Ben's Bites
Posts
AI proteins

AI proteins

Ben Tossell
December 14, 2022

It's Tool Tuesday Wednesday, oh, erm, anyway, todays email is a little tool-heavy because Assembly just announced their hackathon winners. But we like the tools don't we?!

*everyone cheers: 'yeahhhh we love the toooooools'*

I'm getting deal-flow from AI companies in their pre-seed which I'm going to invest in. If you are an accredited investor and you'd potentially want to invest in the space in some capacity (angel, back a fund etc). Please reply to this email 😊

Let's get to it.

🤌 Ben's Picks

Luma Imagine 3D - A new way to create 3D with text. (link)

World’s first robot lawyer: DoNotPay wants to build an AI to help people fight traffic tickets. (link)
A new Protein Design Era with Protein Diffusion. (link)
Assembly just wrapped up its hackathon and these were the winners:
- Superpaint - A drawing + image editing app powered by AI. (link)
- Toy Story Creator - Bring toys to life for bedtime stories. (link)
- OperatorAI - AI system that helps triage 911 calls and connect callers with emergency services. (link)
- DocSpace - tool for analysing legal documents with AI. (link)
- Pupil-ai - Enhance online self-learning with the power of generative AI. (link)
- Supercut - an autonomous podcast editing assistant, that automatically creates short social video clips and tweets from your podcast's video recording. (link)
- Photostudio - Create stunning photoshoots for your products in seconds. (link)
- Scaling Potato - AI-powered technical interview prep. (link)
- MaiBook - A fun and personalised learning platform for kids. (link)
- MidJournal - Effortless and enjoyable journaling with AI. (link)
- Assembly AI hackathon project gallery (link)

🛠️ Cool Tools

One More AI - Unsplash for AI-generated images from multiple sources. (link)
SheetGod - GPT-3.5 powered tool that helps you create excel formulas, macros, and code snippets from plain text. (link)
XP1 - GPT-based assistant with access to your browser tabs to help you with your daily tasks. (link)
Ad Auris Play - Turn any article into audio and create a playlist to listen to as podcasts. (link)
ChatGPT Everywhere - Chrome extension to get ChatGPT on every page of your browser. (link)
ClipDrop Replace Background - Create any background for your product with AI. (link)
Native Swift app for Stable Diffusion - App that shows how to integrate Apple's Core ML Stable Diffusion implementation in a native Swift UI application. (link)
Tome - Generative storytelling app, in BETA. (link)
Bedtimestory - Personalised, instant bedtime stories created in seconds. (link)
Unofficial Hugging Face integration for Zapier. (link)
Creative Reality Studio - Platform to combine GPT-3, Stable Diffusion and D-ID’s unique face animation technology to create future-facing videos. (link)
‍AI Data Sidekick - Write SQL, documentation and more 10x faster for free. (link)
Mindy - Brainstorm and flesh out ideas through mind mapping and chatting with AI. (link)

Join Discord | Hackathon | Partner with us | Host or join a meetup

🎓 Learn

Create 100 programmatic SEO pages in 20 minutes with OpenAI + Webflow + Whalesync. (link)
How to replace, reimagine and remix any background, in any photo with Runway's Backdrop Remix. (link)
How to build a book recommendation site with ChatGPT & Webflow. (link)

🔬 Research

Data2vec 2.0: a new general self-supervised algorithm built by Meta AI for speech, vision & text. (link)
Robotics Transformer 1 (RT-1), a multi-task model that tokenises robot inputs and outputs actions to enable efficient inference at runtime. (link)
Robosuite - A modular simulation framework and benchmark for robot learning. (link)
Seekart MEGA, a general-use "anything" model that significantly improves on SD 1.5 across dozens of styles. (link)
DifFace, for blind face restoration that is capable of dealing with unseen and complex degradations without complicated loss designs. (link)
Neural technique for learning to select a local sub-region around a point which can be used for mesh parameterisation. (link)
Structured 3D features for reconstructing relightable and animatable avatars. (link)
GPViT: a high-resolution non-hierarchical vision transformer with group propagation. (link)
FastMIM: expediting masked image modelling pre-training for vision. (link)
Structure-guided image completion with image-level and object-level semantic discriminators. (link)
DiffStack, a differentiable and modular stack for prediction, planning, and control. (link)
ScanEnts3D: exploiting phrase-to-3d-object correspondences for improved Visio-linguistic models in 3d scenes. (link)
You only need a good embedding extractor to fix spurious correlations. (link)
Structured Prompting: scaling in-context learning to 1,000 examples. (link)
What do vision transformers learn? A visual exploration. (link)
ERNIE-Code: beyond English-centric cross-lingual pretraining for programming languages. (link)

👋 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.

🤓 Everything else

VCs weigh OpenAI tender offer as Microsoft investment talks remain ongoing. (link)
What is ChatGPT and will AI replace programmers and content creators? (link)
How to use AI to teach some of the hardest skills. (link)
TechCrunch article on “OpenAI’s attempts to watermark AI text hits limits.” (link)
Explaining the New AI Paradigm. (link)
How Neeva will change the search experience. (link)
StableDiffusion REST API now supports multi/ negative prompting. (link)
GPT Index: a simple, flexible interface between your external data and LLMs. (link)
A subreddit for entrepreneurs to stay in the loop about how advances in AI can make running or starting a business easier. (link)
Yann LeCun elaborates on their latest research and critiques the chain of citations in a latest interview. (link)
A thread explaining RT-1, a robotic model that can execute over 700 instructions in the real world at 97% success rate. (link)
The New Yorker article, “The world-changing race to develop the quantum computer.” (link)