Hugging Face collaborates with Microsoft for brand spanking new AI-powered service

Contemporary off a $100 million funding round, Hugging Face, which gives hosted AI providers and a community-driven portal for AI instruments and information units, at present introduced a brand new product in collaboration with Microsoft. Known as Hugging Face Endpoints on Azure, Hugging Face co-founder and CEO Clément Delangue described it as a approach to flip Hugging Face-developed AI fashions into “scalable manufacturing options.”

“The mission of Hugging Face is to democratize good machine studying,” Delangue mentioned in a press launch. “We’re striving to assist each developer and group construct high-quality, machine learning-powered purposes which have a constructive impression on society and companies. With Hugging Face Endpoints, we’ve made it easier than ever to deploy state-of-the-art fashions, and we will’t wait to see what Azure prospects will construct with them.”

The demand for AI stays excessive. According to a latest McKinsey survey, almost two-thirds of firms plan to extend their investments in AI over the subsequent two years. However implementing AI from scratch may be difficult. Furthermore, many firms have strict efficiency, safety, compliance and privateness necessities that require internet hosting fashions on tightly managed infrastructure.

Hugging Face Endpoints is Hugging Face’s answer to the issue.

Obtainable via Azure Machine Learning Services, Hugging Face Endpoints permits prospects to faucet Hugging Face fashions with a couple of clicks or strains of Microsoft Azure SDK code. After deciding on a mannequin and a process kind, prospects can deploy the mannequin wherever they select on inner infrastructure, whether or not for an app, web site or backend service. 

“Transformer fashions have eternally modified how firms construct expertise with machine studying,” Jeff Boudier, product director at Hugging Face, informed TechCrunch through electronic mail. “Endpoints are going to make Transformers … simply accessible to Azure prospects. As of at present, Hugging Face Endpoints helps all Transformers pure language processing duties. We’re going so as to add assist for speech and pc imaginative and prescient duties, together with computerized speech recognition and picture classification, over the subsequent couple of weeks.”

Hugging Face’s fashions give attention to textual content evaluation — particularly duties like summarizing and producing textual content, extracting info and robotically answering questions. They’re largely based mostly on the Transformer, the identical mannequin structure underpinning OpenAI’s GPT-3 and numerous different highly effective AI programs.

Hugging Face Endpoints is launching in beta at present — already, it’s being utilized by Customary Financial institution, a big South African financial institution and monetary providers group. It’ll be free throughout the preview interval, with prospects solely having to pay for the underlying Azure compute, however the pricing mannequin for basic availability will probably be usage-based.

Boudier hinted that it’s solely the start of Hugging Face’s work with Microsoft. 

“That is the beginning of the Hugging Face and Azure collaboration we’re asserting at present as we work collectively to convey our options, our machine studying platform, and our fashions accessible and make it straightforward to work with on Azure. Hugging Face Endpoints on Azure is our first answer obtainable on the Azure Market, however we’re working laborious to convey extra Hugging Face options to Azure,” he mentioned. “We’ve got acknowledged [the] roadblocks for deploying machine studying options into manufacturing and began to collaborate with Microsoft to resolve the rising curiosity in a easy off-the-shelf answer.”

Source link






Leave a Reply

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