Marc Lozano Bosch’s Post

Nvidia CEO Jensen Huang, in his commencement speech at National Taiwan University (NTU) emphasized the transformative power of artificial intelligence (AI) and shared his journey with Nvidia, highlighting key moments of failure and recovery that shaped the company. In his speech at National Taiwan University, Jensen Huang, CEO of Nvidia, discussed “Earth-2,” a groundbreaking initiative aimed at addressing climate change through advanced computational models and AI. Earth-2 is designed to create a digital twin of the Earth, enabling high-resolution climate and weather simulations. This capability is crucial for understanding and mitigating the impacts of climate change, providing policymakers, researchers, and industries with detailed and actionable climate insights, models like FourCastNet have successfully predicted extreme weather events, demonstrating the platform’s potential in forecasting and preparing for climate-related challenges. It was a powerful call to action for all that to embrace AI, learn from failures, and relentlessly pursue their goals, underlining the transformative potential of AI and the importance of resilience and strategic thinking. “Run, not walk” by Jensen Huang. #amazing #revolution #AI #factories #Taiwan #deeplearning #generation #applications #change #pushtothelimits #perseverance #successtraits #interestingspeech

Mind-opening presentation by Nvidia Jensen Huang on Compute Taipei 2024 With the advancement of technology that opens up new era of generative ai, every square meters of earth can be modeled to be earth's digital twin and it will help on weather prediction and many other use cases Many other interesting possibilities presented: digital human, NIM (Nvidia inference micro systems) https://lnkd.in/dB5GSyN6 #genai

Learn how NVIDIA and our partners take center stage in #COMPUTEX2024. Together with our partners ASRock Rack Inc. ASUS GIGABYTE Ingrasys 鴻佰科技股份有限公司 Inventec Pegatron QCT Supermicro Wistron Wiwynn we are ushering in the next industrial revolution through the power of #AI and #acceleratedcomputing.

NVIDIA Generative AI Certifications launches at GTC today! Get NVIDIA certified at GTC and validate your NVIDIA expertise. We'll have 4 certification exams available. -Generative AI LLM -Generative AI Multimodal -AI in the Data Center -InfiniBand https://lnkd.in/gRMk-V3h #NVIDIA #GTC #GenerativeAI #AI #GenAI

Are supercomputers the next big edge in AI innovation? NVIDIA's Eos and Elon Musk's Grok are leveraging high-performance computing to lead the AI race. Leaders like Musk and institutions like Oak Ridge National Laboratory are redefining AI potential through supercomputing powerhouses like Summit and Frontier. #Supercomputing #AIInnovation #TechLeadership

If you are headed to NVIDIA's #gtc2024 conference, don’t miss the keynote on "A Culture of Open and Reproducible Research, in the Era of Large AI Generative Models " by Joelle Pineau. She leads one of the most innovative research orgs in the world. Meta's FAIR has always been about advancing the science of AI -- by not only enabling Research Scientist to do the Fundamental AI Research (hence the moniker FAIR, ...ok ok, I know the "F" was once for Facebook!)... but also releasing the AI models under open licenses, and including sample datasets and code to enable reproducibility. That's how open science actually moves the needle. Should be an interesting talk, which you can catch online or live. https://lnkd.in/gp4YtFhU If you are there live, even more of a bonus. Make sure you catch the breakout session from Alban Desmaison and Jason Ansel on the "State of #PyTorch". There is so many advances around distributed training in every modality (model-, data-, and tensor-parallel) that tools like PyTorch are making it so that every AI model developer can configure their training run uniquely for their model architecture. Wow! I hope they also cover the many innovations in the framework for model inference in the datacenter (enabled by the fast Triton-based compile path) and on the edge (via Executorch). Lets see! https://lnkd.in/gKrCfpHC I’m sure there will be really interesting news at GTC given Nvidia is a key innovator in the industry. Perhaps something we share in common is the focus to move fast. And thanks to Pierre Roux for making sure our collaboration with Nvidia is not only fast, but fruitful. #ai partnerships

NVIDIA GTC kicks off March 18th and it’s the must-not-miss event of 2024. If you’re looking for key sessions to attend, I highly recommend the “Transforming AI panel”. Jensen is hosting all eight authors of "Attention Is All You Need”, the groundbreaking paper that introduced the Transformer (one of the most cited papers among AI researchers). Every major AI model and product in the headlines today—ChatGPT, GPT-4, Midjourney, Stable Diffusion, GitHub Copilot —is built using transformers. Andy Jassy used to say “there is no compression algorithm for experience” but there are cheat codes and this panel is one of them! Add to this session to your GTC Schedule now: https://lnkd.in/gHvq5YjJ

Nvidia GTC '24 has been incredible for so many reasons. Bit nostalgic, given I've kinda grown up coming to it and def has the focus and experts to back it up. Day 3 big event, I had the pleasure to sit on the panel entitled "Strategies for Safer AI" [S62292], chaired by long time Nvidian Jonathan Cohen NVIDIA, and panelists Iftach Orr from ActiveFence , Nicol Turner Lee from The Brookings Institution , Seraphina Goldfarb-Tarrant of Cohere and finally, myself, Steven Eliuk IBM.. ;-). We spoke a lot about magical wishes, educated a bit and dove into to practices we can all easily employ to help prevent black swan events in AI. I was just thrilled to sit next to such esteemed colleagues and be discussing these topics. Just want to remind all, we have several verticals that have had excellent practices, look to financial sector and anything health / medical related in AI and u will see get some great AI governance / risk guidance. We just need to practice these techniques more broadly... I've heard several suggest, "we only do internal AI and not customer focusing so these techniques aren't really needed", these techniques still apply as they produce better outcomes, reduce Ethical & Technical debt and keep AI sustainable and truly Operationalized. Remember, a internal AI decision can be both hugely enlightening and incredibly dangerous... it's all about the instrumentation and validating the models / AI is doing what we expect it to. Don't think AI governance is a burden, it's a requirement and will produce better outcomes. #aiethics #sustainableAI https://lnkd.in/gEtrYrTM

🚀 NVIDIA has made a game-changing move in AI with the launch of its latest open-source model, NVLM 1.0 – a powerhouse with 70 billion parameters. This new model not only surpasses well-established leaders like GPT-4 and Claude 3.5 Sonnet, but it also brings a new level of accessibility for developers and businesses. 💡 Try NVLM 1.0 for free on Hugging Face: https://lnkd.in/d3yJEEyp Key Features of NVLM 1.0 • Superior Performance: NVLM 1.0 outperforms the competition on multiple benchmarks, including an 85.0 score on the Arena Hard test, underscoring its strength in language comprehension and generation. • Open-Source Flexibility: NVIDIA’s decision to make NVLM 1.0 open-source empowers the AI community to innovate freely. The model weights are already available on Hugging Face, with training code soon to follow. • Reinforcement Learning: Trained with reinforcement learning, NVLM 1.0 continues to improve its performance through dynamic learning, making it highly effective in real-world applications. • Multimodal Capabilities: NVLM 1.0 goes beyond text and excels with multimodal inputs, such as images and charts, making it a versatile tool for tasks like visual reasoning and coding. Impact on Businesses NVLM 1.0 is a game-changer for businesses eager to integrate advanced AI. Its open-source availability levels the playing field, giving smaller teams access to the kind of technology once limited to large corporations. NVIDIA’s shift towards software development solidifies its role as a key player in AI, challenging traditional models and expanding AI adoption across industries. This release underscores NVIDIA’s dedication to driving AI innovation while setting a new benchmark for performance and accessibility in the field. Read more: https://lnkd.in/exvu6fET #ArtificialIntelligence #OpenSource #MachineLearning #NVIDIA #AIInnovation #LargeLanguageModels #TechForGood

I'm excited about what's happening with NVIDIA AI Workbench. It's a tool for GenAI and machine learning developers to get started quickly - and also achieve faster results. Example projects provide a great starting point for a variety of use cases. If you are a novice or expert developer, I'd encourage you to check out this session at the NVIDIA GTC conference this week: Breaking Barriers: How NVIDIA AI Workbench Makes AI Accessible to All [S62135] https://lnkd.in/gt5Mv3Y5