Fastai's ImageDataBunch to PyTorch's DataLoader

I have trained a model using the default steps of fastai - ImageList.from_folder and cnn_learner. I have stored the model using ‘torch.save’ to store the entire model architecture.

Now I need to use the same model trained in fastai for a native PyTorch inference environment. To be more specific, I will load the model using
model = torch.load(saved_model)
model.load_state_dict(torch.load(saved_model))
model.train(False)

Now, I have to inference the PyTorch DataLoaders using this model. But I am getting an error when I try to pass the PyTorch DataLoader into this model for inference. Am I following the right approach? Is it possible to pass and inference the native pytorch dataloaders onto the fastai model for inference?

Hi, have you been able to figure this out?
I am looking into your exact same approach and would love some heads up before diving in :grin: