save checkpoint

[gguang10]

when I use tutorial_train.py to train my own dataset, I notice that the checkpoint folder is empty, and hparams.yaml is {}, the code save a version_x every 300 steps, is the checkpoint folder empty because I haven't trained an epoch yet?