Best way to save a trained model in PyTorch

Question

I was looking for alternative ways to save a trained model in PyTorch  So far  I have found two alternatives    torch save   to save a model and torch load   to load a model  model state dict   to save a trained model and model load state dict   to load the saved model    I have come across to this discussion where approach 2 is recommended over approach 1   My question is  why the second approach is preferred  Is it only because torch nn modules have those two function and we are encouraged to use them

User · Answer

A common PyTorch convention is to save models using either a  pt or  pth file extension   Save Load Entire Model Save   path    username directory lstmmodelgpu pth  torch save trainer  path     Load   Model class must be defined somewhere  model   torch load PATH  model eval

User · Answer

The pickle Python library implements binary protocols for serializing and de-serializing a Python object   When you import torch  or when you use PyTorch  it will import pickle for you and you don t need to call pickle dump   and pickle load   directly  which are the methods to save and to load the object    In fact  torch save   and torch load   will wrap pickle dump   and pickle load   for you    A state dict the other answer mentioned deserves just few more notes   What state dict do we have inside PyTorch  There are actually two state dicts   The PyTorch model is torch nn Module has model parameters   call to get learnable parameters  w and b   These learnable parameters  once randomly set  will update over time as we learn  Learnable parameters are the first state dict   The second state dict is the optimizer state dict  You recall that the optimizer is used to improve our learnable parameters  But the optimizer state dict is fixed  Nothing to learn in there   Because state dict objects are Python dictionaries  they can be easily saved  updated  altered  and restored  adding a great deal of modularity to PyTorch models and optimizers   Let s create a super simple model to explain this   import torch import torch optim as optim  model   torch nn Linear 5  2     Initialize optimizer optimizer   optim SGD model parameters    lr 0 001  momentum 0 9   print  Model s state dict    for param tensor in model state dict        print param tensor    t   model state dict   param tensor  size     print  Model weight        print model weight   print  Model bias        print model bias   print  ---   print  Optimizer s state dict    for var name in optimizer state dict        print var name    t   optimizer state dict   var name     This code will output the following   Model s state dict  weight   torch Size  2  5   bias     torch Size  2   Model weight  Parameter containing  tensor    0 1328   0 1360   0 1553  -0 1838  -0 0316             0 0479   0 1760   0 1712   0 2244   0 1408    requires grad True  Model bias  Parameter containing  tensor   0 4112  -0 0733   requires grad True  --- Optimizer s state dict  state       param groups        lr   0 001   momentum   0 9   dampening   0   weight decay   0   nesterov   False   params    140695321443856  140695321443928      Note this is a minimal model  You may try to add stack of sequential   model   torch nn Sequential            torch nn Linear D in  H             torch nn Conv2d A  B  C            torch nn Linear H  D out               Note that only layers with learnable parameters  convolutional layers  linear layers  etc   and registered buffers  batchnorm layers  have entries in the model s state dict    Non learnable things  belong to the optimizer object state dict  which contains information about the optimizer s state  as well as the hyperparameters used   The rest of the story is the same  in the inference phase  this is a phase when we use the model after training  for predicting  we do predict based on the parameters we learned  So for the inference  we just need to save the parameters model state dict     torch save model state dict    filepath    And to use later     model load state dict torch load filepath       model eval     Note  Don t forget the last line model eval   this is crucial after loading the model   Also don t try to save torch save model parameters    filepath   The model parameters   is just the generator object   On the other side  torch save model  filepath  saves the model object itself  but keep in mind the model doesn t have the optimizer s state dict  Check the other excellent answer by  Jadiel de Armas to save the optimizer s state dict

User · Answer

If you want to save the model and wants to resume the training later   Single GPU  Save   state              epoch   epoch           state dict   model state dict             optimizer   optimizer state dict      savepath  checkpoint t7  torch save state savepath    Load   checkpoint   torch load  checkpoint t7   model load state dict checkpoint  state dict    optimizer load state dict checkpoint  optimizer    epoch   checkpoint  epoch     Multiple GPU  Save  state              epoch   epoch           state dict   model module state dict             optimizer   optimizer state dict      savepath  checkpoint t7  torch save state savepath    Load   checkpoint   torch load  checkpoint t7   model load state dict checkpoint  state dict    optimizer load state dict checkpoint  optimizer    epoch   checkpoint  epoch     Don t call DataParallel before loading the model otherwise you will get an error  model   nn DataParallel model   ignore the line if you want to load on Single GPU

User · Answer

I ve found this page on their github repo  I ll just paste the content here   Recommended approach for saving a model There are two main approaches for serializing and restoring a model  The first  recommended  saves and loads only the model parameters  torch save the model state dict    PATH   Then later  the model   TheModelClass  args    kwargs  the model load state dict torch load PATH    The second saves and loads the entire model  torch save the model  PATH   Then later  the model   torch load PATH   However in this case  the serialized data is bound to the specific classes and the exact directory structure used  so it can break in various ways when used in other projects  or after some serious refactors

User · Answer

It depends on what you want to do   Case   1  Save the model to use it yourself for inference  You save the model  you restore it  and then you change the model to evaluation mode  This is done because you usually have BatchNorm and Dropout layers that by default are in train mode on construction   torch save model state dict    filepath    Later to restore  model load state dict torch load filepath   model eval     Case   2  Save model to resume training later  If you need to keep training the model that you are about to save  you need to save more than just the model  You also need to save the state of the optimizer  epochs  score  etc  You would do it like this   state          epoch   epoch       state dict   model state dict         optimizer   optimizer state dict              torch save state  filepath    To resume training you would do things like  state   torch load filepath   and then  to restore the state of each individual object  something like this    model load state dict state  state dict    optimizer load state dict state  optimizer      Since you are resuming training  DO NOT call model eval   once you restore the states when loading   Case   3  Model to be used by someone else with no access to your code  In Tensorflow you can create a  pb file that defines both the architecture and the weights of the model  This is very handy  specially when using Tensorflow serve   The equivalent way to do this in Pytorch would be   torch save model  filepath     Then later  model   torch load filepath    This way is still not bullet proof and since pytorch is still undergoing a lot of changes  I wouldn t recommend it

[python] Best way to save a trained model in PyTorch?

Model class must be defined somewhere

Examples related to python

Examples related to serialization

Examples related to deep-learning

Examples related to pytorch

Examples related to tensor