Function build_dataloader

detrsmpl/data/datasets/builder.py:46–115

Build PyTorch DataLoader. In distributed training, each GPU/process has a dataloader. In non-distributed training, there is only one dataloader for all GPUs. Args: dataset (:obj:`Dataset`): A PyTorch dataset. samples_per_gpu (int): Number of training samples on each GPU

(dataset: Dataset,
                     samples_per_gpu: int,
                     workers_per_gpu: int,
                     num_gpus: Optional[int] = 1,
                     dist: Optional[bool] = True,
                     shuffle: Optional[bool] = True,
                     round_up: Optional[bool] = True,
                     seed: Optional[Union[int, None]] = None,
                     persistent_workers: Optional[bool] = True,
                     **kwargs)

def build_dataloader(dataset: Dataset,
                     samples_per_gpu: int,
                     workers_per_gpu: int,
                     num_gpus: Optional[int] = 1,
                     dist: Optional[bool] = True,
                     shuffle: Optional[bool] = True,
                     round_up: Optional[bool] = True,
                     seed: Optional[Union[int, None]] = None,
                     persistent_workers: Optional[bool] = True,
                     **kwargs):
    """Build PyTorch DataLoader.

    In distributed training, each GPU/process has a dataloader.
    In non-distributed training, there is only one dataloader for all GPUs.

    Args:
        dataset (:obj:`Dataset`): A PyTorch dataset.
        samples_per_gpu (int): Number of training samples on each GPU, i.e.,
            batch size of each GPU.
        workers_per_gpu (int): How many subprocesses to use for data loading
            for each GPU.
        num_gpus (int, optional): Number of GPUs. Only used in non-distributed
            training.
        dist (bool, optional): Distributed training/test or not. Default: True.
        shuffle (bool, optional): Whether to shuffle the data at every epoch.
            Default: True.
        round_up (bool, optional): Whether to round up the length of dataset by
            adding extra samples to make it evenly divisible. Default: True.
        persistent_workers (bool): If True, the data loader will not shutdown
            the worker processes after a dataset has been consumed once.
            This allows to maintain the workers Dataset instances alive.
            The argument also has effect in PyTorch>=1.7.0.
            Default: True
        kwargs: any keyword argument to be used to initialize DataLoader

    Returns:
        DataLoader: A PyTorch dataloader.
    """
    rank, world_size = get_dist_info()
    if dist:
        sampler = DistributedSampler(dataset,
                                     world_size,
                                     rank,
                                     shuffle=shuffle,
                                     round_up=round_up)
        shuffle = False
        batch_size = samples_per_gpu
        num_workers = workers_per_gpu
    else:
        sampler = None
        batch_size = num_gpus * samples_per_gpu
        num_workers = num_gpus * workers_per_gpu

    init_fn = partial(
        worker_init_fn, num_workers=num_workers, rank=rank,
        seed=seed) if seed is not None else None

    data_loader = DataLoader(dataset,