Simple way to get the best scores at the end of a training?

[jgsch]

What is your question?

After a training, is there an easy way to get the best scores returned by the validation_end function? In order to use a hyperparameters optimizer like Tune.

Code example

model = CoolSystem()
trainer = Trainer()    
trainer.fit(model)   

best_scores = ???
print(best_scores)

What's your environment?

• conda 4.7.10
• PyTorch 1.3.0
• Lightning 0.5.3.1

[williamFalcon]

Some options:

• print the scores in training_end
• log them to the logger
• access trainer.tqdm_metrics