3.) Run the command fc-cache -fv to manually rebuild the font cache

DiegoSalazar

#!/bin/sh

. git-sh-setup  # for die 
git-diff-index -p -M --cached HEAD -- | grep '^+' | grep -n '# debug' && die 'Blocking commit: debug detected in code'
: