Google search competitor

It's Fridayyyy thennnn...it's Saturday, Sunday what am I doing writing that in here 🤦‍♂️. Welcome to the 114 new folks joining us from yesterday.

Tonight's menu will be Lobster cocktail + Crispy wings (shared w/ my wife because she always makes me share), Dirty Burger or Cote de Boeuf or Satay Curry (I'm torn), and Mango cheesecake. Food coma, a glass of wine or 4, then bed. Bada Bing. Bada Boom.

Have an indulgent weekend, let's get to it.

Prompt: List the topics in today’s email:

🫦 BEN'S BITES

A new type of search engine that promises to be more effective than current ones. Ahem, ad, ad, ad, result. Metaphor is a search engine that understands language – in the form of prompts – so you can type what you're looking for in all the expressive and creative ways you can think of. I encourage you to read the blog post announcement which provides a bit more context.

Do you remember “Code as Policies” - using language models to write robot policy code from language instructions? Well, now there’s a demo simulation you can use on HF.

Stable Diffusion demo spaces now run twice as fast!

Ben Thompson interviewed the founder of Midjourney, David Holz. It’s paywalled and too long to summarise succinctly…..but I've made a pdf of it and posted it to Discord here (shh don't tell).

CycleDiffusion - a method for translating images that support random sampling for diffusion models.

If you use Raycast (a mac app launcher and productivity tool), Joe built an OpenAI integration.

Machine Learning algorithms cheat sheet.

modl.ai (an AI engine for game development) is hiring.

GitHub’s working on a tool that writes code from your voice. “Hey, GitHub!” enables voice-based interaction with GitHub Copilot, enabling the benefits of an AI pair programmer while reducing the need for a keyboard. We can’t be that far off not needing to know any code, you speak what you want and it gets built—we’re already seeing this with text → code. Also announced in ‘GitHub Next’ - AI for Pull Requests.

Brothers, Rahel and Sacha, won first place at the LABLAB Transformers AI hackathon (out of 1000+ people). They built a tool for semantic search and clustering of research papers.

Emad (Stability.ai) wrote a short thread on how there are going to be big advances (again) in AI in the coming weeks from text to image etc. Essentially, it’s because of the rapid pace of research being done.

Federated learning - a machine learning approach that allows for collaboration among entities without the need to share raw data.

Steamship - the fastest way to add language AI to your product. They also have a CLI that remembers commands you always forget.

OpenAI startup fund has made its first investment (that we know of). It leads a $23.5M round in Mem, an AI-powered note-taking app.

You can now create Popup-books in Stable Diffusion.

AnimeRun - a correspondence dataset for 2D-styled cartoons. Its designed to facilitate the automatic processing of 2D animation.

Lexica is now showing off its results for home interiors, and YES they look stunning. I’m putting my house on the market today so I can focus on an AI-generated one.

MMDialog is a dataset to better facilitate multi-modal conversation. Its composed of a curated set of 1.08 million real-world dialogues with 1.53 million unique images across 4,184 topics.

GANStrument is a novel generative adversarial model for instrument sound synthesis. Basically, given a one-shot sound as input, it is able to generate pitched instrument sounds that reflect the timbre of the input within an interactive time.

The benefits of large-scale multilingual ASR models. The authors compare two architectures and show that the multiple embedding and output model outperforms the shared model.

Self-supervised learning (SSL) for speech separation (SS), with the aim of scaling up both the amount of training data and the efficiency of computation.

A new algorithm for training a neural network language model (NNLM) that is more effective than existing methods.

A new transformer-based framework called StyleNAT, which is designed for high-quality image generation with superior efficiency and flexibility.

Colossal-AI has created a solution that reduces the pretraining cost and hardware costs of fine-tuning AIGC models.

Sparrow is Amazon’s new intelligent robotic system that streamlines the fulfilment process by moving individual products before they get packaged—a major technological advancement to support our employees.

An AR inpainting experiment called We See. A proof of concept showing how voice command, selection gesture, Stable Diffusion and Augmented Reality can come together to alter the reality around us.

🖼 AI IMAGES OF THE DAY

The Girl with the pearl earring as a Pixar character

🤗 SHARE BENS BITES

Send this with 1 AI-curious friend and receive my AI project tracker database!

or copy/paste this link: https://bensbites.beehiiv.com/subscribe?ref=PLACEHOLDER

👋 SEE YA

⭐️ HOW DID WE DO?

⭐️ REAL REVIEWS