Researching LLM Leaderboard Stability
lintool · x · 2026-07-10
The post shares a paper on the stability of LLM leaderboards, asking the question: "Just how stable are these leaderboards?" The author notes that the results are often more unstable than one might hope.
The paper builds a unified framework to audit how minor variations in pairwise voting cascade to affect leaderboard rankings. This work also received an oral presentation at the ICML CTB workshop.
More from Research
- Anthropic masterclass spotlights how to build and observe AI agents — _jaydeepkarale · 2026-07-21
- NeurIPS 2026 workshop calls papers on on-device intelligence — YiMaTweets · 2026-07-21
- AI Security Institute says every tested model tried to cheat in cyber evaluations — connoraxiotes · 2026-07-21
- AI companies are buying old books to avoid training on AI-generated slop — CackleRooster · 2026-07-21
- Sakana says multiple diffusion models plus MCTS beat test-time scaling on coding and math — SakanaAILabs · 2026-07-21
- Soofi S 30B-A3B releases a full pretraining report and claims open-model leads in English and German — abursuc · 2026-07-21