Text-to-3D

Hey everyone! Welcome to the 210 new folks joining since Friday.

I'm testing a new format today so you'll notice that I've split things into sections; Ben's picks - for any major stuff I think is worth checking out, cool tool - self-explanatory, research - any arXiv or scientific breakthroughs, and then everything else.

This is to help skim-ability and not overwhelm. Making it easier to pay attention to the stuff you want.

Please reply with 'better' or 'worse' so I can see your feedback :)

 🤌 Ben's Picks

• Magic3D from Nvidia: High-resolution text-to-3D content creation. It can create high-quality 3D textured mesh models from input text prompts and utilizes a coarse-to-fine strategy that leverages both low- and high-resolution diffusion priors for learning the 3D representation of the target content. Magic3D synthesizes 3D content with 8× higher-resolution supervision than DreamFusion while also being 2× faster. (link)

• Buildspace wrote a guide: Prompt Engineering 101. You can probably guess what it’s about 😉 but their images are so dope I’d check it out just to attempt to replicate similar. (link)

• MultiRay by Meta: a new platform for running state-of-the-art AI models at scale. It allows multiple models to run on the same input and shares the majority of the processing costs while incurring only a small per-model cost. (link)

• Canva text-to-image. Canva, the multi-billion dollar design app, is now able to generate images from text. It makes a lot of sense and was only a matter of time. (link)

🛠️ Cool Tools

• Whisper Memos is an app that records your voice and sends you an email with the transcription a few minutes later. (link)

• Alpaca - a next-generation design platform powered by generative models. (link)

• Get answers for CLI commands from GPT3 right from your terminal. (link)

• Getimg.ai - a suite of AI-powered image-generation tools. (link)

🔬 Research

• TART: follow human-written instructions to find the best documents for a given query. (link)

• The Twitter-to-arXiv pipeline for GPT-3 discoveries. (link)

• Ask4Help enables agents to request and use expert assistance. The policy is designed to efficiently train agents without modifying the original agent's parameters, and to learn a desirable trade-off between task performance and the amount of requested help. (link)

• Efficient exploration in reinforcement learning, called random curiosity with general value functions (RC-GVF). (link)

• Explaining the behaviour of AI systems. (link)

• CNeRV is designed to improve the reconstruction of visual data. (link)

• A new machine learning model, Distilled DeepConsensus - for genome sequencing. The model is faster and more accurate than the standard HMM-based methods and improves downstream applications of genomic sequence analysis. (link)

• CITADEL - efficient and effective multi-vector retrieval. CITADEL is designed to reduce the computation cost while maintaining high accuracy. (link)

• A new framework for efficient inference of MoE models with conditional execution of sparsely activated layers - Who Says Elephants Can't Run. This framework enables faster computation of sparse models and reduces memory consumption significantly. (link)

• SmoothQuant - a new method for quantizing large language models (LLMs). (link)

• Program-Aided Language models (PaL) use a language model to understand natural language problems and generate programs as the intermediate reasoning steps, but offloads the solution step to a programmatic runtime such as a Python interpreter. (link)

• GENIUS - a conditional text generation model that uses sketches as input. GENIUS is pre-trained on a large-scale textual corpus, and is able to generate diverse and high-quality texts given sketches. (link)

• Image completion that incorporates explicit structural guidance. (link)

• Renderdiffusion: image diffusion for 3d reconstruction, inpainting and generation. (link)

🤓 Everything else

• AssemblyAI released its Playground. You can test its transcription by using any YouTube video link, local audio file, or local video file. (link)

• Want to know what Deep Reinforcement Learning is? Here’s a quick thread. (link)

• Snorkel AI released its Data-centric Foundation Model Development, a new paradigm for enterprises to use foundation models to solve complex, real-world problems. (link)

• Interview with generative artist Gene Kogan - Between art and engineering: AI and expanding what it means to create. (link)

• If anyone is actually using Google Chat, get AI summaries of your conversations. (link)

• Scott Galloway wrote about his thoughts on AI. He’s become a bit of a meme for addressing technology and getting it very very wrong. Luckily here there aren’t any outlandish claims. (link)

• An overview of the United States vs China chips situation. (link)

• AltDiffusion is a multilingual text-image generation model built on Stable Diffusion. Currently supports English, Chinese, Spanish, French, Japanese, Korean, Arabic, Russian and Italian. (link)

• DiffusionDB: A large-scale text-to-image prompt gallery dataset based on Stable Diffusion. (link)

• I wrote a thread linking a bunch of educational resources to learn Machine Learning. (link)

• Accuracy and performance testing of OpenAI's transcription software. (link)

🧑‍💻 Who's hiring in AI

VEED.IO - Simple Online Video Editing. VEED is hiring AI / ML engineers to level up its creative toolkit and make it more magical.

Buildspace - where builders, build! They're looking for an ML/AI instructor to build their new course.

Add your company here | Join the Talent pool

🖼 AI IMAGES OF THE DAY

Woody Cage

🤗 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