Metraipolis › archive › dc_a1bbe576fff2
Target-Aware Calibration Data Selection for Preserving Uncertainty in Quantized Language Models — arXiv:2608.21019
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.21019
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T10:18:28.322Z
- SHA-256 of the archived text
a4c5702f85e3c4f4b43af381fe83f2e01d810a2074531007edcd801d08b2fa8e- Stable URL of this page
https://metraipolis.com/documento/dc_a1bbe576fff2
How to cite
Target-Aware Calibration Data Selection for Preserving Uncertainty in Quantized Language Models — arXiv:2608.21019 — arXiv (en), https://arxiv.org/abs/2608.21019 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a1bbe576fff2 (Metraipolis archive dc_a1bbe576fff2, sha256 a4c5702f85e3c4f4b43af381fe83f2e01d810a2074531007edcd801d08b2fa8e)
BibTeX
@misc{dca1bbe576fff2,
title = {Target-Aware Calibration Data Selection for Preserving Uncertainty in Quantized Language Models — arXiv:2608.21019},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.21019},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a1bbe576fff2; sha256 a4c5702f85e3c4f4b43af381fe83f2e01d810a2074531007edcd801d08b2fa8e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.21019. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Quantization is widely used to deploy large language models, but its effect on uncertainty behavior, such as confidence, margins, and abstention, is rarely treated as a primary objective. We frame calibration-data selection for quantization as a target-dependent uncertainty-preservation problem. Different deployments emphasize different regions of the input distribution, yet prior work mainly optimizes accuracy-oriented compression metrics or adjusts scores after quantization. We formalize this goal with distributional and boundary preservation risks, and provide a simple mixture-mismatch argument explaining why no single calibration recipe should be expected to fit all targets. We introduce Doubt-Preserving Quantization (DPQ), a lightweight pre-quantization recipe family that uses full-precision predictions to construct target-aligned calibration mixtures of high-doubt examples and generic anchors. Across 8 language models, 9 NLP benchmarks, and 22 comparison methods, the leading fixed recipe changes with the preservation target: DPQ-r75 leads on SQuAD2 answerability-boundary preservation, while milder or single-signal variants, including DPQ-r50, confidence-only, and entropy-only, better preserve broad multiple-choice QA behavior. These results show that calibration data should be selected for the specific full-precision score behavior a deployment needs to preserve, rather than treated as a fixed quantization detail. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-24 in cs.CL). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.21019 — 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 2026-10-18T10:17:37.592Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).