Metraipolis › archive › dc_d3f3ebb4839e
Behind Aggregate Scores: What Systematically Varying Arena-Hard's LLM Judge Reveals — arXiv:2610.30512
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.30512
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-23T15:14:57.725Z
- SHA-256 of the archived text
a5bfe13f9118cef58b4b0e411f2b85528e92c1df99b12366407ecbe0eeaf18f4- Stable URL of this page
https://metraipolis.com/documento/dc_d3f3ebb4839e
How to cite
Behind Aggregate Scores: What Systematically Varying Arena-Hard's LLM Judge Reveals — arXiv:2610.30512 — arXiv (en), https://arxiv.org/abs/2610.30512 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d3f3ebb4839e (Metraipolis archive dc_d3f3ebb4839e, sha256 a5bfe13f9118cef58b4b0e411f2b85528e92c1df99b12366407ecbe0eeaf18f4)
BibTeX
@misc{dcd3f3ebb4839e,
title = {Behind Aggregate Scores: What Systematically Varying Arena-Hard's LLM Judge Reveals — arXiv:2610.30512},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.30512},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d3f3ebb4839e; sha256 a5bfe13f9118cef58b4b0e411f2b85528e92c1df99b12366407ecbe0eeaf18f4},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.30512. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
LLM judges are used to determine which of two answers is better. Trusting these verdicts requires knowing how repeatable they are and how much they depend on the judge's instructions. We show that similar aggregate scores can conceal substantial changes in individual verdicts, both when the same setup is repeated and when the instructions change. Using 500 Arena-Hard v2.0 prompts with two models' answers held fixed, we run GPT-4.1 as the judge under all eight combinations of criteria, output format and task framing, and repeat every setup unchanged at temperature zero on a pinned model version. Rerunning the default setup moves the aggregate score by about one point but changes the win/tie/loss outcome on 21.2% of prompts. Instruction changes introduce further sensitivity: in one setup, which asks for the verdict without reasoning and also drops several other instructions, outcomes change on 28.2 percentage points more prompts than the two setups' unchanged reruns do on average, and the aggregate score falls by 4.09 points. In another, where the criteria prioritise accuracy, outcomes change on 6.7 percentage points more prompts by the same comparison, while the aggregate score barely moves. A stable aggregate score therefore does not establish that individual verdicts are repeatable. This distinction matters when those verdicts are used as preference labels or to assess individual responses. Evaluations that rely on individual verdicts should repeat each setup they use and report both aggregate scores and prompt-level agreement. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Computers and Society (cs.CY). Pagina del lavoro: https://arxiv.org/abs/2610.30512 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-23T15:09:08.527Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).