Focus AI # 112
#StableDiffusion #DeepMind #Shutterstock #Karpathy
Focus AI #112, the latest AI news in 3 minutes
ps: if you like this content, feel free to share it or to contribute with a small (or large) gift subscription)
This week:
🙈 DeepFace: State-of-the-Art Face Attribute Analysis in Python
📸 Facebook and Shutterstock partner to train gen models
🧠 DeepMind makes a general-purpose RL algorithm
👨⚖️ Copilot class action lawyers prepare lawsuit against StableDiffusion
And about code:
🔐 AIJack: Security and Privacy Risk Simulator for Machine Learning
🐶 nanoGPT
🐻❄️ Modern Polars
🔥 Let's build GPT: from scratch, in code, spelled out.
== News ==
🙈 DeepFace: State-of-the-Art Face Attribute Analysis in Python
In this tutorial we go over how to use DeepFace. Deepface is a lightweight face recognition and facial attribute analysis (age, gender, emotion and race) framework for python. Hope you learn something new!
📸 Facebook and Shutterstock partner to train gen models
Facebook and Shutterstock have extended their partnership, giving the social network a greater ability to use Shutterstock's vast archive of images to train machine learning models. This follows Shutterstock earlier partnering with OpenAI and also LG AI Research.
🧠 DeepMind makes a general-purpose RL algorithm
Researchers with DeepMind and the University of Toronto have built DreamerV3, a "general and scalable [RL] algorithm based on world models that outperforms previous approaches across a wide variety of domains with fixed hyperparameters". In other words, it's one system which you can train on different tasks without too much fiddling - and it works well!
👨⚖️ Copilot class action lawyers prepare lawsuit against StableDiffusion
Matthew Butterick, the lawyer and programmer who instigated the class action suit against Microsoft, GitHub, and OpenAI over Github Copilot, has now filed a class-action complaint against Stability AI, DeviantArt, and Midjourney over the 'Stable Diffusion' AI art model.
The gist of the lawsuit is that "Stable Diffusion contains unauthorized copies of millions—and possibly billions—of copyrighted images.
🌦️ Machine Learning in Weather & Climate
The objective of the MOOC is to train a wider community on the impact and use of Machine Learning in numerical weather and climate predictions. It brings together experts throughout Member and Co-operating States and beyond to provide a shared vision across the communities of Earth system sciences, high-performance computing and Machine Learning.
== Code & Tools ==
🔐 AIJack: Security and Privacy Risk Simulator for Standard/Distributed Machine Learning
AIJack allows you to assess the privacy and security risks of machine learning algorithms such as Model Inversion, Poisoning Attack, Evasion Attack, Free Rider, and Backdoor Attack. AIJack also provides various defense techniques like Differential Privacy, Homomorphic Encryption, and other heuristic approaches.
🐶 nanoGPT
The simplest, fastest repository for training/finetuning medium-sized GPTs. It is a rewrite of minGPT that prioritizes teeth over education
🐻❄️ Modern Polars
This is a side-by-side comparison of the Polars and Pandas dataframe libraries, based on Modern Pandas by Tom Augsburger.
(In case you haven’t heard, Polars is a very fast and elegant dataframe libary that does the same kinds of things Pandas does.)
🔥 Let's build GPT: from scratch, in code, spelled out.
With Andrej Karpathy, we build a Generatively Pretrained Transformer (GPT), following the paper "Attention is All You Need" and OpenAI's GPT-2 / GPT-3.
⚖️ Bringing "balance" to your data
A python package for balancing biased data samples.
#StableDiffusion #DeepMind #Shutterstock #Karpathy
A big thanks to our sources: https://jack-clark.net/, https://www.actuia.com/, https://thevariable.com/news/, https://techcrunch.com/, https://read.deeplearning.ai/the-batch/
What about you? Have you noticed something else?
Don’t miss the next news!
Have a good week-end,
Maxime 🙃 from Toulouse, France with 🌺