Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities
[NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.
