Cerebras breaks record for largest AI models trained on a single device

Cerebras said this can reduce the engineering time to run large natural language processing models from months to minutes, making it more cost-effective.

Cerebras Systems claims it has achieved a new feat, by training an AI model with up to 20bn parameters on a single CS-2 system.

The AI ​​company said using a single system can reduce the engineering time and effort when training natural language processing (NLP) models from months to minutes.

NLP can enable computers to process human language in either text or voice data and understand its full meaning.

Cerebras CEO Andrew Feldman said bigger NLP models are shown to be more accurate, but this means only select companies have the resources and expertise to do the “painstaking work” required.

“As a result, only very few companies could train large NLP models – it was too expensive, time-consuming and inaccessible for the rest of the industry,” Feldman said.

Cerebras said its latest result will help eliminate one of the “most painful aspects” of training large NLP models, which usually involves spreading it across hundreds or thousands of different GPUs.

The company added that the process of partitioning a model across GPUs is unique to each network compute cluster pair, so the work cannot be ported to different clusters or across neural networks.

Cerebras said the ability to train a large model on a single device is possible thanks to its WSE-2 processor, which the company said is the largest processor ever built. It is 56 times larger, has 2.55trn more transistors and has 100 times as many compute cores as the largest GPU.

Last year, Cerebras said it was using the WSE-2 processor to power a new chip cluster that it said can “unlock brain-scale neural networks”.

The AI ​​company said that this processor means that a single CS-2 could support models with hundreds of billions or even trillions of parameters. Intersect360 Research CRO Dan Olds said this could give organizations an “easy and inexpensive on-ramp to major league NLP”.

“Cerebras’ ability to bring large language models to the masses with cost-efficient, easy access opens up an exciting new era in AI,” Olds said. “It will be interesting to see the new applications and discoveries CS-2 customers make as they train GPT-3 and GPT-J class models on massive datasets.”

10 things you need to know direct to your inbox every weekday. Sign up for the Daily BriefSilicon Republic’s digest of essential sci-tech news.


Original Article reposted fromSource link

Disclaimer: The website autopost contents from credible news sources and we are not the original creators. If we Have added some content that belongs to you or your organization by mistake, We are sorry for that. We apologize for that and assure you that this won’t be repeated in future. If you are the rightful owner of the content used in our Website, please mail us with your Name, Organization Name, Contact Details, Copyright infringing URL and Copyright Proof (URL or Legal Document) aT spacksdigital @ gmail.com

I assure you that, I will remove the infringing content Within 48 Hours.

Leave a Reply

Your email address will not be published.

Tech

Asus ROG Phone 6, ROG Phone 6 Pro launched in India, check out the price, specs and availability

FP StaffJul 06, 2022 11:24:51 IST Asus launched two new smartphones under its ROG Phone series, the ROG Phone 6 and the ROG Phone 6 Pro in India as well as the rest of the world. Powered by the Snapdragon 8+ Gen 1 chipset these devices come with some seriously insane specs. As far as […]

Read More
Tech

Polestar’s KOJA is a Secret, Minimalistic Tree House That Looks Like a Spaceship

EV maker Polestar’s KOJA is a secret, minimalistic tree house that could be mistaken for a spaceship in the Finnish forest. It was designed by Kristian Talvitie to be a micro space that is accessible for people who would otherwise take a much longer trip to experience the wilderness. The tree house encapsulates both Polestar […]

Read More
Tech

Voyantis, which develops AI-powered tools to help marketing teams predict a customer's lifetime value, emerges from stealth with a $19M seed (Kyle Wiggers/TechCrunch)

Kyle Wiggers / TechCrunch: Voyantis, which develops AI-powered tools to help marketing teams predict a customer’s lifetime value, emerges from stealth with a $ 19M seed– Voyantis, a startup developing tools to estimate a customer’s lifetime value, today emerged from stealth with $ 19 million in seed funding from Target Global… Original Article reposted fromSource […]

Read More