Ben's Bites
Posts
GPT-3 playground

GPT-3 playground

🛝 A GPT-3 playground 📃 Upload any paper → get an explanation 💬 Sequoia AI office hours 🤖 Stable Diffusion API

Ben Tossell
October 25, 2022

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 filterable+searchable database.

Hey folks, welcome to the 179 new readers who joined us yesterday. We're close to 2k!

Some of you have asked about a discord group to connect and discuss new going's on in AI. If you'd be interested in something like that. Let me know by clicking below.

Are you interested in a Discord group?

Yes | Double yes

Let's get to it!

Prompt: List the topics in today’s email:

🛝 A GPT-3 playground📃 Upload any paper → get an explanation💬 Sequoia AI office hours🤖 Stable Diffusion API

🫦 BEN'S BITES

A playground for GPT-3; folders, team support, and more. If you store your prompts in your notes like me, now you can use Every Prompt. I’ve been a beta tester and spoke with the founders - you should check it out.

Review pull requests, automatically. Your AI-powered code review assistant, What The Diff.

Upload a paper → get an explanation. Explainpaper (mentioned previously) has now released a way to upload any paper and get an explanation and ask for follow-ups. Finally, I can pretend to better understand all these research papers now!

Scale (a huge data AI company) hosted a virtual conference and got a ton of big names to speak all about AI. You can now view all of the recordings here. Some notable speakers include; Eric Schmidt, Nat Friedman, Emad Mostaque, Greg Brockman and Mike Schroepfer.

I tried very hard to watch and summarise this fireside chat (Emad Mostaque, CEO of Stability.AI - Hosted by Elad Gil), but the bloody builders thought it was a great idea to drill and cut all morning. And I've got a funeral to attend so instead, here's a list of topics they discussed; *Stability.AIs journey *Stable Diffusion and Generative Models *Image gen use cases & startup opportunities *Other types of generative models *Open source & AI - what does the world look like in 5 years? *AI safety *Global AI. .

Google’s former CEO Eric Schmidt was appointed to the National Security Commission on AI. He made direct investments in AI startups. There’s no indication that he did anything wrong but government ethics advisors say his investments presented a huge conflict of interest.

Behind the scenes of that crazy video with masks, and effects using Runway that I wrote about last week.

👋 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.

GPT-3 creating both sides of an Interactive Fiction transcript.

Bumble open-sourced its AI that detects unsolicited nudes. I wonder if they got that bug fixed which meant I never got any matches?

The Value Of Artificial Intelligence Isn’t Just Insights.

Making sure robots don’t bump into each other (in factories etc), is a core building block. This model can generate collision-free, smooth motion from just a single depth camera observation.

Sequoia has (quickly) updated its Gen AI market map. Here’s a V2. They’re also hosting AI office hours which you can sign up for here.

[Por]TrAIts: AI Characters + Their Microstories [Book 1] has been released. It’s a fusion of AI-generated art and microstories written in the English/code-hybrid language termed Mezangelle. Created by a human and DALL-E.

Speech to image using Stable Diffusion and Whisper. Demo on HF, code on GitHub.

A letter to Mark Zuckerberg and the Meta Board. Brad believes the future is AI and Meta should get after it, but also keep the spending ‘reasonable’ ($20-25Bn) 😳.

The form tool with AI baked in (mentioned last week), has posted a ‘how-to’ guide for using AI to improve user onboarding as they did.

Wanna know what kinds of biases are hidden in your large language model? Hugging Face has made some tools to help!

Lachy Groom has created bleeding edge—a feed of noteworthy developments in AI.

Compositionality in language is important. However, many phrases aren't semantically compositional. How do language models deal with this? Moreover, how do humans deal with this?

FreeCodeCamp has released a course "The basics of Machine Learning".

Sam Lessin wrote a screenshot essay on “Why you should be ‘default out’ on investing in ‘Generative AI’ companies for the next few years.” - I like to throw in some against-the-grain points every now and then. And Sam makes some good ones here.