After I call trainer.fit
and training finishes, I’d like to do:
best_accuracy = trainer....
Is there a way to get that value, either from the trainer or the Tensorboard logger?
After I call trainer.fit
and training finishes, I’d like to do:
best_accuracy = trainer....
Is there a way to get that value, either from the trainer or the Tensorboard logger?
For now, I’m just creating an instance variable, best_validation_accuracy
, and then setting it in validation_epoch_end
I use the following code:
trainer.callback_metrics["val_acc"]