Search papers, labs, and topics across Lattice.
University of South Dakota
3
0
6
Lightweight models can achieve superior polyp segmentation performance by effectively harnessing foundation model induction with minimal computational overhead.
Freezing large LLMs while using a lightweight diagnostic-prior framework can significantly enhance performance in 3D CT report generation with minimal training overhead.
Swap out quadratic cross-attention for linear-time inference in vision-language models without sacrificing fine-grained understanding.
