Overview
Explore the current trend of increasingly larger AI models in this keynote address from ODSC East 2022. Delve into the research and engineering efforts supporting this trend, its product and engineering impact at Microsoft, and the implications for other companies. Learn about AirScale Vision, the explosion of data, supervised learning, transformer architecture, foundational models, and scaling laws of AI. Discover how these advancements are empowering a wider range of tasks in language, vision, and multi-modality spaces with growing levels of capability. Gain insights into practical applications in services like Sharepoint, Outlook, Dynamics, and Power Apps, as well as the development of multimodal models and CoPilot. Understand the infrastructure and resources required to support these large-scale AI models and their potential impact on various industries.
Syllabus
Introduction
AirScale Vision
The Past Two Years
Research
Explosion of Data
Airscale
supervised learning
transformer architecture
foundational models
scaling laws of AI
size of models
progress over time
other models
Albert Einstein
Transformer Models
Examples
Sharepoint
Outlook
Dynamics
Power Apps
CoPilot
multimodal
new images
raccoon astronaut
services
infrastructure
resources
Taught by
Open Data Science