C3W2 Manual feature Engineering Taxi Fare Ungraded lab

In the taxi fare assignment, we have a .repeat() function that repeats our train dataset infinitely. So in training, we have steps_per_epoch which limits the number of batches that can be trained in one epoch. Why do we need the train dataset to repeat infinitely or what is the intuition behind it, or am I getting something wrong here?

Hi @Athekunal-1
Welcome to Discourse!
Please take a look at the following link. I found there some interesting intuitions about the usage of shuffle and repeat methods.
Hope this can help

Thank you @fabioantonini , this explains my query

Thanks to you to bring this topic up. It’s really important to raise all doubts and questions ! :+1:

1 Like