Metraipolis › archive › dc_14746aa3095e
The Shape of the DPO Hyperparameter Landscape: What to Measure, How It Moves with Scale, and When Not to Transfer It — arXiv:2610.27895
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.27895
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-21T07:08:07.776Z
- SHA-256 of the archived text
f74238eac5e225c52c6e40042cf63ecf8e4fa7f875db99b4c829273360dc6af2- Stable URL of this page
https://metraipolis.com/documento/dc_14746aa3095e
How to cite
The Shape of the DPO Hyperparameter Landscape: What to Measure, How It Moves with Scale, and When Not to Transfer It — arXiv:2610.27895 — arXiv (en), https://arxiv.org/abs/2610.27895 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_14746aa3095e (Metraipolis archive dc_14746aa3095e, sha256 f74238eac5e225c52c6e40042cf63ecf8e4fa7f875db99b4c829273360dc6af2)
BibTeX
@misc{dc14746aa3095e,
title = {The Shape of the DPO Hyperparameter Landscape: What to Measure, How It Moves with Scale, and When Not to Transfer It — arXiv:2610.27895},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.27895},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_14746aa3095e; sha256 f74238eac5e225c52c6e40042cf63ecf8e4fa7f875db99b4c829273360dc6af2},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.27895. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Direct preference optimisation (DPO) is widely used for language model post-training, but its performance depends strongly on the choice of its hyperparameters, such as the learning rate and the reference-policy regularisation parameter $\beta$. To better understand the DPO hyperparameter landscape, we conduct extensive experiments using a dense hyperparameter grid across different model scales from the same family. This allows us to analyse how the optimal DPO hyperparameters evolve with model size and the number of update steps, with intermediate checkpoints providing an inexpensive signal to accelerate the hyperparameter optimisation process. Furthermore, by evaluating our models on both static and dynamic benchmarks, we identify which benchmarks provide the strongest signal for hyperparameter selection. Finally, we investigate how different state-of-the-art hyperparameter optimisation strategies perform in this setting. Our experiments show that the optimal learning rate varies with model size and that hyperparameters are not transferable across training datasets. We also find that validation loss is a poor proxy for hyperparameter optimisation, as hyperparameters that minimize validation loss only weakly correlate with optimal hyperparameters for downstream benchmarks. Therefore, DPO hyperparameters need to be tuned specifically for the target model, dataset, and evaluation objective. We also show that short candidate training runs can substantially reduce the cost of this selection. We release all our training configurations and evaluation results to support reproducible analysis of selection quality and training cost. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.27895 — 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 2026-10-21T07:05:43.606Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).