Fix issue #1 by updating Example to latest Keras API

[jasonbunk]

Fix issue #1 by updating the sentiment LSTM in the Example folder to use the latest Keras API. Uses the Theano backend to fix self._predict_stochastic.

Can be tested by inserting print(str(MC_model_output)) after line 87 of callbacks.py and comparing the difference between K.learning_phase(): np.uint8(1) and K.learning_phase(): np.uint8(0) to see the difference between deterministic test-phase dropout and MC dropout.

[yaringal]

Hi @jasonbunk, thanks for the pull request. I can merge it if you want, but I do not intend to maintain this repo in the long term. It's mostly for demonstration purposes. The main code in the repo has been implemented into Keras, a TensorFlow example, and a Torch package.