i trained and validated my seq2seq with pointer generator network for text2sql generation and every component (embedding, encoder, decoder , attentionEncoder...) is saved with .model extension.
I tried to load them with torch.load and every model is a dictionary of weights; but i couldn't figure out how to use them to deploy the seq2seq model(got many errors in inputs and outputs of every model component(although the training, testing, and validation were smooth)).the results of loading the .model components
