Hi folks! 👋🏻 This is The Prompt! We are like a juicy BLT sandwich - we deliver crispy and fresh AI news to start your day. 🥓

Here's what we're serving today:

Control over AI image generation 👩‍🎨

The most annoying thing when generating images with Stable Diffusion (or any other diffusion model)?

Control over the outputs.

Before ControlNet, there was no good way to tell an AI model exactly what to do.

Now, they introduced 13 (!) implementations that control image-generation outputs.

The tech is quote powerful and produces very accurate results.

ControlNet + Stable Diffusion

Here are some examples of how ControlNet can control Stable Diffusion results:

Edge model: takes an image+text+edges as an input, and generates new images

Pose model: takes a pose+text using OpenPose and generates new images

Sketch model: takes a sketch+text and generates new images

Want to try it?

Here are the links:

1. Paper

2. Official Implementation

3. Video tutorial

4. Web UI on Hugging face (free)

5. Colab notebook (free)

💥 The Latest

-

Amazon to compete with ChatGPT 🤺

Amazon just made a partnership with HuggingFace to make, and I quote " faster and easier to build, train, and deploy the latest ML models in the cloud". And word on the street is that AWS will be powering an open-sourced ChatGPT competitor, which we kinda know is Bloom. But, the deal is not exclusive so HuggingFace can also work with other providers.

-

OpenAI x Coca-Cola partnership 🥤

OpenAI announced a strategic partnership with Bain to provide leading ML models & strategic advice to companies who want to jump on the AI Bandwagon. The first one is The Coca-Cola Company, and their goal is to make people more efficient & to create creative content at speed.

-

200+ e-books on Amazon that list ChatGPT as author 📚

Amazon now has a whole sub-genre dedicated to books about using ChatGPT or written by Chatgpt. With over 200 ebooks to choose from, it looks like there's a real chance for folks to make a killing in this new career path. Link to the article (no paywall)

📚 Learning lounge

• [Interesting read] We found a neuron in GPT-2

• [Long Thought provoking article] Text is all you need

• [Twitter thread] Build simple bot like ChatGPT with a few lines of code

🛼 Toolbox

• YoBulk: GPT3 powered 1-click CSV importer for your SaaS (OpenSource)

• DatingbyAI: AI helps you find your perfect match

• PressHook: Generate press releases with AI

• Fathom - AI meeting assistant that transcribes and summarizes your meetings

• ReadEasy.ai - Make text on your websites easy to read

• AskAI - Build your own AI Q&A, with any content in minutes.

• CMD+J: Use CMD + J to call ChatGPT everywhere

🤓 Latest AI papers

• ChatGPT: Jack of all trades, master of none

• Hyena Hierarchy: Towards Larger Convolutional Language Models

• Zero-Shot Information Extraction via Chatting with ChatGPT

• PC2: Projection-Conditioned Point Cloud Diffusion for Single-Image 3D Reconstruction

• Diffusion Models and Semi-Supervised Learners Benefit Mutually with Few Labels

• RealFusion 360◦ Reconstruction of Any Object from a Single Image

• Learning 3D Photography Videos via Self-supervised Diffusion on Single Images

📸 AI Photo of the day

Anatomy according to MidJourney

❤️ If you like The Prompt, and want to support my work:

• Share The Prompt with a friend, and invite them to subscribe here.

• Book an ad in The Prompt (reply to this email if you’re interested)

• Check out my latest tool: VoicePen.ai (audio to blog post)

Thank you for reading! ✌🏼