Comment on Every Model Learned by Gradient Descent Is Approximately a Kernel MachineparentComments−sdenton45yEpochs are more about the training data than the model... If you've got a big enough dataset, one epoch or less is fine!
Comments
Epochs are more about the training data than the model... If you've got a big enough dataset, one epoch or less is fine!