Pretrain-where pretraining datasets

Dataset Card for ‘Pretrain-where’ pretraining datasets (SatMAE) These datasets were used to pretrain SatMAE to study the impact of spatial context of a pretraining dataset on downstream performance (Main paper: Dataset Details Dataset Description The collection consists of 7 pretraining datasets, each with data from a specific spatial context, namely Africa only (af), Asia only (as), North America only (na), South America only… See the full description on the dataset page:

Source: Hugging Face Hub (Amandeep10/pretrain-where-pretraining-datasets). Metadata imported from the dataset’s Hub tags.