Search papers, labs, and topics across Lattice.
Zhejiang University
2
0
4
Over 80% of real-world LLM applications leak sensitive prompts, but a new defense, AREA, not only mitigates this risk but also boosts usability by over 33%.
Hallucinations in video models can be mitigated by intelligently patching visual tokens, leading to a 50.60% improvement in response accuracy.
