DGurgurov/

Nemotron-Multilingual-Reasoning

Multilingual reasoning dataset derived from the NVIDIA Llama-Nemotron-Post-Training-Dataset (science subset).

This dataset extends the English science reasoning traces to German, French, Spanish, and Italian through machine translation while preserving the original reasoning structure.

	Dataset Statistics




	

Language Code Samples Tokens

English en 708,920 1,321,433,620

German de 708,920 1,927,825,529

French fr… See the full description on the dataset page: https://huggingface.co/datasets/DGurgurov/Nemotron-Multilingual-Reasoning.