Research Visualization
Creating best-in-class visualizations, demos, and tools that amplify our AI research efforts.
ReViz is a cross-functional team within AI2 that creates beautiful, informative, interactive applications and demos that help ourselves and others to understand, share, and build upon our institute’s research. ReViz also contributes to the development of tools and platforms that assist researchers in efficiently building datasets and training models, advancing research efforts across the organization.
Digital Experiences
ReViz partners with researchers and engineers to produce compelling, educational demos of our latest AI advancements. ReViz also enables AI2 researchers to build their own interfaces for data collection, visualization, and presentation. We aim to increase the quantity and quality of research that AI2 shares with the world through engaging, informative, stable, and secure websites and web applications.
Communication
ReViz team members develop the content and channels through which AI2 communicates about our work. We create engaging stories and experiences to educate, inspire, and promote AI2’s work to the wider research community and beyond.
Design
ReViz creates and manages the AI2 brand and its implementation across digital and physical spaces while also developing and managing many of the sub-brands in the AI2 family. ReViz also creates, maintains, and implements an internal AI2 design system that makes it easy for our teams to build their own polished experiences.
Experimentation
ReViz provides AI2 researchers with key infrastructure and platforms to enable, streamline, and share their work. Beaker is a collaborative internal platform ReViz operates that allows for the seamless deployment, scripting, observation, debugging, reproducing, and sharing of machine learning experiments that utilize vast compute resources. ReViz also operates the AI2 Leaderboard platform, a place for the community to keep up to date with the latest research challenges and state-of-the-art models.
Delphi is intended to demonstrate what state-of-the-art models can accomplish today on machine ethics as well as to highlight their limitations.
Try the demo
Delphi is intended to demonstrate what state-of-the-art models can accomplish today on machine ethics as well as to highlight their limitations.
Try the demo
Unified-IO is the first neural model to perform a large, diverse set of AI tasks from computer vision to natural language processing.
Try the demoUnified-IO is the first neural model to perform a large, diverse set of AI tasks from computer vision to natural language processing.
Try the demoMacaw is a high-performance question-answering (QA) model capable of outperforming other popular current language models, all while being an order of magnitude smaller. This demo allows you to explore Macaw's answers and compare them to those of the popular GPT-3 language model on a benchmark set of questions.
Try the demo
Macaw is a high-performance question-answering (QA) model capable of outperforming other popular current language models, all while being an order of magnitude smaller. This demo allows you to explore Macaw's answers and compare them to those of the popular GPT-3 language model on a benchmark set of questions.
Try the demo
In new joint work at AI2 and UW, we study how often popular NLP components produce problematic content, what might trigger this neural toxic degeneration from a given system, and whether or not it can be successfully avoided. We also study how much toxicity is present in the web text that these systems learned…
Try the demoIn new joint work at AI2 and UW, we study how often popular NLP components produce problematic content, what might trigger this neural toxic degeneration from a given system, and whether or not it can be successfully avoided. We also study how much toxicity is present in the web text that these systems learned…
Try the demo
