Metraipolis › archive › dc_55f6ce70dade
Token Distribution versus Data Volume: Domain Balancing in Multi-Domain Meeting Summarisation — arXiv:2608.15935
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.15935
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T18:58:50.621Z
- SHA-256 of the archived text
4c40bfee36a67b98634382dae7a4f77f4d9a6c171dc0f3a0121e658868b9b335- Stable URL of this page
https://metraipolis.com/documento/dc_55f6ce70dade
How to cite
Token Distribution versus Data Volume: Domain Balancing in Multi-Domain Meeting Summarisation — arXiv:2608.15935 — arXiv (en), https://arxiv.org/abs/2608.15935 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_55f6ce70dade (Metraipolis archive dc_55f6ce70dade, sha256 4c40bfee36a67b98634382dae7a4f77f4d9a6c171dc0f3a0121e658868b9b335)
BibTeX
@misc{dc55f6ce70dade,
title = {Token Distribution versus Data Volume: Domain Balancing in Multi-Domain Meeting Summarisation — arXiv:2608.15935},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.15935},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_55f6ce70dade; sha256 4c40bfee36a67b98634382dae7a4f77f4d9a6c171dc0f3a0121e658868b9b335},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.15935. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Jointly fine-tuning an LLM on meeting-summarisation corpora of widely varying size raises a question that prior work leaves confounded: when a domain-balanced training mixture helps, is the gain due to the distribution of tokens across domains, or merely to the volume of data seen? We disentangle these factors by constructing balanced and natural (native-proportional) token mixtures at matched token budgets (2-32M) over five English meeting corpora, fine-tuning Mistral-7B with QLoRA, and evaluating per domain. Balancing redistributes quality, improving the data-scarce minority domains at a low cost to the data-rich ones. The trade favours balancing whenever the minority domains matter: their share under proportional allocation is fixed at 1-2% regardless of budget, so matching balanced quality on those domains requires far more total data. We further find that pruning low-value transcript lines removes ~15% of tokens from the conversational corpora at no measurable cost, and that balancing by tokens is not the same as balancing by examples. Fine-tuning one model per domain is competitive only on the data-rich domains and falls below the zero-shot model on the data-scarce ones. A two-annotator study of 741 judge-labelled facts validates our fact-level evaluation. Together these results give practitioners a basis for deciding when to balance an imbalanced multi-domain mixture, and on what unit. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2608.15935 — 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 2026-10-18T18:06:57.871Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).