Projects JARVIS-1JARVIS-1: Open-world Multi-task Agents with Memory-Augmented Multimodal Language Models GROOTGROOT: Learning to Follow Instructions by Watching Gameplay Videos
