Hello
I am trying to save my trained model after i have finetuned the pretrained Inception using my dataset.I am able to successfully save the model.When i run the model on my test dataset after the training it works fine.However when i load the model in a separate file and evaluate it on the same test data I am getting a constant target label.
I have used
torch.save(model.state_dict(),filepath) and torch.save(model,filepath) to try to save the model
and
model.load_state_dict(torch.load(filepath).state_dict()) to load the model
I have also changed the model to eval() before testing so that is not the problem
Is there a different method to save for inception net model?