evolution of hyperparameter values going down

YukiN
Dataiker
evolution of hyperparameter values going down

As shown in the image, sometimes the score is adjusted in the direction of going down, how does this work? I understood that in most cases the score would go up.

図2.png

0 Kudos
1 Reply
AlexT
Dataiker

Hi,
Dataiku auto-selects the best-performing model during the training not the latest 

As explained here:
DSS only shows the best score found so far, so the graph will show “ever-improving” results, even though the latest evaluated model might not be good
https://doc.dataiku.com/dss/latest/machine-learning/advanced-optimization.html#visualization-of-hype...

You can see the selected model has 0.947, even if at the end of the graph the score is <0.9