Metraipolis › archive › dc_1290ef660a8d

Different Perturbations, Different Mechanisms: Understanding Continued Pre-training for Zero-Shot Dialect Robustness — arXiv:2608.05510

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.05510
License (as archived)
CC0-1.0
Archived
2026-10-25T21:27:09.195Z
SHA-256 of the archived text
930be90e5a992ac8dfdfd4e164e4a64f32700d1e1eb86fcae67f1d7ace343fcb
Stable URL of this page
https://metraipolis.com/documento/dc_1290ef660a8d

How to cite

Different Perturbations, Different Mechanisms: Understanding Continued Pre-training for Zero-Shot Dialect Robustness — arXiv:2608.05510 — arXiv (en), https://arxiv.org/abs/2608.05510 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_1290ef660a8d (Metraipolis archive dc_1290ef660a8d, sha256 930be90e5a992ac8dfdfd4e164e4a64f32700d1e1eb86fcae67f1d7ace343fcb)
BibTeX
@misc{dc1290ef660a8d,
  title = {Different Perturbations, Different Mechanisms: Understanding Continued Pre-training for Zero-Shot Dialect Robustness — arXiv:2608.05510},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.05510},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_1290ef660a8d; sha256 930be90e5a992ac8dfdfd4e164e4a64f32700d1e1eb86fcae67f1d7ace343fcb},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.05510. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Dialectal variation remains a major challenge for multilingual language models. Perturbation-based continued pre-training (CPT) has emerged as a promising approach to improving robustness, yet existing work largely evaluates individual perturbation strategies in isolation and provides limited insight into why they work. We present a systematic study of perturbation-based CPT for multilingual dialect robustness in LLMs, comparing six training conditions across nine German, Italian, and Arabic dialect tasks. Perturbation-based CPT, especially character-noised CPT, consistently improves zero-shot dialect robustness while largely preserving standard variety performance. More importantly, we show that methods with similar downstream performance induce distinct mechanisms of robustness, exhibiting different patterns of language model adaptation, representational alignment, and prediction repair. Our results provide a more complete understanding of how synthetic surface variation improves robustness and offer practical guidance for selecting CPT strategies in multilingual and dialectal settings.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-07 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.05510 — 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 2026-10-25T21:24:38.192Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).