nexa-collaboration/fineweb_filtered_scorepass_10B

该数据集包含多个字段，如文本内容(text)、唯一标识符(id)、语言(language)、语言评分(language_score)、词数(token_count)等。数据集分为训练集(train)，共有约4388万条样本，总大小约为51.67GB。提供了默认配置(default)，用于指定训练数据的文件路径。

The dataset contains multiple fields such as text content (text), unique identifier (id), language (language), language score (language_score), number of words (token_count), etc. The dataset is split into a training set (train) with a total of approximately 43.88 million samples and a total size of about 51.67GB. A default configuration (default) is provided to specify the file path for the training data.