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.

rnag

from __future__ import annotations

import builtins
from collections import deque
from dataclasses import field as dataclass_field, Field, MISSING
from logging import getLogger


log = getLogger(__name__)

myobie

on:
  push:
    branches:
    - master
  pull_request:
    types:
    - opened
    - reopened
    - synchronize
    - ready_for_review

ohmeqwertreysd

[colors]
accent = #fba922

[bar/top]
enable-ipc = true
monitor = LVDS
width = 100%
height = 27

;background = #222

durbon

1. Development environment

• How would you rate the development environment in team? (Use the [1-5] scale)
• How would you rate the project Settings needed to start to work? (Use the [1-5] scale)
• How would you rate documentation of the necessary tools? (Use the [1-5] scale)
• How do you rate the team’s current onboarding process? (Use the [1-5] scale)
• Do you think we have enough tools to work with? (Use the [1-5] scale)
• What tools are you missing in order to be more productive at work?

2. Code

• What is the level of satisfaction with the current codebase? (Use the [1-5] scale)

dshoreman

i3

Get Active Window ID

xdotool getactivewindow

Get Current Desktop ID

xdotool get_desktop_for_window "$(xdotool getactivewindow)"

bast

# adapted from https://web.archive.org/web/20190522110737/https://antergos.com/wiki/uncategorized/create-a-working-live-usb/

# first find out the device of your usb
sudo fdisk -l

# now that you know the letter, replace the "X" below
# typically "X" is replaced by "b" or "c" but it can be different on your system
# make sure it's not the device of your hard drive, otherwise you wipe it
sudo dd bs=4M if=/path/to/image.iso of=/dev/sdX status=progress
sync

GeorgeWeb

#include <chrono>
#include <vector>
#include <CL/sycl.hpp>
namespace sycl = cl::sycl;

using wall_clock_t = std::chrono::high_resolution_clock;
using time_point_t = std::chrono::time_point<wall_clock_t>;
template <typename T, class Period>
using time_interval_t = std::chrono::duration<T, Period>;

mgoldey

# install mkl
RUN apt update && apt install -y --force-yes apt-transport-https && \
  wget https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS-2019.PUB && \
  apt-key add GPG-PUB-KEY-INTEL-SW-PRODUCTS-2019.PUB && \
  sh -c 'echo deb https://apt.repos.intel.com/mkl all main > /etc/apt/sources.list.d/intel-mkl.list' && \
  apt-get update && DEBIAN_FRONTEND=noninteractive apt-get -y install cpio intel-mkl-64bit-2018.3-051 && \
  (find /opt/intel -name "ia32*" -exec rm -rf {} \; || echo "removing ia32 binaries") ; \
  (find /opt/intel -name "examples" -type d -exec rm -rf {} \; || echo "removing examples") ; \
  (find /opt/intel -name "benchmarks" -exec rm -rf {} \; || echo "removing benchmarks") ; \
  (find /opt/intel -name "documentation*" -exec rm -rf {} \; || echo "removing documentation") ; \

project-tuva

# -*- coding: utf-8 -*-
"""
Created on Wed Feb 14 16:17:38 2018
This handler is used to deal with logging with mpi4py in Python3.

@author: cheng

@reference: 
    https://cvw.cac.cornell.edu/python/logging
    https://groups.google.com/forum/#!topic/mpi4py/SaNzc8bdj6U