Search papers, labs, and topics across Lattice.
6
0
6
3
Current action-conditioned world models are limited by their reliance on visual patterns, failing to generalize physical dynamics across different robot embodiments.
Geometry-derived features in HERMES boost N-stage accuracy by 4.3% while simplifying the model's complexity.
Achieving high-fidelity stylized 3D asset generation in under 10 seconds could revolutionize content creation in gaming and animation.
Action-only decoding in GigaWorld-Policy-0.5 slashes inference latency to 85 ms, revolutionizing real-time robot control efficiency.
FlowWAM achieves a remarkable 92.94% success rate in manipulation tasks by harnessing optical flow as a video-native action representation.
Achieve SOTA joint audio-video generation with JavisDiT++ using just 1M public training examples, rivaling performance of models trained on proprietary datasets.
