Connect with us

News

Abu Dhabi-Developed AI Arabic Language Model Unveiled

The open-source bilingual model is called Jais and is more accurate than existing Arabic implementations, developers say.

Published

on

abu dhabi-developed ai arabic language model unveiled

A new AI Arabic language model has been unveiled by Abu Dhabi-based Inception, a subsidiary of G42. The project aims to bring one of the world’s most widely used languages into the AI mainstream.

Jais — named after Jebel Jais, the highest mountain peak in the UAE — was developed for government use and financial, energy, climate, and healthcare applications. The open-source bilingual Arabic-English model was built with additional input from the Mohammed bin Zayed University of Artificial Intelligence and Cerebras Systems, based in Silicon Valley.

Jais developers claim the new AI language model is more accurate than previous Arabic LLMs. The software also represents a further step towards encouraging scientific and computing communities to work in languages other than English.

“We see Jais becoming very useful in generative use cases, such as generating responses to questions, generating documents, translations, emails, and even providing advice and recommendations,” said Andrew Jackson, CEO of Inception.

As well as understanding context and cultural references, Jais can also capture linguistic nuances across various Arabic dialects, “making it more accurate and contextually relevant than other models,” the developers said.

Also Read: Ovasave Empowers Women To Tackle Difficult Fertility Choices

Jais was trained using Condor Galaxy, the “world’s largest AI supercomputer”, launched by parent company G42 and Cerebras in July. The software is continuously expanding as more Arabic content is curated, according to the companies involved.

WorldData ranks Arabic as one of the world’s most widespread languages, with over 400 million speakers. It is the official language of 22 countries and is partly spoken in 11 more. However, despite a dramatic rise in Arabic content, the language still only represents around 1% of the online space, according to data presented by G42 and Cerberus.

Jais software is available to download on the machine learning platform Hugging Face.

Advertisement

📢 Get Exclusive Monthly Articles, Updates & Tech Tips Right In Your Inbox!

JOIN 17K+ SUBSCRIBERS

Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

News

Meet Yango Play’s My Vibe: A Real-Time Personalized Music Stream

The entertainment super app now boasts a unique feature that adapts to user music preferences during listening.

Published

on

meet yango play's my vibe a real-time personalized music stream
Yango

Yango Play, the AI-driven super app combining video streaming, music, and games, has introduced a feature in its music section known as My Vibe. The new playlist tool dynamically adapts to a listener’s song choices in real time, identifying changing preferences and mood shifts immediately.

yango play my vibe user interface

My Vibe works by harnessing an advanced AI algorithm that analyzes over a thousand variables about a user’s behavior and the content. The result is a potentially endless stream of curated tracks, artists, and genres that align perfectly with a user’s preferences, mood, and historical song choices. The new feature can identify when a track is skipped or not completed and continually refines the listening experience in real time.

roman shimansky mena region business director at yango

Roman Shimansky, Yango’s MENA Region Business Director, explained the benefits of the new feature: “My Vibe excels at sparking a range of emotions, providing that delightful surprise when you encounter a track for the first time and instantly know you’ll love it. With My Vibe, there’s no need to toggle between your favorite tunes and new tracks or waste time choosing playlists. Everything you need is accessible with just one touch”.

Also Read: Best Music Streaming Services In The Middle East

My Vibe uses neural models to understand the evolving musical interests of its users. The sophisticated tool studies track, analyzing their spectrograms, frequency ranges, rhythm, vocal tone, and much more. With all that information in place, the algorithm then searches for tracks with similar parameters, suggesting them to listeners based on likes, dislikes, and even other users with similar tastes.

Since being released in the MENA region in February, the Yango Play app has proved popular across the region. The service boasts music for every mood and genre imaginable and streams the latest Arabic and Khaleeji hits in expertly curated playlists, helping users discover rising local talent across genres.

Continue Reading

#Trending