xlnstorch.convert_model.build_lns_sequential

xlnstorch.convert_model.build_lns_sequential#

xlnstorch.convert_model.build_lns_sequential(model, *, keep_containers=False)#

Create a new xlnstorch.nn.LNSSequential object where each layer is instantiated from xlnstorch.nn if available, otherwise from torch.nn.

Parameters:

  • model (torch.nn.Sequential | torch.nn.Module) – The original model to replicate.

  • keep_containers (bool, optional) – If True, the function will also replicate the explicit Sequential containers that appear in parse_sequential (they have no arguments). If False (default) those entries are ignored and a flat ordering of the leaf layers is returned.

Returns:

A new xlnstorch.nn.LNSSequential object containing the layers from the original model.

Return type:

xlnstorch.nn.LNSSequential

Raises:

RuntimeError – If a layer in the original model is not an instance of xlnstorch.nn.LNSModule but has parameters, indicating that it cannot be converted to LNS.

Contents