An Empirical Study into Clustering of Unseen Datasets with Self-Supervised Encoders