Tesla presents artificial intelligence supercomputer training in what it hopes will one day be a true self-driving car • The Register


In short If you’re wondering what it takes to develop a self-driving car, know that Tesla uses a 1.8 exaFLOP AI supercomputer with 5,760 GPUs that train neural networks that he hopes will one day power autonomous vehicles.

The machine was described by the automaker’s senior AI director Andrej Karpathy at an online college computer vision conference this week. It is being used to develop Tesla’s super cruise control system autopilot, as well as what could be a fully autonomous system when completed. Tesla has been pursuing the dream of the autonomous vehicle for years; technology has so far proved elusive.

“It’s a really amazing supercomputer,” Karpathy said. “I actually think in terms of FLOPS it’s about the fifth supercomputer in the world.”

It should be noted that the prowess of a supercomputer is usually measured in FP64 precision. It is not known how precisely Tesla’s computational 1.8 exaFLOPs work.

It has 780 compute nodes, each containing up to eight 80GB Nvidia A100 GPUs. The super also has 10PB of NVMe storage. Tesla’s AI models run through millions of ten-second sequences of driving sequences recorded at 36 frames per second during training. “Computer vision is the bread and butter of what we do and enables autopilot. For this to work you have to train a massive neural network and experiment a lot, ”Karpathy added. “This is why we have invested a lot in computing. “

Can AlphaFold help discover new drugs?

DeepMind has partnered with a nonprofit research organization to use its AlphaFold AI protein folding model to develop drugs that fight parasitic diseases.

Specifically, the Drugs for Neglected Diseases Initiative will use DeepMind’s machine learning software to determine whether new drugs can treat Chagas disease and leishmaniasis. These are prevalent in tropical Latin American climates and develop after people have been bitten by triatomine bugs and sandflies. They can be fatal if left untreated.

“AI can be a game-changer: By predicting protein structures for previously insoluble protein structures, AlphaFold is opening up new avenues of research,” said Ben Perry, DNDi discovery manager, according to the Beeb. “It is encouraging to see a powerful drug with cutting edge discovery technologies making it possible to work on some of the world’s most neglected diseases.

Google Cloud AI Tool for Detecting Defects in Goods

Google has launched a machine learning software tool on its cloud platform that manufacturing customers can use to automatically spot damaged products or packaging during manufacturing.

Visual inspection AI is designed to process images of items passing through assembly lines and identify faulty or problematic equipment. Customers will need to refine the pattern on the goods they are selling and the type of defects they want to identify. By using the tool, customers can better get rid of broken items or fix problems on products before they are shipped, or at least that’s the hope.

“The advantage of a dedicated solution [like Visual Inspection AI] is that it essentially gives you ease of deployment and the peace of mind of being able to run it on the shop floor, ”Dominik Wee, general manager of manufacturing and industry, Google Cloud told VentureBeat.

“You don’t have to run the cloud,” he added. “At the same time, it gives you the power of AI and Google analytics. What we’re basically trying to do is put the capacity for AI at scale in the hands of manufacturers. “®


Source link

Previous Triple-play plan is dead, but this surprisingly popular plan may just stop the bleeding of cable companies
Next China's cyberpower is at least a decade behind the United States, new study finds

No Comment

Leave a reply

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