Super-Powered AI Advances: Zuck’s Llama 3.1, Musk’s Cluster and More

Happy Wednesday, AI Enthusiasts!

In today’s edition: 

• The Most Powerful AI Training of GROK 3 Supercluster Is on Its Way

• Meta Unveils World's Largest Open AI Model: Llama 3.1

• Accenture Advances Custom Llama LLM Models with NVIDIA’s AI Foundry

• Mark Zuckerberg Unveils Meta AI's Latest Feature

- Naseema Perveen

WHAT CAUGHT OUR ATTENTION MOST

The Most Powerful AI Training of GROK 3 Supercluster Is on Its Way

Elon Musk announced the commencement of training for the GROK 3 supercluster, described as "the most powerful AI cluster in the world," using 100,000 NVIDIA H100 GPUs at Memphis. 

This milestone marks the start of a significant venture for Musk's xAI company, leveraging NVIDIA's top-tier data center GPUs to develop advanced AI capabilities. The training began at 4:20 am local time, with Musk proudly highlighting the collaborative effort of the xAI, X, and NVIDIA teams in achieving this feat.

The project aims to build an AI system surpassing previous capabilities, with GROK 2 expected to launch next month and GROK 3 by December 2024. This ambitious timeline follows the recent decision by xAI to end its $10 billion server deal with Oracle, opting instead to build its supercomputer infrastructure. 

This choice was made despite NVIDIA's impending release of the newer H200 GPUs, which promise up to 45% better response times. Musk's goal is to complete the training of this advanced large language model by the end of 2024, positioning it as the fastest AI system globally.

IN PARTNERSHIP WITH QUANTUM TECH

Join Quantum.Tech Europe on September 23^rd - 25th, held at Twickenham Stadium

As the only dedicated event that brings together the full quantum ecosystem, join us at Quantum.Tech Europe, Twickenham Stadium, alongside 400+ quantum thought leaders!

Don’t miss your opportunity to immerse yourself in the cutting edge developments of 2024 with the brand new EXPO ONLY PASS giving you access to explore the buzzing exhibition floor including the Innovation and Ecosystem Stage.

From Gen AI, NISQ to the Fault Tolerant era, with your conference pass, sit in and learn from government figures, global enterprises and Fortune 500 organizations debating, discuss and presenting the strategies your business should be implementing to benefit from the power of quantum computers.

Use code AIJOURNAL20 for 20% off! Sign up today!

KEEP YOUR EYE ON IT

Meta Unveils World's Largest Open AI Model: Llama 3.1

Meta has introduced the Llama 3.1 model, which it claims is the "world’s largest and most capable openly available AI foundation model." This new release aims to compete with major closed-source models like OpenAI’s ChatGPT and Google's Gemini, offering advanced capabilities in general knowledge, mathematics, and multilingual translation. Meta views this launch as a transformative moment, asserting that open-source models are now leading in performance and capability.

The Llama 3.1 405B model is set to be publicly available starting today, and Meta anticipates it will set new standards for open-source AI. With over 300 million downloads of its Llama models to date, Meta highlights that this new model also excels in synthetic data generation, a feature previously unmatched in open-source AI. Meta's latest Llama models are reported to be highly competitive, testing well against flagship models from OpenAI and Anthropic.

Accenture Advances Custom Llama LLM Models with NVIDIA’s AI Foundry

Accenture has launched the Accenture AI Refinery™ framework, powered by NVIDIA AI Foundry, to empower organizations in developing custom LLM models using the newly released Llama 3.1 collection. This framework allows businesses to fine-tune Llama models with their data, creating tailored AI systems that address specific industry needs. According to Julie Sweet, Accenture's chair and CEO, this initiative leverages generative AI's transformative power, enabling companies to build unique AI applications that support their strategic goals.

Jensen Huang, founder and CEO of NVIDIA, highlighted the critical role of Meta's Llama models in advancing enterprise AI capabilities. The Accenture AI Refinery offers end-to-end generative AI services, including domain model customization, data indexing, and autonomous AI system functionality. Available across all major cloud platforms, this solution reflects Accenture's commitment to driving digital transformation and delivering comprehensive, value-driven solutions to its clients and partners worldwide.

Mark Zuckerberg Unveils Meta AI's Latest Feature

Meta CEO Mark Zuckerberg is thrilled to announce a new feature in Meta AI that allows users to create personalized images of themselves in various settings and styles. Now available in several new regions and languages, this exciting update enables users to see themselves in imaginative scenarios, from ancient gladiators to modern-day boy band members.

Zuckerberg shared his enthusiasm by posting a video on Instagram showcasing the feature's capabilities. He scanned his face with Meta AI and experimented with images of himself as a gladiator and a boy band member adorned with a gold chain. He was visibly excited about the feature, praising its ability to generate highly convincing yet distinctly AI-quality images. This new tool offers a fun and creative way for users to explore different personas and settings, reflecting Zuckerberg's genuine delight in bringing this innovation to life.

🎤VOICE YOUR OPINION

ICYMI

• Wiz declined Google’s offer of a $23bn buyout. The cybersecurity company’s CEO said in an employee update that it will instead be targeting an IPO.

• Apple has announced plans to integrate OpenAI's ChatGPT into the Siri voice assistant and iOS 18.

• The International Energy Agency reports global data centers have grown to over 7,000, up from 3,600 in 2015.

• Elon Musk says Tesla's plan to deploy humanoid robots next year.

• Disney's new robot navigates unstructured terrain effortlessly, using expressive movements to convey emotions and engage with audiences.

MONEY MATTERS

• OpenAI doubles its annual revenue to $3.4 billion

• Amazon is reportedly set to invest over $100 billion in AI and data centers over the next decade

• Tesla stands out with a remarkable +21.3% increase in stocks since July 1st, with Microsoft and Alphabet down approximately -2.3% and -2.2%, respectively

• Cohere, a generative AI startup co-founded by ex-Google researchers, secures $500M to compete with other rivals in the generative AI space

• The Monetary Authority of Singapore allocated $74.3M to advance quantum computing and AI in financial institutions.

LINKS WE’RE LOVIN’

✅ Podcast: The host Lex Fridman discusses the dangers of super-intelligent AI with Roman Yampolskiy

✅ Help guide: A cloud server cheatsheet by Bytebyte go

✅ Course: AI For Everyone by Andrew Ng

✅ Whitepaper: Find out how to Harness The Power Of Your Unstructured Data With NLP

✅ Watch: SuperSaf reviews the Samsung Galaxy Z Fold 6