Metraipolis › archive › dc_8fa573489883

Likelihood Ranking doesn't Scale Like Prompting in LLMs — arXiv:2609.29390

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.29390
License (as archived)
CC0-1.0
Archived
2026-10-08T04:49:30.312Z
SHA-256 of the archived text
081402283cf970d599c2d7656f85ed13eb6ee7274e5ead32c9da5ba1a886c4dc
Stable URL of this page
https://metraipolis.com/documento/dc_8fa573489883

How to cite

Likelihood Ranking doesn't Scale Like Prompting in LLMs — arXiv:2609.29390 — arXiv (en), https://arxiv.org/abs/2609.29390 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_8fa573489883 (Metraipolis archive dc_8fa573489883, sha256 081402283cf970d599c2d7656f85ed13eb6ee7274e5ead32c9da5ba1a886c4dc)
BibTeX
@misc{dc8fa573489883,
  title = {Likelihood Ranking doesn't Scale Like Prompting in LLMs — arXiv:2609.29390},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.29390},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_8fa573489883; sha256 081402283cf970d599c2d7656f85ed13eb6ee7274e5ead32c9da5ba1a886c4dc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.29390. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

LLM evaluation is commonly performed either by prompting models to produce answers or by scoring candidate outputs with likelihood-based metrics. In multiple-choice QA, however, standard likelihood-based scoring is still conditioned on the question and answer set, and can therefore leverage the same task-conditioned answer-selection interface used in prompting. We study a complementary protocol based on likelihood ranking of declarative statements constructed from the same question--answer pairs. Across 95 decoder-only models, ranging from 0.1B to 104B parameters, and 10 MCQA datasets, we find a systematic divergence between declarative-statement likelihood ranking and prompted answering. Statement-likelihood accuracy remains comparatively stable across scale, whereas prompted answering improves sharply with scale and instruction-tuning. These results suggest that likelihood preferences over controlled declarative alternatives and task-conditioned answer selection probe distinct aspects of model behavior, and should not be treated as interchangeable.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-25 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.29390 — 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 2026-10-08T04:27:20.051Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).