Will I have to repeat the training procedure if I close the Jupyter notebook after the model has finished training and I've saved it?

Question

The model has been trained, tested, and I have saved both the checkpoint and the model. After the training is complete, I load the stored model in the code. Do I need to retrain the model if I close the Jupyter Notebook?

Answer 1

The short answer is... no! Actually saving a model (under format .pt basically) will store all the trained parameters under a pickle dictionary somewhere in persistent memory (CD or SSD). Closing a notebook only clean allocated memory (RAM or video RAM) but not persistent memory so you will lose an hypothetical variable model but not the saved parameters you trained. Note that the architecture is not store (only the weights and the name of the layers)