Artificiality Bites 💊 Issue #44

Hello Human! This is a new issue from my weekly newsletter, holding a tiny compilation made of interesting articles from last week, projects, tutorials and tools; all related to Data, Artificial Intelligence and adjacent topics. Nyd dit måltid!

📝 Interesting publications this week

• EleutherAI claims new NLP model approaches GPT-3-level performance 5'
  EleutherAI has released GPT-J-6B (aka GPT-J), a model the group claims performs nearly on par with an equivalent-sized GPT-3 model on various tasks.
• Improving Language Model Behavior by Training on a Curated Dataset 6'
  OpenAI researchers have found they can improve language model behavior with respect to specific behavioral values (reducing bias and toxicity) by fine-tuning on a curated dataset of <100 examples of those values.
• DeepMind scientists: Reinforcement learning is enough for general AI 13'
  These researchers suggest that reward maximization and trial-and-error experience are enough to develop behavior that exhibits the kind of abilities associated with intelligence.
• How long before AI can ‘understand’ animals? 8'
  AI might grant us the ability to reliably translate animals in the next decade or so.

Catherine Falls Commercial via Getty Images

• What the Heck is a Data Mesh?! 10'
  What's a Data Mesh and what's not.
• Tabular Data: Deep Learning is Not All You Need 11p
  This paper explores whether deep learning models should be a recommended option for tabular data, by rigorously comparing the new deep models to XGBoost on a variety of datasets.

🔧 Tutorials

• Python Data Viz Libraries Compared 14'
  A notebook containing 8 Popular Graphs Made with pandas, matplotlib, seaborn, and plotly.express.

📦 Repositories

• amundsen-io/amundsen
  Amundsen is a data discovery and metadata engine.
• IBM/UQ360
  The Uncertainty Quantification 360 toolkit provides a diverse set of algorithms to quantify uncertainty.
• jina-ai/jina
  Jina allows you to build deep learning-powered search-as-a-service in just minutes.
• wellecks/naturalproofs
  NaturalProofs is a multi-domain corpus of mathematical statements and their proofs, written in natural mathematical language.
• facebookresearch/flores
  Flores-101 is a Many-to-Many multilingual translation benchmark dataset for 101 languages, consisting of 3000 sentences extracted from English Wikipedia and carefully translated to the rest. More info here.
• graph4ai/graph4nlp
  Graph4NLP is a library for R&D at the intersection of Deep Learning on Graphs and Natural Language Processing.
• PrithivirajDamodaran/Gramformer
  Gramformer is a framework for detecting, highlighting and correcting grammatical errors on natural language text.

🎓 Courses / Events / Books

• Data Science at the Command Line 📕
  A public draft for the second edition of this book written by Jeroen Janssens.

• OpenCV Python Course 3h 📹
  Learn how to use OpenCV for Computer Vision and AI in this basic course for beginners.
• Hugging Face course 🤗
  A completely free course made by Hugging Face engineers that will teach you about natural language processing using libraries from the Hugging Face ecosystem.

👋 See you next week!