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.

dhh

# THIS LINUX SETUP SCRIPT HAS MORPHED INTO A WHOLE PROJECT: HTTPS://OMAKUB.ORG
# PLEASE CHECKOUT THAT PROJECT INSTEAD OF THIS OUTDATED SETUP SCRIPT.
#
#

# Libraries and infrastructure
sudo apt update -y
sudo apt install -y \
	docker.io docker-buildx \
	build-essential pkg-config autoconf bison rustc cargo clang \

lucahammer

// ==UserScript==
// @name         TweetXer
// @namespace    https://github.com/lucahammer/tweetXer/
// @version      0.9.3
// @description  Delete all your Tweets for free.
// @author       Luca,dbort,pReya,Micolithe,STrRedWolf
// @license      NoHarm-draft
// @match        https://x.com/*
// @match        https://mobile.x.com/*
// @match        https://twitter.com/*

huytd

How to use:

./wordle.sh

Or try the unlimit mode:

skeeto

Moved to skeeto/scratch/aidrivers

riveSunder

"""
Simple MLP demo using [autograd](https://github.com/HIPS/autograd)
With l1 and l2 regularization. 

Depends on autograd and scikit-learn (the latter for the mini digits dataset)

pip install autograd scikit-learn
"""

from autograd import numpy as np

IanColdwater

Mute these words in your settings here: https://twitter.com/settings/muted_keywords

ActivityTweet
generic_activity_highlights
generic_activity_momentsbreaking
RankedOrganicTweet
suggest_activity
suggest_activity_feed
suggest_activity_highlights
suggest_activity_tweet

harrison-drift

// This prints a list of buttons that on tap will fire a different type of haptic vibration

import SwiftUI

struct ContentView: View {
    let generator = UINotificationFeedbackGenerator()
    
    var body: some View {
        VStack(alignment: .center, spacing: 30.0) {
            Button(action: {

bmjstau

function [rgb] = wesAnderson(N, which)
% wesAnderson returns N colors taken from Wes Anderson's films, which
% can be used as color maps for hipster line plots. Colors have been 
% adapted from https://github.com/karthik/wesanderson, who has them from 
% http://wesandersonpalettes.tumblr.com/.
% 
% Inputs:
% INT N         Number of colors. Is limited by color palette.
% CHAR which    Name of color palette. See 
%               https://github.com/karthik/wesanderson for previews.

konrad

#!/usr/bin/env bash

# Returns a BibTeX (www.bibtex.org) entry for one or more given DOIs
# (https://www.doi.org/).
#
# Call it like this:
#
# $ doi2bibtex.sh 10.1093/bioinformatics/btu533
#
# Can also be used for several DOIs at once: