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

jnsahaj

# Create a new clone and branch for parallel development.
# Usage: ga <branch-name> [base-branch]
ga() {
  if [[ -z "$1" ]]; then
    echo "Usage: ga <branch-name> [base-branch]"
    return 1
  fi
  local branch="$1"
  local repo_name="$(basename "$PWD")"
  local repo_url="$(git remote get-url origin)"

briandk

# The difference between the `sep` and `collapse` arguments
# in paste can be thought of like this:
# 
# paste can accept multiple *vectors* as input, and will
# concatenate the ith entries of each vector pairwise 
# (or tuplewise), if it can.
# 
# When you pass paste multiple vectors, sep defines what
# separates the entries in those tuple-wise concatenations.
#