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.

cansik

Computer Vision with Python on M1 Apple Silicon (arm64)

When switching over to a new MacBook with an M1 Apple Silicon chip, I noticed that many python packages I use daily are not compatible with M1 chips. Sometimes there was just no prebuilt wheel, or there were still bugs in the code not allowing the package to build at all. So I decided to create drop-in replacement packages with prebuilt wheels for the Apple Mac M1 (Apple Silicon / arm64).

Every package contains a build script to automatically build it in your own enviroment and a Github CI workflow to build it on Github (if possible).

Here you find a list of all the prebuilt packages. Sometimes the official maintainer have updated their binaries too, so please first check if there is an official build ready for M1.

pyrealsense2 for macOSX

A librealsense2 python wrapper built for all modern MacOSX versions (intel / silicon) because Intel does only

antiboredom

import argparse
import re
import os
import subprocess
from moviepy.editor import VideoFileClip, concatenate_videoclips


def get_silences(filename, db, duration):
    args = "ffmpeg -nostats -i {} -af silencedetect=noise={}dB:d={} -f null -".format(
        filename, db, duration

MAKIO135

d3indepth

license: gpl-3.0
height: 220
border: no

widmerin

license: apache-2.0
border: no
height: 900

steveharoz

license: gpl-3.0
height: 1030
scrolling: yes

mbostock

license: gpl-3.0
height: 200
redirect: https://observablehq.com/@d3/beeswarm

mchow01

import tweepy
import MySQLdb
import sys

# Tweepy API doc here: http://pythonhosted.org/tweepy/html/api.html

# Keys
consumer_key = ''
consumer_secret = ''
access_token = ''

postspectacular

Toxi's Lebkuchen (plain "Elisen" style)

Spice blend

• 3 whole star anise
• 15 whole cardamon pods
• 2 tblsp ground cinnamon (or ~0.75 stick)
• 1.5 tblsp ground cloves
• 0.5 tsp ground coriander
• 0.75 tsp ground ginger