cupy.load(file, mmap_mode=None, allow_pickle=None)[source]#

Loads arrays or pickled objects from .npy, .npz or pickled file.

This function just calls numpy.load and then sends the arrays to the current device. NPZ file is converted to NpzFile object, which defers the transfer to the time of accessing the items.

  • file (file-like object or string) – The file to read.

  • mmap_mode (None, 'r+', 'r', 'w+', 'c') – If not None, memory-map the file to construct an intermediate numpy.ndarray object and transfer it to the current device.

  • allow_pickle (bool) – Allow loading pickled object arrays stored in npy files. Reasons for disallowing pickles include security, as loading pickled data can execute arbitrary code. If pickles are disallowed, loading object arrays will fail. Please be aware that CuPy does not support arrays with dtype of object. The default is False. This option is available only for NumPy 1.10 or later. In NumPy 1.9, this option cannot be specified (loading pickled objects is always allowed).


CuPy array or NpzFile object depending on the type of the file. NpzFile object is a dictionary-like object with the context manager protocol (which enables us to use with statement on it).

See also