Using a CNN to Tell Tiffany and Ivanka Trump Apart

(Phil Weslow) #1

Following @jeremy suggestion in the lectures, I have written my first blog post related to the course. As this is my first post on the subject of deep learning, I would definitely welcome feedback!

(Phil Weslow) #2

I am trying to figure out why the log scale learning rate graph in my Jupyter Notebook is not displaying any info.

Clearly, there are far fewer iterations than the original Dogs vs. Cats problem (Lesson 1), 6 vs 304. However, I am not exactly sure what is determining the number of iterations. My first assumption would be the respective sizes of the datasets. I am also guessing that batch size might play into it, but I’m just not sure.


It was talked about on the forum
experiment with changing batch size


(Phil Weslow) #4

Yep, setting the batch size equal to eight worked like a charm! I never noticed before that the batch size is not explicitly set in the code for Lesson One. Thanks.