This message was deleted.

Hi @Manuel. It almost sure means that the training reached the maximum number of training iterations, which can occur in the middle of an epoch. We prefer to define the training in terms of steps with the

max_steps

hyperparameter, but you can also use the

max_epochs

.