Unified-IO 2: Scaling Autoregressive Multimodal Models with Vision, Language, Audio, and Action Paper • 2312.17172 • Published Dec 28, 2023 • 26