dataloaders

Bundle trn/val/test datasets together, and add helpful functionality for viewing data

torch.set_printoptions(precision=2, linewidth=140, sci_mode=False)
mpl.rcParams['image.cmap'] = 'gray'
logging.disable(logging.WARNING)

set_seed(42)

---

source

get_dataloaders

 get_dataloaders (train_dataset, valid_dataset, batch_size,
                  shuffle:Optional[bool]=None, sampler:Union[torch.utils.d
                  ata.sampler.Sampler,Iterable,NoneType]=None, batch_sampl
                  er:Union[torch.utils.data.sampler.Sampler[Sequence],Iter
                  able[Sequence],NoneType]=None, num_workers:int=0,
                  collate_fn:Optional[Callable[[List[~T]],Any]]=None,
                  pin_memory:bool=False, drop_last:bool=False,
                  timeout:float=0,
                  worker_init_fn:Optional[Callable[[int],NoneType]]=None,
                  multiprocessing_context=None, generator=None,
                  prefetch_factor:int=2, persistent_workers:bool=False,
                  pin_memory_device:str='')

---

source

collate_dataset_dict

 collate_dataset_dict (dataset)

---

source

DataLoaders

 DataLoaders (train, valid, n_inp=1)

Initialize self. See help(type(self)) for accurate signature.

---

source

sample_dataset_dict

 sample_dataset_dict (dataset, sample_sizes=(500, 500))

xmean,xstd = 0.28, 0.35

@inplace
def transformi(b): b['image'] = [(TF.to_tensor(o)-xmean)/xstd for o in b['image']]

_dataset = load_dataset('fashion_mnist').with_transform(transformi)
_dataset = sample_dataset_dict(_dataset)
dls = DataLoaders.from_dataset_dict(_dataset, 64, num_workers=4)

100%|██████████| 2/2 [00:00<00:00, 364.98it/s]

_dataset['train']

Dataset({
    features: ['image', 'label'],
    num_rows: 500
})

dls.train.dataset

Dataset({
    features: ['image', 'label'],
    num_rows: 500
})

dls.show_batch()