Prefer a chat interface with context about you and your work?
Oscar: Object-Semantics Aligned Pre-training for Vision-Language Tasks