TF1 COURSE 1 WEEK 2_BEYOND_HELLO_WORLD: Loss represent average loss?

Since the model is doing multi-label classification , does that mean the loss per each epoch as shown in the image above is the average loss obtained in classifying all the images ?

also what do the 1875 and 313 represent ?

Yeah this is the average loss for the epoch for all images. I am wondering gif you are using early stopping so that a certain loss is achieved than the model training stops, and that is probably why you have 313 instead of 1875 as in the previous epochs.

Okay, thanks . but I am running 5 epochs with no early stopping.

I see, I am puzzled too. Maybe its doing validation or somekind of testing for some subset of images…

Yeah , seen it. It is testing after training . Thanks !


  1. You need to take 10 epochs. It’s given in the hint :

You must set your model to train for 10 epochs and the callback should fire before the 9th epoch for you to pass this assignment.

  1. and 1875 is the number of steps per epoch.

With regards,
Nilosree Sengupta

1 Like