---

Dataset Details

This dataset includes all the data used to fine-tune Luth-0.6B-Instruct and Luth-1.7B-Instruct, enhancing their French capabilities on tasks such as instruction following, mathematics, and general knowledge. The models also improved in English thanks to knowledge transfer between the two languages.

It contains ~338M tokens in French. Our data scripts are available on GitHub.

Dataset Sources

Scholar

By Kurakura AI: Dataset Link.
Built from scraped subjects of French Baccalauréat and Preparatory Class (CPGE) entrance exams in mathematics, computer science, and physics.

Tulu 3 Persona Instruct

By AllenAI: Dataset Link.
Translated prompts to French, generated new answers with Qwen3-32B, then filtered the dataset.

Tulu 3 Persona Math

By AllenAI: Dataset Link.
Translated prompts to French, generated new answers with Qwen3-32B, then filtered the dataset.

Smoltalk2

By HuggingFaceTB: Dataset Link.
Extracted French samples only.

Aya Dataset

By CohereLabs: Dataset Link.
Extracted French samples only.

OpenHermes-fr

By legmlai: Dataset Link.
Filtered the dataset.

CroissantLLM

By Manuel Faysse: Dataset Link.
Extracted French samples only.

Citation

@misc{luth2025kurakurai,
  title        = {Luth: Efficient French Specialization for Small Language Models and Cross-Lingual Transfer},
  author       = {Lasbordes, Maxence and Gad, Sinoué},
  year         = {2025},
  howpublished = {\url{https://arxiv.org/abs/2510.05846}},
  note         = {arXiv:2510.05846}
}