🪟🪟 DALL-E 2 is Coming to Azure! and Other Exciting Microsoft’s ML Announcements
Weekly news digest curated by the industry insiders
📝 Editorial
Ignite is one of the most exciting annual events in the Microsoft ecosystem. Every year, Microsoft uses this event to unveil some of its most exciting product announcements, and ML is always front and center of the new releases. This year was no exception, but there was one announcement that dominated the headlines. OpenAI’s text-to-image super model DALL-E 2 is going to be available through the Azure OpenAI Service. This effectively represents the first programmable interface available for DALL-E 2, which means a major partnership validation between Microsoft and OpenAI.
The DALL-E 2 API availability was not the only ML announcement at Ignite. The Azure Cognitive Services API suite saw major updates with new models for language, computer vision and speech. Azure ML Registries is a new catalog that allows the discovery of ML models and related artifacts. The support for PyTorch in Azure also got a boost with the release of Azure Container for PyTorch and Azure Data Science Virtual Machines for PyTorch. In an effort to make ML models safer, Microsoft released the Responsible AI scorecard which allows data scientists to access model scorecards that highlight factors such as biases, fairness and others. Other ML capabilities announced at Ignite include improved AutoML services, Spark integration, and upgraded SDKs.
Certainly, a strong ML week for Microsoft.
🔺🔻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#235: we explain Meta AI’s Make-A-Scene; discuss Meta’s Make-A-Scene Paper; explore LAION, one of the most complete training datasets for text-to-image synthesis models.
Edge#236: we go inside DeepMind’s AlphaTensor, the new super model that can discover new math algorithms.
🎧 Listen Now!
Join us in listening to the new show – Real-Life AI podcast.
The host of the show, Tigran Petrosyan (Co-founder & CEO at SuperAnnotate), invites prominent experts from both industry and academia to dive into real-life applications of AI nowadays.
The guest of the first episode is Professor Trevor Darrell from UC Berkeley, who touched on challenges in computer vision such as autonomous driving, SAR imagery, cashierless checkout systems, and more. He also shared his experience with AI initiatives that emerged as a result of close collaboration between the industry and academia.
Now, let’s review the most important developments in the AI industry this week
🔎 ML Research
AI Model Perception
DeepMind published a paper proposing a benchmark to evaluate perception in AI systems →read more
Meta-Learning and Time Series
Salesforce Research published a paper detailing DeepTime, a technique that approaches time-series forecasting as a meta-learning problem →read more
GNNs for Product Recommendation
Amazon Science published a paper discussing a GNN technique for recommending related products →read more
🤖 Cool AI Tech Releases
DALL-E 2 and Azure OpenAI Service
At the Ignite conference, Microsoft announced the limited availability of DALLE-E 2 as part of the Azure OpenAI service →read more
Microsoft ML Announcements
Microsoft made several other ML announcements at its Ignite conference →read more
Crossmodal-3600
Google Research open-sourced Crossmodal-3600, a dataset to train multilingual image captioning models →read more
PyTorch FSDP
PyTorch released an implementation of the popular Fully Sharded Data Parallel (FSDP) for scaling ML models running on TPUs →read more
🛠 Real World ML
State of Data Science
Kaggle published its annual report about the state of ML and data science →read more
💸 Money in AI
Security and privacy startup IronVest raised a $23 million seed funding round led by Accomplice. Hiring in the US and Israel.
Revenue growth management platform Insite AI raised a $19 million Series A funding round led by NewRoad Capital and M12, Microsoft’s Venture Fund. Hiring remote in the US.
Startup incubator NumberOne AI raised $13 million to fund predictive AI/ML startups.
No-code AI workflow automation platform Levity raised an $8.3 million seed round co-led by Balderton Capital and Chalfen Ventures. Hiring remote or in Berlin/Germany.
Analytics operating system Redbird raised a $7.6 million seed round. Hiring remote.
Sustainability intelligence platform Continue AI raised a $5.7 million seed funding round led by Grove Ventures and Maple Capital.
Job simulation platform TaTiO raised a $5.3 million seed round led by Mensch Capital Partners and Cresson Management.
Data and intelligence company Stears raised a $3.3 million seed round led by MaC Venture Capital. Hiring remote.