In this activity, you will need to play with a seq2seq based chatbot, originally an assignment of Stanford CS20si.
You are not requested to make any improvement of the implementation. You can ignore the submission guideline of the original assignment.
But please make at least 5 conversation exchanges with your trained bot and submit a screen capture.
Note that the implementation does not work out of the box with tensorflow version 1.01. You can simply run it with version 0.12. Alternatively, an extra 1% bonus will be given if you make it work with version 1 and submit your code as well.
As usual, 3% bonus will be given to the earliest submitter. As the winner of Activity 1 submission, Siraj is excluded with this round though.
For the convenience of tracking your submission, I have again setup a page for uploading your homework.