Search papers, labs, and topics across Lattice.
Sapienza University of Rome
3
0
4
Tail-shape estimation in LLM evaluations is more prone to false positives than the literature suggests, challenging the reliability of current methodologies.
