MCPcopy Index your code
hub / github.com/MotrixLab/MotionDiffuse / build_dataloader

Function build_dataloader

text2motion/datasets/dataloader.py:56–121  ·  view source on GitHub ↗

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)

Source from the content-addressed store, hash-verified

54
55
56def build_dataloader(dataset: Dataset,
57 samples_per_gpu: int,
58 workers_per_gpu: int,
59 num_gpus: Optional[int] = 1,
60 dist: Optional[bool] = True,
61 shuffle: Optional[bool] = True,
62 round_up: Optional[bool] = True,
63 seed: Optional[Union[int, None]] = None,
64 persistent_workers: Optional[bool] = True,
65 **kwargs):
66 """Build PyTorch DataLoader.
67
68 In distributed training, each GPU/process has a dataloader.
69 In non-distributed training, there is only one dataloader for all GPUs.
70
71 Args:
72 dataset (:obj:`Dataset`): A PyTorch dataset.
73 samples_per_gpu (int): Number of training samples on each GPU, i.e.,
74 batch size of each GPU.
75 workers_per_gpu (int): How many subprocesses to use for data loading
76 for each GPU.
77 num_gpus (int, optional): Number of GPUs. Only used in non-distributed
78 training.
79 dist (bool, optional): Distributed training/test or not. Default: True.
80 shuffle (bool, optional): Whether to shuffle the data at every epoch.
81 Default: True.
82 round_up (bool, optional): Whether to round up the length of dataset by
83 adding extra samples to make it evenly divisible. Default: True.
84 persistent_workers (bool): If True, the data loader will not shutdown
85 the worker processes after a dataset has been consumed once.
86 This allows to maintain the workers Dataset instances alive.
87 The argument also has effect in PyTorch>=1.7.0.
88 Default: True
89 kwargs: any keyword argument to be used to initialize DataLoader
90
91 Returns:
92 DataLoader: A PyTorch dataloader.
93 """
94 rank, world_size = get_dist_info()
95 if dist:
96 sampler = DistributedSampler(
97 dataset, world_size, rank, shuffle=shuffle, round_up=round_up)
98 shuffle = False
99 batch_size = samples_per_gpu
100 num_workers = workers_per_gpu
101 else:
102 sampler = None
103 batch_size = num_gpus * samples_per_gpu
104 num_workers = num_gpus * workers_per_gpu
105
106 init_fn = partial(
107 worker_init_fn, num_workers=num_workers, rank=rank,
108 seed=seed) if seed is not None else None
109
110 data_loader = DataLoader(
111 dataset,
112 batch_size=batch_size,
113 sampler=sampler,

Callers 1

trainMethod · 0.90

Calls 1

DistributedSamplerClass · 0.85

Tested by

no test coverage detected