Serialization semantics ¶

This note describes how you can save and load PyTorch tensors and module states in Python, and how to serialize Python modules so they can be loaded in C++.

Table of Contents

Serialization semantics

Saving and loading tensors ¶

torch.save() and torch.load() let you easily save and load tensors:

>>> t = torch.tensor([1., 2.])
>>> torch.save(t, 'tensor.pt')
>>> torch.load('tensor.pt')
tensor([1., 2.])

By convention, PyTorch files are typically written with a ‘.pt’ or ‘.pth’ extension.

torch.save() and torch.load() use Python’s pickle by default, so you can also save multiple tensors as part of Python objects like tuples, lists, and dicts:

>>> d = {'a': torch.tensor([1., 2.]), 'b': torch.tensor([3., 4.])}
>>> torch.save(d, 'tensor_dict.pt')
>>> torch.load('tensor_dict.pt')
{'a': tensor([1., 2.]), 'b': tensor([3., 4.])}

Custom data structures that include PyTorch tensors can also be saved if the data structure is pickle-able.

Saving and loading tensors preserves views ¶

Saving tensors preserves their view relationships:

>>> numbers = torch.arange(1, 10)
>>> evens = numbers[1::2]
>>> torch.save([numbers, evens], 'tensors.pt')
>>> loaded_numbers, loaded_evens = torch.load('tensors.pt')
>>> loaded_evens *= 2
>>> loaded_numbers
tensor([ 1,  4,  3,  8,  5, 12,  7, 16,  9])

Behind the scenes, these tensors share the same “storage.” See Tensor Views for more on views and storage.

When PyTorch saves tensors it saves their storage objects and tensor metadata separately. This is an implementation detail that may change in the future, but it typically saves space and lets PyTorch easily reconstruct the view relationships between the loaded tensors. In the above snippet, for example, only a single storage is written to ‘tensors.pt’.

In some cases, however, saving the current storage objects may be unnecessary and create prohibitively large files. In the following snippet a storage much larger than the saved tensor is written to a file:

>>> large = torch.arange(1, 1000)
>>> small = large[0:5]
>>> torch.save(small, 'small.pt')
>>> loaded_small = torch.load('small.pt')
>>> loaded_small.storage().size()
999

Instead of saving only the five values in the small tensor to ‘small.pt,’ the 999 values in the storage it shares with large were saved and loaded.

When saving tensors with fewer elements than their storage objects, the size of the saved file can be reduced by first cloning the tensors. Cloning a tensor produces a new tensor with a new storage object containing only the values in the tensor:

>>> large = torch.arange(1, 1000)
>>> small = large[0:5]
>>> torch.save(small.clone(), 'small.pt')  # saves a clone of small
>>> loaded_small = torch.load('small.pt')
>>> loaded_small.storage().size()
5

Since the cloned tensors are independent of each other, however, they have none of the view relationships the original tensors did. If both file size and view relationships are important when saving tensors smaller than their storage objects, then care must be taken to construct new tensors that minimize the size of their storage objects but still have the desired view relationships before saving.

Saving and loading torch.nn.Modules ¶

In PyTorch, a module’s state is frequently serialized using a ‘state dict.’ A module’s state dict contains all of its parameters and persistent buffers:

>>> bn = torch.nn.BatchNorm1d(3, track_running_stats=True)
>>> list(bn.named_parameters())
[('weight', Parameter containing: tensor([1., 1., 1.], requires_grad=True)),
 ('bias', Parameter containing: tensor([0., 0., 0.], requires_grad=True))]

>>> list(bn.named_buffers())
[('running_mean', tensor([0., 0., 0.])),
 ('running_var', tensor([1., 1., 1.])),
 ('num_batches_tracked', tensor(0))]

>>> bn.state_dict()
OrderedDict([('weight', tensor([1., 1., 1.])),
             ('bias', tensor([0., 0., 0.])),
             ('running_mean', tensor([0., 0., 0.])),
             ('running_var', tensor([1., 1., 1.])),
             ('num_batches_tracked', tensor(0))])

Instead of saving a module directly, for compatibility reasons it is recommended to instead save only its state dict. Python modules even have a function, load_state_dict(), to restore their states from a state dict:

>>> torch.save(bn.state_dict(), 'bn.pt')
>>> bn_state_dict = torch.load('bn.pt')
>>> new_bn = torch.nn.BatchNorm1d(3, track_running_stats=True)
>>> new_bn.load_state_dict(bn_state_dict)
<All keys matched successfully>

Note that the state dict is first loaded from its file with torch.load() and the state then restored with load_state_dict().

Even custom modules and modules containing other modules have state dicts and can use this pattern:

# A module with two linear layers
>>> class MyModule(torch.nn.Module):
      def __init__(self):
        super().__init__()
        self.l0 = torch.nn.Linear(4, 2)
        self.l1 = torch.nn.Linear(2, 1)

      def forward(self, input):
        out0 = self.l0(input)
        out0_relu = torch.nn.functional.relu(out0)
        return self.l1(out0_relu)

>>> m = MyModule()
>>> m.state_dict()
OrderedDict([('l0.weight', tensor([[ 0.1400, 0.4563, -0.0271, -0.4406],
                                   [-0.3289, 0.2827, 0.4588, 0.2031]])),
             ('l0.bias', tensor([ 0.0300, -0.1316])),
             ('l1.weight', tensor([[0.6533, 0.3413]])),
             ('l1.bias', tensor([-0.1112]))])

>>> torch.save(m.state_dict(), 'mymodule.pt')
>>> m_state_dict = torch.load('mymodule.pt')
>>> new_m = MyModule()
>>> new_m.load_state_dict(m_state_dict)
<All keys matched successfully>

Serialized file format for `torch.save`¶

Since PyTorch 1.6.0, torch.save defaults to returning an uncompressed ZIP64 archive unless the user sets _use_new_zipfile_serialization=False.

In this archive, the files are ordered as such

checkpoint.pth
├── data.pkl
├── byteorder  # added in PyTorch 2.1.0
├── data/
│   ├── 0
│   ├── 1
│   ├── 2
│   └── …
└── version

The entries are as follows:

data.pkl is the result of pickling the object passed to torch.save excluding torch.Storage objects that it contains
byteorder contains a string with the sys.byteorder when saving (“little” or “big”)
data/ contains all the storages in the object, where each storage is a separate file
version contains a version number at save time that can be used at load time

When saving, PyTorch will ensure that the local file header of each file is padded to an offset that is a multiple of 64 bytes, ensuring that the offset of each file is 64-byte aligned.

Note

Tensors on certain devices such as XLA are serialized as pickled numpy arrays. As such, their storages are not serialized. In these cases data/ might not exist in the checkpoint.

Serializing torch.nn.Modules and loading them in C++¶

ScriptModules can be serialized as a TorchScript program and loaded using torch.jit.load(). This serialization encodes all the modules’ methods, submodules, parameters, and attributes, and it allows the serialized program to be loaded in C++ (i.e. without Python).

The distinction between torch.jit.save() and torch.save() may not be immediately clear. torch.save() saves Python objects with pickle. This is especially useful for prototyping, researching, and training. torch.jit.save(), on the other hand, serializes ScriptModules to a format that can be loaded in Python or C++. This is useful when saving and loading C++ modules or for running modules trained in Python with C++, a common practice when deploying PyTorch models.

To script, serialize and load a module in Python:

>>> scripted_module = torch.jit.script(MyModule())
>>> torch.jit.save(scripted_module, 'mymodule.pt')
>>> torch.jit.load('mymodule.pt')
RecursiveScriptModule( original_name=MyModule
                      (l0): RecursiveScriptModule(original_name=Linear)
                      (l1): RecursiveScriptModule(original_name=Linear) )

Traced modules can also be saved with torch.jit.save(), with the caveat that only the traced code path is serialized. The following example demonstrates this:

# A module with control flow
>>> class ControlFlowModule(torch.nn.Module):
      def __init__(self):
        super().__init__()
        self.l0 = torch.nn.Linear(4, 2)
        self.l1 = torch.nn.Linear(2, 1)

      def forward(self, input):
        if input.dim() > 1:
            return torch.tensor(0)

        out0 = self.l0(input)
        out0_relu = torch.nn.functional.relu(out0)
        return self.l1(out0_relu)

>>> traced_module = torch.jit.trace(ControlFlowModule(), torch.randn(4))
>>> torch.jit.save(traced_module, 'controlflowmodule_traced.pt')
>>> loaded = torch.jit.load('controlflowmodule_traced.pt')
>>> loaded(torch.randn(2, 4)))
tensor([[-0.1571], [-0.3793]], grad_fn=<AddBackward0>)

>>> scripted_module = torch.jit.script(ControlFlowModule(), torch.randn(4))
>>> torch.jit.save(scripted_module, 'controlflowmodule_scripted.pt')
>>> loaded = torch.jit.load('controlflowmodule_scripted.pt')
>> loaded(torch.randn(2, 4))
tensor(0)

The above module has an if statement that is not triggered by the traced inputs, and so is not part of the traced module and not serialized with it. The scripted module, however, contains the if statement and is serialized with it. See the TorchScript documentation for more on scripting and tracing.

Finally, to load the module in C++:

>>> torch::jit::script::Module module;
>>> module = torch::jit::load('controlflowmodule_scripted.pt');

See the PyTorch C++ API documentation for details about how to use PyTorch modules in C++.

Saving and loading ScriptModules across PyTorch versions ¶

The PyTorch Team recommends saving and loading modules with the same version of PyTorch. Older versions of PyTorch may not support newer modules, and newer versions may have removed or modified older behavior. These changes are explicitly described in PyTorch’s release notes, and modules relying on functionality that has changed may need to be updated to continue working properly. In limited cases, detailed below, PyTorch will preserve the historic behavior of serialized ScriptModules so they do not require an update.

torch.div performing integer division ¶

In PyTorch 1.5 and earlier torch.div() would perform floor division when given two integer inputs:

# PyTorch 1.5 (and earlier)
>>> a = torch.tensor(5)
>>> b = torch.tensor(3)
>>> a / b
tensor(1)

In PyTorch 1.7, however, torch.div() will always perform a true division of its inputs, just like division in Python 3:

# PyTorch 1.7
>>> a = torch.tensor(5)
>>> b = torch.tensor(3)
>>> a / b
tensor(1.6667)

The behavior of torch.div() is preserved in serialized ScriptModules. That is, ScriptModules serialized with versions of PyTorch before 1.6 will continue to see torch.div() perform floor division when given two integer inputs even when loaded with newer versions of PyTorch. ScriptModules using torch.div() and serialized on PyTorch 1.6 and later cannot be loaded in earlier versions of PyTorch, however, since those earlier versions do not understand the new behavior.

torch.full always inferring a float dtype ¶

In PyTorch 1.5 and earlier torch.full() always returned a float tensor, regardless of the fill value it’s given:

# PyTorch 1.5 and earlier
>>> torch.full((3,), 1)  # Note the integer fill value...
tensor([1., 1., 1.])     # ...but float tensor!

In PyTorch 1.7, however, torch.full() will infer the returned tensor’s dtype from the fill value:

# PyTorch 1.7
>>> torch.full((3,), 1)
tensor([1, 1, 1])

>>> torch.full((3,), True)
tensor([True, True, True])

>>> torch.full((3,), 1.)
tensor([1., 1., 1.])

>>> torch.full((3,), 1 + 1j)
tensor([1.+1.j, 1.+1.j, 1.+1.j])

The behavior of torch.full() is preserved in serialized ScriptModules. That is, ScriptModules serialized with versions of PyTorch before 1.6 will continue to see torch.full return float tensors by default, even when given bool or integer fill values. ScriptModules using torch.full() and serialized on PyTorch 1.6 and later cannot be loaded in earlier versions of PyTorch, however, since those earlier versions do not understand the new behavior.

Utility functions ¶

The following utility functions are related to serialization:

torch.serialization.register_package(priority, tagger, deserializer)[source]¶

Registers callables for tagging and deserializing storage objects with an associated priority. Tagging associates a device with a storage object at save time while deserializing moves a storage object to an appropriate device at load time. tagger and deserializer are run in the order given by their priority until a tagger/deserializer returns a value that is not None.

To override the deserialization behavior for a device in the global registry, one can register a tagger with a higher priority than the existing tagger.

This function can also be used to register a tagger and deserializer for new devices.

Parameters

priority (int) – Indicates the priority associated with the tagger and deserializer, where a lower value indicates higher priority.
tagger (Callable[[Union[Storage, TypedStorage, UntypedStorage]], Optional[str]]) – Callable that takes in a storage object and returns its tagged device as a string or None.
deserializer (Callable[[Union[Storage, TypedStorage, UntypedStorage], str], Optional[Union[Storage, TypedStorage, UntypedStorage]]]) – Callable that takes in storage object and a device string and returns a storage object on the appropriate device or None.

Returns

None

Example

>>> def ipu_tag(obj):
>>>     if obj.device.type == 'ipu':
>>>         return 'ipu'
>>> def ipu_deserialize(obj, location):
>>>     if location.startswith('ipu'):
>>>         ipu = getattr(torch, "ipu", None)
>>>         assert ipu is not None, "IPU device module is not loaded"
>>>         assert torch.ipu.is_available(), "ipu is not available"
>>>         return obj.ipu(location)
>>> torch.serialization.register_package(11, ipu_tag, ipu_deserialize)

torch.serialization.get_default_load_endianness()[source]¶

Get fallback byte order for loading files

If byteorder mark is not present in saved checkpoint, this byte order is used as fallback. By default, it’s “native” byte order.

Returns: Optional[LoadEndianness]
Return type: default_load_endian

torch.serialization.set_default_load_endianness(endianness)[source]¶

Set fallback byte order for loading files