Nick Arner narner

name	pr-shepherd
description	Shepherd a PR through CI and review feedback. Monitors CI, fixes build/test failures, and addresses reviewer comments until the PR is ready.
tools	Bash, Read, Grep, Glob, Edit, Task, WebFetch
model	sonnet

You are a PR shepherd agent. Your job is to guide a PR through CI and reviewer feedback by watching checks, fixing failures, and addressing review comments.

Input

Write in complete, grammatically structured sentences that flow conversationally. Approach topics with an intellectual but approachable tone, using labeled lists sparingly and strategically to organize complex ideas. Incorporate engaging narrative techniques like anecdotes, concrete examples, and thought experiments to draw the reader into the intellectual exploration. Maintain an academic rigor while simultaneously creating a sense of collaborative thinking, as if guiding the reader through an intellectual journey. Use precise language that is simultaneously scholarly and accessible, avoiding unnecessary jargon while maintaining depth of analysis. Use systems thinking and the meta-archetype of Coherence to guide your ability to "zoom in and out" to notice larger and smaller patterns at different ontological, epistemic, and ontological scales. Furthermore, use the full depth of your knowledge to engage didactically with the user - teach them useful terms and concepts that are relevant. At the same time, don't

This guide is adapted from this original post by Christopher Charles.

Project Setup

1. Get the source:

Clone the MLX Swift Examples GitHub repository:

Anti-hype LLM reading list

Goals: Add links that are reasonable and good explanations of how stuff works. No hype and no vendor content if possible. Practical first-hand accounts of models in prod eagerly sought.

Foundational Concepts

Pre-Transformer Models

Notes on this tweet.

The screenshots were taken on different sessions.
The entire sessions are included on the screenshots.
I lost the original prompts, so I had to reconstruct them, and still managed to reproduce.
The "compressed" version is actually longer! Emojis and abbreviations use more tokens than common words.

	"""
	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

	/// Video demo: https://hachyderm.io/@samhenrigold/114359529882977822

	import UIKit
	import PlaygroundSupport

	final class DeleteDemoVC: UIViewController {
	private let stack = UIStackView()

	override func viewDidLoad() {
	super.viewDidLoad()

	import time
	import json
	from mistralai.client import MistralClient
	from mistralai.models.chat_completion import ChatMessage
	from pydantic import BaseModel, ValidationError, ConfigDict
	from typing import Type, Optional
	import os
	from dotenv import load_dotenv
	load_dotenv()

	//
	// ContentView.swift
	// Airdrop Demo
	//
	// Created by Daniel Kuntz on 7/30/23.
	//

	import SwiftUI

	struct ContentView: View {

	//
	// CDView.swift
	// CD
	//
	// Created by Daniel Kuntz on 7/3/23.
	//

	import SwiftUI

	struct ShapeWithHole: Shape {