Iain Kirkpatrick iainkirkpatrick

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

In case anyone else wants to play with Zig on webassembly, here's what you need to do to make it work on a mac today.

1. Get LLVM 7 compiled with webassembly support.

You'll need LLVM to output to the WASM target. This has just been added by default in trunk, so if LLVM >7 is available, you might be able to just brew install llvm.

If you have wasm support already you should see:

$ llc --version

Exponent - ex-navigation

This is for now, for my personal use only, things might not be correctly explained here. For the official docs please check: https://github.com/exponentjs/ex-navigation/blob/master/README.md

Navigation bar configuration

On every screen you can use the built-in navigation bar, you can add a title, left button, right button or change navigation bar’s style. All you need to do is pass appropriate params to navigationBar in the route configuration:

import React, { Component } from 'react';

Why we dropped Lerna from PouchDB

We dropped Lerna from our monorepo architecture in PouchDB 6.0.0. I got a question about this from @reconbot, so I thought I'd explain our reasoning.

First off, I don't want this post to be read as "Lerna sucks, don't use Lerna." We started out using Lerna, but eventually outgrew it because we wrote our own custom thing. Lerna is still a great idea if you're getting started with monorepos (monorepi?).

Backstory:

https://pouchdb.com/2016/01/13/pouchdb-5.2.0-a-better-build-system-with-rollup.html

What is the difference between Cerebral and Redux?

Cerebral and Redux were built to solve different problems

Redux was developed to achieve hot reloading global state and state changing logic. To achieve that it was necessary for state changes to be run with pure functions and the state has to be immutable. Now you can change the logic inside your reducer and when the application reloads Redux will put it in its initial state and rerun all the actions again, now running with the new state changing logic.

Cerebral had no intention of achieving hot reloading. Cerebral was initially developed to give you insight into how your application changes its state, using a debugger. In the Redux debugger you see what actions are triggered and how your state looks after the action was handled. In Cerebral you see all actions fired as part of a signal. You see asynchronous behaviour, paths taken based on decisions made in your state changing flow. You see all inputs and outputs produced during the flow and you even

The introduction to Reactive Programming you've been missing

(by @andrestaltz)

This tutorial as a series of videos

If you prefer to watch video tutorials with live-coding, then check out this series I recorded with the same contents as in this article: Egghead.io - Introduction to Reactive Programming.

	/**
	* # `<Image>`
	*
	* This component is a merge between `next/image` and `Chakra-ui`.
	* - last updated on 2023-08-08 with `next/image` 13.4.13 and `chakra-ui/react` 2.8.0
	* - https://github.com/vercel/next.js/blob/v13.4.13/packages/next/src/client/image-component.tsx
	* - https://github.com/vercel/next.js/blob/canary/packages/next/src/client/image-component.tsx
	* - https://github.com/vercel/next.js/commits/canary/packages/next/src/client/image-component.tsx
	* - https://github.com/vercel/next.js/compare/v13.4.4...canary
	*

	const mapValuesAsync = (obj, asyncFn) => {
	const keys = Object.keys(obj);
	const promises = keys.map(k => {
	return asyncFn(obj[k]).then(newValue => {
	return {key: k, value: newValue};
	});
	});
	return Promise.all(promises).then(values => {
	const newObj = {};
	values.forEach(v => {

	{ fontWeight: '100' }, // Thin
	{ fontWeight: '200' }, // Ultra Light
	{ fontWeight: '300' }, // Light
	{ fontWeight: '400' }, // Regular
	{ fontWeight: '500' }, // Medium
	{ fontWeight: '600' }, // Semibold
	{ fontWeight: '700' }, // Bold
	{ fontWeight: '800' }, // Heavy
	{ fontWeight: '900' }, // Black

	/**
	* Fancy ID generator that creates 20-character string identifiers with the following properties:
	*
	* 1. They're based on timestamp so that they sort after any existing ids.
	* 2. They contain 72-bits of random data after the timestamp so that IDs won't collide with other clients' IDs.
	* 3. They sort lexicographically (so the timestamp is converted to characters that will sort properly).
	* 4. They're monotonically increasing. Even if you generate more than one in the same timestamp, the
	* latter ones will sort after the former ones. We do this by using the previous random bits
	* but "incrementing" them by 1 (only in the case of a timestamp collision).
	*/