1. mnist_deep.py read (where is the training, where is the test?)
2. Find the code for the training and save the model locally, called model.ckpt
Docker：
3. app.py
（1）restore model.ckpt, assuming model f（volume）docker run -v
（2）Given any graph x，f(x) = digit
（3）Deploy f(x) to Flask
(4)Write the current time, file name, and predicted number into Cassandra
Level-0：Implement the above code locally
My test1 always identify wrong, my test2 is not working. I need help to fix these two. And then finish step 3 and level-0
Im using Pycharm, Macbook pro. Please make sure I can run your code in mac environment.
