🔥 PyTorch is Getting Serious About the Enterprise
Is TensorFlow domination over?
PyTorch and TensorFlow have quickly become the two most dominant frameworks for deep learning research and development. While researchers love PyTorch’s flexible programming model, TensorFlow by far dominates the production deployments in real-world systems. This has mostly been the result of the enterprise-grade capabilities and toolsets built by TensorFlow over the years and the massive support by Google. However, for machine learning to go mainstream in the enterprise, we need these other frameworks to develop the same type of enterprise-grade capabilities.
Well, PyTorch is starting to catch up.
Over the last two years, PyTorch has been steadily incorporating new features optimized for enterprise deployments. This week we saw a massive step in that direction with the announcement of the PyTorch Enterprise Support program enabled by a partnership between Microsoft and Facebook. As its name indicates, the program will provide support for enterprise users building production applications in PyTorch. As part of the program, Microsoft announced the release of PyTorch Enterprise on Microsoft Azure, which delivers the enterprise-grade feature to PyTorch users. From a market perspective, Microsoft could be the ideal partner to expand PyTorch into enterprise environments. The Redmon giant certainly has the enterprise sales, support and infrastructure capabilities to catalyze the adoption of any machine learning platform. For now, this is a clear indication that PyTorch is getting serious about the enterprise.
🔺🔻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#93: we explain what Q-Learning models are; we explore how Google SEED RL architecture enables highly scalable RL tasks; we discuss Facebook’s ReAgent that is used for building reinforcement learning systems.
Edge#94: deep dive into Determined.ai, which tackles the monster challenge of distributed training.
Now, let’s review the most important developments in the AI industry this week
🔎 ML Research
Microsoft Research published an insightful blog post detailing DeepSpeed Inference, which adapts its large-scale training framework to inference models ->read more on Microsoft Research blog
Contrastive Learning for Image Generation
Google Research published a paper detailing a text-to-image generation method that learns by maximizing mutual information between datasets using contrastive learning ->read more on Google Research blog
Relational Graph Learning
Uber published a detailed blog post about their use of relational graph learning methods to detect users committing fraud in the platform->read more on Uber engineering blog
❗️Call for writers ❗️
TheSequence is looking for freelance writers who cover machine learning and artificial intelligence space in-depth. Send the links to your blog or published articles to email@example.com
🤖 Cool AI Tech Releases
Facebook AI Research (FAIR) open sourced Dynaboard, a platform for standardizing the evaluation of NLP models ->read more on FAIR blog
Databricks open-sourced Delta Sharing, a protocol for sharing data in a secure way ->read more on Databricks blog
Amazon unveiled RedShift ML, a platform that allows the creation of ML models directly in a RedShift cluster ->read more on AWS team blog
GPT-3 in PowerApps
Microsoft incorporated GPT-3 into its Power Apps Studio to streamline the creation of apps using natural language ->read more on Power Apps team blog
PyTorch Enterprise Support Program
Microsoft and Facebook partnered to launch a new program to provide support to data science teams building enterprise solutions using PyTorch->read more on Team Pytorch blog
🙌 Let’s connect
Follow us on Twitter to receive our Edges in compact threads and refresh what you read before
💸 Money in AI
OpenAI announced the $100 million OpenAI Startup Fund to “help AI companies have a profound, positive impact on the world.” If you plan to push the boundaries of today’s artificial intelligence using OpenAI’s API – they want to hear from you.
DataOps startup Atlan raised $16.5 million in a funding round led by Insight Ventures. By combining data cataloging, QA, and lineage, it equips data teams with the tools of the trade needed to build, ship, and maintain outputs.
AI-as-a-service company Faculty.ai raised $42.5. million in growth funding from the Apax Digital Fund (ADF). Faculty works with a broad range of problems for both public and private sector organizations. It enables customers to customize powerful AI solutions to their needs.
AI-chip startup Eys3D Microelectronics raised $7 million in a Series A round led by strategic partners Arm IoT Capital, WI Harper Group, and Marubun Corporation. Eys3D focuses on end-to-end software and hardware systems for computer vision technology.
Financial research platform provider Sentieo raised $20 million in a Series B round led by Ten Coves Capital. It leverages AI to combine all the elements of fundamental research into one seamless workflow, making it easier to discover financial insights.
API discovery and vulnerability detection platform Salt Security raised $70 million in a Series C funding round led by Advent International. Using a combination of AI and ML, it analyzes a copy of the traffic from different APIs technologies and creates a baseline of normal behavior to be able to identify anomalies.
AI-powered personalized learning startup Riiid raised$175 million in a funding round coming from SoftBank’s Vision Fund 2.