I’m seeing a lot of people plotting the training and validation error value vs epoch but both https://www.dataquest.io/blog/learning-curves-machine-learning/ and Andrew Ng talk about plotting the error vs the size of the training set.
scikit-learn provides a nice function called
learning_curve() which plots the error curves vs training set size. How can we do the same in pytorch? Is there a convenience class/function that does this?