Search papers, labs, and topics across Lattice.
Technical University of Munich
1
0
2
LLMs can misclassify harmless rephrasings as critical errors, but a new lightweight metric outperforms larger models in clinical significance evaluation.
