Introducing 𝗦𝘂𝗽𝗲𝗿 𝗝𝗦𝗢𝗡 𝗠𝗼𝗱𝗲, a framework for low latency structured output generation from LLMs.

Generate JSON up to 𝟮𝟬𝘅 𝗳𝗮𝘀𝘁𝗲𝗿 from OpenAI and open source models.

❌ No need to threaten the model, tip the AI, etc ❌

Built with @derhacobian 🔧

🧵👇 1/ Suppose we want to extract the following characteristics from a house listing using an LLM:
- `address`
- `square footage`
- `number of bedrooms`
- `number of bathrooms`
- `price`

We could naively prompt the model to fill in this schema according to the description.

I connected ChatGPT to my personal health data on my iPhone.

Now, I can have a conversation with my digital health history.

The code is also public.

Say hello to HealthGPT 👋 I used it earlier this week to check in on my weekly gym goals.

Imagine a future where everyone has an personalized AI health assistant that can:
- track your data in a privacy-first manner
- understand your body's unique needs
- offer tailored advice towards your fitness goals

A short guide to engineering better GPT-3 prompts.

Here's how I redesigned GraphGPT's prompt to become faster, more efficient, and handle larger inputs: Some quick background:

GraphGPT is a tool that generates knowledge graphs from unstructured text using OpenAI's GPT-3 large language model.

https://twitter.com/varunshenoy_/status/1620511932930490372?s=20

Can LLMs extract knowledge graphs from unstructured text?

Introducing GraphGPT!

Pass in any text (summary of a movie, passage from Wikipedia, etc.) to generate a visualization of entities and their relationships.

A quick example: Here's GraphGPT creating a knowledge graph from the synopsis of Veep:

Are you building in the zero knowledge space?

Do you care about speed, energy usage, and cost efficiency?

Then, you NEED to understand hardware acceleration.

This is the only thread you will ever need to read about making ZKPs faster using hardware 🧵 1/ Some quick review:

Why do we care about ZKPs?

ZKPs enable one party to prove to another that a statement is true via a short proof without sharing more information than necessary.

Check out this thread for more info.

https://twitter.com/varunshenoy_/status/1571949645085155328

Confused about how zero knowledge proofs work under the hood? Don't know where to start?

I've got you covered.

Spoiler: you don't need a degree in math to understand zk-SNARKs 😉

Let's dive in 👇 1/ In this thread, we will cover some basic cryptography and build up to a high-level overview of zk-SNARK systems.

I recommend saving this thread for future use!

Zero knowledge proofs (ZKPs) will revolutionize how we think about privacy and scaling computation.

And they're just starting to take off.

Don't want to get left behind?

Here's a quick primer, just for you👇 1/ Despite becoming popular only recently in the context of crypto, ZKPs have a rich history.

In 1987, @JamesGleick published an article in The New York Times titled "A NEW APPROACH TO PROTECTING SECRETS IS DISCOVERED".

nytimes.com/1987/02/17/sci…