English

Post-Training Language Models for Crosslingual Consistency

Computation and Language 2026-05-29 v3 Artificial Intelligence

Abstract

Language models often respond inconsistently to translation-equivalent prompts across languages, undermining the reliability of multilingual systems. To quantify this, we give an information-theoretic definition of crosslingual consistency as a divergence bound between a model's response distribution and its round-trip pushforward across languages. We then introduce penalized consistency optimization (PCO), a post-training procedure that couples this divergence with a Kullback-Leibler penalty to a fixed reference language model. Because direct optimization of PCO requires expensive on-policy roll-outs, we propose a tractable surrogate, direct consistency optimization (DCO), which can be optimized off-policy. Across diverse language models and 26 languages, DCO significantly improves crosslingual consistency, outperforms existing methods, and enables targeted alignment of low-resource languages.

Keywords

Cite

@article{arxiv.2603.04678,
  title  = {Post-Training Language Models for Crosslingual Consistency},
  author = {Tianyu Liu and Jirui Qi and Mrinmaya Sachan and Ryan Cotterell and Raquel Fernández and Arianna Bisazza},
  journal= {arXiv preprint arXiv:2603.04678},
  year   = {2026}
}

Comments

ICML 2026. The first two authors contributed equally. Codes available at: https://github.com/Betswish/ConsistencyRL

R2 v1 2026-07-01T11:04:05.532Z