General AI Systems and Interaction Design / GUI Semantic Modeling and Task Representation

How does Screen2Vec perform in broad applicability across multi-task and multi-domain settings?