Search for a command to run...
Lens: Rethinking Training Efficiency for Foundational Text-to-Image Models