Search papers, labs, and topics across Lattice.
University of Canberra
1
0
3
GRPO's hierarchical penalty reduces vulnerability prediction errors by nearly 28%, outperforming larger models under challenging conditions.
