Metraipolis › archive › dc_364df696bcbe

Quantiles, Not Points: Streaming Quantile Matching for Ordinal-Aware LLM-as-a-Judge Calibration — arXiv:2611.00447

Source: arXiv (en)
Original: https://arxiv.org/abs/2611.00447
License (as archived): CC0-1.0
Archived: 2026-11-03T18:11:57.943Z
SHA-256 of the archived text: d59af8eda37293d1b91145af80fdcc62672c7bc7eba07815af8b293bedef992d
Stable URL of this page: https://metraipolis.com/documento/dc_364df696bcbe

How to cite

Quantiles, Not Points: Streaming Quantile Matching for Ordinal-Aware LLM-as-a-Judge Calibration — arXiv:2611.00447 — arXiv (en), https://arxiv.org/abs/2611.00447 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_364df696bcbe (Metraipolis archive dc_364df696bcbe, sha256 d59af8eda37293d1b91145af80fdcc62672c7bc7eba07815af8b293bedef992d)

BibTeX

@misc{dc364df696bcbe,
  title = {Quantiles, Not Points: Streaming Quantile Matching for Ordinal-Aware LLM-as-a-Judge Calibration — arXiv:2611.00447},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.00447},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_364df696bcbe; sha256 d59af8eda37293d1b91145af80fdcc62672c7bc7eba07815af8b293bedef992d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00447. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

LLM-as-a-judge scores are often distributionally shifted from human ratings on any given evaluation task, and post-hoc calibration (mapping LLM to human scores in a small labeled anchor set) has become a popular method to correct for this shift. In this paper we derive the calibration from two positions: that LLMaJ scores are ordinal ratings drawn from a distribution rather than isolated points. We present streaming quantile matching (SQM), a version of a classical method not previously bench marked for LLMaJ calibration, which is built on the two positions by computing the optimal-transport map between the LLM and human score distributions. We develop an evaluation framework and within it we show that unlike point-wise calibrations, SQM respects the ordinality of the human anchor set and achieves the best distributional fidelity and the top overall rank across models and data. We also show that in general, post-hoc calibration narrows the raw error gap between smaller and larger judge models, which opens the possibility of more efficient LLM judges.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2611.00447 — 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 2026-11-03T18:09:53.283Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).