Search papers, labs, and topics across Lattice.
4
0
6
Memory consistency in video generation models falters significantly when objects disappear, with state-of-the-art models struggling to recover updated states upon reappearance.
Contextual tunneling in LLMs can be overcome, leading to more reliable and physically grounded materials discovery through the innovative ARIA framework.
Generating a single, effective "Figure 1" from a research paper proves surprisingly difficult for even the best Vision-Language Models, revealing a gap in their ability to couple scientific understanding with visual synthesis.
A 1000x larger video reasoning dataset reveals early signs of emergent generalization, offering a new foundation for training and evaluating spatiotemporal AI.
