minor typos

Author: mhjensen

doc/pub/week16/html/week16-bs.html

@@ -412,7 +412,7 @@ <h2 id="maximizing-reward" class="anchor">Maximizing reward </h2>
 tackle otherwise intractable generative problems. As the generator improves with
  training, the discriminator's performance gets worse because it cannot easily
  tell the difference between real and fake. If the generator ends up succeeding
- perfectly, the the discriminator will do no better than random guessing i.e.
+ perfectly, then the discriminator will do no better than random guessing i.e.
  50\%.
 </p>
 

doc/pub/week16/html/week16-reveal.html

@@ -407,7 +407,7 @@ <h2 id="maximizing-reward">Maximizing reward </h2>
 tackle otherwise intractable generative problems. As the generator improves with
  training, the discriminator's performance gets worse because it cannot easily
  tell the difference between real and fake. If the generator ends up succeeding
- perfectly, the the discriminator will do no better than random guessing i.e.
+ perfectly, then the discriminator will do no better than random guessing i.e.
  50\%.
 </p>
 </section>

doc/pub/week16/html/week16-solarized.html

@@ -368,7 +368,7 @@ <h2 id="maximizing-reward">Maximizing reward </h2>
 tackle otherwise intractable generative problems. As the generator improves with
  training, the discriminator's performance gets worse because it cannot easily
  tell the difference between real and fake. If the generator ends up succeeding
- perfectly, the the discriminator will do no better than random guessing i.e.
+ perfectly, then the discriminator will do no better than random guessing i.e.
  50\%.
 </p>
 

doc/pub/week16/html/week16.html

@@ -445,7 +445,7 @@ <h2 id="maximizing-reward">Maximizing reward </h2>
 tackle otherwise intractable generative problems. As the generator improves with
  training, the discriminator's performance gets worse because it cannot easily
  tell the difference between real and fake. If the generator ends up succeeding
- perfectly, the the discriminator will do no better than random guessing i.e.
+ perfectly, then the discriminator will do no better than random guessing i.e.
  50\%.
 </p>