Ewen Wang Ewen2015

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.

	function generate_ssh_key () {
	read -p "Enter github email : " email
	echo "Using email $email"

	ssh-keygen -t ed25519 -C "$email"
	eval "$(ssh-agent -s)"
	ssh-add ~/.ssh/id_ed25519

	pub=`cat ~/.ssh/id_ed25519.pub`
	echo "$pub"

	pip install jupyter lab
	jupyter notebook --generate-config

	cd ~/.jupyter
	vi jupyter_notebook_config.py

	# # The IP address the notebook server will listen on.
	# c.NotebookApp.ip = '0.0.0.0' # default value is 'localhost'
	# c.NotebookApp.open_browser = False # default value is True

	import argparse
	import psutil
	import tensorflow as tf
	from typing import Dict, Any, Callable, Tuple

	## Data Input Function
	def data_input_fn(data_param,
	batch_size:int=None,
	shuffle=False) -> Callable[[], Tuple]:
	"""Return the input function to get the test data.

	<snippet>
	<content><![CDATA[
	# =============================================================
	# Basic
	import os

	# Fundamental Libraries for Scientific Computing
	import numpy as np
	import pandas as pd
	import scipy as sp

	require(caret)

	trans = preProcess(X, c("BoxCox", "center", "scale"))
	predictorsTrans = data.frame(trans = predict(trans, X))

	import plotly.io as pio
	pio.renderers.default = 'iframe'