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.

Snarp

/*
<https://stackoverflow.com/questions/40296831/is-it-possible-to-force-a-copy-of-a-protected-google-doc>


NOTE - 2021-05-24
-----------------

The script below isn't the fastest way to copy-and-paste from a protected 
Google Doc. Before trying it, I'd suggest following MikoFrosty's advice from 
the comments:

dtomasi

server {
    listen       80;
    server_name  localhost;
    root       /Users/YOUR_USERNAME/Sites;
 
    access_log  /Library/Logs/default.access.log  main;
 
    location / {
        include   /usr/local/etc/nginx/conf.d/php-fpm;
    }

acdha

# encoding: utf-8
"""Experimental Solr Grouping / Field Collapsing backend for Haystack 2.0"""
# NOTE: You must be running the latest Pysolr master - no PyPI release yet!
# See https://gist.github.com/3750774 for the current version of this code
# See http://wiki.apache.org/solr/FieldCollapsing for the Solr feature documentation
from __future__ import absolute_import

import logging

from django.db.models.loading import get_model