Search papers, labs, and topics across Lattice.
School of Artificial Intelligence, University of Chinese Academy of Sciences
3
0
6
0
LLM safety filters can be bypassed by strategically fragmenting and camouflaging malicious intent across multiple turns, achieving a 26% improvement in jailbreak success rate on GPT-5-mini.
Achieve state-of-the-art MRI super-resolution without paired training data by encoding physical properties into 3D Gaussians and rendering them efficiently.
Generate impressive cover songs with fewer parameters: SongEcho uses a novel modulation technique to reinterpret existing melodies with new vocals and accompaniment, outperforming existing methods while using less compute.
