Search papers, labs, and topics across Lattice.
IQuest Research, University of Manchester
1
0
2
25
LLM-generated rubrics can nearly match human evaluation standards, but they often miss the mark with excessive detail and scoring bias.
