The saved dataset is saved in multiple file "shards". By default, the dataset output is divided to shards in the spherical-robin trend but custom sharding is usually specified by using the shard_func purpose. One example is, It can save you the dataset to employing one shard as follows:Key routines of SCF could be divided into three regions: 1) I