mapping_networks.trainers.BatchAdapter

class mapping_networks.trainers.BatchAdapter

Unpack dataloader batches into (model_inputs, targets).

Different datasets and dataloaders produce batches in different formats. A batch adapter normalizes them into the (inputs, targets) tuple that the trainer loop expects.

__init__()

Methods

__init__()
unpack(batch)	Return ((input_arg1, ...), targets) from a dataloader batch.