This dataset contains 150,000 high-quality English-Romanian parallel translation pairs derived from the Magpie-Reasoning dataset, specifically designed for training and evaluating machine translation models with a focus on technical, mathematical, and code-related content.

Source Datasets

This dataset is created by aligning:

English: Magpie-Align/Magpie-Reasoning-V1-150K Romanian: OpenLLM-Ro/ro_sft_magpie_reasoning… See the full description on the dataset page: https://huggingface.co/datasets/Yxanul/English-Romanian-Magpie-Reasoning.