In today's weekly newsletter, we will cover some more updates from the world of AI.

Object Recognition Has an Income Problem
Object recognition neural networks are only as good as the data they’re trained on. And that data is heavy on images from high-income countries in Europe and North America. So, when confronted with everyday items from lower-income countries, they get it right as litte as 20 percent of the time.
IEEE SpectrumSamuel K. Moore
Transformer models: an introduction and catalog
In the past few years we have seen the meteoric appearance of dozens ofmodels of the Transformer family, all of which have funny, but notself-explanatory, names. The goal of this paper is to offer a somewhatcomprehensive but simple catalog and classification of the most popularTransformer models…
arXiv.orgXavier Amatriain
Prompt Engineering Lecture
A comprehensive overview of prompt engineering for language models
Elvis's Blogelvis
Smash or pass? This computer can tell: AI offers insight into conversations using physiology alone
Could an app tell if a first date is just not that into you? Engineers say the technology might not be far off. They trained a computer to identify the type of conversation two people were having based on their physiological responses alone.
ScienceDaily
MarioGPT: Open-Ended Text2Level Generation through Large Language Models
Procedural Content Generation (PCG) algorithms provide a technique togenerate complex and diverse environments in an automated way. However, whilegenerating content with PCG methods is often straightforward, generatingmeaningful content that reflects specific intentions and constraints remainsch…
arXiv.orgShyam Sudhakaran
Why you shouldn’t trust AI search engines
Plus: The original startup behind Stable Diffusion has launched a generative AI for video.
MIT Technology ReviewMelissa Heikkilä
Toolformer: Language Models Can Teach Themselves to Use Tools
Language models (LMs) exhibit remarkable abilities to solve new tasks fromjust a few examples or textual instructions, especially at scale. They also,paradoxically, struggle with basic functionality, such as arithmetic or factuallookup, where much simpler and smaller models excel. In this paper,…
arXiv.orgTimo Schick
An Overview of Epistemic Uncertainty in Deep Learning
In this article, we explored a broad overview of epistemic uncertainty in deep learning classifiers, and develop intuition about how an ensemble of models can be used to detect its presence for a particular image instance.
Paperspace BlogOreolorun Olu-Ipinlaye
How enterprises can use ChatGPT and GPT-3
Chatbot platforms like ChatGPT and GPT-3 can be valuable tools to automate functions, help with creative ideas, and even suggest new code and fixes for broken apps. But precautions are needed before companies move too fast.
ComputerworldLucas Mearian
A Dive into Vision-Language Models
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
adirik Alara Dirik
Towards Efficient Visual Adaption via Structural Re-parameterization
Parameter-efficient transfer learning (PETL) is an emerging research spotaimed at inexpensively adapting large-scale pre-trained models to downstreamtasks. Recent advances have achieved great success in saving storage costs forvarious vision tasks by updating or injecting a small number of parame…
arXiv.orgGen Luo
The AI industrial revolution puts middle-class workers under threat this time
In the past, leaps in technology replaced low-paid jobs with a greater number of higher-paid jobs. This time, it may be different
The GuardianLarry Elliott
The Illustrated Stable Diffusion
Translations: Vietnamese. (V2 Nov 2022: Updated images for more precise description of forward diffusion. A few more images in this version) AI image generation is the most recent AI capability blowing people’s minds (mine included). The ability to create striking visuals from text descriptions h…
Jay Alammar
URCDC-Depth: Uncertainty Rectified Cross-Distillation with CutFlip for Monocular Depth Estimation
This work aims to estimate a high-quality depth map from a single RGB image.Due to the lack of depth clues, making full use of the long-range correlationand the local information is critical for accurate depth estimation. Towardsthis end, we introduce an uncertainty rectified cross-distillation b…
arXiv.orgShuwei Shao

Videos this week

Events

Signup to stay updated

Over the next year, I will be holding multiple online meetups, each focused on a specific topic, such as:

Generative CMS
Generative LMS
Productivity Software with AI
Coding for search and social
Insights and analytics with AI
Generative AI for Sales and Marketing
Generative AI for Customer Service and Human Resource
Generative AI for Finance and Supply Chain

The agenda for each meetup will be to:
- Explain how current tools can be enhanced using AI
- Share code and build open-source tools that harness the power of generative AI
- Quickly revise some of the previous posts on AI and discuss what is coming up next

I am excited to bring this initiative to the community and I hope that you will join us in this learning journey. The first meetup is scheduled for Saturday, 25 Feb between 11:30 - 12:30 GMT. If you would like to join, please head over to https://everydayseries.com/events/.

Just so that it comes in handy, link for the kick-off meetup:

Everyday Event Series - Kickoff · Zoom · Luma
Welcome to the Everyday Series meetup where we talk about generative AI, new technology, research and much more. If you are not a member yet, please subscribe to...
Register

In case you missed

Last Week's Posts

PEFT
Large Language Models (LLMs) like GPT, T5, and BERT are very effective in processing natural language. They are trained on a large amount of data and then fine-tuned for specific tasks to achieve better performance. However, as these models become more considerable, full fine-tuning becomes impracti…
Everyday SeriesEveryday Series
GPT-1, GPT-2 and GPT-3
GPT-3, or Generative Pre-trained Transformer 3, is an artificial intelligence language model developed by OpenAI and is one of the most advanced language models ever created. But do you know the difference between the three models?
Everyday SeriesEveryday Series
Improving Language Understanding by Generative Pre-Training (GPT-1)
GPT 1
Everyday SeriesEveryday Series
Faster and Simpler Caching Commands for Streamlit
Streamlit has introduced two new caching commands: st.cache_data and st.cache_resource. These commands are faster and simpler than st.cache and will be used going forward as a replacement for it.
Everyday SeriesEveryday Series

We research, curate and publish daily updates from the field of AI.
Consider becoming a paying subscriber to get the latest!

Subscribe Now

Do follow us on our social media where we post micro-byte information daily.

YouTube: http://youtube.com/@everydayseries
TikTok: https://www.tiktok.com/@everydayseries
Instagram: http://instagram.com/everydayseries_
Twitter: http://twitter.com/everydaseries_