Metraipolis › archive › dc_fc4e33bc375c

Predicting Human-Gain Curves under Local Proxy Optimization from Repeated Ratings — arXiv:2610.41114

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.41114
License (as archived): CC0-1.0
Archived: 2026-11-03T07:50:35.171Z
SHA-256 of the archived text: e5883283956257ab001eed9388303511c2dddf7451e26f9d89370ad0d87727ef
Stable URL of this page: https://metraipolis.com/documento/dc_fc4e33bc375c

How to cite

Predicting Human-Gain Curves under Local Proxy Optimization from Repeated Ratings — arXiv:2610.41114 — arXiv (en), https://arxiv.org/abs/2610.41114 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fc4e33bc375c (Metraipolis archive dc_fc4e33bc375c, sha256 e5883283956257ab001eed9388303511c2dddf7451e26f9d89370ad0d87727ef)

BibTeX

@misc{dcfc4e33bc375c,
  title = {Predicting Human-Gain Curves under Local Proxy Optimization from Repeated Ratings — arXiv:2610.41114},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.41114},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fc4e33bc375c; sha256 e5883283956257ab001eed9388303511c2dddf7451e26f9d89370ad0d87727ef},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41114. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Optimizing a proxy objective can improve human-perceived quality, but the resulting gain can vary with optimization strength. Average proxy-human correlation alone does not reveal how human-perceived quality changes as optimization strength increases. We propose PACE (Proxy Alignment and Curvature Expansion), a pre-optimization diagnostic that uses repeated human ratings to predict the human-gain curve induced by a fixed proxy. The method models proxy optimization as a KL-regularized exponential tilt. At KL radius $d$, the local gain is approximated by $A\sqrt{d}+Bd$, where $A$ measures first-order proxy-human alignment and $B$ captures the leading second-order correction along the proxy-optimization path. We estimate $A$ and $B$ from repeated human ratings on calibration items and test the resulting gain-curve prediction on held-out evaluation items. Across SummEval, analysis-eligible WMT MQM clusters, an OpenMEVA-MAGS ROC/WP repeated-rating slice, and a GPT-4o judge proxy experiment, PACE consistently reduces held-out gain-prediction error relative to $A$-only, zero-gain, and reliability-only baselines. With the same pre-pool information, it is competitive with flexible proxy-conditional calibration, and a LoRA-DPO experiment provides a modest bridge to actual optimization. Transfer experiments across datasets and evaluation slices show that source coefficients are useful priors, but target calibration further improves prediction. The resulting diagnostic is local and one-sided: it characterizes gentle optimization along the positive tilt branch of a fixed item pool.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.41114 — 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 2026-11-03T07:49:31.034Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).