rntc/biomed-fr-v3-enriched-softmin-leger

biomed-fr-v3-enriched-softmin-leger数据集是对rntc/biomed-fr-v3-enriched数据集进行质量上采样后的版本，使用了soft-min瓶颈采样方法。该数据集保留了原始数据集的2941107个样本，并针对教育性评分、内容丰富性、术语精确性和写作质量四个维度进行了质量筛选。数据集的预处理包括soft-min计算、权重计算和重采样，确保了样本的质量和分布。此数据集适用于文本生成任务，特别是在医学和生物医学领域。

The biomed-fr-v3-enriched-softmin-leger dataset is a quality-upsampled version of the rntc/biomed-fr-v3-enriched dataset, utilizing soft-min bottleneck sampling. It retains the original datasets 2941107 samples and has been qualityfiltered based on educational score, content richness, terminology precision, and writing quality. The datasets preprocessing involves soft-min calculation, weight computation, and resampling, ensuring sample quality and distribution. This dataset is suitable for text generation tasks, especially in the medical and biomedical fields.