C3_W2_Assignment 2_Content based Filtering & TensorFlow

  1. In TensorFlow default batch size is 32. So for a dataset with 40707 samples there should be 40707/32 ~ 1272 batches.

But in this lab it is showing 40707 batches. Kindly help me understand.

Thank you

Hi @Debatreyo_Roy,
Here, model.fit()'s progress bar is printing out not the number of batches, but the number of samples (i.e. 40707/40707 means 40707 out of 40707 samples have been completed). One hint that it’s using number of samples is the other line it prints out “Train on 40707 samples”

I suspect it’s just the version of keras we’re using in this assignment is older than what you’ve seen in other assignments or work you’ve done on your own, and the keras folks have now changed their print statements to use number of batches. You can try checking the version using:

print(tf.__version__)