lab4d.engine package

lab4d.engine.model

lab4d.engine.train_utils

class lab4d.engine.train_utils.DataParallelPassthrough(module, device_ids=None, output_device=None, dim=0, broadcast_buffers=True, process_group=None, bucket_cap_mb=25, find_unused_parameters=False, check_reduction=False, gradient_as_bucket_view=False, static_graph=False, delay_all_reduce_named_params=None, param_to_hook_all_reduce=None, mixed_precision: _MixedPrecision | None = None, device_mesh=None)

Bases: DistributedDataParallel

For multi-GPU access, forward attributes to the inner module.

lab4d.engine.train_utils.get_local_rank()

lab4d.engine.trainer