add some comments

Author: maxim5

hyperengine/examples/1_1_hello_word_with_mnist.py

@@ -6,9 +6,10 @@
 so that hyper-engine could find them. By convention, hyper-engine looks up the following:
 - input
 - label
-- loss
-- accuracy
+- mode     (optional)
 - minimize
+- loss     (optional)
+- accuracy (optional)
 These are the defaults and they are customizable.
 The whole integration is creation and start of the solver.
 """

hyperengine/examples/1_2_getting_started_with_tuning.py

@@ -1,6 +1,15 @@
 #! /usr/bin/env python
 # -*- coding: utf-8 -*-
 
+"""Optimizing a simple CNN model with one hyper-parameter (learning rate).
+
+The model itself isn't different from "Hello World Example", except for the hyper-parameter that
+is now taken from the input dictionary: `params['learning_rate']`.
+
+Also pay attention to `strategy_params` argument that dictates to save Bayesian optimization parameters
+automatically and continue from the last saved state on subsequent runs.
+"""
+
 __author__ = 'maxim'
 
 import tensorflow as tf