khaledmahmoud/spanish_simplification_20k
收藏资源简介:
该数据集提供了20K个西班牙语简化样本,涵盖多个主题,专为有认知挑战的人群设计。样本长度不一,从短到长均有涵盖。该数据集可用于补充更大的训练语料库,以微调小型语言模型(如Gemma 4B),将复杂的西班牙语简化为更简单的西班牙语,或将英语直接翻译为简化的西班牙语。数据集来源于MarcOrfilaCarreras西班牙新闻数据集,通过OpenAI GPT-5.4进行样本提取、清理、简化、翻译和评估。
This dataset provides 20K samples for Spanish simplification across a diverse range of topics, designed for people with cognitive challenges. The samples vary in length, from short to long. This dataset can supplement a larger training corpus for fine-tuning small language models, such as Gemma 4B, to simplify complex Spanish into simpler Spanish or translate English directly into simplified Spanish. The dataset was synthesized from the MarcOrfilaCarreras Spanish News dataset, using OpenAI GPT-5.4 for sample extraction, cleaning, simplification, translation, and evaluation.





