Search papers, labs, and topics across Lattice.
Beihang University
1
0
3
LLM judges in multi-stakeholder settings suffer from "weighting noise" that gets *worse* as you add more stakeholders, but fixing weights upfront can stabilize the process.