Thanks for reading! Ben’s Bites is a roundup of what’s going on in AI. Not subscribed? Do that here. Link overload? Refer 1 friend with this link and get access to the database.

Hey humans! Welcome to the 119 new folks joining us since yesterday.

I'm looking for some images that I can use to promote Ben's Bites. Bonus points for AI-generated (of course!!). Just reply to this email 🤗.

Almost the weekend, let's get to it.

Prompt: List the topics in today’s email:

🛻 Off-road autonomous driving🎬 AI-Images based on movies🔝 List of top AI companies🌅 Get prompts from your images

🫦 BEN'S BITES

• ToolBot³⁷⁹ — an app that generates custom GPT-3 tools using plain English.

• Swag and merch will never be the same again. You can train Stable Diffusion on TV shows or movies. Spider-Verse Diffusion²⁵⁰ is trained on, you guessed it, Into the Spider-Verse. And Arcane Diffusion¹⁶⁹ is trained on the TV show Arcane. There’s also a way to create your own unique characters²¹².

• Million-dollar homepage meets AI. An infinite collaborative canvas for Stable Diffusion³⁸².

• Stable Diffusion Inpainting Tutorial²¹² using the new Runway model

• Madrona released a list of the top private companies building intelligent applications³⁷⁷.

• An audio synthesizer capable of generating many types of sounds²²⁶. The next Calm/Headspace could be built with AI.

👋 Too many links?! I created a database for all links mentioned in these emails. Refer 1 friend using this link and I'll send over the link database.

• Create lofi artwork for your beats³¹⁰ with AI

• DiffusionDB is the first large-scale text-to-image prompt dataset²⁵². It includes 2 million images with their prompts.

• Generate images and stitch them together into a video with StableDiffusion-Interpolation Community Pipeline¹⁶⁹.

• Stable Diffusion’s launch was last week and they compiled a thread¹⁸⁴ of some projects they’re working on. (bonus: launch video¹⁴¹)

• Save highlights from youtube, newsletters and Twitter^271† in seconds.

• Harmonai³³⁵ - open-source generative audio tools to make music production more accessible and fun for everyone. From Stability AI.

• Upload an image → prompts get generated → an image is generated. See how Stable Diffusion would create your image.²⁶²

• Text semantics to ³²¹improve biomedical vision-language processing³²¹.

• Microsoft is hiring ⁸⁷research interns⁸⁷.

• Nvidia is hiring ^94†research interns^94†.

• OpenAI CTO, Mira Murati¹²⁹, talks about DALL·E 2 and the power of AI on The Daily Show¹⁵⁰ - although the YT video isn’t available in the UK.

• An SF-based LLM and AI hacking event^265†. Attendees get early private access to the DALL-E API on Tuesday 11/1/22.

• Jason, a research scientist at Google Brain, posted the slides he presented¹³¹ at NYU - “Large language models are different”.

• Confident Adaptive Language Modeling¹²⁴ (CALM), a framework for dynamically allocating different amounts of compute per input and generation timestep.

• GP-Tutor^185†: transform your Quizlet flashcards into conversation practice using AI! Demo video^135†.

• A way to train off-road autonomous driving without using real data¹²⁰.

• The Gradio demo for Mubert - text-to-music¹⁶⁴ is out on Hugging Face

• Ford, VW-backed Argo AI is shutting down⁹⁴. This thread¹²⁸ is a good explainer of the situation. TL;DR: they're focusing on driver assistance systems and not autonomous vehicle technology.

• Google Research is hosting its first AI event¹⁸⁵: “a showcase of our latest and greatest work in AI” on Nov 2 at 9 am EST.

• Improving speech-to-speech translation through unlabelled text¹³¹.

• New research³⁵⁴ shows how a single transformer can explore and solve tasks using this method without ever updating its parameters.

• Scientists discover material that can be made like plastic but conducts like metal¹³⁰.

• A website that analyzes Reddit posts and comments to find the most popular products using GPT-3.²⁶²

• An introduction to poisson flow generative models¹³⁹

🖼 AI IMAGES OF THE DAY

👋 SEE YA