❇️ NVIDIA Continues Pushing AI’s Boundaries
Weekly news digest curated by the industry insiders
In the last few years, NVIDIA has ranked among the top companies advancing AI research and development. With its unique approach that encompasses both hardware and software, NVIDIA has been at the forefront of several major milestones in the AI industry. NVIDIA conferences are among the most followed events by the data science community, and they rarely disappoint. During its latest GTC conference, NVIDIA announced major releases in both AI software and hardware technologies.
The AI announcements at GTC show NVIDIA’s depth across every area of AI. On the software side, NVIDIA announced new cloud services such as NeMo for large-scale language models, Omniverse ACE for intelligent avatars, and BioNeMo computational biology. On the hardware end, NVIDIA introduced its new ADA architecture, which includes an improved streaming multiprocessor as well as new Tensor Core with the Hopper FP8 Transformer Engine and 1.4 petaflops of Tensor processor power. NVIDIA also announced a number of architectures based on ADA, such as chips optimized for virtual worlds, computational inference, recommendation models, and many others. Continuing its innovation in areas such as self-driving vehicles and robotics, NVIDIA announced Thor, a new 2,000 teraflops of performance and will become the hardware of choice for many applications in those domains.
GTC provided another window to witness NVIDIA’s astonishing reach in the AI space. Many of the innovations uncovered during GTC are already powering mission-critical applications across many industries. These days, we can safely say that NVIDIA is synonymous with large-scale AI.
🔺🔻TheSequence Scope – our Sunday edition with the industry’s development overview – is free. To receive high-quality content about the most relevant developments in the ML world every Tuesday and Thursday, please subscribe to TheSequence Edge 🔺🔻
🗓 Next week in TheSequence Edge:
Edge#229: we introduce VQGAN + CLIP architecture; discuss the original VQGAN+CLIP paper; explore the VQGAN+CLIP implementations.
Edge#230: we deep dive into how Amazon scaled Alexa to 1000 languages
🗺 Take the Survey!
Let’s create an objective landscape of the ML Value Chain together. You’ve probably seen some AI/ML companies’ landscapes before. They are typically assembled by either analyst firms, media, or VC firms. But we trust that TheSequence’s audience only can shape an accurate landscape of the ML Value Chain. Participate and be the first to receive this super helpful research shaped by you!
Now, let’s review the most important developments in the AI industry this week
🔎 ML Research
OpenAI published a paper detailing Whisper, a neural network that achieved human-level accuracy in speech recognition →read more
DeepMind published a paper introducing Sparrow, a reinforcement learning model used to build safer conversational agents →read more
View Synthesis with Transformers
Google Research published a paper discussing a transformer-based model for view synthesis or the ability to recreate a scene from several images →read more
Sounding More Human
IBM Research published a paper detailing a technique that combines speech recognition and storytelling to make conversational agents sound more human →read more
🤖 Cool AI Tech Releases
NeMo LLM Service
NVIDIA announced NeMo LLM Service, a cloud service for building and training mega large language models →read more
NVIDIA announced Omniverse ACE and AI microservices, a platform that enables the development of AI-powered avatars →read more
Other NVIDIA Announcements
There were too many NVIDIA AI announcements this week to cover individually →read the summary of the announcements
Google Research open-sourced TensorStore, a C++ and Python library designed for high dimensional array storage and manipulations →read more
Salesforce Research open-sourced LAVIS, a deep learning library to streamline language-vision model research and implementation →read more
🛠 Real World ML
Vehicle Routing at Walmart
Walmart Global Technology group discusses the ML architecture for vehicle and container routing for grocery delivery →read more
💸 Money in AI
Generative autonomy platform PassiveLogic secured an additional $15 million in an “off-round” strategic investment from Nvidia’s venture arm, nVentures. To date, the company has raised more than $80 million. Hiring in Salt Lake City/US.