The newswire_topics.py module is now well documented. (following proper python docstring conventions)

Upping the maximum number of words from 1000 to 7000 improved the test accuracy well above 80%.

3 key conclusions:

• The hidden layer size of 256 is great! Tried 128, 512, 1024..
• ‘dropout’ probability seemed to be most optimal around the value of 0.5..
• softPLUS activation function aced! (better than logistic/sigmoid, ReLU, softsign, TanH, in this order..)