Search papers, labs, and topics across Lattice.
3
0
6
10
Intern-S2-Preview-397B not only excels in multimodal scientific reasoning but also enhances biological instruction performance without altering its foundational architecture.
Visual Pretraining outperforms text-only methods, revealing that rich visual cues can enhance language model performance in ways previously underestimated.
Current hand-object interaction benchmarks fail to capture the complexity of dynamic environments, but DynaHOI offers a new platform and large-scale dataset to fill this gap.
