metallic.data.benchmarks

metallic.data.benchmarks.get_benchmarks(name: str, root: str, n_way: int = 5, k_shot: int = 1, **kwargs) → Tuple[metallic.data.datasets.base.MetaDataset]

Return a most commonly used benchmark on the given dataset.

Parameters

• name (str) – Name of the dataset

• root (str) – Root directory of the dataset

• n_way (int) – Number of the classes per tasks (same in ‘trian’ / ‘val’ / ‘test’ set)

• k_shot (int) – Number of samples per class (same in ‘support’ / ‘query’ set)

• **kwargs – Other arguments if needed

metallic.data.benchmarks.list_benchmarks()

Return a list of available benchmarks.