Restore model from config and weights for model training

[IsraelAbebe]

• [x] Added cli argument
• [x] Reformatted code to create directories earlier so we would search for them
• [x] Reading config and model weights
• [x] Error handling if config or weights don't exist
• [x] Tested locally

closes #7

Please check and let me know if I should improve anything or if there are other tests I should run.

[nicolafan]

Firstly thank you for this PR!

I think adding a try-except block for reading the files is a good idea. We can change the utility function I made for building the model to make it raise an exception if there are problems in accessing the files, so that the except block can be triggered and show the error message.

Since this project uses the cookie-cutter data science template, error-info messages should be showed using a logger, that can be obtained with logger = logging.getLogger(__name__), after setting up the logger in the if __name__ == "__main__" block (an example can be seen in make_dataset.py.

[nicolafan]

Perfect, I made these little changes and we are ready to go!

[nicolafan]

Thank you for this PR @IsraelAbebe !