This AI newsletter is all you need #10

Author(s): Towards AI Editorial Team

Originally published on Towards AI the World’s Leading AI and Technology News and Media Company. If you are building an AI-related product or service, we invite you to consider becoming an AI sponsor. At Towards AI, we help scale AI and technology startups. Let us help you unleash your technology to the masses.

What happened this week in AI

This week, our attention is still turned to Stable Diffusion. We had to cover it once more with even more fantastic news as an important step for the AI community. Stable diffusion is a fantastic initiative brought to us thanks to an amazing human being named Emad (learn more about Emad in Yannic’s video!) and his company, Stability.Ai. The idea behind Stability AI is to create and make publicly accessible important AI efforts from the community in direct competition with big companies.

Diffusion models recently achieved state-of-the-art results for most image tasks, including text-to-image with DALLE but many other image generation-related tasks too, like image inpainting, style transfer, or image super-resolution. Stable diffusion is based on the same principle, but within the latent space, which makes all the difference in computing needed for us regular people. If you still do not know what Stable Diffusion is or how latent diffusion in general works, we recommend this article for a better understanding of the model.

Hottest News

1. A GitHub repository full of pre-trained diffuser models!
   Thanks to HuggingFace, 🤗 Diffusers provides pretrained diffusion models across multiple modalities, such as vision and audio, and serves as a modular toolbox for inference and training of diffusion models.
2. Stable Diffusion API is out and you can use it!
   The Stable Diffusion model, a collaboration between Stability, LAION, and EleutherAI, has been released open source and is available on HuggingFace! Try it now 🙂
3. A free AI writing assistant built on top of OpenAI’s GPT-3
   The tool, by ParagraphAI, is able to generate articles, essays, and other original texts and is available for free on the Google chrome application store.

Most interesting papers of the week

1. DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation
   “With just a few images (typically 3–5) of a subject (left), DreamBooth — our AI-powered photo booth — can generate a myriad of images of the subject in different contexts (right), using the guidance of a text prompt.”
2. Z-Code++: A Pre-trained Language Model Optimized for Abstractive Summarization
   As the title says, Z-code++ is “a new pre-trained language model optimized for abstractive text summarization.”
   Wonder what is abstractive text summarization?
   “Abstractive summarizers paraphrase the idea of the source documents in a new form, and have a potential of generating more concise and coherent summaries.”
3. Playing for 3D Human Recovery
   They introduce GTA-Human, “a large-scale 3D human dataset generated with the GTA-V game engine, featuring a highly diverse set of subjects, actions, and scenarios. More importantly, [they] study the use of game-playing data and obtain five major insights…”. Get the code.

Enjoy these papers and news summaries? Get a daily recap in your inbox!

We will be at ECCV 2022 — come meet us!

Thanks to our friends at Qwak and their help, our head of community, Louis Bouchard, will be at ECCV 2022 in Tel Aviv (October 23–27). Feel free to reach out if you think you will be attending and would like to have a coffee or chat there!

The Learn AI Together Community section!

Meme of the week!

Featured Community post from the Discord

altryne#7376 created an awesome list of resources for image synthesis (aiArt). Altryne regroups the different text-to-image models, prompt engineering resources, interesting communities and many useful tools for anyone interested in doing AI arts.

Give it a look!

AI poll of the week!

TAI Curated section

Article of the week

Why Should Adam Optimizer Not Be the Default Learning Algorithm? An increasing share of deep learning practitioners are training their models with adaptive gradient methods due to their rapid training time. Adam, in particular, has become the default algorithm used across many deep learning frameworks. These adaptive optimization methods are known to generalize poorly. The author comprehends the findings of empirical studies carried out to mull over the marginal value of adaptive gradient methods.

If you are interested in writing for us at Towards AI, please sign up here and we will publish your blog to our network if it meets our editorial policies and standards. https://contribute.towardsai.net/

Lauren’s Ethical Take on Twitter’s Whistleblower

The news that I want to spotlight this week regards Twitter whistleblower Peiter “Mudge” Zeiko’s recent allegations over serious data mismanagement, security weaknesses, poor tracking of bots, vulnerability to foreign intelligence, and possible FTC violations. We’ve seen big tech whistleblowers before, and some people have become desensitized or resigned to the nature of the beast due to the news cycle or a cynical outlook on how tech should operate in society. The news cycle’s initial spike, talk of big things happening and the eventual simmering down of the hype don’t help the feeling of impermanence. Yet something tells me this story will be here to stay.

The fact that all of these allegations are so egregious if true, and the gravity of the situation may prove that resulting penalties can have serious do-or-die consequences beyond mere warnings. Regulation might move slowly with respect to tech, but it can be powerful. The exposure of multiple foundational flaws that are brittle and have grown weaker over time shows the costs of building fast and growing so large that no one knows what’s happening anymore. The fate of Twitter may serve as a very important lesson for the perils of building an online community unsustainably. More than just what allegations prove true or untrue is how they will be dealt with by both regulators and users. This will be an important story to follow in the coming months, and I look forward to the insights we learn from this experience.

Featured jobs this week

Computer Vision Scientist — Percipient AI (Santa Clara, CA)

Research Scientist — Machine Learning @ DeepMind (London, United Kingdom)

Senior Data Scientist @ EvolutionIQ (Remote)

Senior ML Engineer — Algolia AI @ Algolia (Hybrid remote)

Senior ML Engineer — Semantic Search @ Algolia (Hybrid remote)

Interested in sharing a job opportunity here? Contact [email protected] or post the opportunity in our #hiring channel on discord!

If you are preparing your next machine learning interview, don’t hesitate to check out our leading interview preparation website, confetti!

This AI newsletter is all you need #10 was originally published in Towards AI on Medium, where people are continuing the conversation by highlighting and responding to this story.

Join thousands of data leaders on the AI newsletter. It’s free, we don’t spam, and we never share your email address. Keep up to date with the latest work in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI