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.

cbrgm

####### Shortcut Hotkeys #############
# open terminal
alt - return : open -n /Applications/Alacritty.app

# restart Yabi, SpaceBar, and SKHD
alt + shift - r : \
    launchctl kickstart -k "gui/${UID}/homebrew.mxcl.yabai"; \
    skhd --reload

jairovadillo

Code Architecture

The main concept is to assign and separate the responsibilities of each layer.

Resources:

• Clean architecture by Uncle Bob
• Django clean architecture
• Python clean architecture

joshbuchea

Semantic Commit Messages

See how a minor change to your commit message style can make you a better programmer.

Format: <type>(<scope>): <subject>

<scope> is optional

Example

alexhayes

Awesomely easy virtualenvs on OSX using pyenv and direnv

Never forget to activate that virtualenv or set that environment variable ever again...

Install

1. Install pyenv

    brew install pyenv
   

baraldilorenzo

##VGG19 model for Keras

This is the Keras model of the 19-layer network used by the VGG team in the ILSVRC-2014 competition.

It has been obtained by directly converting the Caffe model provived by the authors.

Details about the network architecture can be found in the following arXiv paper:

Very Deep Convolutional Networks for Large-Scale Image Recognition

K. Simonyan, A. Zisserman

MichalZalecki

/* VT100 terminal reset (<ESC>c) */

console.log('\033c');

/* numbers comparations */

> '2' == 2
true

> '2' === 2

henrik

tmux cheatsheet

As configured in my dotfiles.

start new:

tmux

start new with session name:

baali

from nltk.probability import ELEProbDist, FreqDist
from nltk import NaiveBayesClassifier
from collections import defaultdict

train_samples = {
    'I hate you and you are a bad person': 'neg',
    'I love you and you are a good person': 'pos',
    'I fail at everything and I want to kill people' : 'neg',
    'I win at everything and I want to love people' : 'pos',
    'sad are things are heppening. fml' : 'neg',