Donavan Stanley Donavan

This was just a simple chunked summary using gpt-4 and 5k chunks

The video 'Life in 2323 A.D.' by Isaac Arthur presents a future panorama about technological advancements and lifestyle adaptations three centuries from now, using several fictional characters to emphasize the changing elements of daily life. In the future pictured, sophisticated technologies such as self-maintaining infrastructures and life extension technologies are subtly integrated into daily life. The characters, including Amy, who lives in a technologically advanced, eco-friendly suburban setting, and Becky, a cybernetically augmented great grandmother residing in a self-sufficient arcology, illustrate the far-reaching influence of technology.

Other charaters like Cameron and Duncan opt for a techno-primitive lifestyle, choosing external devices over implants. The video predicts an Earth population between 100 billion and a trillion, sustained by highly automated, climate-controlled greenhouses and an Orbital Ring enabling cheap, quic

pRoMpT eNgInEeRiNg IsN't A tHing!

The first file contains a closed caption transcript of the video Life in 2323 A.D. by Isaac Arhur.
The sceond file contains a garbage summary of said transcript.
The 3rd contains a much better, but still flawed, summary.

Since prompt engineering isn't a thing it should be no problem to reprodce either of them giving the model no information about the content aside from the title of the video and who made it.

Post a gist link in the comments...

Segmentation 101, part 1: Why your strategy matters

I recent did some more exploring with a local LLM tool that would import your documents into a vector store. Given the promising initial results with a handful of docs I wanted to see how it handled more / different data. I decided to copy over the text files containing Expanse trivia and answers I use as a regression suite to test my own "Q&A over documents" process. I wanted to see what types of questions it could answer from that content...

The Problem With Generic Segmentation

The strategy employed by this tool used double newlines as their segmentation boundary condition. A strategy that works well for many types of content however for this content that was a terrible choice as the text in the files are formatted with numbered questions followed by their answers like this:

1. Long winded question with establishing context

Self-Directed Q&A Over Documents

In the expanding universe of machine learning, the task of accurately answering questions based on a corpus of proprietary documents presents an exciting yet challenging frontier. At the intersection of natural language processing and information retrieval, the quest for efficient and accurate "Q&A over documents" systems is a pursuit that drives many developers and data scientists.

While large language models (LLMs) such as GPT have greatly advanced the field, there are still hurdles to overcome. One such challenge is identifying and retrieving the most relevant documents based on user queries. User questions can be tricky; they're often not well-formed and can cause our neatly designed systems to stumble.

In this blog post, we'll first delve into the intricacies of this challenge and then explain a simple yet innovative solution that leverages the new function calling capabilities baked into the chat completion API for GPT. This approach aims to streamline the retrieval

Conversation

conversationId: acfa33fb-d353-4ca0-8fb2-23867ea4514c
endpoint: openAI
title: Python Text Token Counting
exportAt: 19:29:58 GMT-0400 (Eastern Daylight Time)

Options

endpoint: openAI
presetId: null
model: gpt-4

	def __functions(self) -> List[Dict[str, Any]]:
	"""
	Extracts JSON schemas from the objects in the toolchest

	:return: A list of JSON schemas.
	"""
	if self.schemas is not None:
	return self.schemas

	self.schemas = []

	# This code can be executed in selenium via a javascript executor. It will change the source of an image that uses a blob to a data url instead
	# You will need to supply the ID of the image tag, or change to code to look it up some other way.

	# This line needs changed / passed in
	var image = document.getElementById(YOUR_IMAGE_ID);

	var blobUrl = image.src;
	var xhr = new XMLHttpRequest;
	xhr.responseType = 'blob';

	3 Ajani's Pridemate (M19) 5
	10 Plains (M19) 261
	1 Isolated Chapel (DAR) 241
	2 Legion Lieutenant (RIX) 163
	10 Swamp (M19) 269
	2 Skymarch Bloodletter (XLN) 124
	2 Inspiring Cleric (XLN) 16
	3 Call to the Feast (XLN) 219
	2 Epicure of Blood (M19) 95
	1 Herald of Faith (M19) 13

	4 Ghitu Lavarunner (DAR) 127
	4 Wizard's Lightning (DAR) 152
	4 Fanatical Firebrand (RIX) 101
	3 Lightning Strike (XLN) 149
	21 Mountain (RIX) 195
	4 Shock (M19) 156
	4 Viashino Pyromancer (M19) 166
	3 Goblin Chainwhirler (DAR) 129
	1 Risk Factor (GRN) 113
	4 Skewer the Critics (RNA) 115

	1 Conclave Guildmage (GRN) 162
	4 Sunpetal Grove (XLN) 257
	2 Leonin Warleader (M19) 23
	2 Divine Visitation (GRN) 10
	11 Plains (RIX) 192
	8 Forest (RIX) 196
	3 Ajani's Welcome (M19) 6
	4 Impassioned Orator (RNA) 12
	3 Ixalan's Binding (XLN) 17
	2 Healer's Hawk (GRN) 14