Hi folks!👋🏻 This is The Prompt! Get your thinking caps on because we’re here to spill the AI tea! ☕️

Multi-modalities + GPT4 are all the rage this week, so let’s see what’s new 👇🏻

FEATURED

Visual ChatGPT 🔥

VisualChatGPT connects ChatGPT and other visual models to enable sending and editing images in chat. (demo above)

They use some of the models that we’ve already covered like:

• ControlNet: Control AI results

• Stable Diffusion: text-to-image generation

• BLIP: image-to-chat

• InstructPix2Pix: photo editing with prompts

• ClipSeg: image segmentation

Seems like text is becoming the universal interface, and connecting to smaller models like these can be a good path to achieve multimodality.

Multi-modality is all the rage these days, and we also got a bit of a “tea spill” yesterday for GPT4 👇🏻

HYPE OR REAL?

GPT-4 launch next week - What do we know?

According to Heise, Microsoft Germany's CTO, Andreas Braun said that GPT-4 will come next week in a multimodal format.

This means that GPT-4 might be good at many different tasks than text generation, like understanding language, talking to people, and describing pictures.. and videos.

The amazing multimodality work of MSFS is not a secret. We already know that their newest model called Kosmos, can understand images and follow in-context instructions.

🤔 However, it’s a bit weird that the news is coming only from this source, and it’s not reported elsewhere. So, take these news with a grain of salt.

WHAT ELSE IS GOING ON

🦙 Discord announced the launch of its new AI features. Some of the features include AI Chat, AI-powred Mods, summaries, and avatars.

👀 DuckDuckGo launched DuckAssist. It summarizes answers from Wikipedia and displays them at the top of DuckDuckGo’s search results.

🏋🏻‍♀️  ChatGPT is now available in Azure OpenAI Service. With this update, developers can integrate custom AI-powered experiences directly into their own apps.

RESOURCES

The best resources we came across lately that will help you become better at writing prompts & building AI apps.

📚 Unlocking creativity with prompt engineering [a16z podcast]

👋🏻 Make 3D Models using ChatGPT [quick video tutorial]

🎥 ControlNet Full Tutorial [text & video tutorials]

TOOLBOX

The latest AI tools to use or get inspiration from.

• Rework: AI-driven recruitment

• OpenCommit: GPT CLI to auto-generate commits

• FlexClip: text-to-speech video maker

• Second: Developer bots that create new features or apps

• AI testimonial: Free AI testimonial generator

• WonderStudio: a mindblowing tool that automatically animates, lights, and composes CG characters into a live-action scene

  

PROMPT OF THE DAY

TOOL

Midjourney

PROMPT

Sony A1 camera with Sony Logo, intricate demonic black-paint biomechanical Sony A1 camera with Sony logo, scary, beautiful, circuits platemail armor. sleek iridescent details. domed toned glass, octane, volumetric lighting. sharpen, details + professional lighting, photography lighting 

RESULT

LATEST PAPERS

• Open-Vocabulary Panoptic Segmentation with Text-to-Image Diffusion Models

• Scaling up GANs for Text-to-Image Synthesis

• Open-world Instance Segmentation: Top-down Learning with Bottom-up Supervision

• 3D Video Loops from Asynchronous Input

• Cherry-Picking with Reinforcement Learning

• MathPrompter: Mathematical Reasoning using Large Language Models

• Cones: Concept Neurons in Diffusion Models for Customized Generation