Search papers, labs, and topics across Lattice.
Shenzhen University
2
0
6
LLMs can develop more consistent world models by predicting multiple tokens *and* anchoring those predictions to ground-truth hidden state trajectories, mitigating structural hallucinations.
Real-time AI companions can now proactively interact with users thanks to Proact-VL, a framework that balances response latency, content quality, and video understanding.
