Hey there!

Did you crave some Ai bite-sized news over the weekend? We got you covered!

Here’s what we’ll learn about today:

• Meta has released its new Segment Anything Model 🏙️

• Your GhatGPT financial advisor 💸

• OpenAssistant: Free data for AI training! 💻

Reading time: 4 min and 30 seconds

News Reel:

Highlight 1: Meta’s last work is a Segment-Anything-Model

Last week, Meta released its latest project: a Segment-Anything-Model (SAM). But what can this new model truly bring to the table? Let’s find out!💥

• SAM is designed to locate and separate different objects given in an image or video in real-time…even if the objects recognized by the model have not been used to train the model itself!

• When faced with ambiguity about the object being segmented, SAM can output multiple valid options so that the user can choose the correct one.

• You can interact with the model by clicking and selecting objects with a bounding box.

Other than looking like some coloring book for kids 😉, SAM has an infinite number of applications: From instantaneously understanding both the visual and text content of a webpage to assisting in scientific studies of natural occurrences on Earth or space by localizing animals or objects to study and track in video🔬.

And that is not all! This model is open source 😎 and anyone can use it for research purposes. You can find all of these details in Meta’s article here. And you can also try this model with your own images using the Demo released by Meta here.

Highlight 2: BloombergGPT - ChatGPT for finance

For those folks passionate about finance, this tool will be mindblowing: Welcome BloombergGPT! We are talking about a machine-learning model that is designed specifically for finance 🤑. It uses natural language processing (the ability of computers to understand human text and speech) so that it can respond to user queries about finance. BloombergGPT has a vast amount of data at its disposal and can provide users with accurate and up-to-date information about the financial market. Its main applications are:

• Evaluating financial data quickly to help with risk assessments, gauge financial sentiment, and possibly even automate accounting and auditing activities.

• Assisting users with their financial needs by providing them with accurate and up-to-date information about the financial market.

• Helping people make better investment decisions by providing them with insights and analysis that they might not have access to otherwise.

BloombergGPT is a mind-blowing tool! To explain how powerful it is, just know that an AI algorithm is based on a looot of data to try to account for a vast variety of scenarios. BloombergGPT has 50 billion parameters (not very far from the very advanced GPT-3 which uses 175 billion parameters)🤯.

You can find more information about your future financial advisor here. In case, you are interested in getting a bit more technical you can access a fantastic resource on how neural networks like BloombergGPT’s function here 📚.

A.I. Tool showcase:

We all have heard of ChatGPT and subsequently a few other companies such as Microsoft and Google come up with their own version of an AI assistant. However, how does this process work? Can anyone create an equally good and powerful AI algorithm? 🪞

The answer lies in the learning process AI algorithms go through. They are like kids going through school and being exposed to a variety of concepts, formulas, and facts📚. For AI algorithms, such process is called “training” and the equivalents of books are “datasets“

As you can infer, good-quality books/datasets are essential for a good algorithm and that is exactly what OpenAssistant offers! In the last months, this project has been collecting large amounts of written input and feedback to create an incredibly diverse and unique dataset designed specifically for training language models or other AI applications.

The mind-blowing part is that all of these inputs were created by humans 🤩!

Yes, that is right: 13,000 people volunteered to provide the model with 150,000 messages in an immense number of different languages.

As its creator, Yannic Kilcher, explains on Twitter: “With over 600k human-generated data points covering a wide range of topics and styles of writing, this dataset will be an invaluable tool for any developer looking to create state-of-the-art instruction models!” 🤯.

OpenAssistant was released exactly 2 days ago providing an infinity of resources to all of those eager researchers and developers racing to create a new breakthrough in AI. And its novelty is not even the best news…this tool is absolutely free and open for everyone to access 😎.  You can access the presentation video here, and its more technical details here. Enjoy!

A.I. Art of the Week:

If you are like us, not ready to start the week, we hope that these cute AI puppies will make your Monday a little less tough 😍

See you Thursday!

-A.I. Bits