YnezT/gpt2-directions-non_fully_random_walk_with_mid-10M-non-fully-with-mid

该数据集包含用于训练和测试的样本，特征包括input_ids（int32序列）、attention_mask（int8序列）和labels（int64序列）。训练集包含10,000,000个样本，占用17.93GB空间；测试集包含200,000个样本，占用358.6MB空间。整个数据集的下载大小为2.247GB，总大小为18.29GB。This dataset includes samples for training and testing, with features including input_ids (int32 sequence), attention_mask (int8 sequence), and labels (int64 sequence). The training set contains 10,000,000 samples, occupying 17.93GB of space; the test set contains 200,000 samples, occupying 358.6MB of space. The entire dataset has a download size of 2.247GB and a total size of 18.29GB.