Metraipolis › archive › dc_f0a41e885686
SemiAdapt-Instruct: Extensible Instruction Tuning via Latent Domain-Specialised Adapters — arXiv:2608.05161
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.05161
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-25T21:27:01.063Z
- SHA-256 of the archived text
7ef776f29212f1d798cac2aadf163641240d386581b9951b3b2c4fd8dffe313b- Stable URL of this page
https://metraipolis.com/documento/dc_f0a41e885686
How to cite
SemiAdapt-Instruct: Extensible Instruction Tuning via Latent Domain-Specialised Adapters — arXiv:2608.05161 — arXiv (en), https://arxiv.org/abs/2608.05161 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f0a41e885686 (Metraipolis archive dc_f0a41e885686, sha256 7ef776f29212f1d798cac2aadf163641240d386581b9951b3b2c4fd8dffe313b)
BibTeX
@misc{dcf0a41e885686,
title = {SemiAdapt-Instruct: Extensible Instruction Tuning via Latent Domain-Specialised Adapters — arXiv:2608.05161},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.05161},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f0a41e885686; sha256 7ef776f29212f1d798cac2aadf163641240d386581b9951b3b2c4fd8dffe313b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.05161. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Instruction-tuned LLMs are deployed into environments where domains evolve, yet extending a fine-tuned model's capabilities without full retraining remains an unsolved practical challenge. We present SemiAdapt-Instruct, a modular framework that discovers latent instruction domains, trains per-domain LoRA adapters in parallel, and performs parameter-free routing, incorporating new domains via single-adapter training without modifying existing components. SemiAdapt-Instruct outperforms full model fine-tuning across all configurations on both ROUGE-L and LLM-as-a-judge evaluation, while matching single LoRA fine-tuning and delivering extensibility that monolithic approaches cannot provide. We empirically demonstrate this extensibility by showing that updating a single adapter with new domain data outperforms all monolithic baselines. Our study also finds that independent discovery methods converge on the same specialisation-friendly domains. These findings demonstrate that decomposing heterogeneous instruction data into latent domains enables extensible NLP systems where evolving domains require only targeted single-adapter updates, eliminating the need for full model retraining. 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) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2608.05161 — 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 2026-10-25T21:24:38.169Z 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).