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

ivan

Things I might read in 2024.

Now extended into 2025.

• CSS for JavaScript Developers^{/ just the things I don't know and which would be used frequently}
• Jamie Brandon - Reflections of a decade of coding
• John Ousterhout - A Philosophy of Software Design
• Will Larson - the starred posts on https://lethain.com/
• https://notes.andymatuschak.org/
• The Tailwind documentation (even before updated for Tailwind 4 - see changes)

ivan

[This page is best viewed with https://github.com/ludios/expand-everything, which will load all the comnents below.]

Wherein I try to prioritize reading for the limited amount of time I have this year, and to remind myself to read more than just comments on the Internet. Because of problems of time and shifting interests, I will consider this a success if I read a third of the list. I'll reflect on the reading and deviations from the plan in Jan 2024.

{+} = added after initial planning

• Jamie Brandon - Reflections of a decade of coding
• {+} Aleksandr Hovhannisyan's blog
• the SvelteKit tutorial
• the SvelteKit documentation

rondy

FWIW: I (@rondy) am not the creator of the content shared here, which is an excerpt from Edmond Lau's book. I simply copied and pasted it from another location and saved it as a personal note, before it gained popularity on news.ycombinator.com. Unfortunately, I cannot recall the exact origin of the original source, nor was I able to find the author's name, so I am can't provide the appropriate credits.

---

Effective Engineer - Notes

• By Edmond Lau
• Highly Recommended 👍
• http://www.theeffectiveengineer.com/

What's an Effective Engineer?

jboursiquot

[{ "number": 1, "text": "Add some fucking contrast."},{ "number": 2, "text": "Have a clear fucking hierarchy."},{ "number": 3, "text": "Use fucking Helvetica."},{ "number": 4, "text": "Read a fucking book."},{ "number": 5, "text": "Get over your fucking self."},{ "number": 6, "text": "Make it fucking balanced."},{ "number": 7, "text": "Sketch more fucking ideas."},{ "number": 8, "text": "Sans-fucking-serif."},{ "number": 9, "text": "Use a fucking grid system."},{ "number": 10, "text": "Use more fucking white space."},{ "number": 11, "text": "Make it fucking consistent."},{ "number": 12, "text": "Make it fucking asymmetrical."},{ "number": 13, "text": "Fucking simplify."},{ "number": 14, "text": "Use a fucking color scheme."},{ "number": 15, "text": "Have a fucking focal point."},{ "number": 16, "text": "Fix your fucking letterspacing."},{ "number": 17, "text": "Rag your fucking type."},{ "number": 18, "text": "Do your fucking research."},{ "number": 19, "text": "Fucking communicate something."},{ "number": 20