Skip to content

Instantly share code, notes, and snippets.

View DustinAlandzes's full-sized avatar

Dustin Alandzes DustinAlandzes

View GitHub Profile
@karpathy
karpathy / llm-wiki.md
Created April 4, 2026 16:25
llm-wiki

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
karpathy / microgpt.py
Last active September 20, 2026 14:24
microgpt
"""
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
@grahamhelton
grahamhelton / detect-nodes-proxy.sh
Last active April 21, 2026 04:43
Detection script for nodes/proxy
#!/bin/bash
# Detect all subjects with nodes/proxy permissions
# Colors
NOCOLOR=$(tput sgr0)
RED=$(tput setaf 1)
GREEN=$(tput setaf 2)
YELLOW=$(tput setaf 3)
CYAN=$(tput setaf 6)
DIM=$(tput setaf 8)
@levidyrek
levidyrek / django_orm_optimization_cheat_sheet.py
Last active August 25, 2026 11:16
"""
Django ORM Optimization Tips
Caveats:
* Only use optimizations that obfuscate the code if you need to.
* Not all of these tips are hard and fast rules.
* Use your judgement to determine what improvements are appropriate for your code.
"""
# ---------------------------------------------------------------------------
@selfsame
selfsame / Procjam-WFC-tutorial.md
Last active July 9, 2026 05:05

Generating Procedural Game Worlds with Wave Function Collapse

Wave Function Collapse (WFC) by @exutumno is a new algorithm that can generate procedural patterns from a sample image. It's especially exciting for game designers, letting us draw our ideas instead of hand coding them. We'll take a look at the kinds of output WFC can produce and the meaning of the algorithm's parameters. Then we'll walk through setting up WFC in javascript and the Unity game engine.

sprites

The traditional approach to this sort of output is to hand code algorithms that generate features, and combine them to alter your game map. For example you could sprinkle some trees at random coordinates, draw roads with a brownian motion, and add rooms with a Binary Space Partition. This is powerful but time consuming, and your original vision can someti

@selfsame
selfsame / wfc.js
Created October 1, 2017 15:03
//single file version of https://github.com/kchapelier/wavefunctioncollapse
function randomIndice (array, r) {
var sum = 0,
i,
x;
for (i = 0; i < array.length; i++) {
sum += array[i];
}
@pjobson
pjobson / FFMPEG_Notes.md
Last active September 10, 2026 04:33
FFMPEG Notes
@odewahn
odewahn / error-handling-with-fetch.md
Last active March 31, 2026 18:26
Processing errors with Fetch API

I really liked @tjvantoll article Handling Failed HTTP Responses With fetch(). The one thing I found annoying with it, though, is that response.statusText always returns the generic error message associated with the error code. Most APIs, however, will generally return some kind of useful, more human friendly message in the body.

Here's a modification that will capture this message. The key is that rather than throwing an error, you just throw the response and then process it in the catch block to extract the message in the body:

fetch("/api/foo")
  .then( response => {
    if (!response.ok) { throw response }
    return response.json()  //we only get here if there is no error
 })
@karpathy
karpathy / pg-pong.py
Created May 30, 2016 22:50
Training a Neural Network ATARI Pong agent with Policy Gradients from raw pixels
""" Trains an agent with (stochastic) Policy Gradients on Pong. Uses OpenAI Gym. """
import numpy as np
import cPickle as pickle
import gym
# hyperparameters
H = 200 # number of hidden layer neurons
batch_size = 10 # every how many episodes to do a param update?
learning_rate = 1e-4
gamma = 0.99 # discount factor for reward
@Arachnid
Arachnid / upgradeable.sol
Created May 24, 2016 09:49
/**
* Base contract that all upgradeable contracts should use.
*
* Contracts implementing this interface are all called using delegatecall from
* a dispatcher. As a result, the _sizes and _dest variables are shared with the
* dispatcher contract, which allows the called contract to update these at will.
*
* _sizes is a map of function signatures to return value sizes. Due to EVM
* limitations, these need to be populated by the target contract, so the
* dispatcher knows how many bytes of data to return from called functions.