Rate this Page

torch.save#

torch.save(obj,f,pickle_module=pickle,pickle_protocol=2,_use_new_zipfile_serialization=True)[source]#

Saves an object to a disk file.

See also:Saving and loading tensors

SeeLayout Control for more advanced tools to manipulate a checkpoint.

Parameters
  • obj (object) – saved object

  • f (Union[str,PathLike[str],IO[bytes]]) – a file-like object (has to implement write and flush) or a string oros.PathLike object containing a file name

  • pickle_module (Any) – module used for pickling metadata and objects

  • pickle_protocol (int) – can be specified to override the default protocol

Note

A common PyTorch convention is to save tensors using .pt file extension.

Note

PyTorch preserves storage sharing across serialization. SeeSaving and loading tensors preserves views for more details.

Note

The 1.6 release of PyTorch switchedtorch.save to use a newzipfile-based file format.torch.load still retains the ability toload files in the old format. If for any reason you wanttorch.saveto use the old format, pass the kwarg_use_new_zipfile_serialization=False.

Example

>>># Save to file>>>x=torch.tensor([0,1,2,3,4])>>>torch.save(x,"tensor.pt")>>># Save to io.BytesIO buffer>>>buffer=io.BytesIO()>>>torch.save(x,buffer)