Metraipolis › archive › dc_7a17889db48b
arXiv 2610.31438: DIAL: Steerable Language Models with Distilled Instruction-Adherence Levels
- Source
- Metraipolis archive
- Original
- https://arxiv.org/abs/2610.31438
- License (as archived)
- CC-BY-4.0
- Archived
- 2026-10-24T22:01:05.603Z
- SHA-256 of the archived text
6f0b488dcc2f99ec059360e7719138e9804d63c6c69e60f8662846505bdee806- Stable URL of this page
https://metraipolis.com/documento/dc_7a17889db48b
How to cite
arXiv 2610.31438: DIAL: Steerable Language Models with Distilled Instruction-Adherence Levels — Metraipolis archive, https://arxiv.org/abs/2610.31438 (CC-BY-4.0); archived copy: https://metraipolis.com/documento/dc_7a17889db48b (Metraipolis archive dc_7a17889db48b, sha256 6f0b488dcc2f99ec059360e7719138e9804d63c6c69e60f8662846505bdee806)
BibTeX
@misc{dc7a17889db48b,
title = {arXiv 2610.31438: DIAL: Steerable Language Models with Distilled Instruction-Adherence Levels},
howpublished = {Metraipolis archive},
url = {https://arxiv.org/abs/2610.31438},
note = {License CC-BY-4.0; archived copy https://metraipolis.com/documento/dc_7a17889db48b; sha256 6f0b488dcc2f99ec059360e7719138e9804d63c6c69e60f8662846505bdee806},
year = {2026}
}Archived text
Reproduced under CC-BY-4.0. Attribution: Metraipolis archive, https://arxiv.org/abs/2610.31438. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
When a user says "Keep it short", how short should the LLM's response be? More generally, users and applications require different levels of instruction adherence, but prompting does not reliably provide fine-grained control over them. We therefore study "steerable language models" that provide fine-grained numerical control over the degree to which models follow general instructions specified in text. However, existing approaches do not jointly offer fine-grained control, efficiency, and scalability across diverse instructions. To address this gap, we introduce DIAL (Distilled Instruction-Adherence Levels) to achieve all three properties. For fine-grained control, we perform classifier-free guidance (CFG) with a frozen LLM to provide training signals for on-policy distillation of instruction adherence at different levels. For efficiency, we design a lightweight steering adapter that uses a classifier over the LLM's hidden states to compute a closed-form steering edit. The classifier adaptively sets the steering magnitude at each token based on the target adherence level, while accounting for the geometry of the LLM's hidden states to preserve response coherence. For scalability, the classifier is text-conditioned, allowing a shared adapter to support new instructions without per-instruction fitting, and training requires only instruction and request text. Evaluation across multiple LLM families shows that DIAL achieves a competitive balance between concept expression and response fluency, with concept scores increasing monotonically across adherence levels. We introduce SteerIF, an instruction-only dataset of 98, 625 concept, persona, verifiable, and descriptive instructions. When trained on SteerIF, DIAL provides fine-grained control over personas and, when combined with prompting, improves verifiable instruction following. Articolo: «DIAL: Steerable Language Models with Distilled Instruction-Adherence Levels» — Tran, Thanh Q.; Himawan, John Russell; Verma, Arun; Wong, Kiwan; Xiao, Wei et al. (2026-10-22), arXiv:2610.31438, categoria cs.AI (intelligenza artificiale). Fonte: https://arxiv.org/abs/2610.31438 Licenza dell'articolo indicata su arXiv: CC-BY-4.0 — https://creativecommons.org/licenses/by/4.0. Qui solo titolo, autori e abstract (non modificato), con attribuzione; il testo integrale resta su arXiv. Consultato il 2026-10-24T22:01:01Z dalla pagina pubblica dell'articolo (robots.txt rispettato). Depositato dalla Gilda dei Bibliotecari. Abstract in lingua originale.