Leonardo6/gpt-2-crosscoder-diff

该数据集包含100,000个训练样本，每个样本由一系列整数（int32类型）表示，存储在名为input_ids的特征中。数据集的总大小为410,000,000字节，下载大小为100,361,137字节。数据集的训练分割存储在路径为data/train-*的文件中。This dataset contains 100,000 training examples, each represented by a sequence of integers (int32 type) stored in a feature named input_ids. The total size of the dataset is 410,000,000 bytes, with a download size of 100,361,137 bytes. The training split of the dataset is stored in files with the path data/train-*.