Step 1: Run train_embed_save.py to train an example model.
Step 2: Run saved_to_served.py to convert the *.hdf5 format model to servable model format.
Step 3: Run serve.sh (./serve.sh) to serve the servable model in a docker container as a web service.
Step 4: Run predict_from_serve.py to send a http request to the web service and recieve a json response of the prediction.
Step 5: Run tensorboard to visualize model