Metraipolis › archive › dc_85bdff80e257

Preserve to Expand: Source-Aware Adaptation and Model Merging for Multilingual LLM-Based ASR — arXiv:2610.35846

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.35846
License (as archived): CC0-1.0
Archived: 2026-10-27T08:12:31.680Z
SHA-256 of the archived text: 4a16bc55d08417b81526739a0558f7e7cec3d4ac624ebbc4096c8ddef7fb287b
Stable URL of this page: https://metraipolis.com/documento/dc_85bdff80e257

How to cite

Preserve to Expand: Source-Aware Adaptation and Model Merging for Multilingual LLM-Based ASR — arXiv:2610.35846 — arXiv (en), https://arxiv.org/abs/2610.35846 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_85bdff80e257 (Metraipolis archive dc_85bdff80e257, sha256 4a16bc55d08417b81526739a0558f7e7cec3d4ac624ebbc4096c8ddef7fb287b)

BibTeX

@misc{dc85bdff80e257,
  title = {Preserve to Expand: Source-Aware Adaptation and Model Merging for Multilingual LLM-Based ASR — arXiv:2610.35846},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.35846},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_85bdff80e257; sha256 4a16bc55d08417b81526739a0558f7e7cec3d4ac624ebbc4096c8ddef7fb287b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.35846. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Expanding multilingual ASR systems to new languages is challenging when adaptation must preserve previously supported languages without replaying source-language training data. We study this problem in connector-based Speech LLMs, where a trainable connector maps speech representations into a frozen large language model. We evaluate target-update construction and merging strategies for adapting the connector while preserving source-language performance. In single-target expansion, target-only fine-tuning and standard TIES merging learn new languages but cause severe source-language forgetting, while SSU-Wanda+TIES gives the best source-preserving trade-off. In multi-target expansion, update merging provides stable source retention, whereas recursive expansion can improve target adaptation under favorable orders but is highly order-sensitive. These results show that multilingual ASR expansion requires balancing target acquisition, source retention, and interactions among target-language updates.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.35846 — 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 2026-10-27T07:56:00.281Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).