Metraipolis › archive › dc_753e27dd7ba0

Beyond Polarization: The Generative Constraint of Chain-of-Thought in Pointwise Reranking — arXiv:2608.30398

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.30398
License (as archived)
CC0-1.0
Archived
2026-10-17T08:07:32.265Z
SHA-256 of the archived text
aec016f8105deaad7b33586a1768f085a574387928bbb53ad119381779c51f67
Stable URL of this page
https://metraipolis.com/documento/dc_753e27dd7ba0

How to cite

Beyond Polarization: The Generative Constraint of Chain-of-Thought in Pointwise Reranking — arXiv:2608.30398 — arXiv (en), https://arxiv.org/abs/2608.30398 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_753e27dd7ba0 (Metraipolis archive dc_753e27dd7ba0, sha256 aec016f8105deaad7b33586a1768f085a574387928bbb53ad119381779c51f67)
BibTeX
@misc{dc753e27dd7ba0,
  title = {Beyond Polarization: The Generative Constraint of Chain-of-Thought in Pointwise Reranking — arXiv:2608.30398},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.30398},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_753e27dd7ba0; sha256 aec016f8105deaad7b33586a1768f085a574387928bbb53ad119381779c51f67},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30398. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

In pointwise document reranking, Chain-of-Thought models typically underperform direct scoring models. While existing diagnostics attribute this to inferior classification, score polarization, or calibration breakdown, whether targeted training can bridge this gap remains unclear. Our empirical study first confirms that this gap is stable across scales up to 32B parameters, ruling out model and data capacity confounders. We then apply stress tests utilizing reinforcement learning, fine-grained supervision, and architectural decoupling to explicitly repair these deviations. Although these interventions improve classification accuracy and absolute scores, the relative ranking gap persists. These findings suggest that, within the pointwise scoring paradigm, routing continuous relevance semantics through discrete text constrains ranking signal resolution, revealing a bottleneck that is stable and difficult to overcome under current standard methods, rather than an easily resolvable training bias.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Information Retrieval (cs.IR).
Pagina del lavoro: https://arxiv.org/abs/2608.30398 — 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 2026-10-17T07:47:52.812Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).