pytorch_pfn_extras.dataloaders.utils.CollateAsDict#

class pytorch_pfn_extras.dataloaders.utils.CollateAsDict(names, collate_fn=<function default_collate>)#

Bases: object

Creates a collate function that converts inputs to a dict of tensors.

An instantiated callable object can be feeded to torch.utils.data.DataLoader as a collate_fn option.

Parameters:

  • names (list of str) – Names of keys of output dict.

  • collate_fn (function) – A function preprocesses inputs.

Methods

__init__(names[, collate_fn])

__call__(*args, **kwargs)#

Converts inputs the dataset generated to a dictionary of tensors.

Returns (dict of Tensor):

A dictionary with keys that specified as names option, and values as input tensors.

Parameters:

  • args (Any) –

  • kwargs (Any) –

Return type:

Dict[str, Any]

__init__(names, collate_fn=<function default_collate>)#
Parameters:

  • names (Sequence[str]) –

  • collate_fn (Callable[[...], Any]) –

Return type:

None