Leaderboard Hacking: Preference-Based Model Evaluations are Vulnerable to Manipulation

arXiv:2611.14769 · cs.AI · Submitted 2026-11-10 · Read on arXiv

cs.AI

Submitted: 2026-11-10

Updated: 2026-11-10

Code: https://github.com/lmarena/arena-rank

Terminology

Sources

Related papers