You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi,
During the training, how the number of epochs are getting decided ?
As I see, no part of training data has been reserved for evaluation during training of model. So, how can I be assured that model has converged during training ?
The text was updated successfully, but these errors were encountered:
Correct. As in word2vec, there is no such validation set built into the training process, but indeed it would probably be useful to have this (as done when training language models, etc.). I guess that if the train set is large enough then you can look at the train error without too much of a risk of over-fitting.
Hi,
During the training, how the number of epochs are getting decided ?
As I see, no part of training data has been reserved for evaluation during training of model. So, how can I be assured that model has converged during training ?
The text was updated successfully, but these errors were encountered: