tim-lawson/mlsae-pythia-70m-deduped-x16-k32-examples

该数据集包含多个特征，如id、latent、layer、token_id、token、act、token_ids、tokens和acts，每个特征都有其对应的数据类型。数据集包含一个名为train的分割，该分割的大小为333242819字节，包含1422487个样本。数据集的下载大小为70503550字节，总大小为333242819字节。配置信息中指定了默认配置，并指向了训练数据文件的路径。

The dataset contains multiple features such as id, latent, layer, token_id, token, act, token_ids, tokens, and acts, each with its corresponding data type. The dataset includes a split named train, which is 333242819 bytes in size and contains 1422487 examples. The download size of the dataset is 70503550 bytes, and the total size is 333242819 bytes. The configuration information specifies the default config and points to the path of the training data files.