Search papers, labs, and topics across Lattice.
2
0
5
0
Imagine creating high-fidelity, navigable 3D worlds from just a text prompt or a single image – HY-World 2.0 makes it a reality.
Explicitly teaching LVLMs to reason step-by-step with reinforcement learning unlocks state-of-the-art performance on multimodal object-entity relation extraction.
