Meta AI

NVIDIA Blackwell Platform Arrives to Power a New Era of Computing

Retrieved on: 
Montag, März 18, 2024
Cloud, Maker, Hewlett Packard Enterprise, Azure, GTC, Google Cloud, Algorithm, GPU, IBM Cloud, ASUS, B200, QCT, YTL Power, Deep learning, LLM, Statistics, Microsoft, Foxconn, NVLink, Partnership, InfiniBand, SAN, Oracle Cloud, Software, Meta AI, AWS, Ansys, TB, Pegatron, Fabric discography, GIGABYTE, Organization, Synopsys, Acceleration, Library, Grace, Quantum computing, Research, Electronic design automation, Architecture, Amazon Web Services, Dell Technologies, Cisco, Gmail, Machine learning, Game theory, Engineering, HGX, Decompression, OpenAI, Data science, Deep learning super sampling, Drug, Supermicro, DGX, Graphics processing unit, Elasticity, National academy, Dell, TSMC, Video game

The NVIDIA GB200 Grace Blackwell Superchip connects two NVIDIA B200 Tensor Core GPUs to the NVIDIA Grace CPU over a 900GB/s ultra-low-power NVLink chip-to-chip interconnect.

Key Points: 
  • The NVIDIA GB200 Grace Blackwell Superchip connects two NVIDIA B200 Tensor Core GPUs to the NVIDIA Grace CPU over a 900GB/s ultra-low-power NVLink chip-to-chip interconnect.
  • It combines 36 Grace Blackwell Superchips, which include 72 Blackwell GPUs and 36 Grace CPUs interconnected by fifth-generation NVLink.
  • The Blackwell product portfolio is supported by NVIDIA AI Enterprise , the end-to-end operating system for production-grade AI.
  • To learn more about the NVIDIA Blackwell platform, watch the GTC keynote and register to attend sessions from NVIDIA and industry leaders at GTC, which runs through March 21.

Lightning AI Announces Availability of Thunder; A Powerful Source-to-Source Compiler For PyTorch That Speeds Up Training And Serving Generative AI Models Across Multiple GPUs, Built With Support From NVIDIA

Retrieved on: 
Donnerstag, März 28, 2024
Data Management, Technology, Software, Networks, Artificial Intelligence, Internet, Hardware, Deep learning, Lightning, Organization, GPU, Triton, Memory, Meta, Research, AI, Meta AI, Studio, Graphics processing unit, Stable Diffusion, OpenAI, Multimedia, PyTorch, Video game

Thunder is the culmination of two years of research on the next generation of deep learning compilers, built with support from NVIDIA.

Key Points: 
  • Thunder is the culmination of two years of research on the next generation of deep learning compilers, built with support from NVIDIA.
  • Lightning AI leveraged support from NVIDIA for the integration of NVIDIA’s best executors into Thunder.
  • PyTorch Lightning, developed by Falcon, has become the standard for training large-scale AI models such as Stable Diffusion, with over 10,000 companies using it to build AI at scale.
  • Using this knowledge, the team at Lightning AI rethought from the ground up what AI development on the cloud should feel like.

OpenFold Biotech AI Research Consortium releases SoloSeq and Multimer, an integrated protein Large Language Model with 3D structure generation

Retrieved on: 
Montag, Februar 19, 2024
Technology, Research, Health Technology, Software, Biotechnology, Health, Pharmaceutical, Science, Artificial Intelligence, Sequence, Transformer, DeepMind, Meta AI, Internet, Amazon Web Services, Architecture, MSA, Artificial intelligence, Multimedia, Bayer, AF2, De novo, Nature, ChatGPT, Structural biology, LLM, Computation, University, Meta, PDB, ESM, Organization, Protein, Columbia University, Agriculture, Deep learning, Google Brain, AWS

OpenFold , a non-profit artificial intelligence (AI) research consortium, today announced the release of two new tools: 1) SoloSeq, which integrates a new protein Large Language Model (LLM) with its OpenFold structure prediction software, and 2) OpenFold-Multimer software, which creates higher quality models of protein/protein complexes than OpenFold alone.

Key Points: 
  • OpenFold , a non-profit artificial intelligence (AI) research consortium, today announced the release of two new tools: 1) SoloSeq, which integrates a new protein Large Language Model (LLM) with its OpenFold structure prediction software, and 2) OpenFold-Multimer software, which creates higher quality models of protein/protein complexes than OpenFold alone.
  • The new Multimer code is the first fully open source training code system for the generation of protein/protein structures.
  • View the full release here: https://www.businesswire.com/news/home/20240219658831/en/
    OpenFold SoloSeq model prediction vs experimentally resolved structure for PDB protein 4B9Z (Graphic: Business Wire)
    “OpenFold-Multimer and SoloSeq are particularly useful for designed proteins that don't exist in nature.
  • Previous work in protein LLMs, primarily out of Meta AI, has produced a series of valuable LLM standalone models and the ESMFold structure model, which pairs ESM-family LLMs with OpenFold for structure prediction.

Groq® LPU™ Inference Engine Leads in First Independent LLM Benchmark

Retrieved on: 
Dienstag, Februar 13, 2024
Meta AI, RAG, AI, API, LLM, Meta, Latency, Cryptocurrency

MOUNTAIN VIEW, Calif., Feb. 13, 2024 /PRNewswire/ -- Groq®, a generative AI solutions company, is the clear winner in the latest large language model (LLM) benchmark by ArtificialAnalysis.ai, besting eight top cloud providers in key performance indicators including Latency vs. Throughput, Throughput over Time, Total Response Time, and Throughput Variance. The Groq LPU™ Inference Engine performed so well with a leading open-source LLM from Meta AI, Llama 2 70b, that axes had to be extended to plot Groq on the Latency vs. Throughput chart. Groq participated in its first public LLM benchmark in January 2024 with competition-crushing results.

Key Points: 
  • The Groq LPU™ Inference Engine performed so well with a leading open-source LLM from Meta AI, Llama 2 70b, that axes had to be extended to plot Groq on the Latency vs. Throughput chart.
  • Groq participated in its first public LLM benchmark in January 2024 with competition-crushing results.
  • "Inference is critical to achieving that goal because speed is what turns developers' ideas into business solutions and life-changing applications.
  • The LPU Inference Engine is available through the Groq API.

Expedera NPUs Run Large Language Models Natively on Edge Devices

Retrieved on: 
Montag, Januar 8, 2024
Car, Architecture, Meta, Natural language processing, Data analysis, AI accelerator, LLM, AI, Generative, IP, NPU, Silicon, Automation, Meta AI, Communication, Mobile phone

SANTA CLARA, Calif., Jan. 8, 2024 /PRNewswire/ -- Expedera, Inc, a leading provider of customizable Neural Processing Unit (NPU) semiconductor intellectual property (IP), announced today that its Origin NPUs now support generative AI on edge devices. Specifically designed to handle both classic AI and Generative AI workloads efficiently and cost-effectively, Origin NPUs offer native support for large language models (LLMs), including stable diffusion. In a recent performance study using the open-source foundational LLM, Llama-2 7B by Meta AI, Origin IP demonstrated performance and accuracy on par with cloud platforms while achieving the energy efficiency necessary for edge and battery-powered applications.

Key Points: 
  • Specifically designed to handle both classic AI and Generative AI workloads efficiently and cost-effectively, Origin NPUs offer native support for large language models (LLMs), including stable diffusion.
  • LLMs bring a new level of natural language processing and understanding capabilities, making them versatile tools for enhancing communication, automation, and data analysis tasks.
  • Additionally, running LMM-based applications on edge devices improves reliability, reduces latency, and provides a better user experience.
  • "Edge AI designs require a careful balance of performance, power consumption, area, and latency," said Da Chuang, co-founder and CEO of Expedera.

"ChatGPT for Music": AI Music Producer Creates Hip-Hop Album in 24 Hours

Retrieved on: 
Donnerstag, Dezember 7, 2023
Economics, Research, Music therapy, Big Baby Gandhi, Ethics, Multimedia, Religion, KOTD, Music, Intelligence, Meta AI, DMX, OpenAI, Part Two, Media, Pontiac Fiero, ChatGPT, Capitalism, Meta Platforms, AI, Meta, Nightclub, Cassette tape, Film industry, San Francisco Bay Area

OAKLAND, Calif., Dec. 7, 2023 /PRNewswire/ -- East Oakland hip-hop artist Nimrod Cain ("Nimsins") has achieved a milestone of recording the first studio album produced by generative AI, using experimental MusicGen technology developed by Meta (formerly known as Facebook). The album recording process, consisting of 8 songs recorded over a 24-hour period, has been catalogued in a two-part documentary "Sensory Overload: The AI Music Producer," available now via Slouchy Media.

Key Points: 
  • The album recording process, consisting of 8 songs recorded over a 24-hour period, has been catalogued in a two-part documentary " Sensory Overload: The AI Music Producer ," available now via Slouchy Media.
  • The MusicGen language model was trained using over 20,000 hours of music owned and licensed by Meta.
  • Nimsins was introduced to MusicGen by Vinay Pai, an acclaimed Bay Area music producer moonlighting as an engineer at a San Francisco-based AI startup.
  • The documentary "The AI Music Producer" is a capsule of a historic moment in music, as generative AI raises high-stakes questions about the ethics and economics of art, and the uncertain future of legacy media industries.

Groq Sets New Large Language Model Performance Record of 300 Tokens per Second per User on Meta AI Foundational LLM, Llama-2 70B

Retrieved on: 
Dienstag, November 7, 2023
LLM, Meta, Human, LPU, Meta AI, Running, Language, AI, Graphics processing unit, Acceleration, Customer service, Cryptocurrency

The Groq Language Processing Unit™ system is the AI assistance enablement technology poised to provide real-time, "low lag" experiences for users with its inference performance.

Key Points: 
  • The Groq Language Processing Unit™ system is the AI assistance enablement technology poised to provide real-time, "low lag" experiences for users with its inference performance.
  • MOUNTAIN VIEW, Calif., Nov. 7, 2023 /PRNewswire/ -- Groq , an AI solutions company, announced it still holds the foundational Large Language Model (LLM) performance record for speed and accuracy amidst emerging market competition.
  • Groq has set a new performance bar of more than 300 tokens per second per user on Meta AI's industry-leading LLM, Llama-2 70B, run on its Language Processing Unit™ system.
  • An LPU™ system is built for the sequential and compute-intensive nature of GenAI language processing.

Comet Launches Course on Building with LLMs Taught by Elvis Saravia

Retrieved on: 
Donnerstag, Oktober 26, 2023
Electronic Design Automation, Engineering, Data Analytics, Technology, Other Retail, Logistics, Supply Chain Management, Manufacturing, Mobile, Wireless, Specialty, Security, Retail, Small Business, Audio, Video, Finance, Artificial Intelligence, Professional Services, Online Retail, Transport, Software, Networks, Internet, PyTorch, LLM, Education, Elastic, Savaria, Meta, Mathematics, Etsy, FAIR, Data science, Research, MLOps, Multimedia, Uber, Galactica, Curriculum, Meta AI, Netflix, AI, Bias, Metadata, Prompt engineering, Online shopping, Comet, Shopify

The course also reveals the latest trends in LLMs and shows participants how to operationalize them.

Key Points: 
  • The course also reveals the latest trends in LLMs and shows participants how to operationalize them.
  • View the full release here: https://www.businesswire.com/news/home/20231026965707/en/
    “LLMs are changing the world, and data scientists and software engineers must keep up,” said Saravia.
  • “We designed a course that will teach participants exactly how to build modern software with LLMs, using the newest tools and techniques.
  • It’s everything you need to know to take a big leap forward.”
    Comet introduced the industry’s only course completely focused on building real-world applications with LLMs.

Groq to Feature World's Fastest GenAI Inference Performance for Foundational LLMs at Supercomputing '23 on Its LPU™ Systems

Retrieved on: 
Donnerstag, Oktober 26, 2023
Cloud, LLM, Meta, HPC, Research, LPU, Artificial intelligence, Intel, Qualcomm, Falcon, Meta AI, Language, AI, Medical device, Engineering, Broadcom Inc.

Subject matter experts from Groq will be presenting four sessions during the conference on a range of HPC, AI, and research-related topics.

Key Points: 
  • Subject matter experts from Groq will be presenting four sessions during the conference on a range of HPC, AI, and research-related topics.
  • The LPU™ accelerator is Groq's response to the next level of processing power required by enterprise-scale AI.
  • At Groq we are setting a new standard with our LPU™-based systems that improve performance, power, and scale when serving a large customer base.
  • With a clear market need for a purpose-built and software-driven processor, the Groq LPU accelerator will power LLMs for the exploding GenAI market.

Tome Adds Key Machine Learning and Engineering Leaders To Shape AI-Powered Communication

Retrieved on: 
Donnerstag, Oktober 5, 2023
OpenSea, Tome, Multimedia, Google Brain, Engineering, Character, Student, Machine learning, Personality, Gen, AI, Carnegie Mellon University, Meta, Meta AI, Google Assistant, Lightspeed Venture Partners, Netflix, Natural language processing, Enterprise, Mathematics, Growth, Sunlight, ML, Marissa Mayer, SAN, Generation Z, Roberta, Sachs, Brown University, Silicon, Research, Bard, Medical device, Mobile phone, Management, Online shopping, Video game, Sankar

SAN FRANCISCO, Oct. 5, 2023 /PRNewswire/ -- Tome — The AI-powered medium for shaping and sharing ideas, announced today the hire of two key machine learning leaders, Sachs, former head of natural language processing and generative AI at Robinhood, and Archana Sankaranarayanan, former engineering leader at Google and Netflix.

Key Points: 
  • Outlier Partners: Sarah Sachs, former generative AI leader at Robinhood, and Archana Sankaranarayanan, former engineering leader at Google Assistant and Netflix, have joined Tome to lead its world-class machine learning and engineering team.
  • The leaders join a world-class artificial intelligence team, which includes some of Silicon Valley's top machine learning engineers, including Ves Stoyanov.
  • Freelancers, creatives, government leaders, students, and educators have also embraced Tome as a new way to share ideas.
  • The company's latest hires will further drive the innovative machine learning and engineering that is making Tome's platform increasingly refined and personalized.