sparselearning.tests

sparselearning.tests.test_data

Try testing salient dataset features:

1. Is it downloaded?

2. Does the loader work

3. Does the loader have data in your desired format?

sparselearning.tests.test_data.test_get_loaders(dataset)

Test dataloader

Parameters

dataset (str) – Dataset to use

sparselearning.tests.test_data.test_registry(dataset)

Test get_dataset functions

Parameters

dataset (str) – Dataset to use

sparselearning.tests.test_data.test_splitter()

Test data splitting using DatasetSplitter

sparselearning.tests.test_mask_loading_saving

sparselearning.tests.test_mask_loading_saving.test_save_load()

1. Initialise

2. Save

3. Load

   Assert if equal

4. Perform optim step

sparselearning.tests.test_struct_sparse

sparselearning.tests.test_struct_sparse.is_channel_sparse(mask: sparselearning.core.Masking) → bool

Checks if the conv mask is channel-wise sparse.

Parameters

mask (Masking) – Masking instance

Returns

True if channel-wise sparse

Return type

bool

sparselearning.tests.test_struct_sparse.test_struct_init(init_scheme: str)

Test structured sparsity for various init schemes

Parameters

init_scheme (str) – Random/ER/ERK

sparselearning.tests.test_struct_sparse.test_struct_prune_growth(prune_mode, growth_mode)

Test structured sparsity across prune, growth modes. See sparselearning.funcs.prune,growth

Parameters

• prune_mode (str) – prune mode

• growth_mode (str) – growth mode