Metraipolis › archive › dc_33eb15c91007

Task Format as a Factor in Adaptation: A Controlled Comparison of LoRA and Full Fine-Tuning for Prompt-Safety Classification — arXiv:2610.33813

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.33813
License (as archived)
CC0-1.0
Archived
2026-10-27T08:11:26.691Z
SHA-256 of the archived text
5ee5cb4a791267c4ffd36fa02391320c606af8efe8140891a620064deab9c6ac
Stable URL of this page
https://metraipolis.com/documento/dc_33eb15c91007

How to cite

Task Format as a Factor in Adaptation: A Controlled Comparison of LoRA and Full Fine-Tuning for Prompt-Safety Classification — arXiv:2610.33813 — arXiv (en), https://arxiv.org/abs/2610.33813 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_33eb15c91007 (Metraipolis archive dc_33eb15c91007, sha256 5ee5cb4a791267c4ffd36fa02391320c606af8efe8140891a620064deab9c6ac)
BibTeX
@misc{dc33eb15c91007,
  title = {Task Format as a Factor in Adaptation: A Controlled Comparison of LoRA and Full Fine-Tuning for Prompt-Safety Classification — arXiv:2610.33813},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.33813},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_33eb15c91007; sha256 5ee5cb4a791267c4ffd36fa02391320c606af8efe8140891a620064deab9c6ac},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.33813. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Fine-tuning a language model for classification involves many design decisions. Comparisons of low-rank adaptation (LoRA) with full fine-tuning are common but often confounded, and task format is rarely studied. We compare both regimes on prompt-safety classification with a matched 400M encoder-decoder pair (Ettin) across three task formats, with up to 15 seeds per cell and pre-specified analyses. At the pre-specified learning rates, full fine-tuning leads LoRA by 2-3 pp in every architecture and format trained on the label alone. On the encoder, the gap closes once both rates are chosen by the same search. Training LoRA to write a reason after its verdict adds 4.0 pp and closes the gap, but a content-free sentence of equal length does not. Trained models score above a prompted 8B in domain but fall behind it on two of three out-of-distribution corpora, trailing by 8-23 pp on adversarial jailbreaks and flagging 27-66% of benign but harmful-looking prompts as unsafe, against the 8B's 2-3%. Together, the results suggest that the adaptation regime matters less than how each model is trained, its learning rate, and training target, and that in-domain accuracy alone is an incomplete guide to robustness.

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/2610.33813 — 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 2026-10-27T07:56:00.211Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).