number of passes 3579, is it correct?
It looks like there is something wrong in the code, as it is training with stochastic gradient descent and not mini-batch, when I run this, or other training in tensorflow, I see that the number of passes is number of observations/batch_size, so when I run the code it is 36 passes. I had the same issue at some point, using colab, it runs ok, however, this time I had the same results running on my PC and in colab.
Please let me know if I am wrong.
3 answers ( 0 marked as helpful)
Hi, did you manage to find a solution? I am getting exactly the same results as you.
Hi Karlo,
it is actually ok what we get, when they run it in the course is wrong.
6 months later: same problem.