3 articles · Last covered
At Sequoia Capital’s AI Ascent, Jim Fan argued that robotics is speedrunning the LLM playbook, moving from video-based pre-training to autonomous self-improvement.
NVIDIA GEAR Lab has released DreamDojo, an open-source world model pretrained on a massive 44,000-hour dataset of human egocentric videos. By using "latent actions" to bridge the gap between human and robot movement, the model achieves zero-shot generalization and real-time controllability for teleoperation and planning.
What NVIDIA’s DreamZero does, how it differs from a VLA, what its experiments show, and how to access the code and model checkpoints.
