loader

class transfer_nlp.loaders.loaders.DatasetSplits(train_set: torch.utils.data.Dataset, train_batch_size: int, val_set: torch.utils.data.Dataset, val_batch_size: int, test_set: torch.utils.data.Dataset = None, test_batch_size: int = None)

This file contains an abstract CustomDataset class, on which we can build up custom dataset classes.

In your project, you will have to customize your data loader class. To let the framework interact with your class, you need to use the decorator @register_dataset, just as in the examples in this file

class transfer_nlp.loaders.loaders.DataFrameDataset(df)

class transfer_nlp.loaders.loaders.DatasetHyperParams(vectorizer: transfer_nlp.loaders.vectorizers.Vectorizer)