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.

talperetz

bsodmike

Following mining and findings performed on EVGA GeForce GTX 1070 SC GAMING Black Edition Graphics Card cards.

First run nvidia-xconfig --enable-all-gpus then set about editing the xorg.conf file to correctly set the Coolbits option.

# /etc/X11/xorg.conf
Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"

FiloSottile

#! /bin/bash
set -euo pipefail

# This script will remove automatic association for all networks not listed in the whitelist
# passed as the first argument. Passwords will NOT be removed from the Keychain.
#
# Alternatively, you can untick "Remember networks" in Network Preferences > Wi-Fi > Advanced,
# but then you won't be able to auto-join networks even temporarily, and you might already
# have a long list to go through.
#

dermesser

# include <stdlib.h>
# include <stdio.h>

# include <sys/stat.h>

# include <pthread.h>

# include <fcgiapp.h>

const char* const sockpath = "/tmp/fcgicpp.sock";