geodesic-research/sfm-midtraining-mix-ai-filtering-results

该数据集名为alignment_filtering_20251126-0344，由geodesic-research提供。数据集包含多个特征字段，包括id、word_filter（布尔类型）、word_filter_metadata（结构化数据，包含keywords和reason字段）、bert_filter（布尔类型）、bert_filter_metadata（结构化数据，包含highest_score、lowest_score和mean_score字段）以及combined_filter（布尔类型）。数据集仅包含一个训练集分割，样本数量为42,781,400个，文件大小为3,425,502,053字节。

The dataset is named alignment_filtering_20251126-0344 and provided by geodesic-research. It includes multiple feature fields such as id, word_filter (boolean), word_filter_metadata (structured data with keywords and reason fields), bert_filter (boolean), bert_filter_metadata (structured data with highest_score, lowest_score, and mean_score fields), and combined_filter (boolean). The dataset contains only a training split with 42,781,400 examples and a file size of 3,425,502,053 bytes.