Microsoft expands its AI partnership with Meta
Mary Jo Foley Microsoft and Meta are extending their ongoing AI partnership, with Meta selecting Azure as “a strategic cloud provider” to accelerate its own AI research and development. Microsoft officers shared extra particulars in regards to the newest on the Microsoft-Meta partnership on Day 2 of the Microsoft Construct 2022 builders convention. Microsoft and … The post Microsoft expands its AI partnership with Meta appeared first on Ferdja.
Microsoft and Meta are extending their ongoing AI partnership, with Meta selecting Azure as “a strategic cloud provider” to accelerate its own AI research and development. Microsoft officers shared extra particulars in regards to the newest on the Microsoft-Meta partnership on Day 2 of the Microsoft Construct 2022 builders convention.
Microsoft and Meta — again when it was nonetheless referred to as Fb — introduced the ONNX (Open Neural Community Trade) format in 2017 within the identify of enabling builders to maneuver deep-learning fashions between completely different AI frameworks. Microsoft open sourced the ONNX Runtime, which is the inference engine for fashions within the ONNX format, in 2018.
In the present day, Meta officers mentioned they will be utilizing Azure to speed up analysis and improvement throughout the Meta AI group. Meta performs to make use of a devoted Azure cluster of 5400 GPUs utilizing the newest digital machine (VM) sequence in Azure (NDm A100 v4 sequence, that includes
NVIDIA A100 Tensor Core 80GB GPUs) for a few of their large-scale AI analysis workloads. Final yr, Meta started utilizing Azure VMs for a few of its large-scale AI analysis, particularly distributed AI coaching.
The pair additionally mentioned at the moment that they’ll collaborate to scale PyTorch adoption on Azure. Final yr, Microsoft dedicated publicly to stepping up its help for enterprise clients which can be utilizing Meta’s PyTorch deep-learning framework on Azure. I feel the brand new piece right here is Microsoft committing to construct “within the coming months” new PyTorch improvement accelerators to facilitate speedy implementation of PyTorch-based options on Azure.
AI improvement has been an enormous theme at Construct this week, with Microsoft saying a brand new Arm-based dev equipment operating on the Qualcomm Snapdragon platform that will likely be accessible someday later this yr. One of many largest claims to fame of the approaching “Undertaking Volterra” dev equipment is its capability to run AI workloads quicker and extra effectively on Arm than on different processors. (Microsoft nonetheless hasn’t launched any tech specs, together with info on the processors that will likely be onboard.)
Microsoft’s newly-announced Energy Apps Categorical Design functionality makes use of AI fashions from Azure Cognitive Providers to show drawings, photographs, PDFs and Figma design recordsdata into purposes.
Microsoft additionally introduced this week that the Azure OpenAI Service is now accessible in preview. These within the preview will have the ability to entry completely different fashions from OpenAI, together with the GPT-3 base sequence (Ada, Babbage, Curie and DaVinci), Codex sequence and embedding fashions, and mix them with the enterprise capabilities of Azure. In 2019, Microsoft invested $1 billion in OpenAI in alternate for commitments from OpenAI to make Microsoft its “most popular accomplice for commercializing new AI applied sciences.”
One other Microsoft Cognitive Service, the Cognitive Service for Language, now presents summarization for paperwork and conversations. This might show helpful for name middle utilization, Microsoft officers advised. Microsoft is making new Language Service capabilities usually accessible, together with custom-named entity recognition, which can assist builders manage and categorize textual content with a buyer’s domain-specific labels, corresponding to a help ticket or bill.
Microsoft additionally introduced the Azure Machine Studying accountable AI dashboard is in preview. This functionality will assist builders and knowledge scientists extra simply implement accountable AI, Microsoft officers mentioned.
Microsoft’s present set of AI fashions, embrace Turing for wealthy language understanding, Z-Code for language translation and Florence for visible recognition. Its household of Azure Cognitive Providers embrace building-block companies for Imaginative and prescient, Speech, Language, Resolution, plus its OpenAI service. And among the many present Azure Utilized AI Providers are Cognitive search, Type Recognizer, Immersive Reader, Bot Service and Video Analyzer.