Publishing models#

Pytorch Hub supports publishing pre-trained models(model definitions and pre-trained weights)to a GitHub repository by adding a simplehubconf.py file;

hubconf.py can have multiple entrypoints. Each entrypoint is defined as a python function(example: a pre-trained model you want to publish).

defentrypoint_name(*args,**kwargs):# args & kwargs are optional, for models which take positional/keyword arguments....

dependencies=['torch']fromtorchvision.models.resnetimportresnet18as_resnet18# resnet18 is the name of entrypointdefresnet18(pretrained=False,**kwargs):""" # This docstring shows up in hub.help()      Resnet18 model      pretrained (bool): kwargs, load pretrained weights into the model      """# Call the model, load pretrained weightsmodel=_resnet18(pretrained=pretrained,**kwargs)returnmodel

ifpretrained:# For checkpoint saved in local GitHub repo, e.g. <RELATIVE_PATH_TO_CHECKPOINT>=weights/save.pthdirname=os.path.dirname(__file__)checkpoint=os.path.join(dirname,<RELATIVE_PATH_TO_CHECKPOINT>)state_dict=torch.load(checkpoint)model.load_state_dict(state_dict)# For checkpoint saved elsewherecheckpoint='https://download.pytorch.org/models/resnet18-5c106cde.pth'model.load_state_dict(torch.hub.load_state_dict_from_url(checkpoint,progress=False))

Loading models from Hub#

Pytorch Hub provides convenient APIs to explore all available models in hubthroughtorch.hub.list(), show docstring and examples throughtorch.hub.help() and load the pre-trained models usingtorch.hub.load().

torch.hub.list(github,force_reload=False,skip_validation=False,trust_repo=None,verbose=True)[source]#

List all callable entrypoints available in the repo specified bygithub.

Parameters

• github (str) – a string with format “repo_owner/repo_name[:ref]” with an optionalref (tag or branch). Ifref is not specified, the default branch is assumed to bemain ifit exists, and otherwisemaster.Example: ‘pytorch/vision:0.10’

• force_reload (bool,optional) – whether to discard the existing cache and force a fresh download.Default isFalse.

• skip_validation (bool,optional) – ifFalse, torchhub will check that the branch or commitspecified by thegithub argument properly belongs to the repo owner. This will makerequests to the GitHub API; you can specify a non-default GitHub token by setting theGITHUB_TOKEN environment variable. Default isFalse.

• trust_repo (bool,str orNone) –

  "check",True,False orNone.This parameter was introduced in v1.12 and helps ensuring that usersonly run code from repos that they trust.

  • IfFalse, a prompt will ask the user whether the repo shouldbe trusted.

  • IfTrue, the repo will be added to the trusted list and loadedwithout requiring explicit confirmation.

  • If"check", the repo will be checked against the list oftrusted repos in the cache. If it is not present in that list, thebehaviour will fall back onto thetrust_repo=False option.

  • IfNone: this will raise a warning, inviting the user to settrust_repo to eitherFalse,True or"check". Thisis only present for backward compatibility and will be removed inv2.0.

  Default isNone and will eventually change to"check" in v2.0.

• verbose (bool,optional) – IfFalse, mute messages about hittinglocal caches. Note that the message about first download cannot bemuted. Default isTrue.

Returns

The available callables entrypoint

Return type

list

Example

>>>entrypoints=torch.hub.list("pytorch/vision",force_reload=True)

torch.hub.help(github,model,force_reload=False,skip_validation=False,trust_repo=None)[source]#

Show the docstring of entrypointmodel.

Parameters

• github (str) – a string with format <repo_owner/repo_name[:ref]> with an optionalref (a tag or a branch). Ifref is not specified, the default branch is assumedto bemain if it exists, and otherwisemaster.Example: ‘pytorch/vision:0.10’

• model (str) – a string of entrypoint name defined in repo’shubconf.py

• force_reload (bool,optional) – whether to discard the existing cache and force a fresh download.Default isFalse.

• skip_validation (bool,optional) – ifFalse, torchhub will check that the refspecified by thegithub argument properly belongs to the repo owner. This will makerequests to the GitHub API; you can specify a non-default GitHub token by setting theGITHUB_TOKEN environment variable. Default isFalse.

• trust_repo (bool,str orNone) –

  "check",True,False orNone.This parameter was introduced in v1.12 and helps ensuring that usersonly run code from repos that they trust.

  • IfFalse, a prompt will ask the user whether the repo shouldbe trusted.

  • IfTrue, the repo will be added to the trusted list and loadedwithout requiring explicit confirmation.

  • If"check", the repo will be checked against the list oftrusted repos in the cache. If it is not present in that list, thebehaviour will fall back onto thetrust_repo=False option.

  • IfNone: this will raise a warning, inviting the user to settrust_repo to eitherFalse,True or"check". Thisis only present for backward compatibility and will be removed inv2.0.

  Default isNone and will eventually change to"check" in v2.0.

Example

>>>print(torch.hub.help("pytorch/vision","resnet18",force_reload=True))

torch.hub.load(repo_or_dir,model,*args,source='github',trust_repo=None,force_reload=False,verbose=True,skip_validation=False,**kwargs)[source]#

Load a model from a github repo or a local directory.

Note: Loading a model is the typical use case, but this can also be used tofor loading other objects such as tokenizers, loss functions, etc.

Ifsource is ‘github’,repo_or_dir is expected to beof the formrepo_owner/repo_name[:ref] with an optionalref (a tag or a branch).

Ifsource is ‘local’,repo_or_dir is expected to be apath to a local directory.

Parameters

• repo_or_dir (str) – Ifsource is ‘github’,this should correspond to a github repo with formatrepo_owner/repo_name[:ref] withan optional ref (tag or branch), for example ‘pytorch/vision:0.10’. Ifref is not specified,the default branch is assumed to bemain if it exists, and otherwisemaster.Ifsource is ‘local’ then it should be a path to a local directory.

• model (str) – the name of a callable (entrypoint) defined in therepo/dir’shubconf.py.

• *args (optional) – the corresponding args for callablemodel.

• source (str,optional) – ‘github’ or ‘local’. Specifies howrepo_or_dir is to be interpreted. Default is ‘github’.

• trust_repo (bool,str orNone) –

  "check",True,False orNone.This parameter was introduced in v1.12 and helps ensuring that usersonly run code from repos that they trust.

  • IfFalse, a prompt will ask the user whether the repo shouldbe trusted.

  • IfTrue, the repo will be added to the trusted list and loadedwithout requiring explicit confirmation.

  • If"check", the repo will be checked against the list oftrusted repos in the cache. If it is not present in that list, thebehaviour will fall back onto thetrust_repo=False option.

  • IfNone: this will raise a warning, inviting the user to settrust_repo to eitherFalse,True or"check". Thisis only present for backward compatibility and will be removed inv2.0.

  Default isNone and will eventually change to"check" in v2.0.

• force_reload (bool,optional) – whether to force a fresh download ofthe github repo unconditionally. Does not have any effect ifsource='local'. Default isFalse.

• verbose (bool,optional) – IfFalse, mute messages about hittinglocal caches. Note that the message about first download cannot bemuted. Does not have any effect ifsource='local'.Default isTrue.

• skip_validation (bool,optional) – ifFalse, torchhub will check that the branch or commitspecified by thegithub argument properly belongs to the repo owner. This will makerequests to the GitHub API; you can specify a non-default GitHub token by setting theGITHUB_TOKEN environment variable. Default isFalse.

• **kwargs (optional) – the corresponding kwargs for callablemodel.

Returns

The output of themodel callable when called with the given*args and**kwargs.

Example

>>># from a github repo>>>repo="pytorch/vision">>>model=torch.hub.load(...repo,"resnet50",weights="ResNet50_Weights.IMAGENET1K_V1"...)>>># from a local directory>>>path="/some/local/path/pytorch/vision">>>model=torch.hub.load(path,"resnet50",weights="ResNet50_Weights.DEFAULT")

torch.hub.download_url_to_file(url,dst,hash_prefix=None,progress=True)[source]#

Download object at the given URL to a local path.

Parameters

• url (str) – URL of the object to download

• dst (str) – Full path where object will be saved, e.g./tmp/temporary_file

• hash_prefix (str,optional) – If not None, the SHA256 downloaded file should start withhash_prefix.Default: None

• progress (bool,optional) – whether or not to display a progress bar to stderrDefault: True

Example

>>>torch.hub.download_url_to_file(..."https://s3.amazonaws.com/pytorch/models/resnet18-5c106cde.pth",..."/tmp/temporary_file",...)

torch.hub.load_state_dict_from_url(url,model_dir=None,map_location=None,progress=True,check_hash=False,file_name=None,weights_only=False)[source]#

Loads the Torch serialized object at the given URL.

If downloaded file is a zip file, it will be automaticallydecompressed.

If the object is already present inmodel_dir, it’s deserialized andreturned.The default value ofmodel_dir is<hub_dir>/checkpoints wherehub_dir is the directory returned byget_dir().

Parameters

• url (str) – URL of the object to download

• model_dir (str,optional) – directory in which to save the object

• map_location (optional) – a function or a dict specifying how to remap storage locations (see torch.load)

• progress (bool,optional) – whether or not to display a progress bar to stderr.Default: True

• check_hash (bool,optional) – If True, the filename part of the URL should follow the naming conventionfilename-<sha256>.ext where<sha256> is the first eight or moredigits of the SHA256 hash of the contents of the file. The hash is used toensure unique names and to verify the contents of the file.Default: False

• file_name (str,optional) – name for the downloaded file. Filename fromurl will be used if not set.

• weights_only (bool,optional) – If True, only weights will be loaded and no complex pickled objects.Recommended for untrusted sources. Seeload() for more details.

Return type

dict[str,Any]

Example

>>>state_dict=torch.hub.load_state_dict_from_url(..."https://s3.amazonaws.com/pytorch/models/resnet18-5c106cde.pth"...)

• torchao
• torchrec
• torchft
• TorchCodec
• torchvision
• ExecuTorch
• PyTorch on XLA Devices