InfoQ Homepage AI, ML & Data Engineering Content on InfoQ
-
Google AI Developed a Language Model to Solve Quantitative Reasoning Problems
Google AI developed a deep learning language model called Minerva which could solve mathematical quantitative problems. Google AI researchers achieved a state-of-the-art deep learning model by training on a large dataset that contains quantitative reasoning with symbolic expressions. The final model, Minerva, could solve quantitative mathematical problems on STEM reasoning tasks.
-
MLGO Framework Brings Machine Learning in Compiler Optimizations
Google’s new Machine Learning Guided Optimization (MLGO) is an industrial-grade general framework for integrating machine-learning (ML) techniques systematically in a compiler and in particular in LLVM. Compiling faster and smaller code can significantly reduce the operational cost of large data-center applications.
-
OpenAI Releases Minecraft-Playing AI VPT
Researchers from OpenAI have open-sourced Video PreTraining (VPT), a semi-supervised learning technique for training game-playing agents. In a zero-shot setting, VPT performs tasks that agents cannot learn via reinforcement learning (RL) alone, and with fine-tuning is the first AI to craft a diamond pickaxe in Minecraft.
-
Google's BigQuery Introduces Column-Level Encryption Functions and Dynamic Masking of Information
Google recently released new features for its SaaS data warehouse BigQuery which include column level encryption functions and dynamic masking of information. Specifically, dynamic masking of information can be used for real-time transactions whereas column level encryption provides additional security for data at rest or in motion where real-time usability is not required.
-
LinkedIn Open-Sourced Its Feature Store to Evangelize Productive Machine Learning
LinkedIn Engineering recently open-sourced its feature store Feathr, which helps engineers to develop machine Learning products by simplifying feature management and usage in production. It defines features, computes them for training and inference purposes, and makes them discoverable by other machine learning developers.
-
Amazon Unveils ML-Powered Coding Assistant CodeWhisperer
Amazon launched CodeWhisperer, an ML-Powered Coding Companion which provides code recommendations based on developers' comments in natural language and their code in the integrated development environment. The machine learning-powered service increases developer productivity.
-
Adobe Researchers Open-Source Image Captioning AI CLIP-S
Researchers from Adobe and the University of North Carolina (UNC) have open-sourced CLIP-S, an image-captioning AI model that produces fine-grained descriptions of images. In evaluations with captions generated by other models, human judges preferred those generated by CLIP-S a majority of the time.
-
AWS and Microsoft Working Together on PyWhy, the New Home of Causal ML Library DoWhy
AWS in a joint effort with Microsoft have established PyWhy as a fresh GitHub organization to integrate AWS algorithms into DoWhy, a casual ML library from Microsoft, which has moved to PyWhy.
-
Stanford University Open-Sources Controllable Generative Language AI Diffusion-LM
Researchers at Stanford University have open-sourced Diffusion-LM, a non-autoregressive generative language model that allows for fine-grained control of the model's output text. When evaluated on controlled text generation tasks, Diffusion-LM outperforms existing methods.
-
Amazon Released Incremental Training Feature in SageMaker JumpStart
AWS recently released a new feature in SageMaker (AWS Machine Learning Service) JumpStart to incrementally retrain machine-learning (ML) models trained with expanded datasets. By using this feature, developers could fine-tune their models for better performance in production with a couple of clicks. This recent feature is among the series of efforts to add more automation to SageMaker JumpStart.
-
Microsoft Limits Public Access to AI-Powered Facial Analysis Features
Microsoft recently announced phasing out public access to AI-powered Facial Analysis features in several Azure services.
-
GitHub Copilot Adopts Paid Model, Still Free for Some Open-Source Maintainers and Students
After almost one year in technical preview, GitHub Copilot is now prime time-ready for students and individual developers, says GitHub, while companies and larger organizations could get access to it before the end of the year.
-
DeepMind Trains 80 Billion Parameter AI Vision-Language Model Flamingo
DeepMind recently trained Flamingo, an 80B parameter vision-language model (VLM) AI. Flamingo combines separately pre-trained vision and language models and outperforms all other few-shot learning models on 16 vision-language benchmarks. Flamingo can also chat with users, answering questions about input images and videos.
-
Microsoft Launches New Storage Optimized VMs with Lasv3 and Lsv3
Recently Microsoft announced the general availability (GA) of new storage-optimized Azure Virtual Machines (VMs). These VMs are the Lasv3 and Lsv3 series designed to run workloads requiring high throughput and IOPS, including big data applications, SQL and NoSQL databases, distributed file systems, and data analytics engines.
-
Google's New Imagen AI Outperforms DALL-E on Text-to-Image Generation Benchmarks
Researchers from Google's Brain Team have announced Imagen, a text-to-image AI model that can generate photorealistic images of a scene given a textual description. Imagen outperforms DALL-E 2 on the COCO benchmark, and unlike many similar models, is pre-trained only on text data.